zsh/fusevm_bridge.rs
1//! fusevm bytecode-VM bridge for ShellExecutor.
2//!
3//! ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
4//! !!! LAST-RESORT FILE — NOT FOR NEW LOGIC !!!
5//!
6//! This file is a **bridge**, not a port. It exists ONLY because zshrs uses
7//! a fusevm bytecode VM where C zsh uses its own wordcode walker (Src/exec.c
8//! `execlist`). Every line here is plumbing that hooks fusevm opcodes onto
9//! the canonical ports in `src/ported/`.
10//!
11//! **Before adding code to this file, STOP and ask:**
12//!
13//! 1. Is this logic that already lives in `src/ported/`?
14//! → Call the canonical fn. Don't reinline.
15//!
16//! 2. Is this logic that SHOULD live in `src/ported/` but isn't ported yet?
17//! → Port it. Add it to `src/ported/<file>.rs` with a `c:` citation.
18//! Then call the canonical fn from here.
19//!
20//! 3. Is this purely fusevm/bytecode plumbing (Op decode, Value conversion,
21//! VM-stack manipulation, thread-local executor pointer, etc.)?
22//! → OK to put it here. Cite the closest C analog in the comment.
23//!
24//! **NEVER:** reinvent paramsubst/expansion/glob/typeset/redirect logic here.
25//! Those have canonical ports in `src/ported/subst.rs`, `src/ported/glob.rs`,
26//! `src/ported/builtin.rs`, etc. The bridge should be SHRINKING over time,
27//! not growing.
28//!
29//! See also: memory `feedback_no_shortcuts_in_porting` (port C bodies
30//! faithfully, no structural shells), `feedback_no_exec_script_from_ported`
31//! (the inverse direction — src/ported must not call back into the bridge).
32//! ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
33//!
34//! **Extension** — has no Src/exec.c counterpart. C zsh's `Src/exec.c::execlist`
35//! (and related routines) implement the native **wordcode VM** that executes
36//! compiler output from `parse.c`. zshrs compiles the parsed AST to fusevm
37//! bytecode and runs it on a stack VM; this
38//! file holds the bridge between fusevm's `ShellHost` trait and our
39//! `ShellExecutor` state, the thread-local executor pointer, all
40//! `BUILTIN_*` opcode constants, and the giant `register_builtins`
41//! handler table that wires zsh builtins onto fusevm CallBuiltin
42//! opcodes.
43
44#![allow(unused_imports)]
45
46use indexmap::IndexMap;
47use std::collections::{HashMap, HashSet};
48use std::env;
49use std::path::PathBuf;
50
51use crate::exec_jobs::JobState;
52use crate::intercepts::Intercept;
53use crate::ported::vm_helper::*;
54use std::io::Write;
55
56// ═══════════════════════════════════════════════════════════════════════════
57// Thread-local executor context for VM builtin dispatch
58// ═══════════════════════════════════════════════════════════════════════════
59
60use crate::ported::options::opt_state_get;
61use crate::ported::zsh_h::{isset, options, ERREXIT, MAX_OPS};
62use fusevm::op::redirect_op as r;
63use fusevm::shell_builtins::*;
64use fusevm::Value;
65use std::cell::{Cell, RefCell};
66use std::cmp::Ordering;
67use std::ffi::CString;
68use std::fs;
69use std::io::BufRead;
70use std::io::Read;
71use std::io::Write as _;
72use std::os::unix::fs::FileTypeExt;
73use std::os::unix::fs::MetadataExt;
74use std::os::unix::fs::PermissionsExt;
75use std::os::unix::io::AsRawFd;
76use std::os::unix::io::IntoRawFd;
77use std::time::Instant;
78use std::time::{SystemTime, UNIX_EPOCH};
79
80thread_local! {
81 /// Mirror of C zsh's `doneps4` local in execcmd_exec
82 /// (Src/exec.c:2517+). Tracks whether PS4 has been emitted
83 /// for the current xtrace line so a coalesced sequence of
84 /// XTRACE_ASSIGN + XTRACE_ARGS produces ONE line:
85 /// `<PS4>a=1 b=2 echo 1 2\n`
86 /// instead of three. Reset to false by XTRACE_ARGS /
87 /// XTRACE_NEWLINE after emitting the trailing `\n`.
88 static XTRACE_DONE_PS4: Cell<bool> = const { Cell::new(false) };
89
90 /// Port of C's `FILE *xtrerr` xtrace stream (Src/exec.c:81). C builds
91 /// each trace line in this stdio buffer — `printprompt4` does
92 /// `fprintf(xtrerr, …)`, then args via `fputs`/`fputc` — and
93 /// `fflush(xtrerr)` writes the WHOLE line to stderr in one syscall
94 /// (makecline c:2122-2123, addvars c:2588, condition c:1372). That
95 /// single flush is exactly why a forked pipeline stage's trace line
96 /// reaches the shared stderr fd atomically and never interleaves with
97 /// a concurrent stage. zshrs previously emitted PS4 and the command
98 /// text as separate `eprint!` writes, which raced under load. Model
99 /// the FILE buffer as this thread-local String (a forked child owns
100 /// its own copy); `xtrerr_fputs` appends, `xtrerr_flush` does the
101 /// single write.
102 static XTRERR: RefCell<String> = const { RefCell::new(String::new()) };
103
104 /// Stack of (RETFLAG, BREAKS, CONTFLAG, EXIT_PENDING) tuples saved
105 /// at try-block exit so the always-arm body can run cleanly even
106 /// when the try-block fired `return` / `break` / `continue` /
107 /// `exit`. Restored right before the post-always re-jump so the
108 /// escape resumes propagation past the construct.
109 /// c:Src/exec.c WC_TRYBLOCK — zsh's wordcode walker handles this
110 /// inline; the zshrs port lifts it into a paired SET / RESTORE
111 /// pair around the always-arm.
112 /// Tuple: `(retflag, breaks, contflag, exit_pending, try_errflag,
113 /// try_interrupt)`. The last two are c:Src/loop.c:762-763's
114 /// `save_try_errflag` / `save_try_interrupt` — the ENCLOSING try
115 /// block's values, restored at c:778-779 so nested
116 /// `{…} always {…}` constructs don't leak `$TRY_BLOCK_ERROR`
117 /// outward.
118 static TRY_ESCAPE_SAVE: RefCell<Vec<(i32, i32, i32, i32, i64, i64)>> =
119 const { RefCell::new(Vec::new()) };
120 /// Re-entry guard for BUILTIN_DEBUG_TRAP. While the DEBUG trap
121 /// body is running, the per-statement DEBUG_TRAP dispatch in the
122 /// trap body must NOT re-fire (otherwise infinite recursion +
123 /// stack overflow). zsh's in_trap counter at Src/signals.c
124 /// serves the same purpose.
125 static DEBUG_TRAP_REENTRY: Cell<bool> = const { Cell::new(false) };
126 /// Stack of (saved_stdout, saved_stderr) tuples pushed by
127 /// `cmd_subst` around its nested-VM run. RUST-ONLY: zsh forks
128 /// each cmdsub so trap output during the cmdsub naturally
129 /// lands on the PARENT's stdout. zshrs's in-process cmdsub
130 /// dups fd 1 → pipe, so a trap firing during cmdsub would
131 /// emit into the captured value. Traps consult this stack
132 /// to route their body output to the topmost saved_stdout
133 /// instead of the cmdsub's fd 1. Bug #56 in docs/BUGS.md.
134 pub static CMDSUBST_OUTER_FDS: RefCell<Vec<(i32, i32)>> =
135 const { RefCell::new(Vec::new()) };
136 /// c:Src/exec.c:5025 getproc (PATH_DEV_FD branch) — the parent
137 /// keeps the `>(cmd)` pipe WRITE end open under `/dev/fd/N`,
138 /// parks it in the job's filelist (`fdtable[fd] =
139 /// FDT_PROC_SUBST; addfilelist(NULL, fd)`), and deletefilelist
140 /// closes it when the consuming job finishes — that close is
141 /// what lets the `>(cmd)` child's reader see EOF. zshrs runs
142 /// commands in-process, so the equivalent is: record
143 /// `(scope_depth, fd)` here and drain after the consuming
144 /// command (external exec or builtin dispatch) completes.
145 static PSUB_PENDING_FDS: RefCell<Vec<(usize, i32)>> = const { RefCell::new(Vec::new()) };
146 /// `(scope_depth, path)` for `=(cmd)` temp files, unlinked at the same
147 /// job-end boundary as the pending fds (c:Src/jobs.c deletefilelist).
148 static PSUB_PENDING_FILES: RefCell<Vec<(usize, String)>> = const { RefCell::new(Vec::new()) };
149 /// Scope depth for PSUB_PENDING_FDS tagging. Incremented around
150 /// nested execution contexts (cmd-subst bodies, shell-function
151 /// bodies) so a command running INSIDE the nested context only
152 /// drains its own `>(cmd)` fds, never the enclosing command's
153 /// (e.g. `tee >(wc) $(print x)` — print must not close tee's
154 /// fd). Mirrors C's per-job filelist ownership.
155 static PSUB_SCOPE_DEPTH: Cell<usize> = const { Cell::new(0) };
156 /// Forked `<(cmd)`/`>(cmd)` child pids awaiting reap. Drained
157 /// non-blockingly (WNOHANG) by note_psub_child so proc-sub children
158 /// don't accumulate as zombies across a shell session.
159 static PSUB_CHILDREN: RefCell<Vec<i32>> = const { RefCell::new(Vec::new()) };
160}
161
162/// Record a proc-sub child pid and best-effort reap any already-exited
163/// proc-sub children (WNOHANG). Non-blocking: still-running children
164/// stay parked and get reaped on a later call.
165pub(crate) fn note_psub_child(pid: i32) {
166 if pid <= 0 {
167 return;
168 }
169 PSUB_CHILDREN.with(|v| {
170 let mut v = v.borrow_mut();
171 v.push(pid);
172 v.retain(|&p| {
173 let mut status = 0;
174 // WNOHANG: reap if exited, else keep parked.
175 let r = unsafe { libc::waitpid(p, &mut status, libc::WNOHANG) };
176 // r == p → reaped; r == 0 → still running (keep); r < 0 →
177 // already gone/not ours (drop).
178 r == 0
179 });
180 });
181}
182
183/// Port of `fputs(s, xtrerr)` / `fprintf(xtrerr, "%s", s)` (Src/exec.c):
184/// append `s` to the buffered xtrace line. Nothing reaches stderr until
185/// [`xtrerr_flush`] (the port of `fflush(xtrerr)`) writes the line whole.
186pub(crate) fn xtrerr_fputs(s: &str) {
187 XTRERR.with(|b| b.borrow_mut().push_str(s));
188}
189
190/// Port of `fflush(xtrerr)` (Src/exec.c:1373/2123/2596): write the
191/// buffered xtrace line to stderr in ONE `write` and clear the buffer, so
192/// the line lands on the shared fd atomically (no interleaving across
193/// concurrent pipeline stages).
194pub(crate) fn xtrerr_flush() {
195 XTRERR.with(|b| {
196 let mut buf = b.borrow_mut();
197 if !buf.is_empty() {
198 use std::io::Write;
199 let _ = std::io::stderr().write_all(buf.as_bytes());
200 buf.clear();
201 }
202 });
203}
204
205/// RAII guard bumping the psub scope depth — see PSUB_SCOPE_DEPTH.
206pub(crate) struct PsubScope;
207
208impl PsubScope {
209 pub(crate) fn enter() -> Self {
210 PSUB_SCOPE_DEPTH.with(|d| d.set(d.get() + 1));
211 PsubScope
212 }
213}
214
215impl Drop for PsubScope {
216 fn drop(&mut self) {
217 PSUB_SCOPE_DEPTH.with(|d| d.set(d.get().saturating_sub(1)));
218 }
219}
220
221/// RAII guard bumping `$ZSH_SUBSHELL` for the duration of an
222/// in-process command substitution.
223///
224/// c:Src/exec.c:1161 — entersubsh() does `zsh_subshell++;` and zsh's
225/// cmdsub FORKS, so the increment dies with the child and the parent's
226/// value is untouched. zshrs runs cmdsubs in-process on a nested VM,
227/// so the visible param must be bumped on entry and restored on exit.
228/// Writes paramtab u_val directly because ZSH_SUBSHELL is PM_READONLY
229/// (same bypass pattern as the subshell-builtin bump below); also
230/// mirrors into ported::exec::zsh_subshell so exec.c:4376-style
231/// `forked | zsh_subshell` reads agree.
232pub(crate) struct CmdSubstSubshellBump {
233 saved_val: i64,
234 saved_str: Option<String>,
235}
236
237impl CmdSubstSubshellBump {
238 pub(crate) fn enter() -> Self {
239 let mut saved_val = 0i64;
240 let mut saved_str = None;
241 if let Ok(mut tab) = crate::ported::params::paramtab().write() {
242 if let Some(pm) = tab.get_mut("ZSH_SUBSHELL") {
243 saved_val = pm.u_val;
244 saved_str = pm.u_str.clone();
245 pm.u_val = saved_val + 1;
246 pm.u_str = Some((saved_val + 1).to_string());
247 pm.node.flags &= !(crate::ported::zsh_h::PM_UNSET as i32);
248 }
249 }
250 crate::ported::exec::zsh_subshell.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
251 CmdSubstSubshellBump {
252 saved_val,
253 saved_str,
254 }
255 }
256}
257
258impl Drop for CmdSubstSubshellBump {
259 fn drop(&mut self) {
260 if let Ok(mut tab) = crate::ported::params::paramtab().write() {
261 if let Some(pm) = tab.get_mut("ZSH_SUBSHELL") {
262 pm.u_val = self.saved_val;
263 pm.u_str = self.saved_str.take();
264 }
265 }
266 crate::ported::exec::zsh_subshell.fetch_sub(1, std::sync::atomic::Ordering::Relaxed);
267 }
268}
269
270/// Port of deletefilelist() from Src/jobs.c (the `>(cmd)` fd arm):
271/// closes every pending proc-subst write end created at or inside
272/// the current scope depth, exactly when C deletes the consuming
273/// job's filelist (getproc parks the fd there via
274/// `addfilelist(NULL, fd)`, Src/exec.c:5025+).
275fn close_pending_psub_fds() {
276 let depth = PSUB_SCOPE_DEPTH.with(|d| d.get());
277 PSUB_PENDING_FDS.with(|v| {
278 v.borrow_mut().retain(|&(d, fd)| {
279 if d >= depth {
280 unsafe { libc::close(fd) };
281 false
282 } else {
283 true
284 }
285 });
286 });
287 // c:Src/jobs.c deletefilelist — `=(cmd)` temp files are unlinked at the
288 // same job-end boundary. `f==(print x); [[ -f $f ]]` is false after the
289 // command line ends.
290 PSUB_PENDING_FILES.with(|v| {
291 v.borrow_mut().retain(|(d, path)| {
292 if *d >= depth {
293 let _ = fs::remove_file(path);
294 false
295 } else {
296 true
297 }
298 });
299 });
300}
301
302/// RAII drain guard — instantiated at the top of the consuming-
303/// command paths (dispatch_builtin, ZshrsHost::exec) so the pending
304/// `>(cmd)` write ends close on every exit path once the command
305/// finished, exactly when C's job filelist would be deleted.
306struct PsubFdGuard;
307
308impl Drop for PsubFdGuard {
309 fn drop(&mut self) {
310 close_pending_psub_fds();
311 }
312}
313
314/// Peek the outermost cmdsub-saved (stdout, stderr) fds, if any.
315/// Returns None when no cmdsub is currently capturing. Used by the
316/// trap dispatcher in `src/ported/signals.rs::dotrap` to route trap
317/// body output to the parent's real stdout (matching zsh's forked
318/// cmdsub behaviour) instead of the cmdsub's pipe-bound fd 1.
319/// Bug #56 in docs/BUGS.md.
320pub fn cmdsubst_outer_stdout() -> Option<i32> {
321 CMDSUBST_OUTER_FDS.with(|s| s.borrow().last().map(|(o, _)| *o))
322}
323
324thread_local! {
325 /// The pipeline fds a stage still has to install onto 0/1, as
326 /// `(input, output)` with -1 meaning "leave this fd alone".
327 ///
328 /// c:Src/exec.c:3720-3724 — the pipe's read/write ends are the
329 /// FIRST entries of the command's multio table:
330 /// /* Add pipeline input/output to mnodes */
331 /// if (input) addfd(forked, save, mfds, 0, input, 0, NULL);
332 /// if (output) addfd(forked, save, mfds, 1, output, 1, NULL);
333 /// and that runs AFTER prefork (c:3304) + globlist (c:3702) have
334 /// expanded the command's argument words. So a `$(...)` inside a
335 /// stage's ARGS reads the shell's original fd 0, not the pipe:
336 /// `print -rl -- c a b | print -r -- "[$(cat)]"` prints `[]`.
337 /// (The stage's own fork at c:3000 happens before the expansion,
338 /// which is why `${x::=v}` in a non-last stage doesn't survive —
339 /// but the fds are still installed after it.)
340 ///
341 /// zshrs's [`BUILTIN_RUN_PIPELINE`] forks per stage, so it parks
342 /// the stage's fds here instead of dup2'ing them itself, and the
343 /// compiled stage chunk installs them via
344 /// [`BUILTIN_PIPE_FDS_INSTALL`] at the C-faithful point: after the
345 /// arg-word ops, before the redirect scope
346 /// (compile_zsh.rs::emit_stage_fds_install). A compound stage
347 /// (`{ … }`, `( … )`, a function) installs at chunk entry — its
348 /// body legitimately reads the pipe.
349 static PENDING_STAGE_FDS: std::cell::Cell<(i32, i32)> = const { std::cell::Cell::new((-1, -1)) };
350}
351
352/// Park the current stage's `(input, output)` pipe fds for the stage
353/// chunk's `BUILTIN_PIPE_FDS_INSTALL` to pick up. Returns the previous
354/// value so a nested pipeline (`print -- "$(a | b)" | c`) can restore
355/// the outer stage's still-uninstalled fds when it finishes.
356fn stage_fds_park(input: i32, output: i32) -> (i32, i32) {
357 PENDING_STAGE_FDS.with(|c| c.replace((input, output)))
358}
359
360/// Take (and clear) the parked stage fds.
361fn stage_fds_take() -> (i32, i32) {
362 PENDING_STAGE_FDS.with(|c| c.replace((-1, -1)))
363}
364
365// Thread-local pointer to the current ShellExecutor.
366// Set before VM execution, cleared after. Used by builtin handlers.
367thread_local! {
368 static CURRENT_EXECUTOR: RefCell<Option<*mut ShellExecutor>> = const { RefCell::new(None) };
369 /// The installed session executor, registered by
370 /// `exec::install_session_executor`. Lets [`with_session_context`]
371 /// establish a VM execution context for STARTUP work that runs
372 /// before the loop's first `execode` enters one (rc-file sourcing in
373 /// `run_init_scripts`, c:1914). Mirrors exec.rs's own
374 /// `SESSION_EXECUTOR`; kept here so the context helper lives next to
375 /// `ExecutorContext`/`CURRENT_EXECUTOR`.
376 static SESSION_EXECUTOR_PTR: std::cell::Cell<Option<*mut ShellExecutor>> = const { std::cell::Cell::new(None) };
377 /// GLOB_ASSIGN eligibility carrier. Set true by BUILTIN_MARK_GLOB_ELIGIBLE
378 /// (emitted by the compiler ONLY when a scalar-assignment RHS carries an
379 /// UNQUOTED glob token — Star/Quest/Inbrack), read+cleared by the next
380 /// BUILTIN_SET_VAR. Matches C zsh's `GLOB_ASSIGN` (Src/exec.c:2554): only
381 /// a literal unquoted glob pattern in the wordcode is globbed; values from
382 /// `$param` / `$(cmd)` / quoted strings are NOT (verified against zsh).
383 /// The runtime SET_VAR value arrives untokenized (the compiler DQ-wraps to
384 /// suppress compile-time globbing), so quoting can no longer be recovered
385 /// from the value bytes — this flag carries the compile-time decision.
386 static SET_VAR_GLOB_ELIGIBLE: std::cell::Cell<bool> = const { std::cell::Cell::new(false) };
387}
388
389/// Register the session executor pointer (called from
390/// `install_session_executor`). See [`with_session_context`].
391pub fn register_session_executor(exec: &mut ShellExecutor) {
392 SESSION_EXECUTOR_PTR.with(|c| c.set(Some(exec as *mut ShellExecutor)));
393}
394
395/// Run `f` with the registered session executor established as
396/// `CURRENT_EXECUTOR`, so code reaching the live executor via
397/// `try_with_executor` works even when no per-command `execode` context
398/// is active yet.
399///
400/// Sole caller: `zsh_main`'s `run_init_scripts()` (c:1914), which
401/// sources `.zshenv`/`.zshrc`/`.zlogin` via `source()` BEFORE the loop's
402/// first `execode`. Without an active context those sourced bodies
403/// `try_with_executor` → `None` → no-op, so the shell silently ignored
404/// the user's dotfiles. The scope is entered once around the startup
405/// sourcing window and dropped before the loop begins — deliberately NOT
406/// Run `f` with the registered session executor established as
407/// `CURRENT_EXECUTOR`, so code reaching the live executor via
408/// `try_with_executor` works even when no per-command `execode` context
409/// is active yet.
410///
411/// Sole caller: `zsh_main`'s `run_init_scripts()` (c:1914), which
412/// sources `.zshenv`/`.zshrc`/`.zlogin` via `source()` BEFORE the loop's
413/// first `execode`. Without an active context those sourced bodies
414/// `try_with_executor` → `None` → no-op, so the shell silently ignored
415/// the user's dotfiles. The scope is entered once around the startup
416/// sourcing window and dropped before the loop begins — deliberately NOT
417/// a global fallback inside `execute_script_zsh_pipeline`, which would
418/// re-enter the executor on nested command substitution and block on
419/// input.
420pub fn with_session_context<R>(f: impl FnOnce() -> R) -> R {
421 let ptr = SESSION_EXECUTOR_PTR.with(|c| c.get());
422 match ptr {
423 // SAFETY: set by install_session_executor to an executor that
424 // outlives the single-threaded interactive session.
425 Some(ptr) => {
426 let _ctx = ExecutorContext::enter(unsafe { &mut *ptr });
427 f()
428 }
429 None => f(),
430 }
431}
432
433/// Merge finished background-compinit results into shell state, callable
434/// from any site (active VM context first, session executor otherwise —
435/// the ZLE completion path runs OUTSIDE a VM frame, where
436/// `try_with_executor` alone is None and $_comps stayed empty forever).
437pub fn drain_compinit_bg_hook() {
438 if try_with_executor(|exec| exec.drain_compinit_bg()).is_some() {
439 return;
440 }
441 let ptr = SESSION_EXECUTOR_PTR.with(|c| c.get());
442 if let Some(ptr) = ptr {
443 // SAFETY: per with_session_context.
444 let _ctx = ExecutorContext::enter(unsafe { &mut *ptr });
445 unsafe { (*ptr).drain_compinit_bg() }
446 }
447}
448
449/// RAII guard that sets/clears the thread-local executor pointer.
450///
451/// Idempotent: calling `enter` when a context is already active is a no-op
452/// for the entry side, and the guard's drop only clears the thread-local if
453/// *this* call was the one that set it. Nested `execute_command` invocations
454/// (e.g. from inside a builtin handler) reuse the outer pointer instead of
455/// stomping it.
456pub(crate) struct ExecutorContext {
457 we_set_it: bool,
458}
459
460impl ExecutorContext {
461 pub(crate) fn enter(executor: &mut ShellExecutor) -> Self {
462 let we_set_it = CURRENT_EXECUTOR.with(|cell| {
463 let mut slot = cell.borrow_mut();
464 if slot.is_some() {
465 false
466 } else {
467 *slot = Some(executor as *mut ShellExecutor);
468 true
469 }
470 });
471 ExecutorContext { we_set_it }
472 }
473}
474
475impl Drop for ExecutorContext {
476 fn drop(&mut self) {
477 if self.we_set_it {
478 CURRENT_EXECUTOR.with(|cell| {
479 *cell.borrow_mut() = None;
480 });
481 }
482 }
483}
484
485/// Access the current executor from a builtin handler.
486/// # Safety
487/// Only call this from within a VM execution context (after ExecutorContext::enter).
488#[inline]
489pub(crate) fn with_executor<F, R>(f: F) -> R
490where
491 F: FnOnce(&mut ShellExecutor) -> R,
492{
493 CURRENT_EXECUTOR.with(|cell| {
494 let ptr = cell
495 .borrow()
496 .expect("with_executor called outside VM context");
497 // SAFETY: The pointer is valid for the duration of VM execution,
498 // and we're single-threaded within the executor.
499 let executor = unsafe { &mut *ptr };
500 f(executor)
501 })
502}
503
504/// Non-panicking variant of [`with_executor`]: runs `f` against the
505/// current executor and returns `Some(result)`, or `None` when no
506/// executor is in scope (`CURRENT_EXECUTOR` unset — e.g. unit tests /
507/// compsys contexts with no fusevm bridge running).
508///
509/// This is the primitive the `crate::ported::exec` accessor wrappers
510/// (array/assoc/dispatch_function_call/execute_script/...) use to
511/// reach the live executor while preserving the exact "no executor →
512/// fall back to the direct param table / default value" semantics that
513/// the deleted `exec_hooks` OnceLock layer encoded via its
514/// "is-the-hook-installed?" check. `CURRENT_EXECUTOR` being set is the
515/// faithful equivalent of "the bridge installed the hooks".
516#[inline]
517pub(crate) fn try_with_executor<F, R>(f: F) -> Option<R>
518where
519 F: FnOnce(&mut ShellExecutor) -> R,
520{
521 CURRENT_EXECUTOR.with(|cell| {
522 let ptr = (*cell.borrow())?;
523 // SAFETY: same contract as with_executor — the pointer is valid
524 // for the duration of VM execution and access is single-threaded.
525 let executor = unsafe { &mut *ptr };
526 Some(f(executor))
527 })
528}
529
530/// Look up a canonical builtin by name in `BUILTINS` and dispatch
531/// via `execbuiltin` (Src/builtin.c:250). NO shadow check — calls the
532/// builtin even if a user function with the same name exists. Used by
533/// the `builtin foo` prefix opcode (which explicitly bypasses function
534/// lookup per zsh semantics) and by internal call sites where shadowing
535/// is unwanted. For zsh's normal name-resolution order (function shadows
536/// builtin), use `dispatch_builtin` instead.
537/// Shell-identifier prefix for diagnostic lines. Reads the canonical
538/// scriptname (`zsh` in `--zsh` parity mode, `zshrs` otherwise) so a
539/// single helper replaces hardcoded `"zshrs:"` literals across the
540/// file's eprintln paths.
541fn shname() -> String {
542 crate::ported::utils::scriptname_get().unwrap_or_else(|| "zshrs".to_string())
543}
544
545/// c:Src/subst.c:505-507 + Src/exec.c:3378-3380 — per-command
546/// CSH_NULL_GLOB outcome check. During this command's word expansion
547/// `expand_glob` accumulated `badcshglob |= 1` per failed glob and
548/// `|= 2` per successful one (Src/glob.c:1871-1875). Exactly 1 —
549/// failures and no successes — is the csh-style error: `no match`,
550/// command skipped, status 1. Any other value (0 = no globs, 2/3 =
551/// at least one matched) is silent. Always resets the counter for
552/// the next command (C resets at prefork entry, subst.rs:1307).
553/// Returns true when the error fired; callers mirror their
554/// glob_failed handling (builtins leave ERRFLAG_ERROR set so the
555/// script aborts, externals clear it so the next sublist runs —
556/// verified against zsh 5.9.1).
557/// Restore the user's GLOB_SUBST after a `${~spec}` carrier flip
558/// (see subst::TILDE_GLOBSUBST_CARRIER). Runs at the same
559/// command-dispatch boundaries that consume glob_failed /
560/// badcshglob — by then every glob op of the current word pipeline
561/// has read the carrier.
562pub(crate) fn consume_tilde_globsubst_carrier() {
563 crate::ported::subst::TILDE_GLOBSUBST_CARRIER.with(|c| {
564 if let Some(saved) = c.take() {
565 crate::ported::options::opt_state_set("globsubst", saved);
566 }
567 });
568}
569
570/// Pop `argc` stack slots for a whole-array assignment: the LAST popped
571/// (deepest pushed) is the param name, the rest are the values in stack
572/// order, with any `Value::Array` flattened to its elements. Mirrors the
573/// Flatten an array-assignment RHS value into scalar strings, descending
574/// through nested `Value::Array`s. zsh arrays are always flat, so recursion
575/// only collapses the wrapper layers the compiler introduces — in particular
576/// the single `Value::Array` built by `Op::MakeArray` for `arr=(...)` literals
577/// (used to dodge `CallBuiltin`'s u8 argc cap), whose own elements may
578/// themselves be arrays from an unquoted `$other_array` expansion. A
579/// one-level flatten would stringify those inner arrays into a single element.
580fn flatten_array_value(v: Value, out: &mut Vec<String>) {
581 match v {
582 Value::Array(items) => {
583 for it in items {
584 flatten_array_value(it, out);
585 }
586 }
587 other => out.push(other.to_str()),
588 }
589}
590
591/// pop/flatten prologue of BUILTIN_SET_ARRAY / BUILTIN_APPEND_ARRAY.
592fn pop_array_args_with_name(vm: &mut fusevm::VM, argc: u8) -> (String, Vec<String>) {
593 let n = argc as usize;
594 let mut popped: Vec<Value> = Vec::with_capacity(n);
595 for _ in 0..n {
596 popped.push(vm.pop());
597 }
598 popped.reverse();
599 let name = popped.pop().map(|v| v.to_str()).unwrap_or_default();
600 let mut values: Vec<String> = Vec::new();
601 for v in popped {
602 flatten_array_value(v, &mut values);
603 }
604 (name, values)
605}
606
607fn consume_badcshglob() -> bool {
608 let v = crate::ported::glob::BADCSHGLOB.swap(0, std::sync::atomic::Ordering::Relaxed);
609 if v == 1 {
610 crate::ported::utils::zerr("no match"); // c:Src/subst.c:507
611 true
612 } else {
613 false
614 }
615}
616
617/// Map a builtin name to the zsh module that owns it, IFF zsh does
618/// not auto-load that builtin on first use. Used by
619/// `dispatch_builtin_raw` to gate `--zsh` mode dispatch behind
620/// `zmodload`, mirroring `zsh -fc <name>` returning 127 for these
621/// names without an explicit module load.
622///
623/// Returns `Some(module_name)` if `name` belongs to a non-auto-load
624/// module per the per-module `Src/Modules/<x>.c` `bintab[]` plus
625/// the auto-load flag set at module-build time. `None` for core
626/// builtins and for auto-loaded module builtins (sched, log, echotc,
627/// echoti, zformat, zparseopts, zregexparse, zstyle, strftime,
628/// private, vared, zle, bindkey, comp*) which work without zmodload.
629fn module_bound_builtin_module(name: &str) -> Option<&'static str> {
630 match name {
631 "zftp" => Some("zsh/zftp"),
632 "zsocket" => Some("zsh/net/socket"),
633 "ztcp" => Some("zsh/net/tcp"),
634 "zstat" => Some("zsh/stat"),
635 "zselect" => Some("zsh/zselect"),
636 "zpty" => Some("zsh/zpty"),
637 "zprof" => Some("zsh/zprof"),
638 "zsystem" | "syserror" => Some("zsh/system"),
639 "clone" => Some("zsh/clone"),
640 "zcurses" => Some("zsh/curses"),
641 "ztie" | "zuntie" | "zgdbmpath" => Some("zsh/db/gdbm"),
642 "pcre_compile" | "pcre_match" | "pcre_study" => Some("zsh/pcre"),
643 "example" => Some("zsh/example"),
644 "cap" | "getcap" | "setcap" => Some("zsh/cap"),
645 "zgetattr" | "zsetattr" | "zdelattr" | "zlistattr" => Some("zsh/attr"),
646 // c:Src/Modules/datetime.c — `strftime` is registered via
647 // partab[] when zsh/datetime loads. Verified by
648 // `zsh -fc 'strftime -s s %Y 0'` → 127 "command not found".
649 "strftime" => Some("zsh/datetime"),
650 _ => None,
651 }
652}
653
654/// Dispatch a zshrs-ORIGINAL builtin by NAME, argv-style. These are
655/// registered as fusevm opcodes in [`register_builtins`] (async, doctor,
656/// peach, …), so a *literal* name compiles to `CallBuiltin` and runs. But
657/// they are absent from the static `BUILTINS` port table and the merged
658/// `builtintab`, so when the command name is resolved only at run time —
659/// `$var` indirection, `builtin NAME` — the ported command-resolution path
660/// never finds them and reports "command not found" / "no such builtin",
661/// even though `whence` (correctly) calls them builtins. The
662/// `register_builtins` closures use the VM only to pop args and then call an
663/// executor method, so the identical dispatch works here from any parent-side
664/// resolver that has an executor — no VM re-entry (which would alias the
665/// running `&mut VM`). Returns `None` for a name that is not one of them, so
666/// the caller falls through to external lookup.
667///
668/// !!! Keep in sync with the matching `vm.register_builtin(...)` closures in
669/// `register_builtins`: both must route a name to the same executor method.
670pub(crate) fn try_run_registered_builtin(name: &str, argv: &[String]) -> Option<i32> {
671 let s = match name {
672 "async" => with_executor(|e| e.builtin_async(argv)),
673 "await" => with_executor(|e| e.builtin_await(argv)),
674 "barrier" => with_executor(|e| e.builtin_barrier(argv)),
675 "peach" => with_executor(|e| e.builtin_peach(argv)),
676 "pmap" => with_executor(|e| e.builtin_pmap(argv)),
677 "pgrep" => with_executor(|e| e.builtin_pgrep(argv)),
678 "intercept" => with_executor(|e| e.builtin_intercept(argv)),
679 "intercept_proceed" => with_executor(|e| e.builtin_intercept_proceed(argv)),
680 "doctor" => with_executor(|e| e.builtin_doctor(argv)),
681 "dbview" => with_executor(|e| e.builtin_dbview(argv)),
682 "profile" => with_executor(|e| e.builtin_profile(argv)),
683 "caller" => with_executor(|e| e.builtin_caller(argv)),
684 "help" => with_executor(|e| e.builtin_help(argv)),
685 "cdreplay" => with_executor(|e| e.builtin_cdreplay(argv)),
686 "zsleep" => crate::extensions::ext_builtins::zsleep(argv),
687 _ => return None,
688 };
689 Some(s)
690}
691
692pub(crate) fn dispatch_builtin_raw(name: &str, args: Vec<String>) -> i32 {
693 // !!! WARNING: RUST-ONLY — NO C COUNTERPART !!!
694 // Native p10k engine intercept (src/extensions/p10k): sourcing
695 // powerlevel10k.zsh-theme activates the Rust segment engine
696 // instead of executing the ~13k-line zsh theme. The user's
697 // `.p10k.zsh` CONFIG is NOT intercepted — it sources normally so
698 // its POWERLEVEL9K_* typesets land in the paramtab, which the
699 // engine reads live at every render. Placed here (the chokepoint
700 // every builtin route funnels through) so `source`, `.`, and
701 // `builtin source` all hit it.
702 if matches!(name, "source" | ".") {
703 if let Some(status) = crate::p10k::maybe_intercept_theme_source(&args) {
704 // Register a `p10k` stub function so `${+functions[p10k]}`
705 // guards in .zshrc templates stay truthy. The body forwards
706 // to the bridge-intercepted `zshrs-p10k-api` name so
707 // `p10k segment` (custom-segment protocol) and the other
708 // API subcommands reach the native engine (p10k_api).
709 try_with_executor(|exec| {
710 let _ = exec.execute_script("function p10k() { zshrs-p10k-api \"$@\" }");
711 });
712 return status;
713 }
714 }
715 // Native p10k API dispatch — the `p10k` stub function forwards
716 // here (see the theme intercept above). Must run before the
717 // generic builtintab lookup: the name is not a real builtin.
718 if let Some(status) = crate::p10k::maybe_intercept_command(name, &args) {
719 return status;
720 }
721 // c:Src/exec.c:2700-2717 — `private` is an autoloaded builtin in
722 // zsh (autofeature b:private of zsh/param/private): first use
723 // runs ensurefeature → require_module → load_module → boot_,
724 // marking the module MOD_INIT_B (what `zmodload -e` reads) and
725 // installing the wrap_private FuncWrap (param_private.c:712).
726 // doshfunc gates the wrapper dispatch on that load state, so
727 // this require_module is what activates private scoping. The
728 // raw dispatcher is the chokepoint every builtin route funnels
729 // through; require_module is idempotent after the first call
730 // (needs_load checks MOD_INIT_B).
731 if name == "private" {
732 if let Ok(mut tab) = crate::ported::module::MODULESTAB.lock() {
733 let _ = crate::ported::module::require_module(&mut tab, "zsh/param/private", None, 0);
734 // c:2710 ensurefeature
735 }
736 }
737 // c:Src/Modules/param_private.c:682-685 setup_ — loading
738 // zsh/param/private REPLACES the `local` builtintab node's
739 // handlerfunc + optstr with bin_private's ("Even more horrible
740 // hack"), so once the module is loaded `local` IS bin_private: it
741 // accepts the -P/-Pa private-scope flags, and without -P delegates
742 // to bin_typeset, which already treats `local` and `private`
743 // identically (is_locallike, builtin.rs:3666). Replicate the swap by
744 // routing `local` through the `private` node only after the module
745 // is loaded — before then, `local -P` still errors "bad option: -P"
746 // exactly like stock zsh. The `private` node carries the augmented
747 // optstr (with P) that the `local` node lacks.
748 if name == "local"
749 && crate::ported::module::MODULESTAB
750 .lock()
751 .map(|t| t.is_bound("zsh/param/private"))
752 .unwrap_or(false)
753 {
754 return dispatch_builtin_raw("private", args);
755 }
756 // c:Bugs #475/#504/#555 — bash-only builtins (`mapfile`,
757 // `readarray`, `compopt`) should emit "command not found" in
758 // `--zsh` mode matching zsh's external-command-lookup miss.
759 // The per-opcode closures for caller/help/complete/compgen
760 // already gate via IS_ZSH_MODE at their registration sites;
761 // names without dedicated opcodes (compopt/mapfile/readarray)
762 // route through this generic builtintab lookup and need the
763 // gate here.
764 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed)
765 && matches!(name, "compopt" | "mapfile" | "readarray")
766 {
767 eprintln!("zsh:1: command not found: {}", name);
768 let _ = args;
769 return 127;
770 }
771 // c:Src/exec.c:2700-2724 resolvebuiltin — autoloaded-builtin stub
772 // (registered by `zmodload -ab MOD NAME`, Src/module.c:426
773 // add_autobin) fires on first use: ensurefeature loads the owning
774 // module, then dispatch proceeds against the real builtin. Must
775 // run BEFORE the module-bound 127 gate below — `zmodload -ab
776 // zsh/zselect zselect; zselect` previously died there with
777 // `command not found` because the gate only checked is_loaded,
778 // never the autoload ledger.
779 if let Some(rc) = crate::ported::module::resolvebuiltin(name) {
780 if rc != 0 {
781 // Load failed or feature undefined — diagnostics already
782 // printed (load_module zwarn / resolvebuiltin zerr).
783 // C's execbuiltin head returns 1 (Src/builtin.c:264-267).
784 return 1;
785 }
786 // Module loaded — fall through; the is_loaded gates below now
787 // pass and the normal dispatch chain runs the real builtin.
788 }
789 // c:Src/Modules/<mod>.c boot_/setup_ chain — module-bound builtins
790 // (zftp, zsocket, ztcp, zstat, etc.) are only registered into
791 // `builtintab` when their module is loaded via `zmodload`. In
792 // zsh `-fc` (the parity test harness's invocation), the modules
793 // are NOT pre-loaded, so each name reports "command not found"
794 // with exit 127. zshrs intentionally pre-loads all module bintabs
795 // in `createbuiltintable` (builtin.rs:131-152) for the default
796 // mode so users can call these without `zmodload`; that auto-load
797 // diverges from zsh's gate behavior. Match zsh's stance only when
798 // the user explicitly asked for parity via `--zsh`.
799 //
800 // The list is the union of builtins from modules that zsh does
801 // NOT auto-load (verified via `zsh -fc <name>` returning 127):
802 // zsh/zftp → zftp
803 // zsh/net/socket → zsocket
804 // zsh/net/tcp → ztcp
805 // zsh/stat → zstat (NOT `stat`; that name resolves to
806 // /bin/stat on PATH per zsh's setup)
807 // zsh/zselect → zselect
808 // zsh/zpty → zpty
809 // zsh/zprof → zprof
810 // zsh/system → zsystem, syserror
811 // zsh/clone → clone
812 // zsh/curses → zcurses
813 // zsh/db/gdbm → ztie, zuntie, zgdbmpath
814 // zsh/pcre → pcre_compile, pcre_match, pcre_study
815 // zsh/example → example
816 // zsh/cap → cap, getcap, setcap
817 // zsh/attr → zgetattr, zsetattr, zdelattr, zlistattr
818 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed)
819 && module_bound_builtin_module(name)
820 .map(|m| {
821 !crate::ported::module::MODULESTAB
822 .lock()
823 .map(|t| t.is_loaded(m))
824 .unwrap_or(false)
825 })
826 .unwrap_or(false)
827 {
828 eprintln!("zsh:1: command not found: {}", name);
829 let _ = args;
830 return 127;
831 }
832 // c:Src/Modules/files.c:806-824 — zsh/files registers `chmod`,
833 // `chown`, `chgrp`, `ln`, `mkdir`, `mv`, `rm`, `rmdir`, `sync`
834 // (plus their `zf_*` aliases) into builtintab on module load.
835 // Without an explicit `zmodload zsh/files`, zsh resolves the
836 // names through PATH lookup — `zsh -fc 'chmod +x f'` runs
837 // `/bin/chmod`, whose argv-parser accepts symbolic modes like
838 // `+x` that bin_chmod's octal-only parser rejects with
839 // "invalid mode `+x'". The shadow-aware wrapper at
840 // `dispatch_builtin` (line 438) already has this gate, but the
841 // direct `dispatch_builtin_raw` path used by fusevm's
842 // CallBuiltin opcode bypasses it. Mirror the gate here so the
843 // low-level dispatch matches C's PATH-fall-through behavior.
844 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed)
845 && module_gated_files_builtin(name)
846 && !crate::ported::module::MODULESTAB
847 .lock()
848 .map(|t| t.is_loaded("zsh/files"))
849 .unwrap_or(false)
850 {
851 // PATH lookup uses the LITERAL name: bare `mkdir` finds
852 // /bin/mkdir; a `zf_*` alias finds nothing and exits 127 —
853 // matching zsh -fc `zf_mkdir d` → "command not found:
854 // zf_mkdir" (the aliases exist ONLY in the loaded module's
855 // builtintab, Src/Modules/files.c:816-824; PATH has no
856 // /bin/zf_rm). The previous zf_-strip silently ran the
857 // system binary instead.
858 let status = with_executor(|exec| exec.execute_external(name, &args, &[])).unwrap_or(127);
859 return status;
860 }
861 // c:Src/Modules/stat.c:637-638 — zsh/stat registers BOTH `stat`
862 // and `zstat`. `zstat` is in the module_bound 127-gate above (no
863 // /usr/bin/zstat exists), but the bare `stat` name must FALL
864 // THROUGH to PATH when zsh/stat isn't loaded — zsh -fc
865 // 'stat -f %Lp f' runs /usr/bin/stat, while bin_stat's parser
866 // rejects stat(1) flags ("bad option: -c"). Same fall-through
867 // shape as the zsh/files gate above.
868 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed)
869 && name == "stat"
870 && !crate::ported::module::MODULESTAB
871 .lock()
872 .map(|t| t.is_loaded("zsh/stat"))
873 .unwrap_or(false)
874 {
875 let status = with_executor(|exec| exec.execute_external(name, &args, &[])).unwrap_or(127);
876 return status;
877 }
878 // c:Src/exec.c:3050-3068 — builtin lookup hits `builtintab` (the
879 // merged table containing module-provided builtins). The previous
880 // port walked only the core `BUILTINS` slice, so per-module
881 // entries like `log` (Src/Modules/watch.c:693 `BUILTIN("log", …,
882 // bin_log, …)`) were registered into builtintab via
883 // createbuiltintable but never reached at dispatch — `log` fell
884 // through to PATH and ran `/usr/bin/log`. Bug #72 in docs/BUGS.md.
885 let tab = crate::ported::builtin::createbuiltintable();
886 if let Some(bn_static) = tab.get(name) {
887 let bn_ptr = *bn_static as *const _ as *mut _;
888 return crate::ported::builtin::execbuiltin(args, Vec::new(), bn_ptr);
889 }
890 1
891}
892
893/// Shadow-aware dispatch matching zsh's name-resolution order:
894/// alias → reserved word → **function (shadows builtin)** → builtin →
895/// external. All `BUILTIN_X` opcode handlers route through here so a
896/// user-defined `cd () { … }` (or `r`, `fc`, `which`, … anything in
897/// fusevm's name→opcode map) takes precedence over the C builtin —
898/// matching `Src/exec.c:execcmd_exec`'s dispatch at c:3050-3068.
899/// Without this, compile-time builtin resolution silently ignored
900/// user wrappers (e.g. ZPWR's `cd () { builtin cd "$@"; … }`).
901/// True for builtins that are bound by zsh/files's boot_/setup_
902/// chain (Src/Modules/files.c:806-824). These are the bare-name
903/// `mkdir`/`rm`/`mv`/`ln`/`chmod`/`chown`/`chgrp`/`sync`/`rmdir`
904/// AND their `zf_*` aliases at c:816-824. Without explicit
905/// `zmodload zsh/files`, the names fall through to PATH lookup
906/// (zsh's `type rm` reports `/bin/rm`). Bug #28.
907fn module_gated_files_builtin(name: &str) -> bool {
908 matches!(
909 name,
910 "mkdir"
911 | "rmdir"
912 | "rm"
913 | "mv"
914 | "ln"
915 | "chmod"
916 | "chown"
917 | "chgrp"
918 | "sync"
919 | "zf_mkdir"
920 | "zf_rmdir"
921 | "zf_rm"
922 | "zf_mv"
923 | "zf_ln"
924 | "zf_chmod"
925 | "zf_chown"
926 | "zf_chgrp"
927 | "zf_sync"
928 )
929}
930
931pub(crate) fn dispatch_builtin(name: &str, args: Vec<String>) -> i32 {
932 // c:Src/exec.c getproc + Src/jobs.c deletefilelist — close any
933 // `>(cmd)` write ends owned by this command once it finishes
934 // (drops on every return path below).
935 let _psub_fds = PsubFdGuard;
936 // c:Src/exec.c — when any redirect in the current scope failed
937 // (e.g. noclobber blocked a `>` overwrite), zsh refuses to
938 // execute the command and exits with status 1. The Rust port
939 // still applied the command (writing to the /dev/null sink
940 // installed by host_apply_redirect's noclobber arm) so the
941 // success status overwrote the intended 1. Short-circuit here
942 // for builtins (the external-exec equivalent lives in
943 // ZshrsHost::exec).
944 let redir_failed = with_executor(|exec| {
945 let f = exec.redirect_failed;
946 exec.redirect_failed = false;
947 f
948 });
949 if redir_failed {
950 // c:Src/exec.c:4367-4386 — POSIX special-builtin escalation:
951 // a failed redirect on a PSPECIAL builtin (set, readonly,
952 // typeset, ...) under POSIX_BUILTINS is FATAL in a
953 // non-interactive shell (`exit(1)` at c:4383). The `command`
954 // prefix resets this (BINF_COMMAND, c:4369) — that path
955 // dispatches through bin_command, not here.
956 if crate::ported::zsh_h::isset(crate::ported::zsh_h::POSIXBUILTINS)
957 && !crate::ported::zsh_h::isset(crate::ported::zsh_h::INTERACTIVE)
958 && builtin_is_pspecial(name)
959 {
960 use std::sync::atomic::Ordering;
961 crate::ported::builtin::EXIT_VAL.store(1, Ordering::Relaxed);
962 crate::ported::builtin::EXIT_PENDING.store(1, Ordering::Relaxed);
963 }
964 return 1;
965 }
966 // c:Src/glob.c:1876-1880 NOMATCH path — when expand_glob() failed
967 // on a no-match glob, zsh aborts the simple command after zerr()
968 // printed "no matches found". In C, this works because zerr()
969 // sets ERRFLAG_ERROR (Src/utils.c) and execcmd_exec()
970 // (Src/exec.c:3050+) checks errflag before invoking the builtin
971 // table. Rust's builtin dispatch doesn't sit on the same errflag
972 // gate, so we explicitly consume the per-command glob-fail cell
973 // and short-circuit with status 1. Mirrors the external-path
974 // guard at host_exec_external (line 5167). Without this:
975 // `echo /never/*` would print empty (silently rolled back to ""
976 // by the empty glob expansion). Parity bug #13.
977 consume_tilde_globsubst_carrier();
978 let glob_failed = with_executor(|exec| {
979 let f = exec.current_command_glob_failed.get();
980 exec.current_command_glob_failed.set(false); // c:1879 cleanup
981 f
982 });
983 if glob_failed {
984 // c:Src/glob.c:1876-1880 + Src/exec.c — NOMATCH zerr sets
985 // ERRFLAG_ERROR (via utils.c:184). For a BUILTIN command the
986 // expansion runs IN the shell process, so errflag stays set
987 // and the rest of the input aborts (zsh -fc 'echo /nope_*;
988 // echo after' prints nothing after the error — verified
989 // against zsh 5.9). The continue-after-nomatch behaviour
990 // belongs ONLY to externals: C forks BEFORE expansion there,
991 // so the child's zerr can't touch the parent's errflag (zsh
992 // -fc 'ls /nope_*; echo after' prints `after`) — that path's
993 // clear lives in fn exec / execute_external. Leave
994 // ERRFLAG_ERROR set here; BUILTIN_ERREXIT_CHECK trigger 4
995 // aborts the remaining script at the next command boundary.
996 return 1; // c:1880 — command aborted, status 1
997 }
998 // c:Src/subst.c:505-507 — CSH_NULL_GLOB sibling of the NOMATCH
999 // gate above: all of this command's globs failed silently (words
1000 // dropped, badcshglob accumulated 1s and no 2s) → `no match`,
1001 // skip the builtin, status 1. Like the NOMATCH path, ERRFLAG
1002 // from zerr stays set for builtins so the rest of the script
1003 // aborts (zsh -fc 'setopt cshnullglob; print *nope* x; print
1004 // after' prints only the error — verified zsh 5.9.1).
1005 if consume_badcshglob() {
1006 // c:Src/exec.c:3380 — `lastval = 1;` so the shell's final
1007 // exit status reflects the aborted command.
1008 with_executor(|exec| exec.set_last_status(1));
1009 return 1;
1010 }
1011 // c:Src/exec.c:4162-4295 — assignment-builtin (BINF_ASSIGN family:
1012 // typeset / declare / local / export / readonly / integer / float /
1013 // private) whose `name=value` postassign arg raised errflag while
1014 // its RHS was preforked (PREFORK_ASSIGN, c:4239-4245) — the classic
1015 // case is a math error in `typeset -F fv=$((1/0))`. The postassign
1016 // loop `break`s on errflag (c:4243) and then `if (!errflag)
1017 // execbuiltin(...)` (c:4287) SKIPS the builtin entirely, so `lastval`
1018 // is left UNCHANGED from before the command (0 fresh, 1 after
1019 // `false`). This differs from a PLAIN assignment `x=$((1/0))`, which
1020 // goes through execsimple c:1375 `lv = errflag ? errflag : cmdoutval`
1021 // → 1, and from a NON-assign builtin `print $((1/0))`, whose main
1022 // args-prefork errflag lands on c:3760 `lastval = 1`. Only the
1023 // assignment-BUILTIN postassign path preserves the prior status.
1024 // Mirror it here: the fusevm reg_passthru dispatch still calls us
1025 // with errflag set (unlike C's pre-invoke gate), so consume that
1026 // state and return the prior LASTVAL instead of running the builtin.
1027 {
1028 use std::sync::atomic::Ordering;
1029 let live = crate::ported::utils::errflag.load(Ordering::Relaxed);
1030 let ef = live & crate::ported::zsh_h::ERRFLAG_ERROR;
1031 let hard = live & crate::ported::zsh_h::ERRFLAG_HARD;
1032 // Only the SOFT recoverable error (math failure like `$((1/0))`,
1033 // ERRFLAG_ERROR without ERRFLAG_HARD) preserves the prior status
1034 // per c:4287. A HARD error (`${var?msg}`, which c:Src/subst.c
1035 // OR's ERRFLAG_HARD onto errflag) is a script-abort that yields
1036 // status 1 regardless of the prior status — leave that to the
1037 // normal dispatch/abort path below (which returns 1 and keeps
1038 // ERRFLAG_HARD set for the downstream errexit gate).
1039 if ef != 0 && hard == 0 && builtin_is_assign_family(name) {
1040 // c:4287 — execbuiltin skipped; lastval unchanged.
1041 return crate::ported::builtin::LASTVAL.load(Ordering::Relaxed);
1042 }
1043 }
1044 if let Some(status) = try_user_fn_override(name, &args) {
1045 // c:Src/jobs.c:1748 waitonejob — canonical single-command
1046 // pipestats update via the no-procs else-branch.
1047 crate::ported::builtin::LASTVAL.store(status, std::sync::atomic::Ordering::Relaxed);
1048 let mut synth = crate::ported::zsh_h::job::default();
1049 crate::ported::jobs::waitonejob(&mut synth);
1050 return status;
1051 }
1052 // c:Src/builtin.c:587 + Src/exec.c:3056 — a builtin disabled via
1053 // `disable <name>` has its `DISABLED` flag set in `builtintab`;
1054 // `builtintab->getnode` (the DISABLED-filtering accessor) returns
1055 // NULL for it at lookup time, so execcmd_exec falls through to
1056 // PATH lookup and runs the external. The Rust port stores the
1057 // disabled set in `BUILTINS_DISABLED`; the previous dispatcher
1058 // only checked the immutable `createbuiltintable` HashMap which
1059 // never reflects disablement — so `disable echo; echo hi` kept
1060 // running the bin_echo builtin. Bug #106 in docs/BUGS.md.
1061 //
1062 // dispatch_builtin (the high-level wrapper used by the BUILTIN_*
1063 // opcode handlers and reg_passthru! callsites) is the correct
1064 // gate: `dispatch_builtin_raw` is the low-level entry point
1065 // used by `bin_builtin` itself which MUST bypass the disabled
1066 // set (man zshbuiltins: `builtin name` runs the builtin
1067 // regardless of disable state). Place the check here so the
1068 // bypass path stays clean.
1069 let disabled = crate::ported::builtin::BUILTINS_DISABLED
1070 .lock()
1071 .map(|s| s.contains(name))
1072 .unwrap_or(false);
1073 if disabled {
1074 let status = with_executor(|exec| exec.execute_external(name, &args, &[])).unwrap_or(127);
1075 crate::ported::builtin::LASTVAL.store(status, std::sync::atomic::Ordering::Relaxed);
1076 let mut synth = crate::ported::zsh_h::job::default();
1077 crate::ported::jobs::waitonejob(&mut synth);
1078 return status;
1079 }
1080 // c:Src/Modules/files.c:806-814 — `mkdir`, `rm`, `mv`, `ln`, `chmod`,
1081 // `chown`, `chgrp`, `sync`, `rmdir` are bound by the `zsh/files`
1082 // module's boot_/setup_ chain. Without explicit `zmodload zsh/files`,
1083 // these bare names fall through to PATH (`/bin/rm`, `/usr/bin/chmod`,
1084 // etc.) in zsh; `type rm` reports `rm is /bin/rm`. The `zf_*`
1085 // aliases (`zf_rm`, `zf_chmod`, …) are bound by the same module
1086 // and gated the same way. Bug #28 in docs/BUGS.md.
1087 if module_gated_files_builtin(name) {
1088 if !crate::ported::module::MODULESTAB
1089 .lock()
1090 .unwrap()
1091 .is_loaded("zsh/files")
1092 {
1093 // PATH lookup uses the literal name. In --zsh parity mode
1094 // `zf_rm` must 127 like zsh -fc (no /bin/zf_rm); default
1095 // zshrs mode keeps the convenience zf_-strip so the alias
1096 // still reaches the system binary.
1097 let path_name = if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed) {
1098 name
1099 } else {
1100 name.strip_prefix("zf_").unwrap_or(name)
1101 };
1102 let status =
1103 with_executor(|exec| exec.execute_external(path_name, &args, &[])).unwrap_or(127);
1104 crate::ported::builtin::LASTVAL.store(status, std::sync::atomic::Ordering::Relaxed);
1105 let mut synth = crate::ported::zsh_h::job::default();
1106 crate::ported::jobs::waitonejob(&mut synth);
1107 return status;
1108 }
1109 }
1110 // c:Src/exec.c:3997 `int q = queue_signal_level();`
1111 // c:Src/exec.c:4231 `dont_queue_signals();`
1112 // c:Src/exec.c:4243 `restore_queue_signals(q);`
1113 //
1114 // C runs EVERY builtin with signal queueing switched OFF. Two
1115 // consequences the zshrs port was missing:
1116 //
1117 // 1. `dont_queue_signals()` DRAINS the pending queue (it calls
1118 // run_queued_signals()), so a signal that arrived while an
1119 // enclosing scope held queue_signals() — doshfunc holds one
1120 // for the whole call, c:Src/exec.c:5835 — fires its trap at
1121 // the NEXT command boundary rather than at function exit.
1122 // 2. While the builtin runs, queueing stays off, so a signal the
1123 // builtin sends to itself (`kill -USR1 $$`) dispatches the
1124 // trap synchronously inside the builtin — which is why zsh
1125 // prints pre/trap/post for
1126 // `f() { print pre; kill -USR1 $$; print post }`.
1127 //
1128 // Without this bracket every trap raised inside a function was
1129 // deferred to the enclosing unqueue_signals() (i.e. script end).
1130 let q = crate::ported::signals_h::queue_signal_level(); // c:3997
1131 crate::ported::signals_h::dont_queue_signals(); // c:4231
1132 let status = dispatch_builtin_raw(name, args);
1133 crate::ported::signals_h::restore_queue_signals(q); // c:4243
1134 // c:Src/jobs.c:1748 waitonejob — canonical single-command pipestats update.
1135 crate::ported::builtin::LASTVAL.store(status, std::sync::atomic::Ordering::Relaxed);
1136 let mut synth = crate::ported::zsh_h::job::default();
1137 crate::ported::jobs::waitonejob(&mut synth);
1138 // c:Src/exec.c:4367-4386 — done: tail. A PSPECIAL builtin that
1139 // raised errflag under POSIX_BUILTINS exits the non-interactive
1140 // shell with status 1 ("hard error in POSIX" — e.g. bin_dot's
1141 // zerrnam at Src/builtin.c:6133). Arm the deferred-exit pair so
1142 // the next ERREXIT_CHECK unwinds; EXIT_VAL=1 matches C's
1143 // hardcoded exit(1), NOT the builtin's own status (dot returns
1144 // 127 but POSIX exits 1).
1145 if crate::ported::zsh_h::isset(crate::ported::zsh_h::POSIXBUILTINS)
1146 && !crate::ported::zsh_h::isset(crate::ported::zsh_h::INTERACTIVE)
1147 && builtin_is_pspecial(name)
1148 && (crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed)
1149 & crate::ported::zsh_h::ERRFLAG_ERROR)
1150 != 0
1151 {
1152 use std::sync::atomic::Ordering;
1153 crate::ported::builtin::EXIT_VAL.store(1, Ordering::Relaxed);
1154 crate::ported::builtin::EXIT_PENDING.store(1, Ordering::Relaxed);
1155 }
1156 status
1157}
1158
1159/// c:Src/zsh.h:1467 BINF_PSPECIAL — true when `name` is a POSIX
1160/// special builtin per the canonical builtin table flags
1161/// (Src/builtin.c:48-129: `.`, `:`, break, continue, declare, eval,
1162/// exit, export, float, integer, local, readonly, return, set,
1163/// shift, source, times, trap, typeset, unset).
1164fn builtin_is_pspecial(name: &str) -> bool {
1165 crate::ported::builtin::createbuiltintable()
1166 .get(name)
1167 .map(|b| (b.node.flags as u32 & crate::ported::zsh_h::BINF_PSPECIAL) != 0)
1168 .unwrap_or(false)
1169}
1170
1171/// c:Src/zsh.h:1486 BINF_ASSIGN — the assignment-builtin family
1172/// (typeset / declare / local / export / readonly / integer / float /
1173/// private). Their `name=value` args are handled as postassigns
1174/// (c:Src/exec.c:4162-4295), whose errflag-abort skips execbuiltin and
1175/// preserves the prior `lastval`. Read the flag straight from the
1176/// builtin table (same pattern as `builtin_is_pspecial`).
1177fn builtin_is_assign_family(name: &str) -> bool {
1178 crate::ported::builtin::createbuiltintable()
1179 .get(name)
1180 .map(|b| (b.node.flags as u32 & crate::ported::zsh_h::BINF_ASSIGN) != 0)
1181 .unwrap_or(false)
1182}
1183
1184// The former `install_exec_hooks()` fn-pointer registry is gone. Code
1185// under `src/ported/` now reaches `ShellExecutor` operations
1186// (array/assoc storage, script eval, function dispatch, command
1187// substitution) through the `crate::ported::exec::*` accessor wrappers,
1188// which resolve the live executor via `try_with_executor`
1189// (`CURRENT_EXECUTOR`). The bridge lives in exec.rs — the sanctioned
1190// fusevm-access exception — per `feedback_no_exec_script_from_ported` /
1191// `feedback_no_shellexecutor_in_ported`.
1192
1193/// Register all zsh builtins with the VM.
1194pub(crate) fn register_builtins(vm: &mut fusevm::VM) {
1195 // src/ported/ reaches the live executor (param store, function
1196 // dispatch, nested script/cmdsubst exec) through the
1197 // `crate::ported::exec::*` accessor wrappers, which read
1198 // `CURRENT_EXECUTOR` via `try_with_executor`. No install step is
1199 // needed: the executor is in scope for the duration of any VM run
1200 // (set by `ExecutorContext::enter`), so the wrappers resolve it
1201 // directly. (Replaces the former `exec_hooks` OnceLock fn-ptr
1202 // registry, now deleted.)
1203 // Engage fusevm's tiered JIT (block + tracing) so hot, fully-eligible
1204 // numeric chunks run in native code and — with the `jit-disk-cache`
1205 // feature (on by default) — persist that native code to
1206 // `~/.cache/fusevm-jit`, letting repeated zsh invocations skip Cranelift
1207 // codegen. fusevm gates the JIT on per-chunk eligibility and warms up by
1208 // an invocation threshold, falling back to the interpreter for any chunk
1209 // it cannot compile (e.g. host-builtin/`Extended` command dispatch), so
1210 // enabling it here never changes observable behaviour — it only caches
1211 // the numeric hot path. Idempotent: re-enabling on each VM is a no-op.
1212 vm.enable_tracing_jit();
1213 // Macro for builtins that user functions are allowed to shadow.
1214 // zsh dispatch order is alias → function → builtin; without the
1215 // try_user_fn_override probe a `cat() { ... }; cat` would silently
1216 // run the C builtin and ignore the user function.
1217 macro_rules! reg_overridable {
1218 ($vm:expr, $id:expr, $name:literal, $method:ident) => {
1219 $vm.register_builtin($id, |vm, argc| {
1220 let args = pop_args(vm, argc);
1221 // c:Src/exec.c getproc + Src/jobs.c deletefilelist —
1222 // close `>(cmd)` write ends owned by this command
1223 // once it finishes (shadows bypass dispatch_builtin
1224 // and ZshrsHost::exec, so they need their own guard:
1225 // `tee >(wc -c) </dev/null` left wc blocked).
1226 let _psub_fds = PsubFdGuard;
1227 if let Some(s) = try_user_fn_override($name, &args) {
1228 return Value::Status(s);
1229 }
1230 // c:Src/exec.c — redirect failure in the current
1231 // scope means the command must NOT run. coreutils
1232 // shadows (cat / head / tail / etc.) take a separate
1233 // dispatch path from dispatch_builtin, so they need
1234 // their own gate. Without this `cat <&3` after a
1235 // closed-fd diagnostic still ran the shadow and
1236 // overwrote $? from the forced 1.
1237 let redir_failed = with_executor(|exec| {
1238 let f = exec.redirect_failed;
1239 exec.redirect_failed = false;
1240 f
1241 });
1242 if redir_failed {
1243 return Value::Status(1);
1244 }
1245 // `[builtins].coreutils_shadows = off` in
1246 // ~/.zshrs/zshrs.toml (or `ZSHRS_NO_COREUTILS_SHADOWS=1`
1247 // env override) bypasses the in-process shadow and
1248 // fork-execs the real /bin/X. Safety valve for any
1249 // script that hits an edge-case divergence between
1250 // the zshrs shadow and system coreutils. Cached
1251 // after first call, so the hot path is one atomic
1252 // load per shadowed-builtin invocation.
1253 if !crate::daemon_presence::coreutils_shadows_enabled() {
1254 return Value::Status(exec_system_command($name, &args));
1255 }
1256 let status = with_executor(|exec| exec.$method(&args));
1257 Value::Status(status)
1258 });
1259 };
1260 }
1261
1262 // Pure-passthru builtin: pops args, routes to canonical
1263 // `dispatch_builtin(name, args)` (which goes via execbuiltin →
1264 // BUILTINS[name] → bin_X). No pre/post bridge work. Used by
1265 // ~25 handlers that were 4-line copy-paste boilerplate.
1266 macro_rules! reg_passthru {
1267 ($vm:expr, $id:expr, $name:literal) => {
1268 $vm.register_builtin($id, |vm, argc| {
1269 let args = pop_args(vm, argc);
1270 // function > builtin: a same-named user function wins over
1271 // the builtin on the normal (CallBuiltin) invocation path.
1272 // The compiler's `user_function_shadow` already routes the
1273 // same-compile-unit case through CallFunction; this probe
1274 // extends that to the cross-unit / interactive case (define
1275 // `zstyle() { … }` on one line, call it on the next). The
1276 // forced `builtin NAME` / `command NAME` paths dispatch
1277 // through their own handlers, not this one, so they still
1278 // reach the builtin as required.
1279 if let Some(s) = try_user_fn_override($name, &args) {
1280 return Value::Status(s);
1281 }
1282 Value::Status(dispatch_builtin($name, args))
1283 });
1284 };
1285 }
1286
1287 // zshrs-original extension builtins (async / peach / doctor / …) that
1288 // route to an ExecutorContext method. Like `reg_overridable!`, they
1289 // probe `try_user_fn_override` FIRST so a user function of the same
1290 // name wins — zsh's alias → function → builtin dispatch order. Without
1291 // the probe, `doctor() { … }; doctor` silently ran the builtin and
1292 // ignored the function (function > builtin violated for these).
1293 macro_rules! reg_ext_overridable {
1294 ($vm:expr, $id:expr, $name:literal, $method:ident) => {
1295 $vm.register_builtin($id, |vm, argc| {
1296 let args = pop_args(vm, argc);
1297 if let Some(s) = try_user_fn_override($name, &args) {
1298 return Value::Status(s);
1299 }
1300 Value::Status(with_executor(|exec| exec.$method(&args)))
1301 });
1302 };
1303 }
1304
1305 // Core builtins
1306 vm.register_builtin(BUILTIN_CD, |vm, argc| {
1307 let args = pop_args(vm, argc);
1308 if let Some(s) = try_user_fn_override("cd", &args) {
1309 return Value::Status(s);
1310 }
1311 let status = dispatch_builtin("cd", args);
1312 // c:Src/builtin.c:1258 — `callhookfunc("chpwd", NULL, 1, NULL)`
1313 // after cd succeeds. The canonical port at
1314 // src/ported/utils.rs:1532 handles both the `chpwd` shfunc
1315 // dispatch AND the `chpwd_functions` array walk.
1316 if status == 0 {
1317 crate::ported::utils::callhookfunc("chpwd", None, 1, std::ptr::null_mut());
1318 }
1319 Value::Status(status)
1320 });
1321
1322 vm.register_builtin(BUILTIN_PWD, |vm, argc| {
1323 let args = pop_args(vm, argc);
1324 if let Some(s) = try_user_fn_override("pwd", &args) {
1325 return Value::Status(s);
1326 }
1327 // Route through the canonical execbuiltin path so the `rLP`
1328 // optstr at BUILTINS["pwd"] is parsed into `ops`.
1329 let status = dispatch_builtin("pwd", args);
1330 Value::Status(status)
1331 });
1332
1333 vm.register_builtin(BUILTIN_ECHO, |vm, argc| {
1334 let args = pop_args(vm, argc);
1335 if let Some(s) = try_user_fn_override("echo", &args) {
1336 return Value::Status(s);
1337 }
1338 // Update `$_` to the last arg before running. C zsh sets
1339 // zunderscore in execcmd_exec for every simple command,
1340 // including builtins.
1341 crate::ported::params::set_zunderscore(&args);
1342 let status = dispatch_builtin("echo", args);
1343 Value::Status(status)
1344 });
1345
1346 vm.register_builtin(BUILTIN_PRINT, |vm, argc| {
1347 let args = pop_args(vm, argc);
1348 if let Some(s) = try_user_fn_override("print", &args) {
1349 return Value::Status(s);
1350 }
1351 crate::ported::params::set_zunderscore(&args);
1352 let status = dispatch_builtin("print", args);
1353 Value::Status(status)
1354 });
1355
1356 reg_passthru!(vm, BUILTIN_PRINTF, "printf");
1357 reg_passthru!(vm, BUILTIN_EXPORT, "export");
1358 reg_passthru!(vm, BUILTIN_UNSET, "unset");
1359 // `source` (Src/builtin.c c:116) wired to bin_dot via BUILTINS.
1360 reg_passthru!(vm, BUILTIN_SOURCE, "source");
1361 reg_passthru!(vm, BUILTIN_DOT, ".");
1362 reg_passthru!(vm, BUILTIN_LOGOUT, "logout");
1363
1364 vm.register_builtin(BUILTIN_EXIT, |vm, argc| {
1365 let args = pop_args(vm, argc);
1366 let status = dispatch_builtin("exit", args);
1367 Value::Status(status)
1368 });
1369
1370 vm.register_builtin(BUILTIN_RETURN, |vm, argc| {
1371 let args = pop_args(vm, argc);
1372 // zsh: bare `return` (no arg) returns with the status of
1373 // the most recently executed command — `false; return`
1374 // returns 1, not 0. Direct port of zsh's bin_break/RETURN.
1375 // The executor's `last_status` is stale here (synced at
1376 // statement boundaries, not after each VM op), so read
1377 // the live `vm.last_status` instead.
1378 let live_status = vm.last_status;
1379 let status = {
1380 // Sync canonical LASTVAL to the VM's view BEFORE
1381 // bin_break("return") reads it for the no-arg fallback.
1382 with_executor(|exec| exec.set_last_status(live_status));
1383 dispatch_builtin("return", args)
1384 };
1385 Value::Status(status)
1386 });
1387
1388 vm.register_builtin(BUILTIN_TRUE, |vm, argc| {
1389 let args = pop_args(vm, argc);
1390 if let Some(s) = try_user_fn_override("true", &args) {
1391 return Value::Status(s);
1392 }
1393 // c:Src/exec.c:1257 — zsh sets `zunderscore` AT THE END of
1394 // each command (the `if (!noerrs)` block runs `zsfree(prev_argv0); …;
1395 // zunderscore = …`). For no-arg `true`, $_ becomes the
1396 // command name itself. Set DIRECTLY (not via pending_underscore)
1397 // so the NEXT command's argv-expansion of `$_` reads "true",
1398 // not the stale prior value — pending_underscore is consumed
1399 // by pop_args which runs AFTER argv expansion, too late.
1400 // c:Src/exec.c:1257 — `zunderscore = …` at end-of-command.
1401 // With args, $_ = args.last(). Without args, $_ = command name.
1402 // Write DIRECTLY to the canonical zunderscore static (the
1403 // underscoregetfn at params.rs:7003 reads from there); the
1404 // paramtab "_" slot is shadowed by lookup_special_var so
1405 // set_scalar on it has no effect on `$_` reads.
1406 if args.is_empty() {
1407 crate::ported::params::set_zunderscore(&["true".to_string()]);
1408 } else {
1409 crate::ported::params::set_zunderscore(&args);
1410 }
1411 // Route through canonical execbuiltin so PS4 xtrace fires
1412 // via the c:442 printprompt4 path.
1413 Value::Status(dispatch_builtin("true", args))
1414 });
1415 vm.register_builtin(BUILTIN_FALSE, |vm, argc| {
1416 let args = pop_args(vm, argc);
1417 if let Some(s) = try_user_fn_override("false", &args) {
1418 return Value::Status(s);
1419 }
1420 // Direct set; see BUILTIN_TRUE above for rationale.
1421 if args.is_empty() {
1422 crate::ported::params::set_zunderscore(&["false".to_string()]);
1423 } else {
1424 crate::ported::params::set_zunderscore(&args);
1425 }
1426 // Route through canonical execbuiltin — see BUILTIN_TRUE
1427 // above for the same rationale (xtrace + fast-path removal).
1428 let status = dispatch_builtin("false", args);
1429 Value::Status(status)
1430 });
1431 vm.register_builtin(BUILTIN_COLON, |vm, argc| {
1432 let args = pop_args(vm, argc);
1433 // Direct set; see BUILTIN_TRUE above for rationale.
1434 if args.is_empty() {
1435 crate::ported::params::set_zunderscore(&[":".to_string()]);
1436 } else {
1437 crate::ported::params::set_zunderscore(&args);
1438 }
1439 let status = dispatch_builtin(":", args);
1440 Value::Status(status)
1441 });
1442
1443 vm.register_builtin(BUILTIN_TEST, |vm, argc| {
1444 let args = pop_args(vm, argc);
1445 // Distinguish `[ … ]` from `test …` by sniffing the trailing
1446 // `]` — `[` requires it (c:Src/builtin.c:7241), `test` rejects
1447 // it. The compile path emits BUILTIN_TEST for both, so the
1448 // dispatch name carries the `[` vs `test` semantic for
1449 // execbuiltin's funcid (BIN_BRACKET=21 vs BIN_TEST=20). Without
1450 // this, bin_test's `if func == BIN_BRACKET` arm (which pops
1451 // the trailing `]`) never fired for `[` calls, so the `]`
1452 // leaked into evalcond as a positional and silently changed
1453 // the result. Bug surfaced via test_test_dashdash_unknown_condition.
1454 let name = if args.last().map(|s| s.as_str()) == Some("]") {
1455 "["
1456 } else {
1457 "test"
1458 };
1459 let status = dispatch_builtin(name, args);
1460 Value::Status(status)
1461 });
1462
1463 // Variable declaration. `local` (Src/builtin.c bin_local) handles
1464 // the scope chain (`pm->old = oldpm` at Src/params.c:1137 inside
1465 // createparam, `pm->level = locallevel` at Src/builtin.c:2576).
1466 // `typeset` / `declare` are aliases — fusevm maps both to
1467 // BUILTIN_TYPESET; compile_zsh special-cases `declare` to keep
1468 // the `declare:` error prefix.
1469 reg_passthru!(vm, BUILTIN_LOCAL, "local");
1470 reg_passthru!(vm, BUILTIN_TYPESET, "typeset");
1471
1472 reg_passthru!(vm, BUILTIN_DECLARE, "declare");
1473 reg_passthru!(vm, BUILTIN_READONLY, "readonly");
1474 reg_passthru!(vm, BUILTIN_INTEGER, "integer");
1475 reg_passthru!(vm, BUILTIN_FLOAT, "float");
1476 reg_passthru!(vm, BUILTIN_READ, "read");
1477 // c:Bug #504 — fusevm reserves BUILTIN_MAPFILE for the bash
1478 // mapfile/readarray builtins. Neither exists in zsh; in --zsh
1479 // parity mode the dispatch must emit "command not found" + rc=127
1480 // matching zsh's external-command-lookup miss. The previous wiring
1481 // left BUILTIN_MAPFILE unregistered, so fusevm's VM treated the op
1482 // as a no-op rc=0 — `mapfile` (and `readarray`) silently succeeded
1483 // in --zsh mode. The host gate in `dispatch_builtin_raw` never
1484 // fired because the compile path emitted `Op::CallBuiltin(31, ..)`
1485 // directly. Register the slot so the gate runs (or a future
1486 // non-zsh mode can wire in a real impl).
1487 vm.register_builtin(fusevm::shell_builtins::BUILTIN_MAPFILE, |vm, argc| {
1488 let args = pop_args(vm, argc);
1489 // The fusevm name→id map collapses both `mapfile` and
1490 // `readarray` to the same opcode; pick the right diagnostic
1491 // by sniffing the user's actual invocation. The xtrace ARGS
1492 // push earlier records the cmd-prefix as the bottom of the
1493 // popped argv, but `args` here excludes the prefix — so we
1494 // can't recover the user-typed name from the stack. Default
1495 // to `mapfile` (the more-common spelling); both produce
1496 // identical diagnostics in any case.
1497 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed) {
1498 eprintln!("zsh:1: command not found: mapfile");
1499 let _ = args;
1500 return Value::Status(127);
1501 }
1502 // Non-zsh modes (bash drop-in): mapfile / readarray reads lines
1503 // from stdin (or `-u fd`) into an array. Handled by the ported
1504 // bash builtin in ext_builtins.
1505 Value::Status(crate::extensions::ext_builtins::readarray(&args))
1506 });
1507 reg_passthru!(vm, BUILTIN_BREAK, "break");
1508 reg_passthru!(vm, BUILTIN_CONTINUE, "continue");
1509 reg_passthru!(vm, BUILTIN_SHIFT, "shift");
1510
1511 vm.register_builtin(BUILTIN_EVAL, |vm, argc| {
1512 // Direct port of `bin_eval(UNUSED(char *nam), char **argv, UNUSED(Options ops), UNUSED(int func))` body from Src/builtin.c:6151:
1513 // `if (!*argv) return 0;`
1514 // `prog = parse_string(zjoin(argv, ' ', 1), 1);`
1515 // `execode(prog, 1, 0, "eval");`
1516 // The execode invocation lives here (not in the canonical
1517 // free-fn) because it must run through the bytecode VM's
1518 // current executor — the same VM that's mid-dispatch.
1519 let mut args = pop_args(vm, argc);
1520 // c:Src/builtin.c:407-411 — generic `--` end-of-options
1521 // strip applied by `execbuiltin` for builtins that have
1522 // NULL optstr AND no BINF_HANDLES_OPTS. `eval` qualifies
1523 // (Src/builtin.c:65 `BUILTIN("eval", BINF_PSPECIAL, ...,
1524 // NULL, NULL)`). The BUILTIN_EVAL fast-path bypasses
1525 // execbuiltin, so we mirror the strip inline. Bug #319.
1526 if args.first().is_some_and(|s| s == "--") {
1527 args.remove(0);
1528 }
1529 if args.is_empty() {
1530 return Value::Status(0); // c:6160
1531 }
1532 let src = args.join(" "); // c:6166
1533 // c:Src/builtin.c:6164-6165 — `if (!ineval) scriptname =
1534 // "(eval)";`. Diagnostics emitted while the eval body runs
1535 // (command-not-found, parse errors, etc.) use scriptname as
1536 // the source-context prefix. Without setting it here the
1537 // BUILTIN_EVAL fast-path leaked the outer "zsh" prefix
1538 // through, breaking the `(eval):N:` convention zsh uses
1539 // for in-eval errors. Bug #420.
1540 // c:Src/builtin.c:6209 — `execode(prog, 1, 0, "eval");`. execode
1541 // (c:Src/exec.c:1245-1266) APPENDS its context argument to
1542 // `zsh_eval_context` for the duration of the body, so code inside
1543 // `eval` sees `cmdarg:eval` (and `cmdarg:shfunc:eval` when the eval is
1544 // in a function). zshrs pushed "shfunc" and, since #1065, "cmdsubst",
1545 // but never "eval". Popped on every return path by the guard, matching
1546 // execode's stack discipline. Bug #1065 (eval leg).
1547 let sync_eval_ctx = |stack: &[String]| {
1548 let joined = stack.join(":");
1549 if let Ok(mut tab) = crate::ported::params::paramtab().write() {
1550 if let Some(pm) = tab.get_mut("zsh_eval_context") {
1551 pm.u_arr = Some(stack.to_vec());
1552 pm.node.flags &= !(crate::ported::zsh_h::PM_UNSET as i32);
1553 }
1554 if let Some(pm) = tab.get_mut("ZSH_EVAL_CONTEXT") {
1555 pm.u_str = Some(joined);
1556 pm.node.flags &= !(crate::ported::zsh_h::PM_UNSET as i32);
1557 }
1558 }
1559 };
1560 if let Ok(mut ctx) = crate::ported::exec::zsh_eval_context.lock() {
1561 ctx.push("eval".to_string());
1562 sync_eval_ctx(&ctx);
1563 }
1564 struct EvalCtxGuard<F: Fn(&[String])>(F);
1565 impl<F: Fn(&[String])> Drop for EvalCtxGuard<F> {
1566 fn drop(&mut self) {
1567 if let Ok(mut ctx) = crate::ported::exec::zsh_eval_context.lock() {
1568 ctx.pop();
1569 (self.0)(&ctx);
1570 }
1571 }
1572 }
1573 let _eval_ctx_guard = EvalCtxGuard(sync_eval_ctx);
1574 // c:Src/builtin.c:6163-6178 — `eval` pushes a funcstack frame named
1575 // "(eval)" (tp = FS_EVAL), gated on `ineval = !isset(EVALLINENO)` /
1576 // `if (!ineval)` — i.e. pushed when EVAL_LINENO is SET, which is the
1577 // zsh default. zshrs already set `scriptname = "(eval)"` (below) but
1578 // never pushed the frame, so `eval '…${#funcstack}'` reported 0 where
1579 // zsh reports 1, and inside a function `${(j:,:)funcstack}` was `f`
1580 // rather than `(eval),f`. Both shells already agreed under
1581 // `unsetopt evallineno` (no frame), so the option gate is load-bearing
1582 // and is mirrored here. Popped on every return path by the guard.
1583 // Bug #1066.
1584 let eval_pushed_frame = crate::ported::zsh_h::isset(crate::ported::zsh_h::EVALLINENO);
1585 if eval_pushed_frame {
1586 let caller = {
1587 let stk = crate::ported::modules::parameter::FUNCSTACK
1588 .lock()
1589 .unwrap_or_else(|e| e.into_inner());
1590 stk.last()
1591 .map(|f| f.name.clone()) // c:6167 funcstack->name
1592 .or_else(|| crate::ported::utils::argzero()) // c:6167 argzero
1593 };
1594 let frame = crate::ported::zsh_h::funcstack {
1595 prev: None, // c:6166 (Vec-stack: index encodes link)
1596 name: "(eval)".to_string(), // c:6166 fstack.name = scriptname
1597 filename: None,
1598 caller,
1599 flineno: 0,
1600 lineno: 0, // c:6169
1601 tp: crate::ported::zsh_h::FS_EVAL, // c:6170
1602 };
1603 crate::ported::modules::parameter::FUNCSTACK
1604 .lock()
1605 .unwrap_or_else(|e| e.into_inner())
1606 .push(frame); // c:6178 funcstack = &fstack
1607 }
1608 struct EvalFuncstackGuard(bool);
1609 impl Drop for EvalFuncstackGuard {
1610 fn drop(&mut self) {
1611 if self.0 {
1612 crate::ported::modules::parameter::FUNCSTACK
1613 .lock()
1614 .unwrap_or_else(|e| e.into_inner())
1615 .pop();
1616 }
1617 }
1618 }
1619 let _eval_fs_guard = EvalFuncstackGuard(eval_pushed_frame);
1620 let oscriptname = crate::ported::utils::scriptname_get();
1621 crate::ported::utils::set_scriptname(Some("(eval)".to_string()));
1622 // Recursion backstop — c:Src/jobs.c:1878-1884. zsh caps eval recursion
1623 // via its job table (every eval'd pipeline grabs a job slot; the table
1624 // caps at MAX_MAXJOBS → "job table full or recursion limit exceeded").
1625 // The fusevm runtime allocates no job per pipeline, and nested evals
1626 // push no funcstack frame (INEVAL suppression, c:6164), so eval nesting
1627 // is invisible to both the job table AND FUNCNEST/FUNCSTACK — runaway
1628 // `eval`-string recursion overflowed the 256 MB main-thread stack →
1629 // uncatchable SIGBUS. Track eval re-entry depth (the Rust proxy for
1630 // held job slots) and refuse at the same MAX_MAXJOBS ceiling.
1631 let eval_depth = crate::vm_helper::EVAL_RECURSION_DEPTH.with(|d| {
1632 let v = d.get() + 1;
1633 d.set(v);
1634 v
1635 });
1636 let mut status = if eval_depth >= crate::ported::jobs::MAX_MAXJOBS {
1637 crate::ported::utils::zerr("job table full or recursion limit exceeded");
1638 1
1639 } else {
1640 with_executor(|exec| {
1641 // c:6175 execode
1642 exec.execute_script(&src).unwrap_or(1)
1643 })
1644 };
1645 crate::vm_helper::EVAL_RECURSION_DEPTH.with(|d| d.set(d.get().saturating_sub(1)));
1646 // c:Src/builtin.c:6211-6212 — `if (errflag && !lastval)
1647 // lastval = errflag;`
1648 // c:Src/builtin.c:6221 — `errflag &= ~ERRFLAG_ERROR;`
1649 // eval is a CONTAINMENT boundary: an error inside the eval
1650 // body (readonly reassign, bad assoc set, ${unset?msg}, …)
1651 // breaks the eval body's lists via errflag, then eval clears
1652 // the flag and returns lastval, and the CALLER's next list
1653 // runs. zsh 5.9: `eval 'assoc=(odd)'; echo "after $?"`
1654 // prints `after 1` in -c, script, and stdin contexts.
1655 {
1656 use std::sync::atomic::Ordering;
1657 let ef = crate::ported::utils::errflag.load(Ordering::Relaxed)
1658 & crate::ported::zsh_h::ERRFLAG_ERROR;
1659 if ef != 0 && status == 0 {
1660 status = ef; // c:6212 lastval = errflag
1661 }
1662 crate::ported::utils::errflag
1663 .fetch_and(!crate::ported::zsh_h::ERRFLAG_ERROR, Ordering::Relaxed);
1664 }
1665 crate::ported::utils::set_scriptname(oscriptname);
1666 Value::Status(status)
1667 });
1668
1669 // `builtin foo args…`: precmd-modifier that forces builtin dispatch,
1670 // bypassing alias AND function lookup. Without this, `builtin cd /`
1671 // inside a user `cd () { … }` wrapper recurses (real-world ZPWR pattern).
1672 // Handler pops argc args from the stack, treats args[0] as the builtin
1673 // name, and dispatches the rest via `dispatch_builtin` → `execbuiltin`
1674 // → `bin_*` directly. No function/alias lookup happens.
1675 vm.register_builtin(BUILTIN_BUILTIN, |vm, argc| {
1676 let args = pop_args(vm, argc);
1677 let Some((name, rest)) = args.split_first() else {
1678 // `builtin` with no args → list builtins (zsh emits nothing,
1679 // exit 0). Match that behavior; the BIN_BUILTIN bin_* in C
1680 // does the same default-list-nothing.
1681 return Value::Status(0);
1682 };
1683 // zshrs extension builtins (daemon z* family: zd, zcache, zjob,
1684 // …) are dispatched by name via try_dispatch instead of living
1685 // in builtintab — but they ARE builtins, so the `builtin`
1686 // precommand must reach them (`builtin zd ping` errored
1687 // "no such builtin: zd" while bare `zd ping` worked).
1688 if crate::daemon::builtins::is_zshrs_builtin(name) {
1689 let argv: Vec<String> =
1690 std::iter::once(name.to_string()).chain(rest.iter().cloned()).collect();
1691 return Value::Status(
1692 crate::daemon::builtins::try_dispatch(name, &argv).unwrap_or(1),
1693 );
1694 }
1695 // c:Src/exec.c:3435-3436 — `builtin NAME` with NAME not in
1696 // builtintab emits `zwarn("no such builtin: %s", cmdarg)`
1697 // and returns 1. zshrs's dispatch_builtin_raw bare-returned 1
1698 // silently. Probe the table here so the diagnostic fires
1699 // before dispatch.
1700 let tab = crate::ported::builtin::createbuiltintable();
1701 if !tab.contains_key(name.as_str()) {
1702 // zshrs-original opcode builtins (async, doctor, peach, …) aren't
1703 // in builtintab; `builtin NAME` must still reach them.
1704 if let Some(status) = try_run_registered_builtin(name, rest) {
1705 return Value::Status(status);
1706 }
1707 // c:Src/exec.c:3436 — `zwarn("no such builtin: %s", cmdarg);`.
1708 // Route through the ported `zwarn` rather than formatting the
1709 // prefix by hand: zwarn emits zsh's `zsh:LINE:` prefix, and the
1710 // hand-rolled `eprintln!` here printed `zshrs:1:` instead. Of the
1711 // twelve error shapes probed this was the ONLY one carrying the
1712 // wrong prefix — the ported twin at exec.rs:9878 already used
1713 // zwarn correctly, so this was a reimplementation shadowing a
1714 // faithful port (same shape as #1027 / #1031 / #1044 / #1050).
1715 // Bug #1063.
1716 crate::ported::utils::zwarn(&format!("no such builtin: {}", name));
1717 return Value::Status(1);
1718 }
1719 // `builtin foo` MUST bypass function shadow — that's the whole
1720 // point of the prefix. Use the _raw helper, not the shadow-aware
1721 // one. Without this, `cd () { builtin cd "$@"; }` recurses.
1722 Value::Status(dispatch_builtin_raw(name, rest.to_vec()))
1723 });
1724
1725 // `command foo args…` — BINF_COMMAND prefix (Src/builtin.c:44). Zsh
1726 // semantic: bypass alias+function lookup, search builtin then $PATH.
1727 // Without this, `cd () { command cd "$@" }` would re-invoke the user
1728 // wrapper (same root cause as the `builtin` bug). Flags `-p`/`-v`/`-V`
1729 // route to bin_whence with BIN_COMMAND funcid; bare `command foo`
1730 // dispatches builtin if present, else external (no fork — direct
1731 // spawn via execute_external since zshrs is non-forking).
1732 // BUILTIN_COMMAND — `command [-p] [-v|-V] cmd args…` BIN_PREFIX
1733 // (Src/builtin.c:45). PURE PASSTHRU: prepend "command" and hand
1734 // to `exec::execcmd_compile_head` (the fusevm-bytecode-time head
1735 // resolver mirroring `Src/exec.c::execcmd_exec` precommand-modifier
1736 // walk at c:3104-3187). That helper already does the -p / -v / -V
1737 // option parsing, surfaces `has_command_vv` for the whence
1738 // redirect, and reports the dispatch shape (is_builtin vs external).
1739 vm.register_builtin(BUILTIN_COMMAND, |vm, argc| {
1740 let args = pop_args(vm, argc);
1741 let mut full = Vec::with_capacity(args.len() + 1);
1742 full.push("command".to_string());
1743 full.extend(args.clone());
1744 let dispatch =
1745 crate::ported::exec::execcmd_compile_head(&full, crate::ported::zsh_h::WC_SIMPLE);
1746 let post = &full[dispatch.precmd_skip..];
1747 // c:Src/builtin.c:4500 — `command -p` resets PATH for the
1748 // exec to the POSIX-defined default (`getconf PATH`), so
1749 // standard utilities resolve even when the caller has
1750 // emptied $PATH. zsh restores the original PATH after the
1751 // command returns. Mirror via a scoped env::set_var.
1752 //
1753 // command's OWN options end at the first non-flag arg —
1754 // everything after the command name belongs to IT. The
1755 // previous `.any()` scan over ALL args stole `-p` from
1756 // `command mkdir -p DIR` (zconvey.plugin.zsh:44), stripping
1757 // the flag before /bin/mkdir ran → "File exists" errors on
1758 // every re-source.
1759 let mut lead = 0usize;
1760 let mut dash_p = false;
1761 let mut kept_flags: Vec<String> = Vec::new();
1762 for a in post.iter() {
1763 let s = a.as_str();
1764 if s == "--" {
1765 lead += 1;
1766 break;
1767 }
1768 if s.starts_with('-')
1769 && s.len() >= 2
1770 && s[1..].chars().all(|c| c == 'p' || c == 'v' || c == 'V')
1771 {
1772 if s.contains('p') {
1773 dash_p = true;
1774 }
1775 // -v / -V drive the whence-style lookup downstream —
1776 // keep them in post (only the PATH-reset `p` is
1777 // consumed here).
1778 let rest: String = s[1..].chars().filter(|c| *c != 'p').collect();
1779 if !rest.is_empty() {
1780 kept_flags.push(format!("-{}", rest));
1781 }
1782 lead += 1;
1783 continue;
1784 }
1785 break;
1786 }
1787 let mut post: Vec<String> = {
1788 let mut v = kept_flags;
1789 v.extend(post[lead..].iter().cloned());
1790 v
1791 };
1792 // c:Src/exec.c:3176-3177 — `BINF_COMMAND` arm strips a single
1793 // leading `--` end-of-options marker.
1794 // `execcmd_compile_head` (src/ported/exec.rs:1042) performs
1795 // this removal on its LOCAL `preargs` Vec but doesn't surface
1796 // the modified args; the caller still sees `--` in `full` and
1797 // tried to dispatch it as the command name. Bug #251. Mirror
1798 // the C strip here so `command -- echo hi` and
1799 // `command -p -- echo hi` route correctly.
1800 if let Some(first) = post.first() {
1801 if first == "--" {
1802 post.remove(0);
1803 }
1804 }
1805 let post = post.as_slice();
1806 let _path_guard = if dash_p {
1807 let saved = env::var("PATH").ok();
1808 let default_path = std::process::Command::new("getconf")
1809 .arg("PATH")
1810 .output()
1811 .ok()
1812 .and_then(|o| String::from_utf8(o.stdout).ok())
1813 .map(|s| s.trim().to_string())
1814 .filter(|s| !s.is_empty())
1815 .unwrap_or_else(|| "/usr/bin:/bin:/usr/sbin:/sbin".to_string());
1816 env::set_var("PATH", &default_path);
1817 crate::ported::params::setsparam("PATH", &default_path);
1818 Some(saved)
1819 } else {
1820 None
1821 };
1822 struct PathGuard {
1823 saved: Option<String>,
1824 active: bool,
1825 }
1826 impl Drop for PathGuard {
1827 fn drop(&mut self) {
1828 if !self.active {
1829 return;
1830 }
1831 match self.saved.take() {
1832 Some(p) => {
1833 env::set_var("PATH", &p);
1834 crate::ported::params::setsparam("PATH", &p);
1835 }
1836 None => {
1837 env::remove_var("PATH");
1838 crate::ported::params::setsparam("PATH", "");
1839 }
1840 }
1841 }
1842 }
1843 let _restore = PathGuard {
1844 saved: _path_guard.unwrap_or(None),
1845 active: dash_p,
1846 };
1847 if dispatch.has_command_vv {
1848 // `-v` / `-V` → bin_whence with BIN_COMMAND funcid.
1849 let mut ops = options {
1850 ind: [0u8; MAX_OPS],
1851 args: Vec::new(),
1852 argscount: 0,
1853 argsalloc: 0,
1854 };
1855 let mut name_pos = 0usize;
1856 let mut flag_byte = b'v';
1857 for (i, a) in post.iter().enumerate() {
1858 if a.starts_with('-') && a.len() >= 2 {
1859 let body = &a.as_bytes()[1..];
1860 if body.contains(&b'V') {
1861 flag_byte = b'V';
1862 }
1863 name_pos = i + 1;
1864 } else {
1865 name_pos = i;
1866 break;
1867 }
1868 }
1869 ops.ind[flag_byte as usize] = 1;
1870 let whence_args: Vec<String> = post[name_pos..].to_vec();
1871 return Value::Status(crate::ported::builtin::bin_whence(
1872 "command",
1873 &whence_args,
1874 &ops,
1875 crate::ported::hashtable_h::BIN_COMMAND,
1876 ));
1877 }
1878 if dispatch.is_empty_command {
1879 return Value::Status(0);
1880 }
1881 let Some((name, rest)) = post.split_first() else {
1882 return Value::Status(0);
1883 };
1884 // c:Src/exec.c:3275-3278 — `execcmd_compile_head` cleared
1885 // hn for the BINF_COMMAND + !POSIXBUILTINS case, surfacing
1886 // is_builtin=false. Run as external. Under POSIXBUILTINS
1887 // dispatch.is_builtin would be true; honour it.
1888 let n = name.clone();
1889 let r = rest.to_vec();
1890 if dispatch.is_builtin
1891 && crate::ported::builtin::BUILTINS
1892 .iter()
1893 .any(|b| b.node.nam == n.as_str())
1894 {
1895 return Value::Status(dispatch_builtin_raw(&n, r));
1896 }
1897 Value::Status(with_executor(|exec| exec.execute_external(&n, &r, &[])).unwrap_or(127))
1898 });
1899
1900 // `exec cmd args…` — BINF_EXEC prefix (Src/builtin.c:45). Zsh
1901 // semantic: replace the current shell process with `cmd`. On Unix
1902 // this is `execvp(2)`; the call only returns on error. zshrs is
1903 // non-forking, so the shell process IS the calling process —
1904 // execvp here directly replaces it. Options `-a name` (override
1905 // argv[0]), `-c` (clean env), `-l` (login shell — prepend `-`)
1906 // ported minimally; advanced redirect-only `exec >file` is handled
1907 // upstream by compile_zsh and never reaches this handler.
1908 vm.register_builtin(BUILTIN_EXEC, |vm, argc| {
1909 let mut args = pop_args(vm, argc);
1910 let mut argv0_override: Option<String> = None;
1911 let mut clean_env = false;
1912 let mut login = false;
1913 let mut i = 0;
1914 // c:Src/builtin.c:1075-1080 — track if any flag was consumed.
1915 // `exec -c`, `exec -l`, `exec -a NAME` without a following
1916 // command emit "exec requires a command to execute" rc=1.
1917 // Bare `exec` (no args at all) is the silent-redirect-apply
1918 // form per POSIX.
1919 let mut saw_flag = false;
1920 while i < args.len() {
1921 let a = &args[i];
1922 if a == "--" {
1923 args.remove(i);
1924 break;
1925 }
1926 // c:Src/builtin.c:42 `BIN_PREFIX("-", BINF_DASH)`. A bare
1927 // `-` is its own BINF_PREFIX builtin (BINF_DASH flag —
1928 // "login shell, prepend `-` to argv[0]"). In the canonical
1929 // precmd-walk at Src/exec.c:3056-3091 a bare `-` after
1930 // `exec` is recognized AS a builtin and stripped from
1931 // preargs (precmd_skip++), accumulating BINF_DASH into
1932 // cflags. The fast-path here bypasses execcmd_compile_head,
1933 // so we mirror the strip locally: bare `-` → set login,
1934 // remove, continue. Without this `exec -` (with no command
1935 // following) tried to exec `-` as a literal command and
1936 // exited the shell. Bug #252.
1937 if a == "-" {
1938 saw_flag = true;
1939 login = true;
1940 args.remove(i);
1941 continue;
1942 }
1943 if !a.starts_with('-') || a.len() < 2 {
1944 break;
1945 }
1946 match a.as_str() {
1947 "-a" => {
1948 saw_flag = true;
1949 args.remove(i);
1950 if i < args.len() {
1951 argv0_override = Some(args.remove(i));
1952 }
1953 }
1954 "-c" => {
1955 saw_flag = true;
1956 clean_env = true;
1957 args.remove(i);
1958 }
1959 "-l" => {
1960 saw_flag = true;
1961 login = true;
1962 args.remove(i);
1963 }
1964 _ => {
1965 // c:Src/exec.c:3196-3208 — when an unrecognized
1966 // `-X`-style arg has NO following arg, the lexer's
1967 // IS_DASH walk hits the "no next node" branch at
1968 // c:3199 before the unknown-flag-letter switch at
1969 // c:3249, so the canonical message is "exec
1970 // requires a command to execute" rc=1 (verified vs
1971 // `/opt/homebrew/bin/zsh -fc 'exec --bad'`).
1972 // Consume the lone flag so the post-loop check
1973 // fires. When a following arg exists, leave the
1974 // unknown-flag arg in place — that arg becomes
1975 // the command name and execution proceeds.
1976 if args.len() == 1 {
1977 saw_flag = true;
1978 args.remove(i);
1979 continue;
1980 }
1981 break;
1982 }
1983 }
1984 }
1985 let Some(cmd) = args.first().cloned() else {
1986 if saw_flag {
1987 // c:Src/builtin.c:1078-1080 — flags consumed but no
1988 // command follows → "exec requires a command to
1989 // execute" rc=1.
1990 eprintln!("zshrs:1: exec requires a command to execute");
1991 return Value::Status(1);
1992 }
1993 // `exec` with no command + no redirects = no-op success.
1994 return Value::Status(0);
1995 };
1996 let rest: Vec<String> = args[1..].to_vec();
1997 let display_argv0 = match argv0_override {
1998 Some(a) => a,
1999 None => {
2000 if login {
2001 format!("-{}", cmd)
2002 } else {
2003 cmd.clone()
2004 }
2005 }
2006 };
2007 // c:Src/exec.c::execcmd — `exec funcname` runs the function
2008 // in-process as the shell's last act, then exits with the
2009 // function's status. zsh's dispatcher falls through from the
2010 // BINF_EXEC prefix into the normal Builtin/External/Function
2011 // resolution and only execvp's if the target ISN'T a
2012 // function. Bug #101 in docs/BUGS.md: zshrs's exec went
2013 // straight to execvp and errored `not found` for shell
2014 // functions.
2015 //
2016 // For both subshell and top-level contexts: dispatch through
2017 // the function/builtin lookup first; only fall through to
2018 // execvp/spawn if the name isn't shell-resolvable.
2019 let has_user_fn = with_executor(|exec| exec.functions_compiled.contains_key(&cmd));
2020 if has_user_fn {
2021 let status =
2022 with_executor(|exec| exec.dispatch_function_call(&cmd, &rest).unwrap_or(127));
2023 // Top-level `exec funcname` — exit the shell with the
2024 // function's status (mirrors C's "exec replaces shell as
2025 // last act"). Subshell `(exec funcname)` — return through
2026 // the EXIT_PENDING path so the subshell body aborts and
2027 // the parent resumes via subshell_end.
2028 let in_subshell_now = with_executor(|exec| !exec.subshell_snapshots.is_empty());
2029 if in_subshell_now {
2030 crate::ported::builtin::EXIT_VAL
2031 .store(status, std::sync::atomic::Ordering::Relaxed);
2032 crate::ported::builtin::EXIT_PENDING.store(1, std::sync::atomic::Ordering::Relaxed);
2033 return Value::Status(status);
2034 }
2035 std::process::exit(status);
2036 }
2037 // c:Src/exec.c — builtin path: `exec builtin` runs the
2038 // builtin in-process and exits.
2039 let bn_in_tab = crate::ported::builtin::createbuiltintable().contains_key(&cmd);
2040 if bn_in_tab {
2041 let status = dispatch_builtin_raw(&cmd, rest.clone());
2042 let in_subshell_now = with_executor(|exec| !exec.subshell_snapshots.is_empty());
2043 if in_subshell_now {
2044 crate::ported::builtin::EXIT_VAL
2045 .store(status, std::sync::atomic::Ordering::Relaxed);
2046 crate::ported::builtin::EXIT_PENDING.store(1, std::sync::atomic::Ordering::Relaxed);
2047 return Value::Status(status);
2048 }
2049 std::process::exit(status);
2050 }
2051 // c:Src/exec.c — `exec` inside a subshell (`(exec cmd)`)
2052 // replaces ONLY the subshell child process; the parent shell
2053 // continues. C zsh always forks for `(...)`, so the actual
2054 // execvp lands in the forked child. zshrs runs subshells via
2055 // a snapshot/restore pattern in the SAME process — calling
2056 // execvp here would replace the parent too. Bug #94 in
2057 // docs/BUGS.md.
2058 //
2059 // Detect subshell context via the non-empty
2060 // `subshell_snapshots` stack. When in a subshell: spawn the
2061 // command as a child, wait for it, then signal the subshell
2062 // body to abort (return Status(N) and the caller's
2063 // subshell_end will pop the snapshot and resume the parent).
2064 let in_subshell = with_executor(|exec| !exec.subshell_snapshots.is_empty());
2065 if in_subshell {
2066 let mut command = std::process::Command::new(&cmd);
2067 command.arg0(&display_argv0);
2068 command.args(&rest);
2069 if clean_env {
2070 command.env_clear();
2071 }
2072 // Queue signals across spawn+wait so the SIGCHLD reaper
2073 // can't reap this child before child.wait() does — see
2074 // ForegroundWaitGuard.
2075 let _wait_guard = ForegroundWaitGuard::enter();
2076 let status = match command.spawn() {
2077 Ok(mut child) => match child.wait() {
2078 Ok(s) => s.code().unwrap_or(127),
2079 Err(_) => 127,
2080 },
2081 Err(e) => {
2082 // c:Src/exec.c:797 — `zerr("%e: %s", lerrno, arg0)`
2083 // when arg0 contains `/`.
2084 // c:872-876 — when arg0 has no `/` (PATH search
2085 // path), C tracks the "good" errno
2086 // via `isgooderr`; if all PATH entries
2087 // were ENOENT-not-good, eno stays 0
2088 // and C emits `command not found: %s`
2089 // instead of strerror.
2090 // %e expands to strerror(errno) with the first
2091 // letter lowercased (unless errno == EIO; see
2092 // Src/utils.c:362-368). `zerr` prepends the
2093 // scriptname:lineno: prefix — matching zsh's
2094 // canonical `zsh:N: <errmsg>: <cmd>` pattern.
2095 // Previously emitted `zshrs: exec: {}: not found`
2096 // (wrong prefix, hardcoded message, missing
2097 // lineno). Bug #140 in docs/BUGS.md.
2098 let errno = e.raw_os_error().unwrap_or(libc::ENOENT);
2099 let has_slash = cmd.contains('/');
2100 if !has_slash && errno == libc::ENOENT {
2101 // c:876 — PATH search exhausted with no good
2102 // errno → `command not found: arg0`.
2103 crate::ported::utils::zerr(&format!("command not found: {}", cmd));
2104 } else {
2105 let mut errmsg = crate::ported::compat::strerror(errno);
2106 if errno != libc::EIO {
2107 if let Some(c) = errmsg.chars().next() {
2108 errmsg = format!(
2109 "{}{}",
2110 c.to_ascii_lowercase(),
2111 &errmsg[c.len_utf8()..]
2112 );
2113 }
2114 }
2115 crate::ported::utils::zerr(&format!("{}: {}", errmsg, cmd));
2116 }
2117 // c:881 — `_exit((eno == EACCES || eno == ENOEXEC) ? 126 : 127);`
2118 if errno == libc::EACCES || errno == libc::ENOEXEC {
2119 126
2120 } else {
2121 127
2122 }
2123 }
2124 };
2125 // Mark the subshell as exec-replaced so subsequent body
2126 // commands skip — mirrors the post-execvp "child process
2127 // is gone" reality in C. EXIT_PENDING + EXIT_VAL drive
2128 // the next ERREXIT_CHECK to unwind to the subshell-end
2129 // patch.
2130 crate::ported::builtin::EXIT_VAL.store(status, std::sync::atomic::Ordering::Relaxed);
2131 crate::ported::builtin::EXIT_PENDING.store(1, std::sync::atomic::Ordering::Relaxed);
2132 return Value::Status(status);
2133 }
2134 let mut command = std::process::Command::new(&cmd);
2135 command.arg0(&display_argv0);
2136 command.args(&rest);
2137 if clean_env {
2138 command.env_clear();
2139 }
2140 use std::os::unix::process::CommandExt;
2141 // `exec` returns the OS error iff exec(2) failed; on success
2142 // it never returns. Match zsh: print the error to stderr with
2143 // the `exec` prefix and exit 127 (cmd not found) or 126 (not
2144 // executable).
2145 let err = command.exec();
2146 // c:Src/exec.c:797 / c:872-876 — same format as in-subshell
2147 // branch. arg0-has-/ → `<strerror>: <cmd>`; arg0-no-/ +
2148 // ENOENT → `command not found: <cmd>`. Lowercase strerror
2149 // first letter unless EIO. Bug #140 in docs/BUGS.md.
2150 let errno = err.raw_os_error().unwrap_or(libc::ENOENT);
2151 let has_slash = cmd.contains('/');
2152 if !has_slash && errno == libc::ENOENT {
2153 crate::ported::utils::zerr(&format!("command not found: {}", cmd));
2154 } else {
2155 let mut errmsg = crate::ported::compat::strerror(errno);
2156 if errno != libc::EIO {
2157 if let Some(c) = errmsg.chars().next() {
2158 errmsg = format!("{}{}", c.to_ascii_lowercase(), &errmsg[c.len_utf8()..]);
2159 }
2160 }
2161 crate::ported::utils::zerr(&format!("{}: {}", errmsg, cmd));
2162 }
2163 // c:881 — `_exit((eno == EACCES || eno == ENOEXEC) ? 126 : 127);`
2164 let code = if errno == libc::EACCES || errno == libc::ENOEXEC {
2165 126
2166 } else {
2167 127
2168 };
2169 std::process::exit(code);
2170 });
2171
2172 reg_passthru!(vm, BUILTIN_LET, "let");
2173
2174 // Job control
2175 reg_passthru!(vm, BUILTIN_JOBS, "jobs");
2176 reg_passthru!(vm, BUILTIN_FG, "fg");
2177 reg_passthru!(vm, BUILTIN_BG, "bg");
2178 reg_passthru!(vm, BUILTIN_KILL, "kill");
2179 reg_passthru!(vm, BUILTIN_DISOWN, "disown");
2180 reg_passthru!(vm, BUILTIN_WAIT, "wait");
2181 reg_passthru!(vm, BUILTIN_SUSPEND, "suspend");
2182
2183 // History — `fc` / `history` / `r` all route to `bin_fc` (zsh
2184 // registers them as aliases of the same builtin per Src/builtin.c).
2185 reg_passthru!(vm, BUILTIN_FC, "fc");
2186 reg_passthru!(vm, BUILTIN_HISTORY, "history");
2187 reg_passthru!(vm, BUILTIN_R, "r");
2188
2189 // Aliases — alias is `BINF_MAGICEQUALS` per Src/builtin.c:50.
2190 // c:Src/exec.c:3298-3304 — when a builtin has BINF_MAGICEQUALS,
2191 // execcmd_exec sets esprefork = PREFORK_TYPESET and calls
2192 // `prefork(args, esprefork, NULL)` on the argv. prefork (subst.c:
2193 // 100) drives `filesub` on each word (c:133), which (c:677-686)
2194 // looks for the assignment Equals and runs `filesubstr` on the
2195 // VALUE side. That's how `alias bad===` triggers equalsubstr's
2196 // "= not found" via the inner Equals after the first `=`.
2197 //
2198 // The fusevm dispatch path doesn't go through execcmd_exec, so
2199 // BUILTIN_ALIAS previously passed args straight to bin_alias with
2200 // no expansion — `alias x=~/foo` stored literal `~/foo` (no tilde
2201 // expand), `alias bad===` stored a broken entry without firing
2202 // the "= not found" diagnostic. The prefork(PREFORK_TYPESET) runs
2203 // per arg word via BUILTIN_MAGIC_EQUALS_PREFORK ops that
2204 // compile_simple emits BEFORE the redirect scope opens (matching
2205 // c:3304 prefork-before-addfd order), so the dispatch here is a
2206 // plain passthrough — re-running prefork would double-fire the
2207 // "= not found" diagnostic.
2208 reg_passthru!(vm, BUILTIN_ALIAS, "alias");
2209 // c:Src/exec.c:3298-3304 — per-word magic-equals prefork; see the
2210 // const doc at BUILTIN_MAGIC_EQUALS_PREFORK. prefork's filesub
2211 // trigger (subst.c:678 `strchr(*namptr+1, Equals)`) looks for the
2212 // EQUALS TOKEN, not literal `=`. The fusevm path delivers args
2213 // already-untokenized, so re-tokenize each element via
2214 // `shtokenize` (the same call C's lexer makes implicitly when
2215 // assembling the word) so prefork sees Equals tokens at `=`
2216 // boundaries and Tilde tokens at `~` starts. After prefork
2217 // expands, untokenize for storage.
2218 vm.register_builtin(BUILTIN_MAGIC_EQUALS_PREFORK, |vm, _argc| {
2219 let raw = vm.pop();
2220 let inputs: Vec<String> = match raw {
2221 Value::Array(items) => items.into_iter().map(|v| v.to_str()).collect(),
2222 other => vec![other.to_str()],
2223 };
2224 let mut as_linklist: crate::ported::linklist::LinkList<String> = Default::default();
2225 for s in &inputs {
2226 let mut tokd = s.clone();
2227 crate::ported::glob::shtokenize(&mut tokd);
2228 as_linklist.push_back(tokd);
2229 }
2230 let mut rf = 0i32;
2231 crate::ported::subst::prefork(
2232 &mut as_linklist,
2233 crate::ported::zsh_h::PREFORK_TYPESET,
2234 &mut rf,
2235 );
2236 let mut expanded: Vec<String> = Vec::with_capacity(inputs.len());
2237 while let Some(s) = as_linklist.pop_front() {
2238 expanded.push(crate::ported::lex::untokenize(&s).to_string());
2239 }
2240 if expanded.len() == 1 {
2241 Value::str(expanded.into_iter().next().unwrap())
2242 } else {
2243 Value::Array(expanded.into_iter().map(Value::str).collect())
2244 }
2245 });
2246
2247 // Options. `setopt` (BIN_SETOPT=0) / `unsetopt` (BIN_UNSETOPT=1)
2248 // share bin_setopt (options.c:580) — funcid bit discriminates
2249 // the polarity via BUILTINS table entries.
2250 reg_passthru!(vm, BUILTIN_SET, "set");
2251 reg_passthru!(vm, BUILTIN_SETOPT, "setopt");
2252 reg_passthru!(vm, BUILTIN_UNSETOPT, "unsetopt");
2253
2254 vm.register_builtin(BUILTIN_SHOPT, |vm, argc| {
2255 let args = pop_args(vm, argc);
2256 let status = crate::extensions::ext_builtins::shopt(&args);
2257 Value::Status(status)
2258 });
2259
2260 reg_passthru!(vm, BUILTIN_EMULATE, "emulate");
2261 reg_passthru!(vm, BUILTIN_GETOPTS, "getopts");
2262 reg_passthru!(vm, BUILTIN_AUTOLOAD, "autoload");
2263 reg_passthru!(vm, BUILTIN_FUNCTIONS, "functions");
2264 reg_passthru!(vm, BUILTIN_TRAP, "trap");
2265 reg_passthru!(vm, BUILTIN_DIRS, "dirs");
2266 // pushd / popd dispatch through canonical bin_cd via execbuiltin
2267 // — the BUILTINS table at src/ported/builtin.rs:9298 wires
2268 // `pushd` to bin_cd with funcid=BIN_PUSHD, and `popd` similarly
2269 // with BIN_POPD. Without these reg_passthru lines the fusevm
2270 // BUILTIN_PUSHD/POPD opcodes had no handler installed, so the
2271 // emitted CallBuiltin(110, …) silently returned a no-op and the
2272 // dirstack/$dirstack/pwd all stayed unchanged.
2273 reg_passthru!(vm, BUILTIN_PUSHD, "pushd");
2274 reg_passthru!(vm, BUILTIN_POPD, "popd");
2275 // type / whence / where / which all route through `bin_whence`
2276 // (canonical port at `src/ported/builtin.rs:3734` of
2277 // `Src/builtin.c:3975`). Each gets its own opcode so funcid +
2278 // defopts come from the BUILTINS table entry — execbuiltin
2279 // applies them correctly via the module-level dispatch_builtin.
2280 reg_passthru!(vm, BUILTIN_WHENCE, "whence");
2281 reg_passthru!(vm, BUILTIN_TYPE, "type");
2282 reg_passthru!(vm, BUILTIN_WHICH, "which");
2283 reg_passthru!(vm, BUILTIN_WHERE, "where");
2284 reg_passthru!(vm, BUILTIN_HASH, "hash");
2285 reg_passthru!(vm, BUILTIN_REHASH, "rehash");
2286
2287 // `unhash`/`unalias`/`unfunction` share `bin_unhash` (Src/builtin.c
2288 // c:4350) but each carries its own funcid (BIN_UNHASH /
2289 // BIN_UNALIAS / BIN_UNFUNCTION) in the BUILTINS table.
2290 reg_passthru!(vm, BUILTIN_UNHASH, "unhash");
2291 vm.register_builtin(BUILTIN_UNALIAS, |vm, argc| {
2292 let args = pop_args(vm, argc);
2293 Value::Status(dispatch_builtin("unalias", args))
2294 });
2295 vm.register_builtin(BUILTIN_UNFUNCTION, |vm, argc| {
2296 let args = pop_args(vm, argc);
2297 Value::Status(dispatch_builtin("unfunction", args))
2298 });
2299
2300 // Completion
2301 vm.register_builtin(BUILTIN_COMPGEN, |vm, argc| {
2302 let args = pop_args(vm, argc);
2303 // c:Bug #475/#555 — `compgen` is a bash-only builtin. In
2304 // `--zsh` mode emit "command not found" matching zsh's
2305 // external-command lookup miss — UNLESS a user FUNCTION of
2306 // that name exists: zsh has no such builtin, so bashcompinit's
2307 // `compgen() {...}` definition wins the dispatch there. The
2308 // unconditional 127 shadowed it and broke every
2309 // bashcompinit-style completion file (zsh-more-completions
2310 // _msync/_gocomplete/_qshell/_cw), spraying "command not
2311 // found: complete" at every deferred compinit load.
2312 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed) {
2313 if crate::ported::utils::getshfunc("compgen").is_some() {
2314 let status = with_executor(|exec| exec.dispatch_function_call("compgen", &args))
2315 .unwrap_or(127);
2316 return Value::Status(status);
2317 }
2318 eprintln!("zsh:1: command not found: compgen");
2319 let _ = args;
2320 return Value::Status(127);
2321 }
2322 let status = with_executor(|exec| exec.builtin_compgen(&args));
2323 Value::Status(status)
2324 });
2325
2326 vm.register_builtin(BUILTIN_COMPLETE, |vm, argc| {
2327 let args = pop_args(vm, argc);
2328 // c:Bug #475 — `complete` is a bash-only builtin. Same gate +
2329 // user-function precedence as BUILTIN_COMPGEN above.
2330 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed) {
2331 if crate::ported::utils::getshfunc("complete").is_some() {
2332 let status = with_executor(|exec| exec.dispatch_function_call("complete", &args))
2333 .unwrap_or(127);
2334 return Value::Status(status);
2335 }
2336 eprintln!("zsh:1: command not found: complete");
2337 let _ = args;
2338 return Value::Status(127);
2339 }
2340 let status = with_executor(|exec| exec.builtin_complete(&args));
2341 Value::Status(status)
2342 });
2343
2344 reg_passthru!(vm, BUILTIN_COMPADD, "compadd");
2345 reg_passthru!(vm, BUILTIN_COMPSET, "compset");
2346
2347 // See the const's doc comment for the contract. Stack (bottom→top):
2348 // base, e1, …, eN — argc = N + 1.
2349 vm.register_builtin(BUILTIN_TYPESET_PAREN_PACK, |vm, argc| {
2350 let mut vals: Vec<Value> = Vec::with_capacity(argc as usize);
2351 for _ in 0..argc {
2352 vals.push(vm.pop());
2353 }
2354 vals.reverse();
2355 let mut it = vals.into_iter();
2356 let mut out = it.next().map(|v| v.to_str()).unwrap_or_default();
2357 for v in it {
2358 match v {
2359 // Array → splice items as separate elements (splat);
2360 // empty array contributes nothing (empty elision).
2361 Value::Array(items) => {
2362 for item in items {
2363 out.push('\u{1f}');
2364 out.push_str(&item.to_str());
2365 }
2366 }
2367 other => {
2368 out.push('\u{1f}');
2369 out.push_str(&other.to_str());
2370 }
2371 }
2372 }
2373 Value::str(out)
2374 });
2375
2376 vm.register_builtin(BUILTIN_TYPESET_PAREN_CLOSE, |vm, _argc| {
2377 let base = vm.pop().to_str();
2378 Value::str(format!("{}\u{1f})", base))
2379 });
2380
2381 vm.register_builtin(BUILTIN_COMPDEF, |vm, argc| {
2382 let args = pop_args(vm, argc);
2383 // ACTUALLY A ZSH FUNCTION: compdef is defined by `compinit`, it is
2384 // never a builtin. Without the completion system set up it is
2385 // command-not-found (127) in every mode — `zsh -f; compdef` prints
2386 // "command not found: compdef". A user/compsys `compdef` FUNCTION
2387 // (autoload compinit → compinit defines compdef) wins and runs the
2388 // fast native impl; otherwise it's command-not-found. Previously the
2389 // extension builtin ran in native mode (bare `compdef` → "I need
2390 // arguments"), diverging from zsh.
2391 // compinit installs a `compdef` function stub (see
2392 // NATIVE_COMPDEF_MARKER) purely so `${+functions[compdef]}` is
2393 // true; route that exact body to the fast native impl instead of
2394 // dispatching the stub. A genuine user/compsys compdef function
2395 // (any other body) still wins via try_user_fn_override below.
2396 let is_native_stub = crate::ported::hashtable::shfunctab_lock()
2397 .read()
2398 .ok()
2399 .and_then(|t| t.get("compdef").and_then(|shf| shf.body.clone()))
2400 .map(|b| b.trim() == crate::extensions::ext_builtins::NATIVE_COMPDEF_MARKER)
2401 .unwrap_or(false);
2402 if is_native_stub {
2403 return Value::Status(with_executor(|exec| exec.builtin_compdef(&args)));
2404 }
2405 if let Some(s) = try_user_fn_override("compdef", &args) {
2406 return Value::Status(s);
2407 }
2408 if with_executor(|exec| exec.function_exists("compdef")) {
2409 return Value::Status(with_executor(|exec| exec.builtin_compdef(&args)));
2410 }
2411 eprintln!("zsh:1: command not found: compdef");
2412 Value::Status(127)
2413 });
2414
2415 vm.register_builtin(BUILTIN_COMPINIT, |vm, argc| {
2416 let args = pop_args(vm, argc);
2417 // ACTUALLY A ZSH FUNCTION: compinit is a contrib FUNCTION (autoloaded
2418 // from $fpath), never a builtin. Without `autoload -Uz compinit` it is
2419 // command-not-found
2420 // (127) in every mode — `zsh -f; compinit` prints
2421 // "command not found: compinit". zshrs previously ran its builtin
2422 // unconditionally, so bare `compinit` succeeded. Gate on a compinit
2423 // function entry existing (which `autoload -Uz compinit` creates);
2424 // once the user has autoloaded/defined it, run zshrs's implementation.
2425 if !with_executor(|exec| exec.function_exists("compinit")) {
2426 eprintln!("zsh:1: command not found: compinit");
2427 let _ = args;
2428 return Value::Status(127);
2429 }
2430 Value::Status(with_executor(|exec| exec.builtin_compinit(&args)))
2431 });
2432
2433 reg_ext_overridable!(vm, BUILTIN_CDREPLAY, "cdreplay", builtin_cdreplay);
2434
2435 // Zsh-specific
2436 reg_passthru!(vm, BUILTIN_ZSTYLE, "zstyle");
2437 reg_passthru!(vm, BUILTIN_ZMODLOAD, "zmodload");
2438 reg_passthru!(vm, BUILTIN_BINDKEY, "bindkey");
2439 reg_passthru!(vm, BUILTIN_ZLE, "zle");
2440 reg_passthru!(vm, BUILTIN_VARED, "vared");
2441 reg_passthru!(vm, BUILTIN_ZCOMPILE, "zcompile");
2442 reg_passthru!(vm, BUILTIN_ZFORMAT, "zformat");
2443 reg_passthru!(vm, BUILTIN_ZPARSEOPTS, "zparseopts");
2444 reg_passthru!(vm, BUILTIN_ZREGEXPARSE, "zregexparse");
2445
2446 // Resource limits
2447 reg_passthru!(vm, BUILTIN_ULIMIT, "ulimit");
2448 reg_passthru!(vm, BUILTIN_LIMIT, "limit");
2449 reg_passthru!(vm, BUILTIN_UNLIMIT, "unlimit");
2450 reg_passthru!(vm, BUILTIN_UMASK, "umask");
2451
2452 // Misc
2453 reg_passthru!(vm, BUILTIN_TIMES, "times");
2454
2455 vm.register_builtin(BUILTIN_CALLER, |vm, argc| {
2456 let args = pop_args(vm, argc);
2457 // c:Bug #475 — `caller` is a bash-only builtin. In `--zsh`
2458 // mode emit the canonical "command not found" diagnostic
2459 // and rc=127 matching zsh's external-command-lookup miss.
2460 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed) {
2461 eprintln!("zsh:1: command not found: caller");
2462 let _ = args;
2463 return Value::Status(127);
2464 }
2465 Value::Status(with_executor(|exec| exec.builtin_caller(&args)))
2466 });
2467
2468 vm.register_builtin(BUILTIN_HELP, |vm, argc| {
2469 let args = pop_args(vm, argc);
2470 // c:Bug #475 — `help` is a bash-only builtin. Same gate as
2471 // BUILTIN_CALLER above.
2472 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed) {
2473 eprintln!("zsh:1: command not found: help");
2474 let _ = args;
2475 return Value::Status(127);
2476 }
2477 Value::Status(with_executor(|exec| exec.builtin_help(&args)))
2478 });
2479
2480 reg_passthru!(vm, BUILTIN_ENABLE, "enable");
2481 reg_passthru!(vm, BUILTIN_DISABLE, "disable");
2482 reg_passthru!(vm, BUILTIN_TTYCTL, "ttyctl");
2483 reg_passthru!(vm, BUILTIN_SYNC, "sync");
2484 reg_passthru!(vm, BUILTIN_MKDIR, "mkdir");
2485 reg_passthru!(vm, BUILTIN_STRFTIME, "strftime");
2486
2487 vm.register_builtin(BUILTIN_ZSLEEP, |vm, argc| {
2488 let args = pop_args(vm, argc);
2489 // function > builtin: a user `zsleep() { … }` wins.
2490 if let Some(s) = try_user_fn_override("zsleep", &args) {
2491 return Value::Status(s);
2492 }
2493 Value::Status(crate::extensions::ext_builtins::zsleep(&args))
2494 });
2495
2496 reg_passthru!(vm, BUILTIN_ZSYSTEM, "zsystem");
2497
2498 // PCRE
2499 reg_passthru!(vm, BUILTIN_PCRE_COMPILE, "pcre_compile");
2500 reg_passthru!(vm, BUILTIN_PCRE_MATCH, "pcre_match");
2501 reg_passthru!(vm, BUILTIN_PCRE_STUDY, "pcre_study");
2502
2503 // Database (GDBM)
2504 reg_passthru!(vm, BUILTIN_ZTIE, "ztie");
2505 reg_passthru!(vm, BUILTIN_ZUNTIE, "zuntie");
2506 reg_passthru!(vm, BUILTIN_ZGDBMPATH, "zgdbmpath");
2507
2508 // Prompt
2509 vm.register_builtin(BUILTIN_PROMPTINIT, |vm, argc| {
2510 let args = pop_args(vm, argc);
2511 // ACTUALLY A ZSH FUNCTION: promptinit is a contrib FUNCTION
2512 // (autoloaded from $fpath), never a builtin. Command-not-found until
2513 // `autoload -Uz promptinit`; once autoloaded, run the native impl.
2514 if !with_executor(|exec| exec.function_exists("promptinit")) {
2515 eprintln!("zsh:1: command not found: promptinit");
2516 let _ = args;
2517 return Value::Status(127);
2518 }
2519 Value::Status(crate::extensions::ext_builtins::promptinit(&args))
2520 });
2521
2522 vm.register_builtin(BUILTIN_PROMPT, |vm, argc| {
2523 let args = pop_args(vm, argc);
2524 Value::Status(crate::extensions::ext_builtins::prompt(&args))
2525 });
2526
2527 // Async / Parallel (zshrs extensions) — all overridable by a
2528 // same-named user function (function > builtin).
2529 reg_ext_overridable!(vm, BUILTIN_ASYNC, "async", builtin_async);
2530 reg_ext_overridable!(vm, BUILTIN_AWAIT, "await", builtin_await);
2531 reg_ext_overridable!(vm, BUILTIN_PMAP, "pmap", builtin_pmap);
2532 reg_ext_overridable!(vm, BUILTIN_PGREP, "pgrep", builtin_pgrep);
2533 reg_ext_overridable!(vm, BUILTIN_PEACH, "peach", builtin_peach);
2534 reg_ext_overridable!(vm, BUILTIN_BARRIER, "barrier", builtin_barrier);
2535
2536 // Intercept (AOP)
2537 reg_ext_overridable!(vm, BUILTIN_INTERCEPT, "intercept", builtin_intercept);
2538 reg_ext_overridable!(
2539 vm,
2540 BUILTIN_INTERCEPT_PROCEED,
2541 "intercept_proceed",
2542 builtin_intercept_proceed
2543 );
2544
2545 // Debug / Profile
2546 reg_ext_overridable!(vm, BUILTIN_DOCTOR, "doctor", builtin_doctor);
2547 reg_ext_overridable!(vm, BUILTIN_DBVIEW, "dbview", builtin_dbview);
2548 reg_ext_overridable!(vm, BUILTIN_PROFILE, "profile", builtin_profile);
2549
2550 reg_passthru!(vm, BUILTIN_ZPROF, "zprof");
2551
2552 // ═══════════════════════════════════════════════════════════════════════
2553 // Coreutils builtins (anti-fork, gated by !posix_mode)
2554 //
2555 // All of these are routinely wrapped by user functions in real
2556 // dotfiles (zpwr, oh-my-zsh, etc.) — `cat() { ... }`, `ls() { ... }`,
2557 // `find() { ... }`. Each handler MUST consult try_user_fn_override
2558 // first (via reg_overridable!) so the user definition wins, matching
2559 // zsh's alias → function → builtin dispatch order.
2560 // ═══════════════════════════════════════════════════════════════════════
2561
2562 reg_overridable!(vm, BUILTIN_CAT, "cat", builtin_cat);
2563 reg_overridable!(vm, BUILTIN_HEAD, "head", builtin_head);
2564 reg_overridable!(vm, BUILTIN_TAIL, "tail", builtin_tail);
2565 reg_overridable!(vm, BUILTIN_WC, "wc", builtin_wc);
2566 reg_overridable!(vm, BUILTIN_BASENAME, "basename", builtin_basename);
2567 reg_overridable!(vm, BUILTIN_DIRNAME, "dirname", builtin_dirname);
2568 reg_overridable!(vm, BUILTIN_TOUCH, "touch", builtin_touch);
2569 reg_overridable!(vm, BUILTIN_REALPATH, "realpath", builtin_realpath);
2570 reg_overridable!(vm, BUILTIN_SORT, "sort", builtin_sort);
2571 reg_overridable!(vm, BUILTIN_FIND, "find", builtin_find);
2572 reg_overridable!(vm, BUILTIN_UNIQ, "uniq", builtin_uniq);
2573 reg_overridable!(vm, BUILTIN_CUT, "cut", builtin_cut);
2574 reg_overridable!(vm, BUILTIN_TR, "tr", builtin_tr);
2575 reg_overridable!(vm, BUILTIN_SEQ, "seq", builtin_seq);
2576 reg_overridable!(vm, BUILTIN_REV, "rev", builtin_rev);
2577 reg_overridable!(vm, BUILTIN_TEE, "tee", builtin_tee);
2578 reg_overridable!(vm, BUILTIN_SLEEP, "sleep", builtin_sleep);
2579 reg_overridable!(vm, BUILTIN_WHOAMI, "whoami", builtin_whoami);
2580 reg_overridable!(vm, BUILTIN_ID, "id", builtin_id);
2581
2582 reg_overridable!(vm, BUILTIN_HOSTNAME, "hostname", builtin_hostname);
2583 reg_overridable!(vm, BUILTIN_UNAME, "uname", builtin_uname);
2584 reg_overridable!(vm, BUILTIN_DATE, "date", builtin_date);
2585 reg_overridable!(vm, BUILTIN_MKTEMP, "mktemp", builtin_mktemp);
2586 // `cp` — zshrs extension (NOT in upstream zsh; upstream's
2587 // zsh/files module ships `ln`/`mv`/`rm`/`chmod`/`chown` but no
2588 // `cp`). In-process implementation in
2589 // `ext_builtins::cp_impl` — recursive copy with -r/-R, -f, -i,
2590 // -n, -p (chown + utimensat), -v. ID 263 is the first slot
2591 // past fusevm's built-in range (260-262) and before BUILTIN_MAX
2592 // (280).
2593 /// `BUILTIN_CP` constant.
2594 pub const BUILTIN_CP: u16 = 263;
2595 reg_overridable!(vm, BUILTIN_CP, "cp", builtin_cp);
2596
2597 // Pipeline execution — bytecode-native fork-per-stage. Pops N sub-chunk
2598 // indices, forks N children with stdin/stdout wired through N-1 pipes,
2599 // each child runs its stage's compiled bytecode and exits. Parent waits
2600 // and returns the last stage's status.
2601 //
2602 // Caveats: post-fork in a multi-threaded program, only async-signal-safe
2603 // ops are POSIX-safe. We violate this (running the bytecode VM after fork
2604 // touches mutexes like REGEX_CACHE). In practice, most pipeline stages
2605 // don't touch shared mutex state — externals fork/exec away, builtins do
2606 // pure I/O. Risks are bounded; if a stage does touch a held mutex, the
2607 // child deadlocks.
2608 vm.register_builtin(BUILTIN_RUN_PIPELINE, |vm, argc| {
2609 let n = argc as usize;
2610 if n == 0 {
2611 return Value::Status(0);
2612 }
2613
2614 // c:Src/exec.c — every pipeline stage forks from the current
2615 // shell state, so each stage observes the pre-pipeline $? until
2616 // it runs its own command. Stage sub-VMs start fresh with
2617 // last_status=0, so seed them with the parent's lastval; without
2618 // this `false; echo $? | cat` prints 0 instead of zsh's 1.
2619 let parent_status = vm.last_status;
2620
2621 // Pop N sub-chunk indices (LIFO → reverse to stage order)
2622 let mut indices: Vec<u16> = Vec::with_capacity(n);
2623 for _ in 0..n {
2624 indices.push(vm.pop().to_int() as u16);
2625 }
2626 indices.reverse();
2627
2628 // Clone each stage's sub-chunk
2629 let stages: Vec<fusevm::Chunk> = indices
2630 .iter()
2631 .filter_map(|&i| vm.chunk.sub_chunks.get(i as usize).cloned())
2632 .collect();
2633 if stages.len() != n {
2634 return Value::Status(1);
2635 }
2636
2637 // Single stage — no pipe, just run inline
2638 if n == 1 {
2639 let stage = stages.into_iter().next().unwrap();
2640 crate::fusevm_disasm::maybe_print_stdout("pipeline:single", &stage);
2641 let mut stage_vm = fusevm::VM::new(stage);
2642 stage_vm.last_status = parent_status;
2643 register_builtins(&mut stage_vm);
2644 let _ = stage_vm.run();
2645 return Value::Status(stage_vm.last_status);
2646 }
2647
2648 // Build N-1 pipes
2649 let mut pipes: Vec<(i32, i32)> = Vec::with_capacity(n - 1);
2650 for _ in 0..n - 1 {
2651 let mut fds = [0i32; 2];
2652 if unsafe { libc::pipe(fds.as_mut_ptr()) } < 0 {
2653 // Cleanup any pipes we already created
2654 for (r, w) in &pipes {
2655 unsafe {
2656 libc::close(*r);
2657 libc::close(*w);
2658 }
2659 }
2660 return Value::Status(1);
2661 }
2662 pipes.push((fds[0], fds[1]));
2663 }
2664
2665 // zsh runs the LAST stage of a pipeline in the CURRENT shell
2666 // (not a forked child) so a trailing `read x` keeps its
2667 // assignment in the parent. Other shells (bash) fork every
2668 // stage. Honor zsh by leaving stage N-1 inline. Forks the
2669 // first N-1 stages with fork(); runs the last in this process
2670 // with stdin dup2'd to the last pipe's read end and stdout
2671 // restored after.
2672 let last_idx = n - 1;
2673 let stages_vec: Vec<fusevm::Chunk> = stages.into_iter().collect();
2674
2675 let mut child_pids: Vec<libc::pid_t> = Vec::with_capacity(n - 1);
2676 for (i, chunk) in stages_vec.iter().take(last_idx).enumerate() {
2677 match unsafe { libc::fork() } {
2678 -1 => {
2679 // fork failed — kill any children we already started
2680 for pid in &child_pids {
2681 unsafe { libc::kill(*pid, libc::SIGTERM) };
2682 }
2683 for (r, w) in &pipes {
2684 unsafe {
2685 libc::close(*r);
2686 libc::close(*w);
2687 }
2688 }
2689 return Value::Status(1);
2690 }
2691 0 => {
2692 // Reset SIGPIPE to default so a broken-pipe write
2693 // kills the child cleanly instead of triggering a
2694 // Rust println! panic. The parent shell ignores
2695 // SIGPIPE so it can handle EPIPE itself, but child
2696 // pipeline stages should die quietly when their
2697 // downstream stage closes early (e.g. `seq | head -3`).
2698 unsafe {
2699 libc::signal(libc::SIGPIPE, libc::SIG_DFL);
2700 }
2701 // c:Src/exec.c — pipeline children are forked
2702 // subshells; their EXIT trap context is reset so
2703 // the parent's `trap '...' EXIT` doesn't fire when
2704 // the child exits. Mirror by dropping EXIT from
2705 // the inherited traps_table inside the child.
2706 if let Ok(mut t) = crate::ported::builtin::traps_table().lock() {
2707 t.remove("EXIT");
2708 }
2709 // c:Src/exec.c:2862 → 1219 — pipeline children run
2710 // entersubsh with ESUB_PGRP, which clears the job
2711 // table (clearjobtab, Src/jobs.c:1780). Without
2712 // this, `sleep 5 & jobs -p | wc -l` reports 1 in
2713 // the forked stage where zsh reports 0. The fork
2714 // already copy-isolates the statics, so mutating
2715 // them here can't leak to the parent.
2716 with_executor(|exec| {
2717 let monitor =
2718 crate::ported::zsh_h::isset(crate::ported::zsh_h::MONITOR) as i32;
2719 crate::ported::jobs::clearjobtab(&mut exec.jobs, monitor);
2720 });
2721 *crate::ported::jobs::THISJOB
2722 .get_or_init(|| std::sync::Mutex::new(-1))
2723 .lock()
2724 .unwrap() = -1;
2725 // c:Src/exec.c:3720-3724 — the stage's own fds go
2726 // onto 0/1 only AFTER its argument words have been
2727 // expanded (prefork c:3304 / globlist c:3702), so
2728 // park them and let the stage chunk's
2729 // BUILTIN_PIPE_FDS_INSTALL do the dup2 at the
2730 // C-faithful point. `print -rl -- c a b |
2731 // print -r -- "[$(cat)]" | cat` therefore prints
2732 // `[]` — the middle stage's `$(cat)` reads the
2733 // shell's stdin, not the pipe.
2734 let in_fd = if i > 0 { pipes[i - 1].0 } else { -1 };
2735 let out_fd = pipes[i].1;
2736 // (Pipe-output MULTIOS marking — c:Src/exec.c:3724 —
2737 // is emitted INTO the stage chunk by compile_pipe
2738 // via BUILTIN_PIPE_OUTPUT_MARK, gated on the stage's
2739 // top-level command actually carrying redirects, so
2740 // a nested `{ echo a > f; } | cat` body redirect
2741 // does not wrongly join the pipe.)
2742 // Close every pipe fd this stage doesn't need. The
2743 // two it does keep are closed by the install op
2744 // right after their dup2.
2745 for (r, w) in &pipes {
2746 unsafe {
2747 if *r != in_fd && *r != out_fd {
2748 libc::close(*r);
2749 }
2750 if *w != in_fd && *w != out_fd {
2751 libc::close(*w);
2752 }
2753 }
2754 }
2755 stage_fds_park(in_fd, out_fd);
2756
2757 // Run this stage's bytecode on a fresh VM
2758 crate::fusevm_disasm::maybe_print_stdout(
2759 &format!("pipeline:child:stage:{i}"),
2760 chunk,
2761 );
2762 let mut stage_vm = fusevm::VM::new(chunk.clone());
2763 stage_vm.last_status = parent_status;
2764 register_builtins(&mut stage_vm);
2765 let _ = stage_vm.run();
2766 // Flush any buffered output before exiting
2767 let _ = std::io::stdout().flush();
2768 let _ = std::io::stderr().flush();
2769 std::process::exit(stage_vm.last_status);
2770 }
2771 pid => {
2772 child_pids.push(pid);
2773 }
2774 }
2775 }
2776
2777 // Parent runs the LAST stage inline. Save stdin, park the last
2778 // pipe's read end for the chunk's BUILTIN_PIPE_FDS_INSTALL
2779 // (c:Src/exec.c:3722 `addfd(..., 0, input, 0, NULL)` — after
2780 // the stage's args are expanded, so `… | print -r -- "[$(cat)]"`
2781 // has its `$(cat)` read the shell's stdin, not the pipe), run
2782 // the chunk, restore stdin. Close every other pipe fd so the
2783 // producer side gets EOF when the last upstream stage exits.
2784 // Shell-internal save — keep it out of the script's fd range (movefd,
2785 // c:Src/exec.c:2425).
2786 let saved_stdin = unsafe { libc::fcntl(libc::STDIN_FILENO, libc::F_DUPFD, 10) };
2787 let last_in_fd = if last_idx > 0 {
2788 pipes[last_idx - 1].0
2789 } else {
2790 -1
2791 };
2792 // Close all pipe fds in the parent except the one the last
2793 // stage still has to install. (Children already have their own
2794 // copies; the install op closes the read end after its dup2.)
2795 for (r, w) in &pipes {
2796 unsafe {
2797 if *r != last_in_fd {
2798 libc::close(*r);
2799 }
2800 libc::close(*w);
2801 }
2802 }
2803 let outer_stage_fds = stage_fds_park(last_in_fd, -1);
2804
2805 // Run the last stage's bytecode on a sub-VM with the host wired up.
2806 // By default (zsh semantics) the sub-VM runs IN THIS PROCESS so the
2807 // last stage's reads/assignments update the parent's state directly
2808 // (`echo x | read v` sets $v; `cmd | mapfile arr` sets arr).
2809 //
2810 // !!! BASH-MODE GATE !!! bash forks EVERY pipeline stage (unless
2811 // `shopt -s lastpipe`), so the last stage runs in a SUBSHELL and its
2812 // variable/array assignments do NOT persist — `echo x | read v; echo
2813 // $v` prints an empty line, `cmd | mapfile arr` leaves arr unset.
2814 // Fork the last stage under `--bash` to match. The parent's existing
2815 // `stage_fds_take()` below closes its `last_in_fd` copy; the forked
2816 // child inherits the parked pipe fd and installs it onto stdin, and
2817 // the writer stages (already forked) supply its input.
2818 let last_stage_status = if crate::dash_mode::bash_mode() {
2819 let last_chunk = stages_vec.into_iter().last().unwrap();
2820 crate::fusevm_disasm::maybe_print_stdout("pipeline:last", &last_chunk);
2821 match unsafe { libc::fork() } {
2822 -1 => 1,
2823 0 => {
2824 // Subshell child: run the last stage, then _exit with its
2825 // status. Reset SIGPIPE + drop the EXIT trap like the
2826 // other pipeline children above.
2827 unsafe { libc::signal(libc::SIGPIPE, libc::SIG_DFL) };
2828 if let Ok(mut t) = crate::ported::builtin::traps_table().lock() {
2829 t.remove("EXIT");
2830 }
2831 let mut stage_vm = fusevm::VM::new(last_chunk);
2832 stage_vm.last_status = parent_status;
2833 register_builtins(&mut stage_vm);
2834 stage_vm.set_shell_host(Box::new(ZshrsHost));
2835 let _ = stage_vm.run();
2836 let st = stage_vm.last_status;
2837 let _ = std::io::stdout().flush();
2838 let _ = std::io::stderr().flush();
2839 unsafe { libc::_exit(st) };
2840 }
2841 pid => {
2842 let mut status: i32 = 0;
2843 unsafe { libc::waitpid(pid, &mut status, 0) };
2844 if libc::WIFEXITED(status) {
2845 libc::WEXITSTATUS(status)
2846 } else if libc::WIFSIGNALED(status) {
2847 128 + libc::WTERMSIG(status)
2848 } else {
2849 1
2850 }
2851 }
2852 }
2853 } else {
2854 let last_chunk = stages_vec.into_iter().last().unwrap();
2855 crate::fusevm_disasm::maybe_print_stdout("pipeline:last", &last_chunk);
2856 let mut stage_vm = fusevm::VM::new(last_chunk);
2857 stage_vm.last_status = parent_status;
2858 register_builtins(&mut stage_vm);
2859 stage_vm.set_shell_host(Box::new(ZshrsHost));
2860 let _ = stage_vm.run();
2861 let _ = std::io::stdout().flush();
2862 let _ = std::io::stderr().flush();
2863 stage_vm.last_status
2864 };
2865
2866 // Reclaim the read end if the stage chunk never reached its
2867 // install op (an expansion error aborted it, or the stage was
2868 // a shape that dispatches without one), then restore the outer
2869 // stage's still-pending fds for a nested pipeline.
2870 let (leftover_in, _) = stage_fds_take();
2871 if leftover_in >= 0 {
2872 unsafe { libc::close(leftover_in) };
2873 }
2874 stage_fds_park(outer_stage_fds.0, outer_stage_fds.1);
2875
2876 // Restore stdin
2877 if saved_stdin >= 0 {
2878 unsafe {
2879 libc::dup2(saved_stdin, libc::STDIN_FILENO);
2880 libc::close(saved_stdin);
2881 }
2882 }
2883
2884 // Wait for all forked stages, capture per-stage statuses for PIPESTATUS.
2885 let mut pipestatus: Vec<i32> = Vec::with_capacity(n);
2886 for pid in child_pids {
2887 let mut status: i32 = 0;
2888 unsafe {
2889 libc::waitpid(pid, &mut status, 0);
2890 }
2891 let s = if libc::WIFEXITED(status) {
2892 libc::WEXITSTATUS(status)
2893 } else if libc::WIFSIGNALED(status) {
2894 128 + libc::WTERMSIG(status)
2895 } else {
2896 1
2897 };
2898 pipestatus.push(s);
2899 }
2900 // Append the in-parent last-stage status so `pipestatus` ends
2901 // with N entries (one per stage).
2902 pipestatus.push(last_stage_status);
2903 // Pipeline exit status: by default, the LAST stage's status.
2904 // With `setopt pipefail` (or `set -o pipefail`), use the
2905 // first non-zero stage status (so failures earlier in the
2906 // pipeline propagate even if the last stage succeeded).
2907 let pipefail_on = with_executor(|exec| opt_state_get("pipefail").unwrap_or(false));
2908 let last_status = if pipefail_on {
2909 pipestatus
2910 .iter()
2911 .copied()
2912 .rfind(|&s| s != 0)
2913 .or_else(|| pipestatus.last().copied())
2914 .unwrap_or(0)
2915 } else {
2916 *pipestatus.last().unwrap_or(&0)
2917 };
2918
2919 // c:Src/params.c:265,438 — only `pipestatus` (lowercase) is the
2920 // zsh special parameter; bash's `PIPESTATUS` doesn't exist in
2921 // zsh's special-params table. Prior port also populated
2922 // `PIPESTATUS` "for portability" — but that's a real divergence
2923 // from zsh: a script doing `[[ -z $PIPESTATUS ]]` to detect
2924 // zsh-vs-bash would mis-classify. Bug #64 in docs/BUGS.md.
2925 with_executor(|exec| {
2926 let strs: Vec<String> = pipestatus.iter().map(|s| s.to_string()).collect();
2927 exec.set_array("pipestatus".to_string(), strs);
2928 });
2929
2930 Value::Status(last_status)
2931 });
2932
2933 // Array→String join. Pops one value; if it's an Array (e.g. from Op::Glob),
2934 // joins string-coerced elements with a single space. Pass-through for
2935 // non-arrays so the op is safe to chain after any String-or-Array producer.
2936 // Scalar coercion of an assembled word: pop a Value; if it's an
2937 // Array (produced by a splice segment like `"$@"` / `"${arr[@]}"`),
2938 // IFS[0]-join it to a single scalar; a scalar passes through. This
2939 // is the assignment-context coercion C zsh applies in multsub when
2940 // the expansion is the RHS of a SCALAR assignment (Src/subst.c
2941 // c:3032 sepjoin under ssub) — `v="$@"` joins the positionals with
2942 // ${IFS[1]} rather than leaving an array whose splat would lose all
2943 // but the first element. Joins via sepjoin so a custom / empty IFS
2944 // is honored (not a hardcoded space).
2945 vm.register_builtin(BUILTIN_ARRAY_JOIN, |vm, _argc| {
2946 let val = vm.pop();
2947 match val {
2948 Value::Array(items) => {
2949 let strs: Vec<String> = items.iter().map(|v| v.to_str()).collect();
2950 Value::str(crate::ported::utils::sepjoin(&strs, None))
2951 }
2952 other => other,
2953 }
2954 });
2955
2956 // `cmd &` background execution. Compile_list emits this for any item
2957 // followed by ListOp::Amp: the job text + the cmd's sub-chunk index are
2958 // pushed, then this builtin pops both, looks up the chunk, forks. The
2959 // child detaches via setsid (so SIGINT to the foreground job doesn't kill
2960 // it), runs the bytecode on a fresh VM with builtins re-registered, exits
2961 // with the last status. The parent registers the job in the canonical
2962 // JOBTAB (c:Src/exec.c::execpline Z_ASYNC arm) and returns Status(0).
2963 vm.register_builtin(BUILTIN_RUN_BG, |vm, _argc| {
2964 // `&|` / `&!` set disown → the job is dropped from the table (no
2965 // `[N] pid` announcement, no `[N] done`), matching C exec.c:1752-1758.
2966 let disown = vm.pop().to_int() != 0;
2967 let sub_idx = vm.pop().to_int() as usize;
2968 let job_text = vm.pop().to_str();
2969 let chunk = match vm.chunk.sub_chunks.get(sub_idx).cloned() {
2970 Some(c) => c,
2971 None => return Value::Status(1),
2972 };
2973
2974 match unsafe { libc::fork() } {
2975 -1 => Value::Status(1),
2976 0 => {
2977 // Child: detach and run.
2978 unsafe { libc::setsid() };
2979 crate::fusevm_disasm::maybe_print_stdout("background_job", &chunk);
2980 let mut bg_vm = fusevm::VM::new(chunk);
2981 register_builtins(&mut bg_vm);
2982 let _ = bg_vm.run();
2983 let _ = std::io::stdout().flush();
2984 let _ = std::io::stderr().flush();
2985 std::process::exit(bg_vm.last_status);
2986 }
2987 pid => {
2988 // Parent: record the PID into `$!` (most recent
2989 // backgrounded job's pid). zsh exposes this for any
2990 // script that needs `wait $!`. Also register the
2991 // bare-pid job so a no-args `wait` can synchronize.
2992 // c:Src/jobs.c:73 — `lastpid = pid;` after a
2993 // background fork. zshrs's `$!` getter
2994 // (params.rs::lookup_special_var "!") reads from
2995 // the same atomic, so a single store here is the
2996 // canonical writer.
2997 crate::ported::modules::clone::lastpid
2998 .store(pid, std::sync::atomic::Ordering::Relaxed);
2999 // c:Src/exec.c:1700 — `thisjob = newjob = initjob()`:
3000 // allocate the canonical jobtab slot. c:Src/exec.c:2950
3001 // zfork path → addproc(pid, text, 0, &bgtime, ...) hangs
3002 // the proc entry (with its display text) off the job.
3003 // c:Src/exec.c:1744-1746 — `clearoldjobtab();
3004 // jobtab[thisjob].stat |= STAT_NOSTTY;` then c:1758
3005 // `spawnjob()` promotes it to curjob (top-level shell
3006 // only), marks STAT_LOCKED and resets thisjob.
3007 {
3008 use crate::ported::jobs;
3009 use std::sync::Mutex;
3010 let table = jobs::JOBTAB.get_or_init(|| Mutex::new(Vec::new()));
3011 let idx = {
3012 let mut tab = table.lock().unwrap_or_else(|e| e.into_inner());
3013 let idx = jobs::initjob(&mut tab); // c:exec.c:1700
3014 jobs::addproc(
3015 &mut tab[idx],
3016 pid,
3017 &job_text,
3018 false,
3019 Some(std::time::Instant::now()),
3020 -1,
3021 -1,
3022 ); // c:exec.c:2950 addproc
3023 tab[idx].stat |= crate::ported::zsh_h::STAT_NOSTTY; // c:exec.c:1746
3024 idx
3025 };
3026 jobs::clearoldjobtab(); // c:exec.c:1744
3027 if let Ok(mut tj) = jobs::THISJOB.get_or_init(|| Mutex::new(-1)).lock() {
3028 *tj = idx as i32;
3029 }
3030 if disown {
3031 // c:exec.c:1752-1755 — `pipecleanfilelist(...);
3032 // deletejob(jobtab + thisjob, 1); thisjob = -1;` — a
3033 // disowned job leaves the table entirely, so neither
3034 // spawnjob's `[N] pid` nor the later `[N] done` prints.
3035 // This is what keeps zinit-turbo's `… &|` completion
3036 // jobs silent (they load inside a `zle -F` handler).
3037 {
3038 let mut tab = table.lock().unwrap_or_else(|e| e.into_inner());
3039 jobs::pipecleanfilelist(&mut tab[idx], false); // c:1753
3040 jobs::deletejob(&mut tab[idx], true); // c:1754
3041 }
3042 if let Ok(mut tj) = jobs::THISJOB.get_or_init(|| Mutex::new(-1)).lock()
3043 {
3044 *tj = -1; // c:1755
3045 }
3046 } else {
3047 jobs::spawnjob(); // c:exec.c:1758
3048 }
3049 }
3050 with_executor(|exec| {
3051 exec.jobs
3052 .add_pid_job(pid, job_text.clone(), JobState::Running);
3053 });
3054 Value::Status(0)
3055 }
3056 }
3057 });
3058
3059 // ── Indexed-array storage ─────────────────────────────────────────────
3060 //
3061 // Stack: pushed values then name (LAST). `arr=(a b c)` → 4 args
3062 // (a, b, c, arr). `arr=($(cmd))` → 2 args (FlatArray, arr).
3063 //
3064 // PURE PASSTHRU: pop name + values, dispatch to canonical
3065 // `setaparam` / `sethparam` (C port of `Src/params.c:3595/3602`).
3066 // assignaparam already handles PM_UNIQUE dedupe, type-flag flip,
3067 // PM_NAMEREF rejection, ASSPM_AUGMENT prepend, and createparam
3068 // for fresh names.
3069 vm.register_builtin(BUILTIN_SET_ARRAY, |vm, argc| {
3070 // `${~spec}` carrier: an assignment statement is a word-
3071 // pipeline boundary too — restore the user's GLOB_SUBST
3072 // before the NEXT word expands (`Z[d]=${~Z[d]}; print
3073 // ${options[globsubst]}` must read the user value).
3074 consume_tilde_globsubst_carrier();
3075 let n = argc as usize;
3076 let mut popped: Vec<Value> = Vec::with_capacity(n);
3077 for _ in 0..n {
3078 popped.push(vm.pop());
3079 }
3080 popped.reverse();
3081 if popped.is_empty() {
3082 return Value::Status(1);
3083 }
3084 let name = popped.pop().unwrap().to_str();
3085 let mut values: Vec<String> = Vec::new();
3086 for v in popped {
3087 flatten_array_value(v, &mut values);
3088 }
3089 // Bash sparse: a full `a=(...)` reassign resets the array to dense
3090 // (drops any prior holes from subscript-assign / unset).
3091 if crate::dash_mode::bash_mode() {
3092 crate::bash_arrays::clear(&name);
3093 }
3094 let blocked = with_executor(|exec| {
3095 // Assoc init `typeset -A m; m=(k v k v ...)` — route to
3096 // canonical sethparam (Src/params.c:3602) which parses the
3097 // flat (k,v) pair list internally.
3098 if exec.assoc(&name).is_some() {
3099 // `[k]=v` / `[k]+=v` elements arrive from the compiler
3100 // as Marker / key / value triples (compile_zsh's port
3101 // of keyvalpairelement, c:Src/subst.c:49-79).
3102 let marker = crate::ported::zsh_h::Marker;
3103 let values = if values.iter().any(|e| e.starts_with(marker)) {
3104 // c:Src/params.c:3544-3560 — under ASSPM_KEY_VALUE
3105 // assocs strictly enforce `[key]=value`: every
3106 // stride-of-3 element must be a Marker. Mixing
3107 // plain pairs with kv triads is an error.
3108 let mut i = 0usize;
3109 while i < values.len() {
3110 if !values[i].starts_with(marker) {
3111 crate::ported::utils::zerr(
3112 "bad [key]=value syntax for associative array",
3113 );
3114 crate::ported::utils::errflag.fetch_or(
3115 crate::ported::zsh_h::ERRFLAG_ERROR,
3116 std::sync::atomic::Ordering::Relaxed,
3117 );
3118 exec.set_last_status(1);
3119 return true;
3120 }
3121 i += 3;
3122 }
3123 if values.len() % 3 != 0 {
3124 // c:Src/params.c:4124-4131 arrhashsetfn — a
3125 // truncated triad leaves an odd non-Marker
3126 // count → "bad set of key/value pairs".
3127 crate::ported::utils::zerr(
3128 "bad set of key/value pairs for associative array",
3129 );
3130 crate::ported::utils::errflag.fetch_or(
3131 crate::ported::zsh_h::ERRFLAG_ERROR,
3132 std::sync::atomic::Ordering::Relaxed,
3133 );
3134 exec.set_last_status(1);
3135 return true;
3136 }
3137 // c:Src/params.c:4136-4168 arrhashsetfn — whole
3138 // assignment builds a FRESH table; a `Marker +`
3139 // triad (`[k]+=v`) appends to the value inserted
3140 // EARLIER IN THIS SAME LITERAL (assignstrvalue
3141 // with eltflags=ASSPM_AUGMENT against the new ht),
3142 // so `h=([k]=a [k]+=b)` yields "ab". Resolve the
3143 // appends here, then hand flat pairs to sethparam.
3144 let mut order: Vec<String> = Vec::new();
3145 let mut map: std::collections::HashMap<String, String> =
3146 std::collections::HashMap::new();
3147 for ch in values.chunks(3) {
3148 let elt_append = ch[0].chars().nth(1) == Some('+');
3149 let k = ch[1].clone();
3150 let v = ch[2].clone();
3151 let nv = if elt_append {
3152 format!("{}{}", map.get(&k).cloned().unwrap_or_default(), v)
3153 } else {
3154 v
3155 };
3156 if !map.contains_key(&k) {
3157 order.push(k.clone());
3158 }
3159 map.insert(k, nv);
3160 }
3161 order
3162 .into_iter()
3163 .flat_map(|k| {
3164 let v = map.get(&k).cloned().unwrap_or_default();
3165 [k, v]
3166 })
3167 .collect()
3168 } else {
3169 values
3170 };
3171 // Odd-count rejection lives in the canonical chain:
3172 // sethparam → setarrvalue (c:3651/c:2920) →
3173 // arrhashsetfn's zerr "bad set of key/value pairs"
3174 // (Src/params.c:4128-4131). zerr sets ERRFLAG_ERROR,
3175 // which aborts the remaining list at the next command
3176 // boundary (BUILTIN_ERREXIT_CHECK trigger 4) —
3177 // matching `zsh -fc 'typeset -A m; m=(odd); print x'`
3178 // printing nothing after the error. Like C's sethparam
3179 // (c:3652-3653 returns v->pm regardless), the Rust
3180 // port returns Some on the odd-count path — the
3181 // failure travels via errflag, so check BOTH.
3182 let pre_err = crate::ported::utils::errflag
3183 .load(std::sync::atomic::Ordering::Relaxed)
3184 & crate::ported::zsh_h::ERRFLAG_ERROR;
3185 let res = crate::ported::params::sethparam(&name, values.clone());
3186 let now_err = crate::ported::utils::errflag
3187 .load(std::sync::atomic::Ordering::Relaxed)
3188 & crate::ported::zsh_h::ERRFLAG_ERROR;
3189 if res.is_none() || (pre_err == 0 && now_err != 0) {
3190 // c:Src/exec.c:2632-2633 addvars — `if
3191 // (!assignaparam(name, arr, myflags)) lastval = 1;`
3192 // — failed assignment sets lastval so the errflag
3193 // abort exits 1 (init.c loop() breaks, zsh_main
3194 // returns lastval).
3195 exec.set_last_status(1);
3196 return true;
3197 }
3198 #[cfg(feature = "recorder")]
3199 if crate::recorder::is_enabled() && exec.local_scope_depth == 0 {
3200 let ctx = exec.recorder_ctx();
3201 let attrs = exec.recorder_attrs_for(&name);
3202 let mut pairs: Vec<(String, String)> = Vec::with_capacity(values.len() / 2);
3203 let mut iter = values.iter().cloned();
3204 while let Some(k) = iter.next() {
3205 if let Some(v) = iter.next() {
3206 pairs.push((k, v));
3207 }
3208 }
3209 crate::recorder::emit_assoc_assign(&name, pairs, attrs, false, ctx);
3210 }
3211 return false;
3212 }
3213 // Indexed-array: setaparam (Src/params.c:3766) wraps
3214 // assignaparam with ASSPM_WARN — handles PM_UNIQUE dedupe,
3215 // type-flag flip, PM_READONLY rejection.
3216 //
3217 // `[k]=v` elements arrive as Marker / key / value triples
3218 // (compile_zsh's keyvalpairelement port). Mirror
3219 // c:Src/exec.c:2552-2553 — `if (prefork_ret &
3220 // PREFORK_KEY_VALUE) myflags |= ASSPM_KEY_VALUE;` — so
3221 // assignaparam runs its kv-resolution block (sparse fill
3222 // for PM_ARRAY, c:3447-3541; strict-triad enforcement for
3223 // special PM_HASHED targets like `options`, c:3544-3560).
3224 let values = values;
3225 let has_kv = values
3226 .iter()
3227 .any(|e| e.starts_with(crate::ported::zsh_h::Marker));
3228 // The tied-array mirror to a PM_TIED scalar
3229 // (`typeset -T PATH path`) lives canonically in
3230 // setarrvalue's dispatch in C zsh; until that wires
3231 // through assignaparam, mirror here so PATH stays in sync
3232 // after `path=(/x)`.
3233 //
3234 // The mirrored value must be the array AS STORED, which for a
3235 // PM_UNIQUE tie means deduped. c:4066-4076 arrsetfn fixes the
3236 // order:
3237 // if (pm->node.flags & PM_UNIQUE) uniqarray(x);
3238 // pm->u.arr = x;
3239 // if (pm->ename && x) arrfixenv(pm->ename, x);
3240 // — the dedupe happens FIRST, so the scalar publishes the same
3241 // list the array holds and the two halves of a tie always agree.
3242 // Mirroring the raw `values` broke exactly that:
3243 // typeset -U path; path=(/a /b /a)
3244 // $path → /a /b (right)
3245 // $PATH → /a:/b:/a (wrong; zsh gives /a:/b)
3246 // i.e. `typeset -U path`, the standard PATH-dedup idiom in
3247 // essentially every .zshrc. assignaparam's own arrfixenv does not
3248 // rescue it: that call is gated on the param having a gsu_a wired,
3249 // and `path` has none.
3250 //
3251 // The dedupe is applied here rather than by reading the array back
3252 // after assignaparam, because this mirror must stay BEFORE it.
3253 // `exec.set_scalar` is heavier than C's arrfixenv — arrfixenv only
3254 // rewrites the environment string, while set_scalar re-derives the
3255 // ARRAY from the scalar. Running it afterwards makes `path=()`
3256 // publish PATH="" and then re-split that back into a one-element
3257 // `path=("")`, where zsh leaves 0 elements.
3258 if let Some((scalar_name, sep)) = exec.tied_array_to_scalar.get(&name).cloned() {
3259 let uniq = crate::ported::params::paramtab()
3260 .read()
3261 .ok()
3262 .and_then(|t| t.get(&name).map(|p| p.node.flags))
3263 .map(|f| (f as u32 & crate::ported::zsh_h::PM_UNIQUE) != 0)
3264 .unwrap_or(false);
3265 let mirror = if uniq {
3266 crate::ported::params::simple_arrayuniq(values.clone()) // c:4068
3267 } else {
3268 values.clone()
3269 };
3270 exec.set_scalar(scalar_name, mirror.join(&sep)); // c:4074-4075
3271 }
3272 // c:Src/exec.c:2632-2633 addvars — `if (!assignaparam(...))
3273 // lastval = 1;` — a failed assignment (bad subscript, bad
3274 // [key]=value syntax, readonly) exits 1 and the errflag
3275 // abort stops the remaining list. Track errflag pre/post
3276 // like the assoc branch above.
3277 let kv_flag = if has_kv {
3278 crate::ported::zsh_h::ASSPM_KEY_VALUE
3279 } else {
3280 0
3281 };
3282 let pre_err = crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed)
3283 & crate::ported::zsh_h::ERRFLAG_ERROR;
3284 let res = crate::ported::params::assignaparam(
3285 &name,
3286 values.clone(),
3287 crate::ported::zsh_h::ASSPM_WARN | kv_flag,
3288 );
3289 let now_err = crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed)
3290 & crate::ported::zsh_h::ERRFLAG_ERROR;
3291 if res.is_none() && pre_err == 0 && now_err != 0 {
3292 exec.set_last_status(1);
3293 return true;
3294 }
3295 // Bash sparse: an explicit-index array literal `a=([2]=x [5]=y)`
3296 // leaves the un-indexed slots as HOLES (bash: count 2, indices
3297 // {2,5}), not dense empties. assignaparam has already placed each
3298 // value at its 0-based index; mark every OTHER slot a hole. Gated
3299 // to a PURE indexed literal (every element a `[idx]=val` triple) —
3300 // a mixed positional/indexed literal (`a=(x [3]=y z)`) needs the
3301 // positional-counter replay we don't model, so it stays dense.
3302 if crate::dash_mode::bash_mode() && has_kv {
3303 let marker = crate::ported::zsh_h::Marker;
3304 let pure_indexed = !values.is_empty()
3305 && values.len() % 3 == 0
3306 && values.chunks(3).all(|ch| ch[0].starts_with(marker));
3307 if pure_indexed {
3308 let mut explicit: std::collections::BTreeSet<usize> =
3309 std::collections::BTreeSet::new();
3310 for ch in values.chunks(3) {
3311 if let Ok(i) = ch[1].trim().parse::<usize>() {
3312 explicit.insert(i);
3313 }
3314 }
3315 let len = exec.array(&name).map(|a| a.len()).unwrap_or(0);
3316 for i in 0..len {
3317 if !explicit.contains(&i) {
3318 crate::bash_arrays::note_unset(&name, i);
3319 }
3320 }
3321 }
3322 }
3323 #[cfg(feature = "recorder")]
3324 if crate::recorder::is_enabled() && exec.local_scope_depth == 0 {
3325 let ctx = exec.recorder_ctx();
3326 let attrs = exec.recorder_attrs_for(&name);
3327 emit_path_or_assign(&name, &values, attrs, false, &ctx);
3328 }
3329 false
3330 });
3331 let status = if blocked { 1 } else { 0 };
3332 // c:Src/jobs.c:1748-1757 waitonejob — in C an array-assignment
3333 // simple command goes through execpline → waitjobs; with no
3334 // procs the else-branch stores `pipestats[0] = lastval;
3335 // numpipestats = 1`. Bare SCALAR assignments never create a
3336 // job (no waitjobs), so this clobber is array/assoc-assignment
3337 // specific: `false|true; x=(1 2); echo $pipestatus` → `0` in
3338 // zsh while `x=1` preserves `1 0`. Bug #373.
3339 crate::ported::builtin::LASTVAL.store(status, std::sync::atomic::Ordering::Relaxed);
3340 let mut synth = crate::ported::zsh_h::job::default();
3341 crate::ported::jobs::waitonejob(&mut synth);
3342 Value::Status(status)
3343 });
3344 // `arr+=(d e f)` — array append. Same calling conventions as SET_ARRAY.
3345 //
3346 // PURE PASSTHRU shape: pop name + values, dispatch through the
3347 // canonical assoc / array setter. assignaparam's ASSPM_AUGMENT
3348 // flag handles the C-source-equivalent "preserve prior value"
3349 // semantics; for now we read the current array, extend with
3350 // new values, write through set_array (which routes to
3351 // setaparam → assignaparam where PM_UNIQUE dedupe lands).
3352 vm.register_builtin(BUILTIN_APPEND_ARRAY, |vm, argc| {
3353 let n = argc as usize;
3354 let mut popped: Vec<Value> = Vec::with_capacity(n);
3355 for _ in 0..n {
3356 popped.push(vm.pop());
3357 }
3358 popped.reverse();
3359 if popped.is_empty() {
3360 return Value::Status(1);
3361 }
3362 let name = popped.pop().unwrap().to_str();
3363 let mut values: Vec<String> = Vec::new();
3364 for v in popped {
3365 flatten_array_value(v, &mut values);
3366 }
3367 let blocked = with_executor(|exec| -> bool {
3368 // Assoc append `m+=(k1 v1 ...)`: merge the (k,v) pairs into
3369 // the existing map and write back via canonical sethparam
3370 // (Src/params.c:3602). The canonical C path would go
3371 // assignaparam(ASSPM_AUGMENT) → arrhashsetfn(ASSPM_AUGMENT)
3372 // at Src/params.c:3850, but the zshrs port of
3373 // arrhashsetfn doesn't yet implement value-storage
3374 // (pending Param.u_hash backend wireup) — until that
3375 // lands, do the augment + write here so the storage
3376 // actually mutates.
3377 if exec.assoc(&name).is_some() {
3378 // `[k]=v` / `[k]+=v` elements arrive as Marker / key /
3379 // value triples (compile_zsh's port of
3380 // keyvalpairelement, c:Src/subst.c:49-79).
3381 let marker = crate::ported::zsh_h::Marker;
3382 let mut map = exec.assoc(&name).unwrap_or_default();
3383 if values.iter().any(|e| e.starts_with(marker)) {
3384 // c:Src/params.c:3544-3560 — strict triad rule.
3385 let mut i = 0usize;
3386 while i < values.len() {
3387 if !values[i].starts_with(marker) {
3388 crate::ported::utils::zerr(
3389 "bad [key]=value syntax for associative array",
3390 );
3391 crate::ported::utils::errflag.fetch_or(
3392 crate::ported::zsh_h::ERRFLAG_ERROR,
3393 std::sync::atomic::Ordering::Relaxed,
3394 );
3395 exec.set_last_status(1);
3396 return true;
3397 }
3398 i += 3;
3399 }
3400 if values.len() % 3 != 0 {
3401 // c:Src/params.c:4124-4131 — odd pair count.
3402 crate::ported::utils::zerr(
3403 "bad set of key/value pairs for associative array",
3404 );
3405 crate::ported::utils::errflag.fetch_or(
3406 crate::ported::zsh_h::ERRFLAG_ERROR,
3407 std::sync::atomic::Ordering::Relaxed,
3408 );
3409 exec.set_last_status(1);
3410 return true;
3411 }
3412 // c:Src/params.c:4133-4168 arrhashsetfn with
3413 // ASSPM_AUGMENT — ht = the EXISTING table, so
3414 // `[k]+=v` appends to the current value
3415 // (assignstrvalue eltflags=ASSPM_AUGMENT,
3416 // c:4144-4150) and `[k]=v` overwrites.
3417 for ch in values.chunks(3) {
3418 let elt_append = ch[0].chars().nth(1) == Some('+');
3419 let k = ch[1].clone();
3420 let v = ch[2].clone();
3421 let nv = if elt_append {
3422 format!("{}{}", map.get(&k).cloned().unwrap_or_default(), v)
3423 } else {
3424 v
3425 };
3426 map.insert(k, nv);
3427 }
3428 } else {
3429 let mut it = values.iter().cloned();
3430 while let Some(k) = it.next() {
3431 if let Some(v) = it.next() {
3432 map.insert(k, v);
3433 }
3434 }
3435 }
3436 exec.set_assoc(name, map);
3437 return false;
3438 }
3439 // Indexed-array append `arr+=(d e f)` — route directly
3440 // through canonical assignaparam with ASSPM_AUGMENT
3441 // (`Src/params.c:3570-3585` append-on-array branch).
3442 // assignaparam reads the prior array internally and
3443 // appends the new values, so the bridge no longer needs
3444 // to pre-concat manually. Marker triples from `[k]=v`
3445 // elements add ASSPM_KEY_VALUE (c:Src/exec.c:2552-2553)
3446 // so the kv sparse-fill block (c:Src/params.c:3447-3541)
3447 // resolves them against the existing elements.
3448 let kv_flag = if values
3449 .iter()
3450 .any(|e| e.starts_with(crate::ported::zsh_h::Marker))
3451 {
3452 crate::ported::zsh_h::ASSPM_KEY_VALUE
3453 } else {
3454 0
3455 };
3456 let pre_err = crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed)
3457 & crate::ported::zsh_h::ERRFLAG_ERROR;
3458 let res = crate::ported::params::assignaparam(
3459 &name,
3460 values.clone(),
3461 crate::ported::zsh_h::ASSPM_AUGMENT | kv_flag,
3462 );
3463 let now_err = crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed)
3464 & crate::ported::zsh_h::ERRFLAG_ERROR;
3465 if res.is_none() && pre_err == 0 && now_err != 0 {
3466 // c:Src/exec.c:2632-2633 — failed assignment → lastval 1.
3467 exec.set_last_status(1);
3468 return true;
3469 }
3470 #[cfg(feature = "recorder")]
3471 if crate::recorder::is_enabled() && exec.local_scope_depth == 0 {
3472 let ctx = exec.recorder_ctx();
3473 let attrs = exec.recorder_attrs_for(&name);
3474 emit_path_or_assign(&name, &values, attrs, true, &ctx);
3475 }
3476 // Tied-scalar mirror — TODO faithful: should live in
3477 // setarrvalue's gsu dispatch once boot_ paramtab wiring
3478 // lands (Task #16). Re-read the canonical post-augment
3479 // array so the joined scalar matches.
3480 let tied_scalar = exec.tied_array_to_scalar.get(&name).cloned();
3481 if let Some((scalar_name, sep)) = tied_scalar {
3482 let merged = exec.array(&name).unwrap_or_default();
3483 let joined = merged.join(&sep);
3484 exec.set_scalar(scalar_name.clone(), joined.clone());
3485 let _ = crate::ported::params::zputenv(&format!("{}={}", &scalar_name, &joined));
3486 // c:Src/params.c:5354
3487 }
3488 false
3489 });
3490 // c:Src/jobs.c:1748-1757 waitonejob — `arr+=(...)` is an
3491 // array-assignment simple command and clobbers pipestats to
3492 // `[lastval]` exactly like `arr=(...)` above. Bug #373.
3493 let status = if blocked { 1 } else { 0 };
3494 crate::ported::builtin::LASTVAL.store(status, std::sync::atomic::Ordering::Relaxed);
3495 let mut synth = crate::ported::zsh_h::job::default();
3496 crate::ported::jobs::waitonejob(&mut synth);
3497 Value::Status(status)
3498 });
3499 // `name[@]=(...)` / `name[*]=(...)` — whole-array SET with the assoc
3500 // guard (c:Src/params.c:3324-3327). Stack: [v0..vn, name].
3501 vm.register_builtin(BUILTIN_SET_ARRAY_AT, |vm, argc| {
3502 let (name, values) = pop_array_args_with_name(vm, argc);
3503 let status = with_executor(|exec| {
3504 if exec.assoc(&name).is_some() {
3505 // c:Src/params.c:3324-3327 — `[@]` (any slice) on a
3506 // PM_HASHED target is an error.
3507 crate::ported::utils::zerr(&format!(
3508 "{}: attempt to set slice of associative array",
3509 name
3510 ));
3511 crate::ported::utils::errflag.fetch_or(
3512 crate::ported::zsh_h::ERRFLAG_ERROR,
3513 std::sync::atomic::Ordering::Relaxed,
3514 );
3515 exec.set_last_status(1);
3516 return 1;
3517 }
3518 exec.set_array(name, values); // whole replace (c:3528 setarrvalue)
3519 0
3520 });
3521 Value::Status(status)
3522 });
3523 // `name[@]+=(...)` / `name[*]+=(...)` — whole-array APPEND (push) with
3524 // the same assoc guard.
3525 vm.register_builtin(BUILTIN_APPEND_ARRAY_AT, |vm, argc| {
3526 let (name, values) = pop_array_args_with_name(vm, argc);
3527 let status = with_executor(|exec| {
3528 if exec.assoc(&name).is_some() {
3529 crate::ported::utils::zerr(&format!(
3530 "{}: attempt to set slice of associative array",
3531 name
3532 ));
3533 crate::ported::utils::errflag.fetch_or(
3534 crate::ported::zsh_h::ERRFLAG_ERROR,
3535 std::sync::atomic::Ordering::Relaxed,
3536 );
3537 exec.set_last_status(1);
3538 return 1;
3539 }
3540 let mut cur = exec.array(&name).unwrap_or_default();
3541 cur.extend(values);
3542 exec.set_array(name, cur); // c:3511-3528 AUGMENT on array → push
3543 0
3544 });
3545 Value::Status(status)
3546 });
3547 vm.register_builtin(BUILTIN_RUN_SELECT, |vm, argc| {
3548 if argc < 2 {
3549 return Value::Status(1);
3550 }
3551 let n = argc as usize;
3552 let mut popped: Vec<Value> = Vec::with_capacity(n);
3553 for _ in 0..n {
3554 popped.push(vm.pop());
3555 }
3556 // popped: [sub_idx, name, word_N, ..., word_1] (popping from top)
3557 let sub_idx_val = popped.remove(0);
3558 let name_val = popped.remove(0);
3559 // c:Src/loop.c — `select` flattens Array values (from `$@`,
3560 // `${arr[@]}`, etc.) into the menu. Without per-element
3561 // splice, `select x do ... done` (bare, iterating $@)
3562 // collapsed all positionals into one joined entry.
3563 let mut words: Vec<String> = Vec::new();
3564 for v in popped.into_iter().rev() {
3565 match v {
3566 Value::Array(items) => {
3567 for item in items {
3568 words.push(item.to_str());
3569 }
3570 }
3571 other => words.push(other.to_str()),
3572 }
3573 }
3574
3575 let sub_idx = sub_idx_val.to_int() as usize;
3576 let name = name_val.to_str();
3577
3578 // c:Src/loop.c:248-252 — `if (!args || empty(args)) {
3579 // state->pc = end; ... return 0; }`. An empty option list
3580 // skips the body entirely; without this gate the prompt loop
3581 // runs indefinitely (or twice on the EOF stdin case before
3582 // exiting). Bug #401.
3583 if words.is_empty() {
3584 return Value::Status(0);
3585 }
3586
3587 let chunk = match vm.chunk.sub_chunks.get(sub_idx).cloned() {
3588 Some(c) => c,
3589 None => return Value::Status(1),
3590 };
3591
3592 let prompt =
3593 with_executor(|exec| exec.scalar("PROMPT3").unwrap_or_else(|| "?# ".to_string()));
3594
3595 let stdin = std::io::stdin();
3596 let mut reader = stdin.lock();
3597 let mut last_status: i32 = 0;
3598
3599 // c:Src/loop.c:264 — `more = selectlist(args, 0);` renders the menu
3600 // ONCE, BEFORE the selection loop. C reprints it ONLY when the user
3601 // enters an EMPTY line (c:290, inside the inner read loop) — never per
3602 // body iteration. This was a single conflated loop that re-rendered at
3603 // the top of every pass, so `printf "1\n2\n" | select x in a b; do
3604 // print $x; done` redrew the list before each prompt where zsh prints
3605 // it once.
3606 //
3607 // (`selectlist` is also ported at src/ported/loop.rs:127, but that copy
3608 // derives its row budget from adjustlines()/adjustcolumns() — an ioctl
3609 // on fd 1 — and so renders nothing when stdout is not a tty, which is
3610 // exactly this path. Keeping the working inline render here.)
3611 let render_menu = || {
3612 // Direct port of zsh's selectlist from
3613 // src/zsh/Src/loop.c:347-409. Layout is column-major
3614 // ("down columns, then across") — NOT row-major. With
3615 // 6 items in 3 cols zsh produces:
3616 // 1 3 5
3617 // 2 4 6
3618 // The previous Rust impl walked row-major which
3619 // produced 1 2 3 / 4 5 6 (visually similar but wrong
3620 // for prompts that mention ordering and breaks scripts
3621 // that rely on column count == ceil(N/rows)).
3622 //
3623 // C variable mapping:
3624 // ct -> word count (n)
3625 // longest -> max item width + 1, then plus digits-of-ct
3626 // fct -> column count
3627 // fw -> per-column width
3628 // colsz -> row count = ceil(ct / fct)
3629 // t1 -> row index, walks 0..colsz
3630 // ap -> item pointer; advances by colsz to step
3631 // DOWN a column.
3632 let term_width: usize = env::var("COLUMNS")
3633 .ok()
3634 .and_then(|v| v.parse().ok())
3635 .unwrap_or(80);
3636 let ct = words.len();
3637 // loop.c:354-363 — find longest item width.
3638 let mut longest = 1usize;
3639 for w in &words {
3640 let aplen = w.chars().count();
3641 if aplen > longest {
3642 longest = aplen;
3643 }
3644 }
3645 // loop.c:365-367 — `longest++` then add digits of `ct`.
3646 longest += 1;
3647 let mut t0 = ct;
3648 while t0 > 0 {
3649 t0 /= 10;
3650 longest += 1;
3651 }
3652 // loop.c:369-373 — fct = (cols - 1) / (longest + 3); if
3653 // 0, fct = 1; else fw = (cols - 1) / fct.
3654 let raw_fct = (term_width.saturating_sub(1)) / (longest + 3);
3655 let (fct, fw) = if raw_fct == 0 {
3656 (1, longest + 3)
3657 } else {
3658 (raw_fct, (term_width.saturating_sub(1)) / raw_fct)
3659 };
3660 // loop.c:374 — colsz = (ct + fct - 1) / fct.
3661 let colsz = ct.div_ceil(fct);
3662 // loop.c:375-395 — for each row t1, walk down columns.
3663 for t1 in 0..colsz {
3664 let mut ap_idx = t1;
3665 while ap_idx < ct {
3666 let w = &words[ap_idx];
3667 let n = ap_idx + 1;
3668 let _ = write!(std::io::stderr(), "{}) {}", n, w);
3669 let mut t2 = w.chars().count() + 2;
3670 let mut t3 = n;
3671 while t3 > 0 {
3672 t2 += 1;
3673 t3 /= 10;
3674 }
3675 // Pad to fw (loop.c:389-390).
3676 while t2 < fw {
3677 let _ = write!(std::io::stderr(), " ");
3678 t2 += 1;
3679 }
3680 ap_idx += colsz;
3681 }
3682 let _ = writeln!(std::io::stderr());
3683 }
3684 };
3685 render_menu(); // c:264 — once, before the loop
3686
3687 'select: loop {
3688 // c:266-290 — inner read loop: prompt and read until a NON-EMPTY
3689 // line arrives; each empty line reprints the menu and re-reads.
3690 let trimmed = loop {
3691 let _ = write!(std::io::stderr(), "{}", prompt);
3692 let _ = std::io::stderr().flush();
3693
3694 let mut line = String::new();
3695 match reader.read_line(&mut line) {
3696 Ok(0) => {
3697 // c:277-285 — EOF (user pressed Ctrl+D): REPLY="",
3698 // a newline to stderr, then leave the construct.
3699 with_executor(|exec| {
3700 exec.set_scalar("REPLY".to_string(), String::new());
3701 });
3702 let _ = writeln!(std::io::stderr());
3703 let _ = std::io::stderr().flush();
3704 break 'select;
3705 }
3706 Ok(_) => {}
3707 Err(_) => break 'select,
3708 }
3709 let t = line.trim_end_matches(['\n', '\r'][..].as_ref()).to_string();
3710 // c:288-289 — `if (*str) break;`
3711 if !t.is_empty() {
3712 break t;
3713 }
3714 // c:290 — `more = selectlist(args, more);` on an empty line.
3715 render_menu();
3716 };
3717 // c:291 `setsparam("REPLY", ztrdup(str));` — REPLY is set once the
3718 // inner loop yields a non-empty line. An empty line never reaches
3719 // here: c:290 reprints and re-reads instead.
3720 with_executor(|exec| {
3721 exec.set_scalar("REPLY".to_string(), trimmed.clone());
3722 });
3723
3724 // c:293 `i = atoi(str);` — atoi(3) reads a LEADING integer:
3725 // optional blanks, optional sign, then digits, ignoring whatever
3726 // trails, and yields 0 when there are no digits at all.
3727 // `parse::<usize>()` is strict and rejected `1 2` / `1abc` / `+2`
3728 // / ` 1`, so a reply with anything after the number selected
3729 // NOTHING where zsh selects the leading number's item.
3730 let i: i64 = {
3731 let b = trimmed.as_bytes();
3732 let mut p = 0;
3733 while p < b.len() && (b[p] == b' ' || b[p] == b'\t') {
3734 p += 1;
3735 }
3736 let neg = p < b.len() && b[p] == b'-';
3737 if p < b.len() && (b[p] == b'-' || b[p] == b'+') {
3738 p += 1;
3739 }
3740 let mut v: i64 = 0;
3741 while p < b.len() && b[p].is_ascii_digit() {
3742 v = v.saturating_mul(10).saturating_add((b[p] - b'0') as i64);
3743 p += 1;
3744 }
3745 if neg {
3746 -v
3747 } else {
3748 v
3749 }
3750 };
3751 // c:294-301 — `if (!i) str = "";` else walk i-1 nodes and take
3752 // that word; running off the end leaves "". A NEGATIVE i walks
3753 // until the list is exhausted (`n && i`), which also lands on "".
3754 let chosen = if i <= 0 {
3755 String::new()
3756 } else {
3757 words.get((i - 1) as usize).cloned().unwrap_or_default()
3758 };
3759
3760 with_executor(|exec| {
3761 exec.set_scalar(name.clone(), chosen);
3762 });
3763
3764 // Reset canonical BREAKS/CONTFLAG before running the body
3765 // so a stale value from a sibling construct doesn't leak in.
3766 crate::ported::builtin::BREAKS.store(0, SeqCst);
3767 crate::ported::builtin::CONTFLAG.store(0, SeqCst);
3768
3769 // c:Src/loop.c — `select` increments LOOPS for the body so
3770 // `break` / `continue` inside the body see loops > 0 and
3771 // don't emit `not in while, until, select, or repeat loop`.
3772 // Mirrors execwhile/execrepeat's `LOOPS.fetch_add` pattern.
3773 // The decrement happens after the body call so a body that
3774 // explicitly returns / errors still leaves the counter
3775 // balanced for the next iteration.
3776 crate::ported::builtin::LOOPS.fetch_add(1, SeqCst);
3777
3778 crate::fusevm_disasm::maybe_print_stdout("select:body", &chunk);
3779 let mut body_vm = fusevm::VM::new(chunk.clone());
3780 register_builtins(&mut body_vm);
3781 let _ = body_vm.run();
3782 last_status = body_vm.last_status;
3783
3784 crate::ported::builtin::LOOPS.fetch_sub(1, SeqCst);
3785
3786 // Drain the canonical BREAKS/CONTFLAG counters. Mirrors
3787 // loop.c:529-534's `if (breaks) { breaks--; if (breaks ||
3788 // !contflag) break; contflag = 0; }` drain pattern.
3789 // The legacy `BREAK_SELECT=1` env-var sentinel is still
3790 // honored for backward compat.
3791 let break_legacy = with_executor(|exec| {
3792 let v = exec.scalar("BREAK_SELECT");
3793 exec.unset_scalar("BREAK_SELECT");
3794 v.map(|s| s != "0" && !s.is_empty()).unwrap_or(false)
3795 });
3796 use std::sync::atomic::Ordering::SeqCst;
3797 let breaks = crate::ported::builtin::BREAKS.load(SeqCst);
3798 if breaks > 0 {
3799 let cont = crate::ported::builtin::CONTFLAG.load(SeqCst);
3800 crate::ported::builtin::BREAKS.fetch_sub(1, SeqCst);
3801 if breaks - 1 > 0 || cont == 0 {
3802 break;
3803 }
3804 crate::ported::builtin::CONTFLAG.store(0, SeqCst);
3805 continue;
3806 }
3807 if break_legacy {
3808 break;
3809 }
3810 }
3811
3812 Value::Status(last_status)
3813 });
3814
3815 // Magic special-parameter assoc lookup. Synthesizes values from
3816 // shell state for zsh's shell-introspection assocs:
3817 // commands, aliases, galiases, saliases, dis_aliases, dis_galiases,
3818 // dis_saliases, functions, dis_functions, builtins, dis_builtins,
3819 // reswords, options, parameters, jobtexts, jobdirs, jobstates,
3820 // nameddirs, userdirs, modules.
3821 // Returns None if `name` isn't a recognized magic name.
3822
3823 // `${arr[idx]}` — pop name, then idx_str. zsh is 1-based for positive
3824 // indices; we honor that. `@`/`*` return the whole array as Value::Array
3825 // so Op::Exec splice produces N argv slots. For `${foo[key]}` where foo
3826 // is an assoc, the idx is a string key — we check assoc_arrays first
3827 // when the idx isn't `@`/`*` and the name has an assoc binding.
3828 // BUILTIN_ARRAY_INDEX — `${name[idx]}` paramsubst dispatch.
3829 // PURE PASSTHRU: pops the idx + name, hands the canonical
3830 // `${name[idx]}` form to `subst::paramsubst` (C port of
3831 // `Src/subst.c::paramsubst`). All subscript-flag dispatch
3832 // ((I)pat / (R)pat / (i)/(r)/(K)/(k), range slices `[N,M]`,
3833 // negative indices, magic-assoc shape lookup, DQ-join collapse)
3834 // lives inside paramsubst → fetchvalue → getarg in params.rs.
3835 //
3836 // Outer-flag dispatch (`(@)` / `(@k)` / `(v)NAME[(I)pat]` / etc.)
3837 // routes through BUILTIN_BRIDGE_BRACE_ARRAY at the compile path
3838 // (canonical paramsubst flag parser owns dispatch at Src/subst.c:2147+),
3839 // so BUILTIN_ARRAY_INDEX receives clean name+key with no sentinel
3840 // prefixes.
3841 vm.register_builtin(BUILTIN_ARRAY_INDEX, |vm, _argc| {
3842 let idx = vm.pop().to_str();
3843 let name = vm.pop().to_str();
3844 array_index_lookup(&name, &idx)
3845 });
3846 // BUILTIN_ARRAY_INDEX_UNBRACED — bare `$name[idx]` (no braces).
3847 // Same subscript dispatch as BUILTIN_ARRAY_INDEX when KSHARRAYS
3848 // is unset, but under KSHARRAYS the UNBRACED form does NOT
3849 // subscript at all:
3850 // c:Src/subst.c:2800-2802 — fetchvalue's bracket-parse arg is
3851 // `(unset(KSHARRAYS) || inbrace) ? 1 : -1`; -1 inhibits
3852 // subscript parsing for the bare form under KSHARRAYS.
3853 // c:Src/subst.c:2867 — the bracket-consuming loop only runs
3854 // `while (v || ((inbrace || (unset(KSHARRAYS) && vunset)) &&
3855 // isbrack(*s)))` — bare + KSHARRAYS leaves `[...]` as literal
3856 // trailing text.
3857 // The bare `$name` expands (first element for identifier-named
3858 // arrays per c:Src/params.c:2293-2296 `v->end = 1, v->isarr = 0`),
3859 // the literal `[idx]` (+ any literal suffix) joins the last word,
3860 // and the word undergoes filename generation: `[...]` is a glob
3861 // char class, so unquoted it hits the c:Src/glob.c:1873-1886
3862 // nomatch/nullglob dispatch (reused via exec.expand_glob).
3863 // Operands: [name, idx, suffix, quoted] — `quoted` set when the
3864 // word carries DQ markers (no filename generation in DQ; zsh 5.9:
3865 // `setopt ksharrays; a=(x y z); print "$a[0]"` → `x[0]`).
3866 // Verbatim zsh 5.9 ground truth for the unquoted form:
3867 // `setopt ksharrays; a=(x y z); print -- $a[0]` →
3868 // stderr `zsh:1: no matches found: x[0]`, rc=1, empty stdout.
3869 vm.register_builtin(BUILTIN_ARRAY_INDEX_UNBRACED, |vm, _argc| {
3870 let quoted = vm.pop().to_str() == "1";
3871 let suffix = vm.pop().to_str();
3872 let idx = vm.pop().to_str();
3873 let name = vm.pop().to_str();
3874 if !crate::ported::zsh_h::isset(crate::ported::zsh_h::KSHARRAYS) {
3875 let v = array_index_lookup(&name, &idx);
3876 if suffix.is_empty() {
3877 return v;
3878 }
3879 // Mirrors the previous compile-shape `ARRAY_INDEX +
3880 // Op::Concat` exactly: Concat stringifies via as_str_cow
3881 // (fusevm value.rs:132-146, arrays join with " ").
3882 return Value::str(format!("{}{}", v.to_str(), suffix));
3883 }
3884 // KSHARRAYS bare form: no subscript. Bare-`$name` words +
3885 // literal `[idx]suffix` glued onto the last word.
3886 let mut words = ksharrays_bare_words(&name);
3887 let last = format!("{}[{}]{}", words.pop().unwrap_or_default(), idx, suffix);
3888 if quoted {
3889 // DQ context — no filename generation, bracket text stays
3890 // literal.
3891 words.push(last);
3892 return Value::str(words.join(" "));
3893 }
3894 // c:Src/glob.c:1873-1886 — expand_glob handles nullglob /
3895 // NOMATCH (zerr "no matches found" + errflag + the
3896 // current_command_glob_failed cell consumed at the command
3897 // dispatch boundary) / literal passthrough for glob-free text.
3898 let matches = with_executor(|exec| exec.expand_glob(&last));
3899 let mut out: Vec<Value> = words.into_iter().map(Value::str).collect();
3900 out.extend(matches.into_iter().map(Value::str));
3901 if out.len() == 1 {
3902 return out.pop().unwrap();
3903 }
3904 Value::Array(out)
3905 });
3906 // BUILTIN_ASSOC_HAS_KEY — `${(k)assoc[name]}` key-existence query.
3907 // Pops [assoc_name, key]; returns key (Str) if present in the
3908 // assoc, empty Str otherwise. Mirrors zsh's `${(k)h[name]}`
3909 // documented semantics in zshparam(1) "Parameter Expansion Flags".
3910 // Distinct from BUILTIN_ARRAY_INDEX (which returns the VALUE) and
3911 // from `${+h[name]}` (which returns "0"/"1"). Bug #145.
3912 vm.register_builtin(BUILTIN_ASSOC_HAS_KEY, |vm, _argc| {
3913 let key = vm.pop().to_str();
3914 let name = vm.pop().to_str();
3915 // c:Src/params.c getindex — the subscript text is substituted
3916 // (singsub) before the lookup; `${(k)H[$k]}` must resolve $k.
3917 // The compiler hands this opcode the RAW subscript text, so a
3918 // dynamic key arrived literally ("$k") and matched nothing.
3919 // singsub is identity for plain keys.
3920 let key = if key.contains('$') || key.contains('`') || key.contains('\u{8c}') {
3921 crate::ported::subst::singsub(&key)
3922 } else {
3923 key
3924 };
3925 // c:Src/params.c:3131 gethkparam covers ordinary PM_HASHED
3926 // paramtab entries only. Special/magic hashes (`parameters`,
3927 // `options`, … — the zsh/parameter module's partab-backed
3928 // params, Src/Modules/parameter.c) aren't in that storage, so
3929 // a None here doesn't mean "no such assoc". Route those
3930 // through paramsubst, whose assoc materialization handles the
3931 // magic hashes and whose getarg port (c:Src/params.c:1591 +
3932 // Src/subst.c:2922) returns the KEY for `(k)` on a plain
3933 // subscript. zsh 5.9: `${(k)parameters[PATH]}` → "PATH".
3934 match crate::ported::params::gethkparam(&name) {
3935 Some(keys) => {
3936 if keys.iter().any(|k| k == &key) {
3937 Value::str(key)
3938 } else {
3939 Value::str("")
3940 }
3941 }
3942 None => paramsubst_to_value(&format!("${{(k){}[{}]}}", name, key)),
3943 }
3944 });
3945 vm.register_builtin(BUILTIN_BRIDGE_BRACE_ARRAY, |vm, _argc| {
3946 // Inner body of `${(...)...}` (already stripped of `${`/`}` by
3947 // the caller). The compiler optionally prefixes Qstring
3948 // (\u{8c}) to signal "expanded in DQ context" — strip it
3949 // here and bump in_dq_context for the paramsubst call so the
3950 // SUB_ZIP and other qt-aware paths fire.
3951 let body = vm.pop().to_str();
3952 let (dq, inner) = if let Some(rest) = body.strip_prefix('\u{8c}') {
3953 (true, rest.to_string())
3954 } else {
3955 (false, body)
3956 };
3957 if dq {
3958 with_executor(|exec| exec.in_dq_context += 1);
3959 }
3960 let v = paramsubst_to_value(&format!("${{{}}}", inner));
3961 if dq {
3962 with_executor(|exec| exec.in_dq_context -= 1);
3963 }
3964 v
3965 });
3966
3967 // BUILTIN_PARAM_FLAG — `${(flags)name}` paramsubst dispatch.
3968 // PURE PASSTHRU: pops sentinel-tagged flags + name, hands the
3969 // canonical `${(flags)name}` form to `subst::paramsubst` (C port
3970 // of `Src/subst.c::paramsubst`). The bridge does no flag
3971 // walking, no DQ-context branching, no array/scalar shape
3972 // selection — all of that lives inside paramsubst. Compile-time
3973 // context (DQ / scalar-assign-RHS) flows through executor cells
3974 // (in_dq_context, in_scalar_assign) bumped by BUILTIN_EXPAND_TEXT.
3975 vm.register_builtin(BUILTIN_PARAM_FLAG, |vm, _argc| {
3976 let flags = vm.pop().to_str();
3977 let name = vm.pop().to_str();
3978 let body = format!("${{({}){}}}", flags, name);
3979 paramsubst_to_value(&body)
3980 });
3981
3982 // `foo[key]=val` — single-key set on an assoc array. Stack: [name, key, value].
3983 // PURE PASSTHRU: assignsparam with `name[key]` form (C port of
3984 // `Src/params.c::assignsparam` subscript path at c:3210-3231)
3985 // already does the indexed-array vs assoc decision, PM_HASHED
3986 // auto-vivification, numeric-subscript bounds handling, and
3987 // PM_READONLY rejection.
3988 vm.register_builtin(BUILTIN_SET_ASSOC, |vm, _argc| {
3989 // `${~spec}` carrier: an assignment statement is a word-
3990 // pipeline boundary too — restore the user's GLOB_SUBST
3991 // before the NEXT word expands (`Z[d]=${~Z[d]}; print
3992 // ${options[globsubst]}` must read the user value).
3993 consume_tilde_globsubst_carrier();
3994 // argc 4 = compile flagged the subscript as DYNAMIC (`H[$k]`):
3995 // an EXPANDED-empty key is then a legal assoc key (C's
3996 // assignsparam isident gate sees the raw `$k` text and the
3997 // empty key stores at getindex time — zinit's
3998 // ZINIT_SICE[$1…$2] relies on it). argc 3 = source-literal
3999 // key; `H[]` stays the "not an identifier" error.
4000 let key_is_dynamic = if _argc == 4 {
4001 vm.pop().to_int() != 0
4002 } else {
4003 false
4004 };
4005 let value = vm.pop().to_str();
4006 let key = vm.pop().to_str();
4007 let name = vm.pop().to_str();
4008 // Bash sparse-array tracking for `a[i]=v` (scalar single-index). A set
4009 // that pads the dense Vec past its old end leaves old_len..i as holes;
4010 // on an undefined array, `a[5]=q` leaves only index 5 (count 1). Only
4011 // for INDEXED arrays (assoc keys are strings), bash mode, numeric key.
4012 // Captured before the assign; applied after (on the array path).
4013 let sparse_track: Option<(String, usize, usize)> =
4014 if crate::dash_mode::bash_mode() && !key.contains(',') {
4015 key.trim().parse::<usize>().ok().and_then(|i| {
4016 with_executor(|exec| {
4017 if !exec.has_assoc(&name) {
4018 let old_len = exec.array(&name).map(|a| a.len()).unwrap_or(0);
4019 Some((name.clone(), old_len, i))
4020 } else {
4021 None
4022 }
4023 })
4024 })
4025 } else {
4026 None
4027 };
4028 if key_is_dynamic && key.is_empty() {
4029 with_executor(|exec| {
4030 let _ = exec;
4031 });
4032 // Mirror assignsparam's PM_HASHED tail directly (the
4033 // textual `name[]` reconstruction can't pass isident).
4034 if let Ok(mut store) = crate::ported::params::paramtab_hashed_storage().lock() {
4035 let entry = store.entry(name.clone()).or_default();
4036 let newval = if let Some(old) = entry.get("") {
4037 // `+=` arrives pre-concatenated by the compile
4038 // read-modify-write; plain `=` overwrites.
4039 let _ = old;
4040 value.clone()
4041 } else {
4042 value.clone()
4043 };
4044 entry.insert(String::new(), newval);
4045 }
4046 return Value::Status(0);
4047 }
4048 with_executor(|exec| {
4049 #[cfg(feature = "recorder")]
4050 if crate::recorder::is_enabled() && exec.local_scope_depth == 0 {
4051 let ctx = exec.recorder_ctx();
4052 let attrs = exec.recorder_attrs_for(&name);
4053 crate::recorder::emit_assoc_assign(
4054 &name,
4055 vec![(key.clone(), value.clone())],
4056 attrs,
4057 true,
4058 ctx,
4059 );
4060 }
4061 let _ = exec;
4062 });
4063 // Build `name[key]=value` shape for assignsparam's subscript
4064 // dispatch. Arith-evaluate numeric subscripts on an existing
4065 // indexed array (`a[i+1]=v` form) before handing off — the
4066 // canonical port currently only handles literal int / string
4067 // keys, so pre-resolve here.
4068 let resolved_key = with_executor(|exec| {
4069 // Existence probes only — use the non-cloning `has_*`
4070 // checks. `exec.assoc()` / `exec.array()` return owned
4071 // clones of the whole map/vector, so probing `.is_some()`
4072 // here copied the entire associative array on every
4073 // `h[k]=v` (O(n) per store → O(n²) for a fill loop). The
4074 // profiler flagged `ShellExecutor::assoc → IndexMap::clone`
4075 // as the dominant cost.
4076 let is_indexed = exec.has_array(&name);
4077 let is_assoc = exec.has_assoc(&name);
4078 let is_scalar = !is_indexed && !is_assoc && exec.has_scalar(&name);
4079 // c:Src/params.c::getindex — `(i)pat` / `(I)pat` / `(R)pat`
4080 // / `(r)pat` subscript flags on an indexed array LHS resolve
4081 // to a numeric index (first / last match of pat). On a
4082 // SCALAR LHS the same flags resolve to a CHAR position
4083 // (1-based first/last match of pat in the scalar string)
4084 // for the c:2748+ char-splice assignment. zshrs's
4085 // read-form `${a[(i)pat]}` already implements both shapes;
4086 // the LHS assignment path silently stored the literal
4087 // "(i)pat" as an assoc key (for scalar: auto-vivified to
4088 // PM_HASHED via the assignsparam unknown-subscript
4089 // fallback). Bug #293 (array) / scalar sibling.
4090 //
4091 // Detect the `(flags)pat` shape and resolve to a numeric
4092 // index before assignsparam.
4093 if is_indexed || is_scalar {
4094 if let Some(rest) = key.strip_prefix('(') {
4095 if let Some(close) = rest.find(')') {
4096 let flags = &rest[..close];
4097 let pat = &rest[close + 1..];
4098 if !flags.is_empty()
4099 && flags
4100 .chars()
4101 .all(|c| matches!(c, 'I' | 'R' | 'i' | 'r' | 'n' | 'e'))
4102 {
4103 // Resolve via the array's contents.
4104 if let Some(arr) = exec.array(&name) {
4105 let return_index = true; // LHS write — index needed
4106 let down = flags.contains('I') || flags.contains('R');
4107 let exact = flags.contains('e');
4108 let iter: Box<dyn Iterator<Item = (usize, &String)>> = if down {
4109 Box::new(arr.iter().enumerate().rev())
4110 } else {
4111 Box::new(arr.iter().enumerate())
4112 };
4113 let mut found: Option<usize> = None;
4114 for (idx, elem) in iter {
4115 let matched = if exact {
4116 elem == pat
4117 } else {
4118 crate::ported::pattern::patcompile(
4119 &{
4120 let mut __pat_tok = (pat).to_string();
4121 crate::ported::glob::tokenize(&mut __pat_tok);
4122 __pat_tok
4123 },
4124 crate::ported::zsh_h::PAT_HEAPDUP as i32,
4125 None,
4126 )
4127 .map_or(false, |p| crate::ported::pattern::pattry(&p, elem))
4128 };
4129 if matched {
4130 found = Some(idx);
4131 break;
4132 }
4133 }
4134 let _ = return_index;
4135 // (i)/(r) return 1-based index of match,
4136 // arr.len()+1 (or 1 for I/R) on miss
4137 // per zsh docs. We mirror the read-form
4138 // semantics from subst.rs.
4139 let idx_1based = match found {
4140 Some(i) => (i + 1) as i64,
4141 None => (arr.len() + 1) as i64,
4142 };
4143 return idx_1based.to_string();
4144 }
4145 // Scalar LHS — resolve to a CHAR position
4146 // (1-based first/last match of pat in the
4147 // string). c:Src/params.c:1411-1418 — the
4148 // scalar path returns the char index from
4149 // sliding-window pattern match against
4150 // pm.u_str. Same algorithm as the read-form
4151 // at subst.rs:5283-5306. Bug (scalar
4152 // sibling of #293): `a=hello; a[(I)l]=X`
4153 // previously auto-vivified `a` into
4154 // PM_HASHED with key "(I)l" instead of
4155 // splicing X at the last 'l' position
4156 // (yielding "helXo").
4157 if is_scalar {
4158 let s = exec.scalar(&name).unwrap_or_default();
4159 let s_chars: Vec<char> = s.chars().collect();
4160 let n = s_chars.len();
4161 let want_last = flags.contains('I') || flags.contains('R');
4162 let exact = flags.contains('e');
4163 let mut found: Option<usize> = None;
4164 'outer: for start in 0..=n {
4165 let lengths: Box<dyn Iterator<Item = usize>> = if want_last {
4166 Box::new((1..=(n - start)).rev())
4167 } else {
4168 Box::new(1..=(n - start))
4169 };
4170 for len in lengths {
4171 let cand: String =
4172 s_chars[start..start + len].iter().collect();
4173 let matched = if exact {
4174 cand == pat
4175 } else {
4176 crate::ported::pattern::patcompile(
4177 &{
4178 let mut __pat_tok = (pat).to_string();
4179 crate::ported::glob::tokenize(&mut __pat_tok);
4180 __pat_tok
4181 },
4182 crate::ported::zsh_h::PAT_HEAPDUP as i32,
4183 None,
4184 )
4185 .map_or(false, |p| {
4186 crate::ported::pattern::pattry(&p, &cand)
4187 })
4188 };
4189 if matched {
4190 found = Some(start);
4191 if !want_last {
4192 break 'outer;
4193 }
4194 break;
4195 }
4196 }
4197 }
4198 // (I)/(R): scan again to find LAST.
4199 if want_last {
4200 let mut last_found: Option<usize> = found;
4201 for start in (0..=n).rev() {
4202 for len in 1..=(n - start) {
4203 let cand: String =
4204 s_chars[start..start + len].iter().collect();
4205 let matched = if exact {
4206 cand == pat
4207 } else {
4208 crate::ported::pattern::patcompile(
4209 &{
4210 let mut __pat_tok = (pat).to_string();
4211 crate::ported::glob::tokenize(
4212 &mut __pat_tok,
4213 );
4214 __pat_tok
4215 },
4216 crate::ported::zsh_h::PAT_HEAPDUP as i32,
4217 None,
4218 )
4219 .map_or(false, |p| {
4220 crate::ported::pattern::pattry(&p, &cand)
4221 })
4222 };
4223 if matched {
4224 last_found = Some(start);
4225 break;
4226 }
4227 }
4228 if last_found.is_some() && last_found.unwrap() >= start {
4229 break;
4230 }
4231 }
4232 found = last_found;
4233 }
4234 let idx_1based = match found {
4235 Some(i) => (i + 1) as i64,
4236 // (i) miss → len+1 (one past end).
4237 None => (n + 1) as i64,
4238 };
4239 return idx_1based.to_string();
4240 }
4241 }
4242 }
4243 }
4244 }
4245 if is_indexed && key.trim().parse::<i64>().is_err() {
4246 crate::ported::math::mathevali(&crate::ported::subst::singsub(&key))
4247 .map(|n| n.to_string())
4248 .unwrap_or(key.clone())
4249 } else {
4250 key.clone()
4251 }
4252 });
4253 // c:Src/params.c getindex — C parses the subscript from the
4254 // TOKENIZED source word, so a `]`/`}` that arrived via `$key`
4255 // expansion is plain data and can never terminate the
4256 // subscript. The textual `name[key]` rebuild below re-parses
4257 // the FLAT string, where an expanded `]` splits the key at the
4258 // first bracket (`c[$k]=5` with k='x]y' stored key "x" and
4259 // spilled junk — zpwr expandstats died on the spill in a later
4260 // math expr). For a PM_HASHED target the compile-time split
4261 // already isolated the exact key: store it directly via the
4262 // canonical hashed storage (same mechanism as the
4263 // dynamic-empty-key arm above / assignsparam's PM_HASHED
4264 // tail), with the readonly guard assignsparam would apply.
4265 let target_flags = with_executor(|exec| exec.param_flags(&name));
4266 // PM_SPECIAL exclusion: the zsh/parameter magic assocs
4267 // (functions / aliases / galiases / saliases / options / …)
4268 // have per-key setfns with SIDE EFFECTS — `functions[x]=body`
4269 // must parse the body into shfunctab (Src/Modules/
4270 // parameter.c:296 setfunction), `aliases[x]=v` must write
4271 // aliastab. The direct hashed-storage store below silently
4272 // swallowed those: zinit's tmp-subst wrappers
4273 // (`functions[autoload]=':zinit-tmp-subst-autoload "$@";'`)
4274 // never became real functions, so every
4275 // `.zinit-tmp-subst-off` spammed `unfunction: no such hash
4276 // table element: autoload/compdef/bindkey/…`. Route specials
4277 // through assignsparam's canonical per-name arms instead.
4278 if (target_flags as u32 & crate::ported::zsh_h::PM_HASHED) != 0
4279 && (target_flags as u32 & crate::ported::zsh_h::PM_SPECIAL) == 0
4280 {
4281 if (target_flags as u32 & crate::ported::zsh_h::PM_READONLY) != 0 {
4282 crate::ported::utils::zerr(&format!("read-only variable: {}", name));
4283 return Value::Status(1);
4284 }
4285 if let Ok(mut store) = crate::ported::params::paramtab_hashed_storage().lock() {
4286 store
4287 .entry(name.clone())
4288 .or_default()
4289 .insert(resolved_key.clone(), value.clone());
4290 }
4291 return Value::Status(0);
4292 }
4293 let subscripted = format!("{}[{}]", name, resolved_key);
4294 crate::ported::params::assignsparam(&subscripted, &value, crate::ported::zsh_h::ASSPM_WARN);
4295 if let Some((nm, old_len, i)) = sparse_track {
4296 crate::bash_arrays::note_subscript_set(&nm, old_len, i);
4297 }
4298 Value::Status(0)
4299 });
4300
4301 // Brace expansion. Routes through executor.xpandbraces (already
4302 // implemented for the pre-fusevm executor). Returns Value::Array.
4303 // BUILTIN_ARRAY_DROP_EMPTY — filter out empty Value::Str entries
4304 // from a Value::Array on the stack. Used by `for x in $@` /
4305 // `for x in $*` unquoted forms which drop empty positionals
4306 // (POSIX-like) but do NOT IFS-split each element internally
4307 // (zsh-specific — scalar word splitting is off by default).
4308 // Distinct from BUILTIN_WORD_SPLIT which routes through
4309 // multsub PREFORK_SPLIT (full IFS-split). Bug #166.
4310 vm.register_builtin(BUILTIN_ARRAY_DROP_EMPTY, |vm, _argc| {
4311 let v = vm.pop();
4312 match v {
4313 Value::Array(items) => {
4314 let filtered: Vec<Value> = items
4315 .into_iter()
4316 .filter(|x| !x.to_str().is_empty())
4317 .collect();
4318 Value::Array(filtered)
4319 }
4320 Value::Str(s) if s.is_empty() => Value::Array(Vec::new()),
4321 other => other,
4322 }
4323 });
4324
4325 // BUILTIN_QUOTEDZPUTS — re-wrap top-of-stack scalar via the
4326 // canonical quotedzputs (Src/utils.c:6464). Non-printable bytes
4327 // come back as `$'…'` C-string form so the cond xtrace prefix
4328 // line preserves the source-quoting form for `[[ -n $'\C-[OP' ]]`
4329 // instead of leaking raw ESC + "OP" bytes through the terminal.
4330 vm.register_builtin(BUILTIN_QUOTEDZPUTS, |vm, _argc| {
4331 let s = vm.pop().to_str();
4332 Value::str(crate::ported::utils::quotedzputs(&s))
4333 });
4334
4335 // BUILTIN_QUOTE_TOKENIZED_OUTPUT — char-aware mirror of
4336 // c:Src/exec.c:2114 `quote_tokenized_output`. The canonical
4337 // port at exec::quote_tokenized_output operates on bytes
4338 // (zsh's metafied encoding); zshrs strings are UTF-8 so
4339 // `\u{87}` Star is `[0xC2, 0x87]`, and a byte walk writes
4340 // 0xC2 raw (invalid UTF-8 lead → U+FFFD on lossy decode).
4341 // Walk by char and dispatch the same switch the byte port
4342 // uses, but with the token chars matching the UTF-8 form.
4343 vm.register_builtin(BUILTIN_QUOTE_TOKENIZED_OUTPUT, |vm, _argc| {
4344 let s = vm.pop().to_str();
4345 let mut out = String::with_capacity(s.len());
4346 let chars: Vec<char> = s.chars().collect();
4347 let mut i = 0;
4348 while i < chars.len() {
4349 let c = chars[i];
4350 // c:2120 — Meta-quoted byte: emit `*++s ^ 32`.
4351 // In UTF-8 strings Meta is `\u{83}`; the next char is
4352 // the metafied payload.
4353 if c == '\u{83}' {
4354 if let Some(&n) = chars.get(i + 1) {
4355 if (n as u32) < 0x80 {
4356 out.push(((n as u8) ^ 32) as char);
4357 } else {
4358 out.push(n);
4359 }
4360 i += 2;
4361 continue;
4362 }
4363 i += 1;
4364 continue;
4365 }
4366 // c:2124 — Nularg: skip.
4367 if c == '\u{a1}' {
4368 i += 1;
4369 continue;
4370 }
4371 // c:2128-2143 — ASCII specials get backslash-prefixed
4372 // then fall through to emit the literal char.
4373 match c {
4374 '\\' | '<' | '>' | '(' | '|' | ')' | '^' | '#' | '~' | '[' | ']' | '*' | '?'
4375 | '$' | ' ' => {
4376 out.push('\\');
4377 out.push(c);
4378 i += 1;
4379 continue;
4380 }
4381 '\t' => {
4382 out.push_str("$'\\t'");
4383 i += 1;
4384 continue;
4385 }
4386 '\n' => {
4387 out.push_str("$'\\n'");
4388 i += 1;
4389 continue;
4390 }
4391 '\r' => {
4392 out.push_str("$'\\r'");
4393 i += 1;
4394 continue;
4395 }
4396 '=' => {
4397 if i == 0 {
4398 out.push('\\');
4399 }
4400 out.push(c);
4401 i += 1;
4402 continue;
4403 }
4404 _ => {}
4405 }
4406 // c:2163 — `if (itok(*s)) putc(ztokens[*s - Pound]);`
4407 // Map zsh token chars (`\u{84}`..`\u{a1}` range, the
4408 // ones the lexer emits for `#$^*()…`) back to their
4409 // source ASCII via the `ztokens` table.
4410 let cp = c as u32;
4411 if (0x84..=0xa1).contains(&cp) {
4412 let idx = (cp - 0x84) as usize;
4413 let ztokens = crate::ported::lex::ztokens.as_bytes();
4414 if idx < ztokens.len() {
4415 out.push(ztokens[idx] as char);
4416 i += 1;
4417 continue;
4418 }
4419 }
4420 out.push(c);
4421 i += 1;
4422 }
4423 Value::str(out)
4424 });
4425
4426 // BUILTIN_WORD_SPLIT — `${=var}` IFS-split runtime.
4427 // PURE PASSTHRU: route through canonical `subst::multsub` with
4428 // PREFORK_SPLIT flag (C port of `Src/subst.c::multsub` at c:544
4429 // — the IFS-split walker with whitespace-vs-non-whitespace
4430 // gating, quote-aware parsing, and empty-field handling).
4431 vm.register_builtin(BUILTIN_WORD_SPLIT, |vm, _argc| {
4432 let s = vm.pop().to_str();
4433 let (_joined, parts, _isarr, _flags) =
4434 crate::ported::subst::multsub(&s, crate::ported::zsh_h::PREFORK_SPLIT);
4435 // Empty single-string special case → empty Array (drop empty arg).
4436 if parts.len() == 1 && parts[0].is_empty() {
4437 return Value::Array(Vec::new());
4438 }
4439 nodes_to_value(parts)
4440 });
4441
4442 // BUILTIN_FORCE_SPLIT — `${=name}` / SH_WORD_SPLIT forced split.
4443 // c:Src/subst.c:3920-3928 —
4444 // if (force_split && !isarr) {
4445 // aval = sepsplit(val, spsep, 0, 1);
4446 // if (!aval || !aval[0]) val = dupstring("");
4447 // else if (!aval[1]) val = aval[0];
4448 // else isarr = nojoin ? 1 : 2;
4449 // }
4450 // with spsep == NULL for the `=` flag, so sepsplit falls through to
4451 // Src/utils.c:3711 spacesplit(s, allownull=0). See BUILTIN_FORCE_SPLIT's
4452 // doc comment for the empty-field rule and the argc contract.
4453 vm.register_builtin(BUILTIN_FORCE_SPLIT, |vm, argc| {
4454 let s = vm.pop().to_str();
4455 let keep_empties = argc == 1;
4456 // c:3921 — `sepsplit(val, spsep, 0, 1)`; spsep NULL → spacesplit.
4457 let raw = crate::ported::utils::sepsplit(&s, None, false);
4458 // c:Src/subst.c:36 `char nulstring[] = {Nularg, '\0'};` — spacesplit
4459 // emits this for an empty field delimited by IFS-NON-whitespace
4460 // (c:Src/utils.c:3732 / :3752); it survives prefork's empty-node
4461 // delete and remnulargs (c:Src/glob.c:3649) turns it back into "".
4462 // A plain "" field (c:3734 / :3757) is what a skipped run of
4463 // IFS-WHITESPACE leaves behind, and prefork DOES delete that one.
4464 let nulstring = crate::ported::zsh_h::Nularg.to_string();
4465 let mut out: Vec<String> = Vec::with_capacity(raw.len());
4466 for w in raw {
4467 if w == nulstring {
4468 out.push(String::new());
4469 } else if w.is_empty() {
4470 if keep_empties {
4471 out.push(String::new());
4472 }
4473 } else {
4474 out.push(w);
4475 }
4476 }
4477 if out.is_empty() {
4478 // c:3922-3923 — `if (!aval || !aval[0]) val = dupstring("");`:
4479 // the split produced nothing, so the value is the empty SCALAR.
4480 // Quoted, that is one empty word (c:4465 `if (qt && !*y) y =
4481 // dupstring(nulstring);` → `a=( "${=v}" )` has one element);
4482 // unquoted, prefork deletes it and the word vanishes.
4483 if keep_empties {
4484 return Value::str(String::new());
4485 }
4486 note_empty_is_scalar(true);
4487 return Value::Array(Vec::new());
4488 }
4489 if out.len() == 1 {
4490 // c:3924 — `else if (!aval[1]) val = aval[0];` — a one-field
4491 // split stays a SCALAR (this is why `${#${(f)v}}` counts
4492 // characters when the split yields a single line).
4493 return Value::str(out.into_iter().next().unwrap());
4494 }
4495 // c:3927 — `isarr = nojoin ? 1 : 2;`
4496 Value::Array(out.into_iter().map(Value::str).collect())
4497 });
4498
4499 vm.register_builtin(BUILTIN_BRACE_EXPAND, |vm, _argc| {
4500 // c:Src/glob.c::xpandbraces — brace expansion runs per word.
4501 // When the upstream produced an array (e.g. `${a:e}` splat),
4502 // expand braces on each element separately so the splat
4503 // survives. `pop().to_str()` would join with space and lose
4504 // the array shape. Parity bug #28 cousin: the BRACE_EXPAND
4505 // emit always fires for any word containing `{` (including
4506 // `${...}` param-expansion braces), so its collapse hit even
4507 // pure-paramsubst args.
4508 let raw = vm.pop();
4509 // c:Src/options.c — `no_brace_expand` (negated braceexpand)
4510 // disables brace expansion entirely. When set, `{a,b}` stays
4511 // literal. Mirror by short-circuiting xpandbraces; pass the
4512 // input through unchanged.
4513 let brace_expand = opt_state_get("braceexpand").unwrap_or(true);
4514 let brace_ccl = opt_state_get("braceccl").unwrap_or(false);
4515 let inputs: Vec<String> = match raw {
4516 Value::Array(items) => items.into_iter().map(|v| v.to_str()).collect(),
4517 other => vec![other.to_str()],
4518 };
4519 if !brace_expand {
4520 return nodes_to_value(inputs);
4521 }
4522 let mut out: Vec<String> = Vec::with_capacity(inputs.len());
4523 for s in inputs {
4524 for w in crate::ported::glob::xpandbraces(&s, brace_ccl) {
4525 out.push(w);
4526 }
4527 }
4528 nodes_to_value(out)
4529 });
4530
4531 // `*(qual)` glob qualifier filter. Stack: [pattern, qualifier].
4532 // Pattern is glob-expanded normally, then each result is filtered by the
4533 // qualifier predicate. Common qualifiers:
4534 // . — regular files only
4535 // / — directories only
4536 // @ — symlinks
4537 // x — executable
4538 // r/w/x — readable/writable/executable
4539 // N — nullglob (no error if no match)
4540 // L+N / L-N — size > N / size < N (in bytes)
4541 // mh-N / mh+N — modified within N hours / older than N hours
4542 // md-N / md+N — modified within N days / older than N days
4543 // on/On — sort by name asc/desc (default)
4544 // oL/OL — sort by length
4545 // om/Om — sort by mtime
4546 // Pop a scalar pattern, run expand_glob, push Value::Array. Used
4547 // by the segment-concat compile path for `$D/*`-style words.
4548 vm.register_builtin(BUILTIN_GLOB_EXPAND, |vm, _argc| {
4549 // c:Src/glob.c:1872 — honour `setopt noglob` / `noglob CMD`
4550 // precommand. When the option is on, the word stays literal
4551 // (zsh skips the glob expansion entirely). Without this, the
4552 // segment-fast-path BUILTIN_GLOB_EXPAND fired even after
4553 // `noglob` set the option, so `noglob echo *.xyz` saw the
4554 // NOMATCH error instead of the literal pass-through.
4555 let raw = vm.pop();
4556 let noglob =
4557 opt_state_get("noglob").unwrap_or(false) || !opt_state_get("glob").unwrap_or(true);
4558 glob_expand_word_value(raw, noglob)
4559 });
4560 // Redirect-target variant of BUILTIN_GLOB_EXPAND. c:Src/glob.c:
4561 // 2161-2167 xpandredir — `prefork(&fake, isset(MULTIOS) ? 0 :
4562 // PREFORK_SINGLE, NULL)` then "Globbing is only done for
4563 // multios.": a redirect target word is only globbed when the
4564 // MULTIOS option is set. With it unset, `echo hi > *.txt`
4565 // creates the literal file `*.txt`, and `wc -c < *.txt` errors
4566 // "no such file or directory: *.txt". Bug #36 follow-up in
4567 // docs/BUGS.md.
4568 vm.register_builtin(BUILTIN_REDIR_GLOB_EXPAND, |vm, _argc| {
4569 let raw = vm.pop();
4570 let noglob =
4571 opt_state_get("noglob").unwrap_or(false) || !opt_state_get("glob").unwrap_or(true);
4572 let multios = opt_state_get("multios").unwrap_or(true);
4573 glob_expand_word_value(raw, noglob || !multios)
4574 });
4575 // Clear the default-word glob-pending carrier before the word's
4576 // expansion runs, so a flag set by a prior word never leaks in.
4577 vm.register_builtin(BUILTIN_DEFAULT_WORD_GLOB_RESET, |_vm, _argc| {
4578 crate::ported::subst::DEFAULT_WORD_GLOB_PENDING.with(|c| c.set(false));
4579 Value::Status(0)
4580 });
4581 // After the word is assembled, run filename generation ONLY if the
4582 // default/alternate paramsubst arm flagged a source-glob default
4583 // (DEFAULT_WORD_GLOB_PENDING). Otherwise pass the word through
4584 // literally — a parameter VALUE must not glob. c:Src/subst.c globlist.
4585 vm.register_builtin(BUILTIN_DEFAULT_WORD_GLOB, |vm, _argc| {
4586 let raw = vm.pop();
4587 let pending = crate::ported::subst::DEFAULT_WORD_GLOB_PENDING.with(|c| {
4588 let v = c.get();
4589 c.set(false); // read + clear
4590 v
4591 });
4592 if !pending {
4593 return raw;
4594 }
4595 let noglob =
4596 opt_state_get("noglob").unwrap_or(false) || !opt_state_get("glob").unwrap_or(true);
4597 glob_expand_word_value(raw, noglob)
4598 });
4599
4600 // `break`/`continue` from a sub-VM body. The compile path emits
4601 // these when the keyword appears at chunk top-level (no enclosing
4602 // for/while in the current chunk's patch lists). Outer-loop
4603 // builtins (BUILTIN_RUN_SELECT and any future loop-via-builtin
4604 // construct) drain canonical BREAKS/CONTFLAG after each iteration.
4605 //
4606 // Writes match `bin_break`'s c:5836+ pattern:
4607 // continue: contflag = 1; breaks++ (Src/builtin.c::bin_break)
4608 // break: breaks++
4609 vm.register_builtin(BUILTIN_SET_BREAK, |_vm, _argc| {
4610 use std::sync::atomic::Ordering::SeqCst;
4611 crate::ported::builtin::BREAKS.fetch_add(1, SeqCst);
4612 Value::Status(0)
4613 });
4614 vm.register_builtin(BUILTIN_SET_CONTINUE, |_vm, _argc| {
4615 use std::sync::atomic::Ordering::SeqCst;
4616 crate::ported::builtin::CONTFLAG.store(1, SeqCst);
4617 crate::ported::builtin::BREAKS.fetch_add(1, SeqCst);
4618 Value::Status(0)
4619 });
4620
4621 // `break N`/`continue N` with a RUNTIME level count. Pops [count,
4622 // name]; math-evaluates count (c:builtin.c:5811 `mathevali`); on
4623 // count <= 0 emits `argument is not positive: N` via zerrnam (sets
4624 // errflag → abort, c:5813) and pushes Int(0) (matches no jump-table
4625 // entry → control falls through to the errflag abort). Otherwise
4626 // pushes Int(count) for the compiled jump table to dispatch on.
4627 vm.register_builtin(BUILTIN_BREAK_COUNT_VALIDATE, |vm, _argc| {
4628 let name = vm.pop().to_str();
4629 let count_s = vm.pop().to_str();
4630 let count = crate::ported::math::mathevali(&count_s).unwrap_or(0);
4631 if count <= 0 {
4632 crate::ported::utils::zerrnam(&name, &format!("argument is not positive: {count}"));
4633 return Value::Int(0);
4634 }
4635 Value::Int(count)
4636 });
4637
4638 // `${arr[*]}` — join array elements with the first IFS char into
4639 // a single string. Matches zsh: in DQ context this preserves the
4640 // join; in array context too the result is one Value::Str.
4641 // Set or clear a shell option directly. Used by `noglob CMD ...`
4642 // precommand wrapping — the compiler emits SET_RAW_OPT to flip the
4643 // option ON before compiling the inner words and OFF after, so glob
4644 // expansion of the inner args sees the temporary state.
4645 vm.register_builtin(BUILTIN_SET_RAW_OPT, |vm, _argc| {
4646 let on = vm.pop().to_int() != 0;
4647 let opt = vm.pop().to_str();
4648 // Pure passthru: canonical port lives in
4649 // src/ported/options.rs::opt_state_set_via_alias and
4650 // handles negation-alias resolution per c:Src/options.c.
4651 crate::ported::options::opt_state_set_via_alias(&opt, on);
4652 Value::Status(0)
4653 });
4654
4655 // c:Src/options.c GLOB_SUBST — runtime glob expansion of
4656 // substituted words. Pop a Value (Str or Array); when
4657 // GLOB_SUBST is ON, run expand_glob on each string element;
4658 // when OFF, pass through unchanged. Bug #119 in docs/BUGS.md.
4659 vm.register_builtin(BUILTIN_GLOB_SUBST_EXPAND, |vm, _argc| {
4660 let raw = vm.pop();
4661 let glob_subst = crate::ported::zsh_h::isset(crate::ported::zsh_h::GLOBSUBST);
4662 if !glob_subst {
4663 return raw;
4664 }
4665 // Collect input strings (Str → vec![s]; Array → multiple).
4666 let inputs: Vec<String> = match raw {
4667 Value::Array(items) => items.into_iter().map(|v| v.to_str()).collect(),
4668 other => vec![other.to_str()],
4669 };
4670 // Run expand_glob on each. Empty matches collapse to a
4671 // single literal pass-through to mirror nullglob-off default.
4672 let mut out: Vec<String> = Vec::with_capacity(inputs.len());
4673 for pattern in inputs {
4674 // c:Src/subst.c — GLOB_SUBST subjects the value to the FULL
4675 // filename-generation pipeline: `filesub` (tilde/`=` expansion)
4676 // BEFORE globbing (prefork runs filesub then globlist). zshrs
4677 // globbed but skipped filesub, so `${~x}` / `setopt globsubst`
4678 // left `~/foo` un-expanded. filesubstr matches the Tilde TOKEN,
4679 // so shtokenize the value first (`~`→Tilde, glob metas active),
4680 // run filesub, then untokenize the surviving glob metas back to
4681 // raw for expand_glob (which re-tokenizes internally).
4682 let pattern = if pattern.contains('~') || pattern.contains('=') {
4683 let mut tok = pattern.clone();
4684 crate::ported::glob::shtokenize(&mut tok);
4685 let fs = crate::ported::subst::filesub(&tok, 0);
4686 crate::ported::lex::untokenize(&fs)
4687 } else {
4688 pattern
4689 };
4690 let matches = with_executor(|exec| exec.expand_glob(&pattern));
4691 if matches.is_empty() {
4692 // No match: keep the literal (like nullglob off).
4693 out.push(pattern);
4694 } else {
4695 for m in matches {
4696 out.push(m);
4697 }
4698 }
4699 }
4700 if out.len() == 1 {
4701 Value::str(out.into_iter().next().unwrap())
4702 } else {
4703 Value::Array(out.into_iter().map(Value::str).collect())
4704 }
4705 });
4706
4707 // c:Src/math.c:337 — `getmathparam` for ArithCompiler pre-load.
4708 // Pop a variable name, return its math-coerced value. Mirrors
4709 // the routing in math::getmathparam: try i64, then f64, then
4710 // recursive arith-eval, else 0. Bug #118 in docs/BUGS.md.
4711 vm.register_builtin(BUILTIN_GET_MATH_VAR, |vm, _argc| {
4712 let name = vm.pop().to_str();
4713 let raw = crate::ported::params::getsparam(&name).unwrap_or_default();
4714 // Empty / unset → 0.
4715 if raw.is_empty() {
4716 return Value::Int(0);
4717 }
4718 // Direct int / float parse.
4719 if let Ok(n) = raw.parse::<i64>() {
4720 return Value::Int(n);
4721 }
4722 if let Ok(f) = raw.parse::<f64>() {
4723 return Value::Float(f);
4724 }
4725 // Recursive arith eval (matches getmathparam fallback at
4726 // Src/math.c:337). If that fails too, return 0 — C's
4727 // mathevall returns 0 with errflag set on parse failure.
4728 match crate::ported::math::mathevali(&raw) {
4729 Ok(n) => Value::Int(n),
4730 Err(_) => Value::Int(0),
4731 }
4732 });
4733
4734 // c:Src/options.c GLOB_SUBST + Src/cond.c:552 cond_match.
4735 // Pop pattern string; when GLOB_SUBST is OFF, escape every glob
4736 // metachar with `\` so the downstream StrMatch + patcompile
4737 // treat them as literals (matching C's tokenization-based
4738 // gate). When GLOB_SUBST is ON, pass through unchanged.
4739 // See BUILTIN_GLOB_SUBST_GUARD docs above for full rationale.
4740 vm.register_builtin(BUILTIN_GLOB_SUBST_GUARD, |vm, _argc| {
4741 let p = vm.pop().to_str();
4742 let glob_subst = crate::ported::zsh_h::isset(crate::ported::zsh_h::GLOBSUBST);
4743 if glob_subst {
4744 return Value::str(p);
4745 }
4746 let mut out = String::with_capacity(p.len() * 2);
4747 for c in p.chars() {
4748 match c {
4749 '*' | '?' | '[' | ']' | '(' | ')' | '|' | '<' | '>' | '#' | '^' | '~' | '\\' => {
4750 out.push('\\');
4751 out.push(c);
4752 }
4753 _ => out.push(c),
4754 }
4755 }
4756 Value::str(out)
4757 });
4758
4759 vm.register_builtin(BUILTIN_ARRAY_JOIN_STAR, |vm, _argc| {
4760 let name = vm.pop().to_str();
4761 let (joined, ifs_full, in_dq) = with_executor(|exec| {
4762 // c:Src/params.c — `"$*"` joins by IFS[0]. zsh
4763 // distinguishes IFS=unset (→ default `" "`) from
4764 // IFS="" (→ EMPTY separator → fields concatenate).
4765 // chars().next() collapsed both into the default, so
4766 // IFS="" was treated as IFS=" ".
4767 let ifs_full = exec.scalar("IFS").unwrap_or_else(|| " \t\n".to_string());
4768 let sep = ifs_full
4769 .chars()
4770 .next()
4771 .map(|c| c.to_string())
4772 .unwrap_or_default();
4773 let in_dq = exec.in_dq_context > 0;
4774 let joined = if let Some(v) = crate::dash_mode::bash_special_array(&name) {
4775 // bash `"${PIPESTATUS[*]}"` / `"${FUNCNAME[*]}"` join.
4776 v.join(&sep)
4777 } else if name == "@" || name == "*" || name == "argv" {
4778 exec.pparams().join(&sep)
4779 } else if let Some(assoc_map) = exec.assoc(&name) {
4780 // c:Src/params.c — assoc-splat values for
4781 // `"${h[@]}"` / `"${h[*]}"`. Bug #109 in
4782 // docs/BUGS.md.
4783 assoc_map.values().cloned().collect::<Vec<_>>().join(&sep)
4784 } else if let Some(arr) = exec.array(&name) {
4785 // bash sparse arrays: `"${a[*]}"` joins only LIVE elements,
4786 // dropping hole slots. No-op in --zsh (no holes tracked).
4787 crate::bash_arrays::compact(&name, arr).join(&sep)
4788 } else {
4789 exec.get_variable(&name)
4790 };
4791 (joined, ifs_full, in_dq)
4792 });
4793 // c:Src/subst.c — UNQUOTED `${name[*]}` (or `$*`) goes
4794 // through the canonical "join via IFS[0], then word-split
4795 // via IFS" pipeline. The fast-path bypassed paramsubst
4796 // entirely so it never word-split, producing one joined
4797 // string instead of N argv entries. Bug #428.
4798 //
4799 // In QUOTED (`"${name[*]}"`) context, the result IS a
4800 // single scalar — return it as Str without splitting.
4801 if in_dq {
4802 return Value::str(joined);
4803 }
4804 if joined.is_empty() {
4805 return Value::Array(Vec::new());
4806 }
4807 // IFS word-split — every IFS char is a separator. Empty
4808 // resulting fields are dropped (the canonical
4809 // "remove empty unquoted words" pass from
4810 // Src/subst.c::prefork c:184-187).
4811 let parts: Vec<String> = joined
4812 .split(|c: char| ifs_full.contains(c))
4813 .filter(|s| !s.is_empty())
4814 .map(String::from)
4815 .collect();
4816 if parts.is_empty() {
4817 Value::Array(Vec::new())
4818 } else if parts.len() == 1 {
4819 Value::str(parts.into_iter().next().unwrap())
4820 } else {
4821 Value::Array(parts.into_iter().map(Value::str).collect())
4822 }
4823 });
4824
4825 vm.register_builtin(BUILTIN_ARRAY_ALL, |vm, _argc| {
4826 let name = vm.pop().to_str();
4827 // c:Src/params.c:2027-2029 — a `[@]`/`[*]` subscript sets
4828 // SCANPM_ISVAR_AT, i.e. `isarr != 0` (c:2915). An empty result is
4829 // therefore an empty ARRAY, and plan9 deletes the whole word
4830 // (c:4362) rather than keeping the surrounding text.
4831 note_empty_is_scalar(false);
4832 // bash `"${PIPESTATUS[@]}"` / `"${FUNCNAME[@]}"` / `"${BASH_VERSINFO[@]}"`
4833 // splat — alias the zsh-native special. No-op in --zsh.
4834 if let Some(v) = crate::dash_mode::bash_special_array(&name) {
4835 return Value::Array(v.into_iter().map(Value::str).collect());
4836 }
4837 with_executor(|exec| {
4838 // Special positional names — splice the positional list.
4839 if name == "@" || name == "*" || name == "argv" {
4840 return Value::Array(exec.pparams().iter().map(Value::str).collect());
4841 }
4842 // c:Src/Modules/parameter.c — funcstack/funcfiletrace/
4843 // funcsourcetrace/functrace are PM_ARRAY|PM_READONLY
4844 // specials backed by the canonical FUNCSTACK Vec.
4845 // `${funcstack[@]}` inside a function call should splat
4846 // the innermost-first names; without this branch the
4847 // runtime fell to the scalar fallback (get_variable
4848 // returns empty for these specials) and `[@]` came out
4849 // empty. Bug #276 in docs/BUGS.md. Mirrors the parallel
4850 // arrays_get handler at src/ported/subst.rs ~10685.
4851 // c:Src/Modules/datetime.c:256 — `epochtime` PM_ARRAY|
4852 // PM_READONLY backed by getcurrenttime(). Same parallel
4853 // arrangement as the FUNCSTACK-backed specials below.
4854 if name == "epochtime" {
4855 let arr = crate::ported::modules::datetime::getcurrenttime();
4856 return Value::Array(arr.into_iter().map(Value::str).collect());
4857 }
4858 if matches!(
4859 name.as_str(),
4860 "funcstack" | "funcfiletrace" | "funcsourcetrace" | "functrace"
4861 ) {
4862 // Route the three trace arrays through the canonical
4863 // ported getfns (Src/Modules/parameter.c:648/:679/:711)
4864 // — the previous inline copy emitted wrong shapes
4865 // (bare filename for funcfiletrace, `name:lineno` for
4866 // functrace instead of `caller:lineno`); same dedup as
4867 // the parallel arrays_get handler in subst.rs.
4868 let vals: Vec<String> = match name.as_str() {
4869 "funcstack" => crate::ported::modules::parameter::FUNCSTACK
4870 .lock()
4871 .map(|f| f.iter().rev().map(|fs| fs.name.clone()).collect())
4872 .unwrap_or_default(),
4873 "funcfiletrace" => crate::ported::modules::parameter::funcfiletracegetfn(
4874 std::ptr::null_mut(),
4875 ),
4876 "funcsourcetrace" => crate::ported::modules::parameter::funcsourcetracegetfn(
4877 std::ptr::null_mut(),
4878 ),
4879 _ => crate::ported::modules::parameter::functracegetfn(std::ptr::null_mut()),
4880 };
4881 return Value::Array(vals.into_iter().map(Value::str).collect());
4882 }
4883 // c:Src/params.c — `${assoc[@]}` enumerates VALUES (per
4884 // params.c:1696-1750 hashparam splat). Check assoc
4885 // storage BEFORE the scalar fallback so an associative
4886 // array named X resolves `${X[@]}` to the values, not
4887 // empty. Bug #109 in docs/BUGS.md: `${h[@]}` on an
4888 // assoc routed through BUILTIN_ARRAY_ALL, which only
4889 // consulted `exec.array(name)` (the indexed-array map)
4890 // — that lookup missed for assocs, fell through to
4891 // `get_variable("h")` (also empty for an assoc-only
4892 // name), and returned `Array(vec![])`. zsh's expected
4893 // behavior is to enumerate values.
4894 if let Some(assoc_map) = exec.assoc(&name) {
4895 return Value::Array(
4896 assoc_map.values().cloned().map(Value::str).collect(),
4897 );
4898 }
4899 match exec.array(&name) {
4900 Some(v) => {
4901 // bash sparse arrays: `"${a[@]}"` splats only LIVE
4902 // elements, dropping hole slots (`a[5]=q` padding,
4903 // `unset a[i]`). No-op in --zsh (no holes tracked).
4904 let v = crate::bash_arrays::compact(&name, v);
4905 Value::Array(v.iter().map(Value::str).collect())
4906 }
4907 None => {
4908 // Fall back to scalar lookup. zsh (unlike bash)
4909 // does NOT IFS-split a scalar variable in a for
4910 // list — `for w in $scalar` iterates ONCE with the
4911 // scalar value. Word-splitting requires either
4912 // sh_word_split option or explicit `${(s.,.)scalar}`.
4913 let val = exec.get_variable(&name);
4914 if val.is_empty() && !exec.has_scalar(&name) && env::var(&name).is_err() {
4915 // c:Src/subst.c:3480-3485 — `${arr[@]}` on a genuinely
4916 // UNSET parameter under NO_UNSET is a "parameter not set"
4917 // error (vunset > 0 && unset(UNSET)), exactly like the
4918 // scalar `$arr`, the `${arr[*]}` splat, and `${arr[1]}`
4919 // — all of which already fire it via GET_VAR. The `[@]`
4920 // splat path returned an empty array silently, so
4921 // `setopt NO_UNSET; print "${arr[@]}"` exited 0 where zsh
4922 // exits 1. A DECLARED-but-empty array (`arr=()`) resolves
4923 // to `Some(vec![])` above and never reaches here, so it
4924 // still splats to nothing without erroring — matching zsh.
4925 if opt_state_get("nounset").unwrap_or(false) {
4926 crate::ported::utils::zerr(&format!("{}: parameter not set", name));
4927 crate::ported::utils::errflag.fetch_or(
4928 crate::ported::zsh_h::ERRFLAG_ERROR,
4929 std::sync::atomic::Ordering::Relaxed,
4930 );
4931 exec.set_last_status(1);
4932 }
4933 // c:Src/subst.c:3480-3485 — an UNSET parameter takes the
4934 // `vunset` arm: `val = dupstring("")` with isarr left at
4935 // 0. That is a SCALAR empty, so plan9 keeps the
4936 // surrounding text (`setopt rcexpandparam;
4937 // print -r -- "[${unset[@]}]"` → `[]`), unlike a
4938 // DECLARED-but-empty array (`arr=()`, matched by the
4939 // `Some(vec![])` arm above), which sets isarr and gets
4940 // the word deleted at c:4362.
4941 note_empty_is_scalar(true);
4942 Value::Array(vec![])
4943 } else if opt_state_get("shwordsplit").unwrap_or(false) {
4944 // c:3921 `aval = sepsplit(val, spsep, 0, 1)` — same
4945 // splitter as `${=name}` (Src/utils.c:3711 spacesplit),
4946 // not a naive `split().filter(non-empty)`: only the
4947 // IFS-WHITESPACE-derived empty fields are elided; the
4948 // `nulstring` ones an IFS-NON-whitespace separator makes
4949 // survive (c:Src/subst.c:36).
4950 let nulstring = crate::ported::zsh_h::Nularg.to_string();
4951 let parts: Vec<Value> =
4952 crate::ported::utils::sepsplit(&val, None, false)
4953 .into_iter()
4954 .filter_map(|w| {
4955 if w == nulstring {
4956 Some(Value::str(String::new()))
4957 } else if w.is_empty() {
4958 None // c:184-187 prefork uremnode
4959 } else {
4960 Some(Value::str(w))
4961 }
4962 })
4963 .collect();
4964 Value::Array(parts)
4965 } else {
4966 Value::Array(vec![Value::str(val)])
4967 }
4968 }
4969 }
4970 })
4971 });
4972
4973 // BUILTIN_ARRAY_FLATTEN(N): pops N values, flattens one level of Array
4974 // nesting, pushes the resulting Array AND its length as a separate Int.
4975 // The two-value return shape lets the caller (for-loop compile path)
4976 // SetSlot the length before SetSlot'ing the array, without re-deriving
4977 // the length from the array via a second builtin call.
4978 // `coproc [name] { body }` — bidirectional pipe to backgrounded body.
4979 // Stack discipline (top first): [name (str, "" for default), sub_idx (int)].
4980 // On success: parent's `executor.arrays[name]` becomes [write_fd, read_fd]
4981 // and Status(0) is returned. The caller writes to the child's stdin via
4982 // write_fd, reads its stdout via read_fd, and closes both when done.
4983 //
4984 // Bash's coproc convention is `${NAME[0]}` = read_fd, `${NAME[1]}` =
4985 // write_fd. We follow that: arrays[name] = [read_fd_str, write_fd_str].
4986 vm.register_builtin(BUILTIN_RUN_COPROC, |vm, _argc| {
4987 let sub_idx = vm.pop().to_int() as usize;
4988 let job_text = vm.pop().to_str();
4989 let raw_name = vm.pop().to_str();
4990 let name = if raw_name.is_empty() {
4991 "COPROC".to_string()
4992 } else {
4993 raw_name
4994 };
4995 let chunk = match vm.chunk.sub_chunks.get(sub_idx).cloned() {
4996 Some(c) => c,
4997 None => return Value::Status(1),
4998 };
4999
5000 // c:Src/exec.c:1710-1712 — starting a new coproc closes the
5001 // previous one's fds FIRST:
5002 // if (coprocin >= 0) { zclose(coprocin); zclose(coprocout); }
5003 // The old coproc child then sees EOF on its stdin and exits on
5004 // its own schedule (its job-table entry stays until it's
5005 // reaped) — zsh does NOT deletejob it here. This is also what
5006 // makes the `exec 4<&p; coproc exit; read -u4` EOF idiom work:
5007 // the replacement coproc closes the shell's write end to the
5008 // old one.
5009 {
5010 use std::sync::atomic::Ordering;
5011 let old_in = crate::ported::modules::clone::coprocin.load(Ordering::Relaxed);
5012 if old_in >= 0 {
5013 let old_out = crate::ported::modules::clone::coprocout.load(Ordering::Relaxed);
5014 unsafe {
5015 libc::close(old_in);
5016 if old_out >= 0 {
5017 libc::close(old_out);
5018 }
5019 }
5020 crate::ported::modules::clone::coprocin.store(-1, Ordering::Relaxed);
5021 crate::ported::modules::clone::coprocout.store(-1, Ordering::Relaxed);
5022 }
5023 }
5024
5025 // (parent_read ← child_stdout)
5026 let mut p2c = [0i32; 2]; // parent writes, child reads
5027 let mut c2p = [0i32; 2]; // child writes, parent reads
5028 if unsafe { libc::pipe(p2c.as_mut_ptr()) } < 0 {
5029 return Value::Status(1);
5030 }
5031 if unsafe { libc::pipe(c2p.as_mut_ptr()) } < 0 {
5032 unsafe {
5033 libc::close(p2c[0]);
5034 libc::close(p2c[1]);
5035 }
5036 return Value::Status(1);
5037 }
5038 // c:Src/exec.c:5160 mpipe — both pipes' fds are moved above
5039 // the user-visible range (movefd → F_DUPFD ≥ 10) so the
5040 // coproc fds never collide with explicit user fds like
5041 // `exec 3>&p`.
5042 for fd in p2c.iter_mut().chain(c2p.iter_mut()) {
5043 *fd = crate::ported::utils::movefd(*fd);
5044 }
5045
5046 match unsafe { libc::fork() } {
5047 -1 => {
5048 unsafe {
5049 libc::close(p2c[0]);
5050 libc::close(p2c[1]);
5051 libc::close(c2p[0]);
5052 libc::close(c2p[1]);
5053 }
5054 Value::Status(1)
5055 }
5056 0 => {
5057 // Child: stdin from p2c[0], stdout to c2p[1]. Close all
5058 // unused fds. setsid so SIGINT to fg doesn't hit us.
5059 unsafe {
5060 libc::dup2(p2c[0], libc::STDIN_FILENO);
5061 libc::dup2(c2p[1], libc::STDOUT_FILENO);
5062 libc::close(p2c[0]);
5063 libc::close(p2c[1]);
5064 libc::close(c2p[0]);
5065 libc::close(c2p[1]);
5066 libc::setsid();
5067 }
5068 crate::fusevm_disasm::maybe_print_stdout("coproc:child", &chunk);
5069 let mut co_vm = fusevm::VM::new(chunk);
5070 register_builtins(&mut co_vm);
5071 let _ = co_vm.run();
5072 let _ = std::io::stdout().flush();
5073 let _ = std::io::stderr().flush();
5074 std::process::exit(co_vm.last_status);
5075 }
5076 pid => {
5077 // Parent: close child ends, store [read_fd, write_fd] in NAME.
5078 unsafe {
5079 libc::close(p2c[0]);
5080 libc::close(c2p[1]);
5081 }
5082 let read_fd = c2p[0];
5083 let write_fd = p2c[1];
5084 with_executor(|exec| {
5085 exec.unset_scalar(&name);
5086 exec.set_array(name, vec![read_fd.to_string(), write_fd.to_string()]);
5087 });
5088 // c:Src/exec.c — `coprocin`/`coprocout` are the
5089 // canonical globals that bin_read's `-p` arm
5090 // (Src/builtin.c:6510) and bin_print's `-p` arm
5091 // (Src/builtin.c:4827) read to find the
5092 // coprocess fds. The Rust port has the atomic
5093 // declarations at src/ported/modules/clone.rs:262
5094 // but the coproc-launch path never updated them,
5095 // so `read -p` / `print -p` always errored with
5096 // "-p: no coprocess" even when a coproc was
5097 // running. Bug #388 in docs/BUGS.md. Update them
5098 // here so the canonical builtins find the live
5099 // pipe.
5100 crate::ported::modules::clone::coprocin
5101 .store(read_fd, std::sync::atomic::Ordering::Relaxed);
5102 crate::ported::modules::clone::coprocout
5103 .store(write_fd, std::sync::atomic::Ordering::Relaxed);
5104 // c:Src/exec.c:1725 — `fdtable[coprocin] =
5105 // fdtable[coprocout] = FDT_UNUSED;`: the two kept ends
5106 // are user-reachable (via `>&p` / `<&p`), so they drop
5107 // the FDT_INTERNAL mark movefd gave them.
5108 crate::ported::utils::fdtable_set(read_fd, crate::ported::zsh_h::FDT_UNUSED);
5109 crate::ported::utils::fdtable_set(write_fd, crate::ported::zsh_h::FDT_UNUSED);
5110 // c:Src/exec.c:2837 — `lastpid = (zlong) pid;`. zsh
5111 // sets the `$!` global to the coproc child's PID so
5112 // subsequent `$!` reads return it. The Rust port at
5113 // exec.rs:6773 mirrors this for regular background
5114 // jobs but the coproc launch path was missing the
5115 // assignment, leaving `$!` at 0 after `coproc cmd`.
5116 crate::ported::modules::clone::lastpid
5117 .store(pid, std::sync::atomic::Ordering::Relaxed);
5118 // c:Src/exec.c:1700-1758 — the coproc rides the SAME
5119 // Z_ASYNC job-table path as `cmd &`: `thisjob = newjob
5120 // = initjob()` (c:1700), addproc hangs the pid+text
5121 // proc entry off the job, `jobtab[thisjob].stat |=
5122 // STAT_NOSTTY` (c:1746), `clearoldjobtab()` (c:1744)
5123 // and `spawnjob()` (c:1758) promote it to curjob. This
5124 // is what makes `jobs` list the coproc as
5125 // `[1] + running cat` and `kill %1` resolve it.
5126 // Mirrors the BUILTIN_RUN_BG parent arm exactly.
5127 {
5128 use crate::ported::jobs;
5129 use std::sync::Mutex;
5130 let table = jobs::JOBTAB.get_or_init(|| Mutex::new(Vec::new()));
5131 let idx = {
5132 let mut tab = table.lock().unwrap_or_else(|e| e.into_inner());
5133 let idx = jobs::initjob(&mut tab); // c:exec.c:1700
5134 jobs::addproc(
5135 &mut tab[idx],
5136 pid,
5137 &job_text,
5138 false,
5139 Some(std::time::Instant::now()),
5140 -1,
5141 -1,
5142 );
5143 tab[idx].stat |= crate::ported::zsh_h::STAT_NOSTTY; // c:exec.c:1746
5144 idx
5145 };
5146 jobs::clearoldjobtab(); // c:exec.c:1744
5147 if let Ok(mut tj) = jobs::THISJOB.get_or_init(|| Mutex::new(-1)).lock() {
5148 *tj = idx as i32;
5149 }
5150 jobs::spawnjob(); // c:exec.c:1758
5151 }
5152 with_executor(|exec| {
5153 exec.jobs
5154 .add_pid_job(pid, job_text.clone(), JobState::Running);
5155 });
5156 Value::Status(0)
5157 }
5158 }
5159 });
5160
5161 vm.register_builtin(BUILTIN_ARRAY_FLATTEN, |vm, argc| {
5162 let n = argc as usize;
5163 let start = vm.stack.len().saturating_sub(n);
5164 let raw: Vec<Value> = vm.stack.drain(start..).collect();
5165 let mut flat: Vec<Value> = Vec::with_capacity(raw.len());
5166 for v in raw {
5167 match v {
5168 Value::Array(items) => flat.extend(items),
5169 other => flat.push(other),
5170 }
5171 }
5172 let len = flat.len() as i64;
5173 // Push the array first; the Int(len) becomes the builtin's return
5174 // value (which CallBuiltin already pushes). Caller consumes in
5175 // reverse: SetSlot(len_slot) pops Int, SetSlot(arr_slot) pops Array.
5176 vm.push(Value::Array(flat));
5177 Value::Int(len)
5178 });
5179
5180 // Shell variable get/set — routes through executor.variables so nested
5181 // VMs (function calls) and tree-walker callers see the same storage.
5182 // GET_VAR / GET_VAR_DQ share one body via `get_var_impl`; the only
5183 // difference is `force_dq`, which the compiler sets for QUOTED simple
5184 // reads (`"$name"`) so an array's empty elements are preserved (the
5185 // `in_dq_context` runtime flag is 0 for these compiler-direct reads).
5186 fn get_var_impl(vm: &mut fusevm::VM, argc: u8, force_dq: bool) -> Value {
5187 let args = pop_args(vm, argc);
5188 let name = args.into_iter().next().unwrap_or_default();
5189 let live_status = vm.last_status;
5190 // `$@` and `$*` need splice semantics — return Value::Array of
5191 // positional params so for-loop's BUILTIN_ARRAY_FLATTEN spreads them
5192 // and pop_args splits them into argv slots. zsh's `"$@"` bslashquote-each-
5193 // word semantics matches: each pos-param becomes its own arg.
5194 // Same for arrays accessed by name (e.g. `$arr` in some contexts).
5195 //
5196 // vm.last_status is authoritative: `subshell_end` now returns
5197 // Some(status) and fusevm's `Op::SubshellEnd` writes it into
5198 // vm.last_status, so a deferred subshell `exit N` is visible
5199 // here. Suppressing this sync (as an older revision did, back
5200 // when the host hook returned nothing) made LASTVAL win over
5201 // any status the VM set AFTER SubshellEnd — which dropped the
5202 // `!` negation of `Src/exec.c:1979-1980`
5203 // if ((slflags & WC_SUBLIST_NOT) && !errflag && !retflag)
5204 // lastval = !lastval;
5205 // for `! (exit 7)` (emit_negate_status' SetStatus updated
5206 // vm.last_status, then `$?` read the stale LASTVAL=7).
5207 let sync_status = |exec: &mut ShellExecutor| {
5208 exec.set_last_status(live_status);
5209 };
5210 if name == "@" || name == "*" {
5211 // Quoting decides empty-word retention (c:Src/subst.c:
5212 // 184-187): the COMPILE site knows it and emits
5213 // BUILTIN_ARRAY_DROP_EMPTY after this read for the
5214 // unquoted form only — in_dq_context is NOT a valid
5215 // discriminator here (the quoted "$@" fast path emits
5216 // GET_VAR directly without an EXPAND_TEXT wrapper).
5217 return with_executor(|exec| {
5218 sync_status(exec);
5219 Value::Array(exec.pparams().iter().map(Value::str).collect())
5220 });
5221 }
5222 // RC_EXPAND_PARAM: when the option is set and `name` refers to
5223 // an array, return Value::Array so the enclosing word's
5224 // BUILTIN_CONCAT_DISTRIBUTE distributes element-wise. Without
5225 // the option, arrays still join to a space-separated scalar
5226 // (zsh's default unquoted-array-as-scalar semantics).
5227 let rc_expand = with_executor(|exec| opt_state_get("rcexpandparam").unwrap_or(false));
5228 // c:Src/subst.c — under KSHARRAYS a bare `$name` (no [@]/[*] subscript;
5229 // this GET_VAR path only handles the bare form) is element 1 ONLY — a
5230 // scalar. RC_EXPAND_PARAM then has a single value to distribute, so
5231 // `$acc` → "p1", NOT the whole array. Skip the whole-array rc_expand
5232 // shortcut when KSHARRAYS is set and fall through to the normal path,
5233 // which applies the element-1 collapse. Without this gate,
5234 // `setopt KSH_ARRAYS rc_expand_param; print -r -- $acc` splatted every
5235 // element while zsh prints just "p1".
5236 let ksh_arrays = crate::ported::zsh_h::isset(crate::ported::zsh_h::KSHARRAYS);
5237 if rc_expand && !ksh_arrays {
5238 let arr_val = with_executor(|exec| {
5239 sync_status(exec);
5240 exec.array(&name)
5241 });
5242 if let Some(arr) = arr_val {
5243 // c:4245 — a real array reference (`isarr != 0`). An empty one
5244 // takes plan9's word-removal path (c:4362), so clear the
5245 // scalar bit; a preceding empty-SCALAR expansion in the same
5246 // word would otherwise leave it set and keep the word alive
5247 // (`empty=''; e=(); a=("$empty"); print -rl -- x$e y`).
5248 note_empty_is_scalar(false);
5249 return Value::Array(arr.into_iter().map(Value::str).collect());
5250 }
5251 }
5252 // Magic-assoc fallback FIRST — `${aliases}` / `${functions}`
5253 // / `${commands}` / etc. should return the value list per
5254 // zsh's bare-assoc semantics. Without this, those names fell
5255 // through to `get_variable` which is empty (they live in
5256 // separate executor tables, not `assoc_arrays`). Return as
5257 // a Value::Array so `arr=(${aliases})` distributes into
5258 // multiple elements, matching zsh's array-context word
5259 // splitting for assoc-bare references.
5260 let magic_vals = with_executor(|exec| {
5261 sync_status(exec);
5262 // Canonical PARTAB dispatch (Src/Modules/parameter.c:2235-
5263 // 2298 + SPECIALPMDEFs in mapfile/terminfo/termcap/system/
5264 // zleparameter): PARTAB_ARRAY entries → whole-array getfn;
5265 // PARTAB entries → scan keys + per-key getpm/scanpm fn
5266 // pointers.
5267 let _ = exec;
5268 if let Some(values) = partab_array_get(&name) {
5269 Some(values)
5270 } else if let Some(keys) = partab_scan_keys(&name) {
5271 Some(
5272 keys.iter()
5273 .map(|k| partab_get(&name, k).unwrap_or_default())
5274 .collect::<Vec<_>>(),
5275 )
5276 } else {
5277 None
5278 }
5279 });
5280 if let Some(vals) = magic_vals {
5281 // Distinguish "name IS a magic-assoc with no entries"
5282 // (return Array(empty)) from "name is unknown — fall
5283 // through to get_variable".
5284 // c:Src/params.c:2293-2296 — KSHARRAYS bare reference
5285 // collapses to the FIRST element in scan order
5286 // (`v->end = 1, v->isarr = 0`). For `options` the scan
5287 // order is optiontab bucket order (OPTIONTAB), so zsh 5.9
5288 // prints `off` (posixargzero) for
5289 // `setopt ksharrays; print $options`.
5290 if opt_state_get("ksharrays").unwrap_or(false) {
5291 return Value::str(vals.into_iter().next().unwrap_or_default());
5292 }
5293 return Value::Array(vals.into_iter().map(Value::str).collect());
5294 }
5295 // Indexed-array path: return Value::Array so pop_args splats
5296 // each element into its own argv slot. Direct port of zsh's
5297 // unquoted `$arr` semantics — each element becomes a separate
5298 // word in command-arg position.
5299 //
5300 // DQ context exception: inside `"...$arr..."`, zsh joins with
5301 // the first char of $IFS (default space) so the DQ word stays
5302 // a single argv slot. Detect via in_dq_context (bumped by
5303 // BUILTIN_EXPAND_TEXT mode 1) and return the joined scalar.
5304 // Direct port of Src/subst.c:1759-1813 nojoin/sepjoin: in DQ
5305 // (qt=1) without explicit `(@)`, sepjoin runs and the result
5306 // is one word.
5307 let arr_assoc_data = with_executor(|exec| {
5308 sync_status(exec);
5309 let in_dq = force_dq || exec.in_dq_context > 0;
5310 // KSH_ARRAYS: bare `$arr` returns ONLY arr[0] (zero-
5311 // based first-element-only semantics). Direct port of
5312 // Src/params.c getstrvalue's KSH_ARRAYS gate which
5313 // returns aval[0] instead of the whole array.
5314 let ksh_arrays = opt_state_get("ksharrays").unwrap_or(false);
5315 if let Some(arr) = exec.array(&name) {
5316 if ksh_arrays {
5317 return Some((vec![arr.first().cloned().unwrap_or_default()], in_dq));
5318 }
5319 return Some((arr.clone(), in_dq));
5320 }
5321 if exec.assoc(&name).is_some() {
5322 // c:Src/params.c:2351-2358 — under KSH EMULATION a bare
5323 // `$assoc` is `${assoc[0]}` (a KEY-"0" lookup), so it is
5324 // EMPTY unless the hash actually has a key "0". This is
5325 // EMULATION-gated, not KSHARRAYS-option-gated: `emulate -L
5326 // ksh; typeset -A h=(a 1 b 2); print $h` is empty, whereas
5327 // `setopt ksharrays; …; print $h` collapses to the bucket-
5328 // first value below.
5329 if crate::ported::zsh_h::EMULATION(crate::ported::zsh_h::EMULATE_KSH) {
5330 let v = crate::ported::subst::assoc_get(&name)
5331 .and_then(|m| m.get("0").cloned())
5332 .unwrap_or_default();
5333 return Some((vec![v], in_dq));
5334 }
5335 // c:Src/hashtable.c scanhashtable — a bare `$assoc` joins its
5336 // VALUES in zsh hash-BUCKET order (the same order `(k)`/`(v)`
5337 // enumerate), NOT sorted or insertion order. `assoc_get`
5338 // rebuilds zsh's bucket layout; use it so `$as` matches
5339 // `${(v)as}` (`as=(zebra 9 apple 1)` → `9 1`, not the
5340 // alphabetical `1 9`). Under KSHARRAYS the bare form collapses
5341 // to the bucket-FIRST value (`9`), matching zsh.
5342 let values: Vec<String> = crate::ported::subst::assoc_get(&name)
5343 .map(|m| m.values().cloned().collect())
5344 .unwrap_or_default();
5345 if ksh_arrays {
5346 return Some((vec![values.into_iter().next().unwrap_or_default()], in_dq));
5347 }
5348 return Some((values, in_dq));
5349 }
5350 None
5351 });
5352 if let Some((items, in_dq)) = arr_assoc_data {
5353 // c:Src/subst.c:184-187 — prefork's `else if (!keep)
5354 // uremnode(list, node)`: UNQUOTED expansion drops empty
5355 // list nodes before they reach argv, so `a=(y '' x);
5356 // print -- $a` passes TWO args in zsh (`y x`), while the
5357 // quoted "${a[@]}" splat keeps the empty slot. The
5358 // paramsubst splat path already does this (Bug #578
5359 // retain); this GET_VAR fast path bypassed it and leaked
5360 // empty argv slots (visible double-space, wrong arg
5361 // counts in `for`/`print -l`).
5362 let items: Vec<String> = if in_dq {
5363 items
5364 } else {
5365 items.into_iter().filter(|s| !s.is_empty()).collect()
5366 };
5367 if in_dq {
5368 // c:Src/utils.c:3936-3945 sepjoin default-sep rule:
5369 // set-but-empty IFS joins with "" (`IFS=""; echo
5370 // "$arr"` concatenates); only unset / space-leading
5371 // IFS yields " ". The previous get_variable read
5372 // couldn't distinguish unset from set-empty.
5373 return Value::str(crate::ported::utils::sepjoin(&items, None));
5374 }
5375 // c:4245 — a real array reference: `isarr != 0`, so an empty
5376 // one takes plan9's word-removal path, not the scalar path.
5377 note_empty_is_scalar(false);
5378 return Value::Array(items.into_iter().map(Value::str).collect());
5379 }
5380 let (val, in_dq, is_known) = with_executor(|exec| {
5381 sync_status(exec);
5382 let v = exec.get_variable(&name);
5383 // For nounset detection: a name is "known" when it has a
5384 // paramtab/array/assoc/env entry. Special chars ($?, $#,
5385 // $@, $*, $-, $$, $!, $_, $0) always count as known
5386 // regardless of value. Pure-digit positional params
5387 // count as known iff index <= $# (set -- has populated
5388 // that slot). c:Src/subst.c:1689 — NOUNSET fires on
5389 // unset positional param too: `set --; echo "$1"` with
5390 // nounset must diagnose.
5391 let is_special_single = name.len() == 1
5392 && matches!(
5393 name.chars().next().unwrap(),
5394 '?' | '#' | '@' | '*' | '-' | '$' | '!' | '_' | '0'
5395 );
5396 let is_pure_digit = !name.is_empty() && name.chars().all(|c| c.is_ascii_digit());
5397 let positional_known = if is_pure_digit {
5398 let idx: usize = name.parse().unwrap_or(0);
5399 if idx == 0 {
5400 true // $0 always set
5401 } else {
5402 idx <= exec.pparams().len()
5403 }
5404 } else {
5405 false
5406 };
5407 let known = !v.is_empty()
5408 || name.is_empty()
5409 || is_special_single
5410 || positional_known
5411 || crate::ported::params::paramtab()
5412 .read()
5413 .ok()
5414 .map(|t| t.contains_key(&name))
5415 .unwrap_or(false)
5416 || env::var(&name).is_ok();
5417 (v, force_dq || exec.in_dq_context > 0, known)
5418 });
5419 // c:Src/subst.c:1689 — NO_UNSET / nounset: reading an unset
5420 // parameter fires "parameter not set" diagnostic and aborts
5421 // the substitution. Direct port of the noerrs gate at c:1689
5422 // (zerr + errflag). Matches `set -u` POSIX semantics.
5423 if !is_known && opt_state_get("nounset").unwrap_or(false) {
5424 crate::ported::utils::zerr(&format!("{}: parameter not set", name));
5425 crate::ported::utils::errflag.fetch_or(
5426 crate::ported::zsh_h::ERRFLAG_ERROR,
5427 std::sync::atomic::Ordering::Relaxed,
5428 );
5429 with_executor(|exec| exec.set_last_status(1));
5430 return Value::str("");
5431 }
5432 // Empty unquoted scalar → drop the arg (zsh "remove empty
5433 // unquoted words" rule). Returning empty Value::Array makes
5434 // pop_args contribute zero items. DQ context keeps the empty
5435 // string so "$a" stays a single empty arg. Direct port of
5436 // subst.c's elide-empty pass.
5437 if val.is_empty() && !in_dq {
5438 // c:1650-1656 / c:4437 — a SCALAR parameter has `isarr == 0`,
5439 // so it never reaches plan9's word-removal at c:4362. Flag the
5440 // empty Array below as a scalar so `setopt rcexpandparam;
5441 // v=; print -rl -- x$v y` still emits `x` (only an empty
5442 // ARRAY deletes the word).
5443 note_empty_is_scalar(true);
5444 return Value::Array(Vec::new());
5445 }
5446 // c:Src/subst.c:1759 SH_WORD_SPLIT — when shwordsplit is set and
5447 // we're in unquoted command-arg position (not DQ), split scalar
5448 // value on IFS into multiple words. Matches BUILTIN_ARRAY_ALL's
5449 // shwordsplit arm (fusevm_bridge.rs:2200). Without this, bare
5450 // `$s` in `print $s` stayed a single arg even with the option
5451 // set, breaking POSIX-style scalar word-splitting.
5452 if !in_dq && opt_state_get("shwordsplit").unwrap_or(false) {
5453 // c:1705 — `spbreak = (pf_flags & PREFORK_SHWORDSPLIT) && !qt`,
5454 // then c:3902 `force_split = !ssub && (spbreak || spsep)` and
5455 // c:3921 `aval = sepsplit(val, spsep, 0, 1)`. SH_WORD_SPLIT runs
5456 // the SAME splitter as `${=name}`, so route it through the same
5457 // port. The previous `split(|c| ifs.contains(c)).filter(non-empty)`
5458 // dropped every empty field, but spacesplit (Src/utils.c:3711)
5459 // only elides the ones a run of IFS-WHITESPACE produces — an
5460 // IFS-NON-whitespace separator preserves them as `nulstring`.
5461 // `IFS=x; v=xaxbx; setopt shwordsplit; print -rl -- $v` is four
5462 // words in zsh (``, a, b, ``), not two.
5463 let raw = crate::ported::utils::sepsplit(&val, None, false); // c:3921
5464 let nulstring = crate::ported::zsh_h::Nularg.to_string(); // c:36
5465 let parts: Vec<Value> = raw
5466 .into_iter()
5467 .filter_map(|w| {
5468 if w == nulstring {
5469 Some(Value::str(String::new()))
5470 } else if w.is_empty() {
5471 // c:184-187 — prefork deletes the truly-empty node.
5472 None
5473 } else {
5474 Some(Value::str(w))
5475 }
5476 })
5477 .collect();
5478 if parts.is_empty() {
5479 // c:3922 — `val = dupstring("")`: an empty SCALAR, not an
5480 // empty array (see EMPTY_EXPANSION_IS_SCALAR).
5481 note_empty_is_scalar(true);
5482 return Value::Array(Vec::new());
5483 } else if parts.len() == 1 {
5484 // c:3924 — `else if (!aval[1]) val = aval[0];`
5485 return parts.into_iter().next().unwrap();
5486 } else {
5487 return Value::Array(parts); // c:3927
5488 }
5489 }
5490 Value::str(val)
5491 }
5492 vm.register_builtin(BUILTIN_GET_VAR, |vm, argc| get_var_impl(vm, argc, false));
5493 vm.register_builtin(BUILTIN_GET_VAR_DQ, |vm, argc| get_var_impl(vm, argc, true));
5494
5495 // `name+=val` (no parens) — runtime dispatch:
5496 // - if `name` is in `arrays` → push `val` as new element
5497 // - if `name` is in `assoc_arrays` → refuse (zsh errors here)
5498 // - else → scalar concat (existing behavior)
5499 // Stack: [name, value].
5500 vm.register_builtin(BUILTIN_APPEND_SCALAR_OR_PUSH, |vm, argc| {
5501 let args = pop_args(vm, argc);
5502 let mut iter = args.into_iter();
5503 let name = iter.next().unwrap_or_default();
5504 let value = iter.next().unwrap_or_default();
5505 with_executor(|exec| {
5506 // Array form: `arr+=elem` pushes a single element.
5507 // Routes through canonical assignaparam(name, [value],
5508 // ASSPM_AUGMENT) — Src/params.c:3357 c:3402-3412 augment
5509 // path prepends prior scalar / appends to existing array.
5510 // Existence probe uses the non-cloning `has_array` — the
5511 // owning `exec.array()` clone here made `arr+=x` in a loop
5512 // O(n²) (see the assoc-store fix above).
5513 if exec.has_array(&name) {
5514 // c:Src/params.c — under KSHARRAYS a bare array name
5515 // addresses element 0 (ksh), so `a+=X` (scalar augment)
5516 // CONCATENATES onto the first element ("firstlast second"),
5517 // it does NOT push a new element. C routes scalar `+=`
5518 // through assignsparam (which targets the elem-0 value);
5519 // zshrs's APPEND_SCALAR_OR_PUSH would otherwise push.
5520 if crate::ported::zsh_h::isset(crate::ported::zsh_h::KSHARRAYS) {
5521 let mut arr = exec.array(&name).unwrap_or_default();
5522 if arr.is_empty() {
5523 arr.push(value.clone());
5524 } else {
5525 arr[0] = format!("{}{}", arr[0], value);
5526 }
5527 exec.set_array(name.clone(), arr);
5528 #[cfg(feature = "recorder")]
5529 if crate::recorder::is_enabled() && exec.local_scope_depth == 0 {
5530 let ctx = exec.recorder_ctx();
5531 let attrs = exec.recorder_attrs_for(&name);
5532 emit_path_or_assign(&name, std::slice::from_ref(&value), attrs, true, &ctx);
5533 }
5534 return;
5535 }
5536 let _ = crate::ported::params::assignaparam(
5537 &name,
5538 vec![value.clone()],
5539 crate::ported::zsh_h::ASSPM_AUGMENT,
5540 );
5541 #[cfg(feature = "recorder")]
5542 if crate::recorder::is_enabled() && exec.local_scope_depth == 0 {
5543 let ctx = exec.recorder_ctx();
5544 let attrs = exec.recorder_attrs_for(&name);
5545 emit_path_or_assign(&name, std::slice::from_ref(&value), attrs, true, &ctx);
5546 }
5547 return;
5548 }
5549 if exec.has_assoc(&name) {
5550 eprintln!("zshrs: {}: cannot use += on assoc without (key val)", name);
5551 return;
5552 }
5553 // Scalar / integer / float form: route through canonical
5554 // assignsparam(name, value, ASSPM_AUGMENT) which
5555 // dispatches PM_TYPE — PM_SCALAR concats, PM_INTEGER
5556 // arith-adds (c:2775-2778), PM_FLOAT float-adds.
5557 let _ = crate::ported::params::assignsparam(
5558 &name,
5559 &value,
5560 crate::ported::zsh_h::ASSPM_AUGMENT,
5561 );
5562 #[cfg(feature = "recorder")]
5563 if crate::recorder::is_enabled() && exec.local_scope_depth == 0 {
5564 let ctx = exec.recorder_ctx();
5565 let attrs = exec.recorder_attrs_for(&name);
5566 // Re-read the canonical value via get_variable for the
5567 // recorder bundle (assignsparam may have transformed it
5568 // through integer/float arithmetic).
5569 let final_val = exec.get_variable(&name);
5570 let lower = name.to_ascii_lowercase();
5571 if matches!(
5572 lower.as_str(),
5573 "path" | "fpath" | "manpath" | "module_path" | "cdpath"
5574 ) {
5575 emit_path_or_assign(&name, std::slice::from_ref(&final_val), attrs, true, &ctx);
5576 } else {
5577 crate::recorder::emit_assign_typed(&name, &final_val, attrs, ctx);
5578 }
5579 }
5580 });
5581 Value::Status(0)
5582 });
5583
5584 // BUILTIN_SET_VAR — `name=value` runtime scalar assignment.
5585 // PURE PASSTHRU: hand to canonical `setsparam` (C port of
5586 // `Src/params.c::setsparam`). That walks assignsparam →
5587 // assignstrvalue which already does:
5588 // - readonly rejection (zerr + errflag at c:2701)
5589 // - PM_INTEGER math evaluation (mathevali at c:3590)
5590 // - PM_EFLOAT / PM_FFLOAT float coercion (c:3608)
5591 // - PM_LOWER / PM_UPPER case fold (via setstrvalue)
5592 // - GSU special-param dispatch (homesetfn / ifssetfn / etc.)
5593 // - allexport env mirror via the PM_EXPORTED setfn
5594 //
5595 // Bridge-only concerns kept here:
5596 // - inline_env_stack (zsh `X=foo cmd` scoped env)
5597 // - recorder emission (PFA-SMR)
5598 // - vm.last_status propagation for `a=$(cmd)` exit-code chaining
5599 // Sets the GLOB_ASSIGN-eligibility flag consumed by the NEXT BUILTIN_SET_VAR.
5600 // Emitted only when a scalar-assign RHS had an unquoted glob token. Takes no
5601 // args; its pushed return is discarded by a following Op::Pop.
5602 vm.register_builtin(BUILTIN_MARK_GLOB_ELIGIBLE, |_vm, _argc| {
5603 SET_VAR_GLOB_ELIGIBLE.with(|c| c.set(true));
5604 fusevm::Value::Int(0)
5605 });
5606 vm.register_builtin(BUILTIN_SET_VAR, |vm, argc| {
5607 // `${~spec}` carrier: an assignment statement is a word-
5608 // pipeline boundary too — restore the user's GLOB_SUBST
5609 // before the NEXT word expands (`Z[d]=${~Z[d]}; print
5610 // ${options[globsubst]}` must read the user value).
5611 consume_tilde_globsubst_carrier();
5612 // Snapshot the raw Values BEFORE pop_args's to_str
5613 // flattening — needed to distinguish Int (arith assignment,
5614 // integer-typed param) from Str (scalar assignment).
5615 let mut raw_values: Vec<fusevm::Value> = Vec::with_capacity(argc as usize);
5616 for _ in 0..argc {
5617 raw_values.push(vm.pop());
5618 }
5619 raw_values.reverse();
5620 let name = raw_values.first().map(|v| v.to_str()).unwrap_or_default();
5621 let value_raw = raw_values.get(1).cloned();
5622 let value = value_raw.as_ref().map(|v| v.to_str()).unwrap_or_default();
5623 // c:Src/params.c — when the bytecode hands us an Int value
5624 // (only the arith assignment paths emit this — `(( X = N ))`
5625 // is the canonical site), route through setiparam so the
5626 // param ends up PM_INTEGER + inherits the math layer's
5627 // `lastbase` for display formatting (`(( X = 16#ff ));
5628 // echo \$X` → `16#FF`). Scalar `X=val` and `$((expr))`
5629 // assignments still take the setsparam path below.
5630 let int_assign = matches!(value_raw, Some(fusevm::Value::Int(_)));
5631 let float_assign = matches!(value_raw, Some(fusevm::Value::Float(_)));
5632 let mut assign_failed = false;
5633 with_executor(|exec| {
5634 // c:Src/params.c assignsparam — PM_READONLY rejection
5635 // BEFORE any env mutation. The inline-env-prefix path
5636 // (`X=2 env`) called env::set_var unconditionally before
5637 // the readonly check fired in setsparam, so the OS env
5638 // got X=2 even though the assignment errored. env then
5639 // inherited the polluted env from fork, leaking the
5640 // attempted override past the readonly guard. Mirror
5641 // C's order: readonly check → zerr → bail; only mutate
5642 // env when the assignment is admissible. Bug #551
5643 // (security-relevant).
5644 if exec.is_readonly_param(&name) {
5645 crate::ported::utils::zerr(&format!("read-only variable: {}", name));
5646 return;
5647 }
5648 // Inline-assignment frame tracking (`X=foo cmd` reverts on
5649 // command return).
5650 if !exec.inline_env_stack.is_empty() {
5651 let prev_var = crate::ported::params::getsparam(&name);
5652 let prev_env = env::var(&name).ok();
5653 exec.inline_env_stack
5654 .last_mut()
5655 .unwrap()
5656 .push((name.clone(), prev_var, prev_env));
5657 let _ = crate::ported::params::zputenv(&format!("{}={}", &name, &value));
5658 // c:Src/params.c:5354
5659 }
5660 // Canonical setsparam handles readonly, integer math, case
5661 // fold, GSU dispatch. For Int values (arith assigns) route
5662 // through setiparam so the param is PM_INTEGER + inherits
5663 // the math layer's lastbase for display formatting. For
5664 // Float (arith assigns producing MN_FLOAT) route through
5665 // setnparam so the param is PM_FFLOAT — `(( b = a * 2 ))`
5666 // with scalar `a="3.14"` should create b as typeset -F,
5667 // not a scalar holding "6.28".
5668 if int_assign {
5669 if let Some(fusevm::Value::Int(i)) = value_raw {
5670 crate::ported::params::setiparam(&name, i);
5671 } else {
5672 assign_failed = crate::ported::params::setsparam(&name, &value).is_none();
5673 }
5674 } else if float_assign {
5675 if let Some(fusevm::Value::Float(f)) = value_raw {
5676 // ArithCompiler returns Value::Float whenever any
5677 // operand came through Str (BUILTIN_GET_VAR yields
5678 // Value::Str even for integer-shaped scalars). To
5679 // avoid forcing every `(( b = a + 3 ))` to PM_FFLOAT
5680 // when `a="5"` (integer-shaped), detect integer-
5681 // valued floats and route through setiparam instead.
5682 // True floats (non-integral) reach setnparam →
5683 // PM_FFLOAT so `typeset -p b` shows `typeset -F …`.
5684 if f.fract() == 0.0 && f.is_finite() && f.abs() <= i64::MAX as f64 {
5685 crate::ported::params::setiparam(&name, f as i64);
5686 } else {
5687 let mnval = crate::ported::math::mnumber {
5688 l: 0,
5689 d: f,
5690 type_: crate::ported::math::MN_FLOAT,
5691 };
5692 crate::ported::params::setnparam(&name, mnval);
5693 }
5694 } else {
5695 assign_failed = crate::ported::params::setsparam(&name, &value).is_none();
5696 }
5697 } else {
5698 // c:Src/exec.c:2554-2567 — GLOB_ASSIGN. When the
5699 // `globassign` option is on and the scalar RHS is a glob
5700 // pattern, glob it and recreate the parameter as a scalar
5701 // (≤1 match) or array (>1) — csh-style assignment. The
5702 // bridge hands `value` UNTOKENIZED, so re-tokenize
5703 // (shtokenize) before haswilds/globlist; zsh's wordcode
5704 // value arrives pre-tokenized via `htok`. The
5705 // `isset(GLOBASSIGN)` gate is first and cheap (option off
5706 // by default), so the common path is unchanged.
5707 let mut globbed = false;
5708 // Only glob the RHS when the compiler flagged an UNQUOTED glob
5709 // token in the literal wordcode (SET_VAR_GLOB_ELIGIBLE). zsh's
5710 // GLOB_ASSIGN (Src/exec.c:2554) globs literal patterns only —
5711 // `x="/tmp/*"`, `x='/tmp/*'`, `x=$param`, `x=$(cmd)` all assign
5712 // verbatim. The value arrives here untokenized (DQ-wrapped by
5713 // the compiler), so this compile-time flag is the only surviving
5714 // signal of whether the pattern was quote-protected.
5715 let glob_eligible = SET_VAR_GLOB_ELIGIBLE.with(|c| c.replace(false));
5716 if glob_eligible && crate::ported::zsh_h::isset(crate::ported::zsh_h::GLOBASSIGN) {
5717 let mut tv = value.clone();
5718 crate::ported::glob::shtokenize(&mut tv);
5719 if crate::ported::pattern::haswilds(&tv) {
5720 // Committed to the glob path: never fall back to
5721 // assigning the literal pattern (zsh errors on
5722 // no-match instead).
5723 globbed = true;
5724 // globlist tokenizes its input internally (for
5725 // haswilds + glob_path) and prints the ORIGINAL
5726 // string verbatim in its "no matches found" error,
5727 // so feed it the UNtokenized value — passing the
5728 // tokenized form would leak the Star/Quest token
5729 // bytes into the error message.
5730 let mut ll: crate::ported::linklist::LinkList<String> = Default::default();
5731 ll.push_back(value.clone());
5732 crate::ported::subst::globlist(&mut ll, 0); // c:2556
5733 if crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed)
5734 == 0
5735 {
5736 let matches: Vec<String> = ll
5737 .nodes
5738 .iter()
5739 .map(|s| crate::ported::lex::untokenize(s).to_string())
5740 .collect();
5741 crate::ported::params::unsetparam(&name); // c:2562
5742 if matches.len() <= 1 {
5743 let v = matches.into_iter().next().unwrap_or_default();
5744 assign_failed =
5745 crate::ported::params::setsparam(&name, &v).is_none();
5746 } else {
5747 crate::ported::params::setaparam(&name, matches);
5748 }
5749 }
5750 // errflag set → globlist already reported
5751 // "no matches found"; leave the param unassigned
5752 // to match zsh's abort.
5753 }
5754 }
5755 // c:Src/exec.c addvars — a NULL return from
5756 // assignsparam (e.g. nameref resolving out of scope,
5757 // createparam refusal at c:1108-1118) fails the
5758 // assignment with status 1.
5759 if !globbed {
5760 assign_failed = crate::ported::params::setsparam(&name, &value).is_none();
5761 }
5762 }
5763 // PM_EXPORTED / allexport env mirror — read AFTER setsparam
5764 // so the flag bit reflects any GSU setfn side-effects.
5765 let allexport = opt_state_get("allexport").unwrap_or(false);
5766 let already_exported =
5767 (exec.param_flags(&name) as u32 & crate::ported::zsh_h::PM_EXPORTED) != 0;
5768 if allexport || already_exported {
5769 let _ = crate::ported::params::zputenv(&format!("{}={}", &name, &value));
5770 // c:Src/params.c:5354
5771 }
5772 #[cfg(feature = "recorder")]
5773 if crate::recorder::is_enabled()
5774 && exec.local_scope_depth == 0
5775 && !matches!(
5776 name.as_str(),
5777 "PPID" | "LINENO" | "ZSH_ARGZERO" | "argv0" | "ARGC" | "?" | "_" | "RANDOM"
5778 )
5779 {
5780 let ctx = exec.recorder_ctx();
5781 let attrs = exec.recorder_attrs_for(&name);
5782 crate::recorder::emit_assign_typed(&name, &value, attrs, ctx);
5783 }
5784 });
5785 Value::Status(vm.last_status)
5786 });
5787
5788 // c:Src/exec.c execfor → Src/params.c:6362 setloopvar — the
5789 // for-loop variable bind. Distinct from BUILTIN_SET_VAR because a
5790 // PM_NAMEREF loop variable REBINDS (new refname) instead of
5791 // assigning through the resolved chain.
5792 vm.register_builtin(BUILTIN_SET_LOOP_VAR, |vm, argc| {
5793 let args = pop_args(vm, argc);
5794 let name = args.first().cloned().unwrap_or_default();
5795 let value = args.get(1).cloned().unwrap_or_default();
5796 if crate::vm_helper::is_nameref(&name) {
5797 let ef_before =
5798 crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed);
5799 crate::ported::params::setloopvar(&name, &value); // c:6362
5800 let ef_after = crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed);
5801 if (ef_after & crate::ported::utils::ERRFLAG_ERROR) != 0 && ef_after != ef_before {
5802 // zerr fired (read-only reference / invalid self
5803 // reference) — abort the loop, status 1 (C errflag).
5804 vm.last_status = 1;
5805 return Value::Bool(false);
5806 }
5807 return Value::Bool(true);
5808 }
5809 // Plain loop var — canonical scalar path (same shape as
5810 // BUILTIN_SET_VAR's setsparam arm).
5811 with_executor(|exec| {
5812 if exec.is_readonly_param(&name) {
5813 crate::ported::utils::zerr(&format!("read-only variable: {}", name));
5814 return;
5815 }
5816 crate::ported::params::setsparam(&name, &value);
5817 let allexport = opt_state_get("allexport").unwrap_or(false);
5818 let already_exported =
5819 (exec.param_flags(&name) as u32 & crate::ported::zsh_h::PM_EXPORTED) != 0;
5820 if allexport || already_exported {
5821 let _ = crate::ported::params::zputenv(&format!("{}={}", &name, &value));
5822 // c:Src/params.c:5354
5823 }
5824 });
5825 Value::Bool(true)
5826 });
5827
5828 // Pre-compiled function registration — used by compile_zsh.rs's
5829 // FuncDef path. Stack: [name, base64-bincode-of-Chunk]. We decode
5830 // the base64, deserialize the Chunk, and store directly in
5831 // executor.functions_compiled. Bypasses the ShellCommand JSON layer.
5832 // BUILTIN_VAR_EXISTS — `[[ -v name ]]` set-test.
5833 // PURE PASSTHRU: build `${+name}` and route through canonical
5834 // `subst::paramsubst` which returns "1" for set / "0" for unset
5835 // (C port of `Src/subst.c::paramsubst` plus-prefix arm).
5836 // paramsubst handles all the shapes the 48-line hand-roll did:
5837 // - bare scalar / array / assoc
5838 // - subscripted `a[N]` / `h[key]`
5839 // - positional params (any digit-only name)
5840 // - env-var fallback (`HOME` set via getsparam → lookup_special_var)
5841 vm.register_builtin(BUILTIN_VAR_EXISTS, |vm, _argc| {
5842 let name = vm.pop().to_str();
5843 // c:Src/cond.c:361 `case 'v': return !issetvar(left)`. `-v` is
5844 // NOT `${+name}` — issetvar (params.c:751) additionally rejects
5845 // trailing chars after the parsed name/subscript (`arr[3]extra`,
5846 // nested `arr[2][1]`) and validates array-slice bounds (an
5847 // out-of-range `(i)`-not-found index is "unset"). `${+}` is
5848 // lenient and reported those as set.
5849 Value::Bool(crate::ported::params::issetvar(&name) != 0)
5850 });
5851
5852 // `time { compound; ... }` — runs the sub-chunk and prints elapsed
5853 // wall-clock time. zsh's full `time` also tracks user/system CPU via
5854 // getrusage on the *child*; we approximate via wall-time only since
5855 // the sub-chunk runs in-process (no fork). Output format matches
5856 // `time simple-cmd` (already implemented elsewhere via exectime).
5857 vm.register_builtin(BUILTIN_TIME_SUBLIST, |vm, argc| {
5858 let sub_idx = vm.pop().to_int() as usize;
5859 // c:Src/jobs.c:1028-1029 — `pn->text` arg to printtime. argc==2
5860 // means the compiler also pushed a desc string (bug #66 fix);
5861 // older callers with argc==1 push only sub_idx and we synthesize
5862 // an empty desc for backward compat with cached bytecode that
5863 // predates the desc-threading patch.
5864 let desc = if argc >= 2 {
5865 vm.pop().to_str().to_string()
5866 } else {
5867 String::new()
5868 };
5869 let chunk_opt = vm.chunk.sub_chunks.get(sub_idx).cloned();
5870 let Some(chunk) = chunk_opt else {
5871 return Value::Status(0);
5872 };
5873 // c:Src/jobs.c:1968 — `getrusage(RUSAGE_CHILDREN, &ti)` before
5874 // and after the timed sublist gives accurate per-stage user/sys
5875 // CPU. Wall-time-only approximation (0.7×/0.1× fudge factors)
5876 // produced bogus user/sys columns and ignored TIMEFMT. Bug #66
5877 // in docs/BUGS.md.
5878 let ru_before: libc::rusage = unsafe {
5879 let mut r: libc::rusage = std::mem::zeroed();
5880 libc::getrusage(libc::RUSAGE_CHILDREN, &mut r);
5881 r
5882 };
5883 // c:Src/jobs.c — zsh's `time` reports only for JOBS (forked
5884 // work). Builtins/brace-groups/functions run in the shell
5885 // process with no job, so `zsh -fc 'time true'` emits NOTHING.
5886 // Snapshot the fork-event counter; report only if the timed
5887 // body forked (external command or subshell).
5888 let forks_before = crate::vm_helper::FORK_EVENTS.load(std::sync::atomic::Ordering::Relaxed);
5889 let start = Instant::now();
5890 crate::fusevm_disasm::maybe_print_stdout("time_sublist", &chunk);
5891 let mut sub_vm = fusevm::VM::new(chunk);
5892 register_builtins(&mut sub_vm);
5893 let _ = sub_vm.run();
5894 let status = sub_vm.last_status;
5895 let elapsed = start.elapsed();
5896 let ru_after: libc::rusage = unsafe {
5897 let mut r: libc::rusage = std::mem::zeroed();
5898 libc::getrusage(libc::RUSAGE_CHILDREN, &mut r);
5899 r
5900 };
5901 // Delta children rusage = timed work's CPU.
5902 let mut delta = ru_after;
5903 let sub = |a: libc::timeval, b: libc::timeval| -> libc::timeval {
5904 let mut sec = a.tv_sec - b.tv_sec;
5905 let mut usec = a.tv_usec as i64 - b.tv_usec as i64;
5906 if usec < 0 {
5907 sec -= 1;
5908 usec += 1_000_000;
5909 }
5910 libc::timeval {
5911 tv_sec: sec,
5912 tv_usec: usec as libc::suseconds_t,
5913 }
5914 };
5915 delta.ru_utime = sub(ru_after.ru_utime, ru_before.ru_utime);
5916 delta.ru_stime = sub(ru_after.ru_stime, ru_before.ru_stime);
5917 let ti = crate::ported::zsh_h::timeinfo::from_rusage(&delta);
5918 // c:Src/jobs.c:808-809 — `s = getsparam("TIMEFMT"); s ||
5919 // DEFAULT_TIMEFMT`. Honor user-set TIMEFMT, fall back to the
5920 // canonical default.
5921 let fmt = crate::ported::params::getsparam("TIMEFMT")
5922 .unwrap_or_else(|| crate::ported::zsh_system_h::DEFAULT_TIMEFMT.to_string());
5923 // c:Src/jobs.c:768 `desc` arg — for the `time { sublist }` /
5924 // `time simple-cmd` keyword path, zsh passes the sublist's
5925 // source text (used by %J via printtime). The compiler now
5926 // threads the rendered source text through as the desc operand
5927 // (compile_zsh.rs Time arm, argc==2 form). Bug #66.
5928 // c:Src/jobs.c — no job, no report (see forks_before above).
5929 let forked = crate::vm_helper::FORK_EVENTS.load(std::sync::atomic::Ordering::Relaxed)
5930 != forks_before;
5931 if forked {
5932 let line = crate::ported::jobs::printtime(elapsed.as_secs_f64(), &ti, &fmt, &desc);
5933 eprintln!("{}", line);
5934 }
5935 Value::Status(status)
5936 });
5937
5938 // `{name}>file` / `{name}<file` / `{name}>>file` — named-fd allocator.
5939 // Stack: [path, varid, op_byte]. Opens path with the appropriate mode
5940 // and stores the resulting fd number in $varid as a string. We use
5941 // a high starting fd (10+) by allocating then dup'ing — matches zsh's
5942 // "fresh fd >= 10" promise so subsequent commands don't collide on
5943 // stdin/out/err.
5944 vm.register_builtin(BUILTIN_OPEN_NAMED_FD, |vm, _argc| {
5945 use std::sync::atomic::Ordering;
5946 let op_byte = vm.pop().to_int() as u8;
5947 let varid = vm.pop().to_str();
5948 let path = vm.pop().to_str();
5949 // Param introspection used by both the open and close forms.
5950 let param_flags = crate::ported::params::paramtab()
5951 .read()
5952 .ok()
5953 .and_then(|t| t.get(&varid).map(|p| p.node.flags));
5954 let param_readonly = param_flags
5955 .map(|f| (f & crate::ported::zsh_h::PM_READONLY as i32) != 0)
5956 .unwrap_or(false);
5957 // `{varid}>&-` / `{varid}<&-` — REDIR_CLOSE with varid.
5958 // Direct port of Src/exec.c:3805-3850.
5959 if matches!(
5960 op_byte,
5961 b if b == fusevm::op::redirect_op::DUP_WRITE
5962 || b == fusevm::op::redirect_op::DUP_READ
5963 ) {
5964 let n = path.trim_start_matches('&');
5965 if n == "-" {
5966 let val = with_executor(|exec| exec.scalar(&varid)).unwrap_or_default();
5967 let fd1 = val.parse::<i32>();
5968 // c:3811-3816 — bad=1: parameter doesn't contain an fd.
5969 let Ok(fd1) = fd1 else {
5970 crate::ported::utils::zwarn(&format!(
5971 "parameter {} does not contain a file descriptor",
5972 varid
5973 ));
5974 with_executor(|exec| exec.redirect_failed = true);
5975 return Value::Status(1);
5976 };
5977 // c:3813-3814 — bad=2: readonly parameter.
5978 if param_readonly {
5979 crate::ported::utils::zwarn(&format!(
5980 "can't close file descriptor from readonly parameter {}",
5981 varid
5982 ));
5983 with_executor(|exec| exec.redirect_failed = true);
5984 return Value::Status(1);
5985 }
5986 // c:3830-3835 — bad=3: fd >= 10 marked FDT_INTERNAL.
5987 if fd1 >= 10
5988 && fd1 <= crate::ported::utils::MAX_ZSH_FD.load(Ordering::Relaxed)
5989 && crate::ported::utils::fdtable_get(fd1) == crate::ported::zsh_h::FDT_INTERNAL
5990 {
5991 crate::ported::utils::zwarn(&format!(
5992 "file descriptor {} used by shell, not closed",
5993 fd1
5994 ));
5995 with_executor(|exec| exec.redirect_failed = true);
5996 return Value::Status(1);
5997 }
5998 // c:3870-3873 — close; report failure (varid form
5999 // always reports, unlike bare `N>&-`).
6000 if crate::ported::utils::zclose(fd1) < 0 {
6001 crate::ported::utils::zwarn(&format!(
6002 "failed to close file descriptor {}: {}",
6003 fd1,
6004 std::io::Error::last_os_error()
6005 ));
6006 return Value::Status(1);
6007 }
6008 return Value::Status(0);
6009 }
6010 // `{varid}>&N` — dup N to a fresh fd >= 10, store in varid.
6011 if let Ok(src) = n.parse::<i32>() {
6012 if param_readonly {
6013 crate::ported::utils::zwarn(&format!(
6014 "can't allocate file descriptor to readonly parameter {}",
6015 varid
6016 ));
6017 with_executor(|exec| exec.redirect_failed = true);
6018 return Value::Status(1);
6019 }
6020 let dup = unsafe { libc::fcntl(src, libc::F_DUPFD, 10) };
6021 if dup < 0 {
6022 crate::ported::utils::zwarn(&format!("{}: bad file descriptor", src));
6023 with_executor(|exec| exec.redirect_failed = true);
6024 return Value::Status(1);
6025 }
6026 // c:2404-2412 addfd varid arm — movefd + FDT_EXTERNAL.
6027 let final_fd = crate::ported::utils::movefd(dup);
6028 crate::ported::utils::fdtable_set(final_fd, crate::ported::zsh_h::FDT_EXTERNAL);
6029 with_executor(|exec| {
6030 exec.set_scalar(varid, final_fd.to_string());
6031 });
6032 return Value::Status(0);
6033 }
6034 return Value::Status(1);
6035 }
6036 // `{varid}<<HERE` / `{varid}<<<str` — op byte 255 (zshrs-side
6037 // contract with compile_redir; fusevm's redirect_op stops at
6038 // 8). C path: gethere/getherestr write the body to a temp
6039 // file (Src/exec.c:4660-4682), then addfd's varid arm moves
6040 // the read fd >= 10, marks FDT_EXTERNAL and sets the param
6041 // (c:2402-2412). `path` carries the BODY text here.
6042 if op_byte == 255 {
6043 if param_readonly {
6044 crate::ported::utils::zwarn(&format!(
6045 "can't allocate file descriptor to readonly parameter {}",
6046 varid
6047 ));
6048 with_executor(|exec| exec.redirect_failed = true);
6049 return Value::Status(1);
6050 }
6051 let body = format!("{}\n", path.trim_end_matches('\n'));
6052 let mut tmpl: Vec<u8> = b"/tmp/zshrs_hd_XXXXXX\0".to_vec();
6053 let write_fd = unsafe { libc::mkstemp(tmpl.as_mut_ptr() as *mut libc::c_char) };
6054 if write_fd < 0 {
6055 crate::ported::utils::zwarn(&format!(
6056 "can't create temp file for here document: {}",
6057 std::io::Error::last_os_error()
6058 ));
6059 return Value::Status(1);
6060 }
6061 let bytes = body.as_bytes();
6062 let mut off = 0;
6063 while off < bytes.len() {
6064 let n = unsafe {
6065 libc::write(
6066 write_fd,
6067 bytes[off..].as_ptr() as *const libc::c_void,
6068 bytes.len() - off,
6069 )
6070 };
6071 if n <= 0 {
6072 unsafe { libc::close(write_fd) };
6073 return Value::Status(1);
6074 }
6075 off += n as usize;
6076 }
6077 unsafe { libc::close(write_fd) };
6078 let read_fd =
6079 unsafe { libc::open(tmpl.as_ptr() as *const libc::c_char, libc::O_RDONLY) };
6080 unsafe { libc::unlink(tmpl.as_ptr() as *const libc::c_char) };
6081 if read_fd < 0 {
6082 return Value::Status(1);
6083 }
6084 let final_fd = crate::ported::utils::movefd(read_fd);
6085 if final_fd < 0 {
6086 return Value::Status(1);
6087 }
6088 crate::ported::utils::fdtable_set(final_fd, crate::ported::zsh_h::FDT_EXTERNAL);
6089 with_executor(|exec| {
6090 exec.set_scalar(varid, final_fd.to_string());
6091 });
6092 return Value::Status(0);
6093 }
6094 // Open form: `{varid}>file` etc.
6095 // c:Src/exec.c:2177-2215 checkclobberparam — gate BEFORE open.
6096 if param_readonly {
6097 // c:2191-2197
6098 crate::ported::utils::zwarn(&format!(
6099 "can't allocate file descriptor to readonly parameter {}",
6100 varid
6101 ));
6102 with_executor(|exec| exec.redirect_failed = true);
6103 return Value::Status(1);
6104 }
6105 // c:2199-2213 — NO_CLOBBER refuses to overwrite a parameter
6106 // already holding an OPEN fd (decimal value, fdtable says
6107 // FDT_EXTERNAL).
6108 if !isset(crate::ported::zsh_h::CLOBBER) && op_byte != fusevm::op::redirect_op::CLOBBER {
6109 if let Some(val) = with_executor(|exec| exec.scalar(&varid)) {
6110 if let Ok(fd) = val.parse::<i32>() {
6111 if fd >= 0
6112 && fd <= crate::ported::utils::MAX_ZSH_FD.load(Ordering::Relaxed)
6113 && crate::ported::utils::fdtable_get(fd)
6114 == crate::ported::zsh_h::FDT_EXTERNAL
6115 {
6116 crate::ported::utils::zwarn(&format!(
6117 "can't clobber parameter {} containing file descriptor {}",
6118 varid, fd
6119 ));
6120 with_executor(|exec| exec.redirect_failed = true);
6121 return Value::Status(1);
6122 }
6123 }
6124 }
6125 }
6126 let path_c = match CString::new(path.clone()) {
6127 Ok(c) => c,
6128 Err(_) => return Value::Status(1),
6129 };
6130 let flags = match op_byte {
6131 b if b == fusevm::op::redirect_op::READ => libc::O_RDONLY,
6132 b if b == fusevm::op::redirect_op::WRITE || b == fusevm::op::redirect_op::CLOBBER => {
6133 libc::O_WRONLY | libc::O_CREAT | libc::O_TRUNC
6134 }
6135 b if b == fusevm::op::redirect_op::APPEND => {
6136 libc::O_WRONLY | libc::O_CREAT | libc::O_APPEND
6137 }
6138 b if b == fusevm::op::redirect_op::READ_WRITE => libc::O_RDWR | libc::O_CREAT,
6139 _ => return Value::Status(1),
6140 };
6141 let fd = unsafe { libc::open(path_c.as_ptr(), flags, 0o666) };
6142 if fd < 0 {
6143 // c:Src/exec.c:3790-3795 — report the open failure and mark the
6144 // redirect failed so the command is SKIPPED, matching the numeric-fd
6145 // (`3< file`) and non-varid (`< file`) paths. Previously this
6146 // silently returned Status(1) with no diagnostic and no
6147 // redirect_failed flag, so `{fd}< /nonexistent` ran the command
6148 // anyway with exit 0 (zsh errors "no such file or directory" +
6149 // skips the command). `%e: %s` = strerror(errno) : filename.
6150 let e = std::io::Error::last_os_error();
6151 let msg = redir_errno_msg(&e);
6152 crate::ported::utils::zwarn(&format!("{}: {}", msg, path));
6153 with_executor(|exec| exec.redirect_failed = true);
6154 return Value::Status(1);
6155 }
6156 // c:2404-2412 addfd varid arm — `fd1 = movefd(fd2);
6157 // fdtable[fd1] = FDT_EXTERNAL; setiparam(varid, fd1);`.
6158 // FDT_EXTERNAL (not INTERNAL): the user owns this fd — the
6159 // NO_CLOBBER gate above and `{fd}>&-` close both key off it.
6160 let final_fd = crate::ported::utils::movefd(fd);
6161 if final_fd < 0 {
6162 crate::ported::utils::zerr(&format!(
6163 "cannot move fd {}: {}",
6164 fd,
6165 std::io::Error::last_os_error()
6166 ));
6167 return Value::Status(1);
6168 }
6169 crate::ported::utils::fdtable_set(final_fd, crate::ported::zsh_h::FDT_EXTERNAL);
6170 let _ = Ordering::Relaxed;
6171 with_executor(|exec| {
6172 exec.set_scalar(varid, final_fd.to_string());
6173 });
6174 Value::Status(0)
6175 });
6176
6177 // BUILTIN_SET_TRY_BLOCK_ERROR — capture the try-block's exit
6178 // status into `__zshrs_try_block_saved_status` (a scratch
6179 // scalar) so the always-arm can later restore it. Also set
6180 // `TRY_BLOCK_ERROR` per zsh semantics: it stays at -1 unless
6181 // the try-block fired an explicit error (errflag), per
6182 // c:Src/exec.c execlist's WC_TRYBLOCK arm.
6183 vm.register_builtin(BUILTIN_SET_TRY_BLOCK_ERROR, |vm, _argc| {
6184 use std::sync::atomic::Ordering;
6185 let vm_status = vm.last_status;
6186 // c:Src/exec.c WC_TRYBLOCK — the always-arm runs with a
6187 // clean escape state. Snapshot RETFLAG / BREAKS / CONTFLAG /
6188 // EXIT_PENDING here and clear them; RESTORE_TRY_BLOCK_STATUS
6189 // re-applies them at always-arm exit so the propagation jump
6190 // emitted by compile_zsh fires correctly.
6191 let ret_save = crate::ported::builtin::RETFLAG.swap(0, Ordering::Relaxed); // c:769-770
6192 let brk_save = crate::ported::builtin::BREAKS.swap(0, Ordering::Relaxed); // c:771-772
6193 let cont_save = crate::ported::builtin::CONTFLAG.swap(0, Ordering::Relaxed); // c:773-774
6194 let exit_save = crate::ported::builtin::EXIT_PENDING.swap(0, Ordering::Relaxed);
6195 // c:Src/loop.c:762-763 — `save_try_errflag = try_errflag;
6196 // save_try_interrupt = try_interrupt;`. Restored at c:778-779
6197 // by RESTORE_TRY_BLOCK_STATUS so a nested try block doesn't
6198 // clobber the enclosing one's `$TRY_BLOCK_ERROR`.
6199 let try_err_save = crate::ported::r#loop::try_errflag.load(Ordering::Relaxed); // c:762
6200 let try_int_save = crate::ported::r#loop::try_interrupt.load(Ordering::Relaxed); // c:763
6201 TRY_ESCAPE_SAVE.with(|s| {
6202 s.borrow_mut().push((
6203 ret_save,
6204 brk_save,
6205 cont_save,
6206 exit_save,
6207 try_err_save,
6208 try_int_save,
6209 ));
6210 });
6211 // c:Src/loop.c:764-766 — `try_errflag = (zlong)(errflag &
6212 // ERRFLAG_ERROR); try_interrupt = (zlong)((errflag &
6213 // ERRFLAG_INT) ? 1 : 0);`. Both are the RAW FLAG BITS, not the
6214 // try-list's exit status: ERRFLAG_ERROR is 1 (zsh.h:2972), so
6215 // `$TRY_BLOCK_ERROR` is 1-or-0 in zsh regardless of what the
6216 // failing command's `$?` was. The try-list's status is carried
6217 // separately in `__zshrs_try_block_saved_status`.
6218 let live_errflag = crate::ported::utils::errflag.load(Ordering::Relaxed);
6219 let try_err = (live_errflag & crate::ported::zsh_h::ERRFLAG_ERROR) as i64; // c:765
6220 let try_int = if (live_errflag & crate::ported::zsh_h::ERRFLAG_INT) != 0 {
6221 1i64
6222 } else {
6223 0i64
6224 }; // c:766
6225 crate::ported::r#loop::try_errflag.store(try_err, Ordering::Relaxed); // c:765
6226 crate::ported::r#loop::try_interrupt.store(try_int, Ordering::Relaxed); // c:766
6227 // c:Src/loop.c:755 — `endval = lastval ? lastval : errflag;`.
6228 // The status of the WHOLE `{…} always {…}` construct, captured
6229 // BEFORE the always-list runs (exectry returns it at c:801) and
6230 // deliberately including the errflag fallback: a try-list that
6231 // failed with `lastval == 0` but raised errflag still reports 1.
6232 let endval = if vm_status != 0 {
6233 vm_status
6234 } else {
6235 live_errflag
6236 }; // c:755
6237 with_executor(|exec| {
6238 // flags=0 (not setsparam's ASSPM_WARN): VM-internal scratch —
6239 // must never surface as a WARN_CREATE_GLOBAL diagnostic inside
6240 // a user function running `{...} always {...}` (f-sy-h's
6241 // `_zsh_highlight` does exactly that under warncreateglobal).
6242 crate::ported::params::assignsparam(
6243 "__zshrs_try_block_saved_status",
6244 &endval.to_string(),
6245 0,
6246 );
6247 let _ = exec;
6248 // Mirror into paramtab so `${parameters[TRY_BLOCK_ERROR]}`
6249 // and the PM_INTEGER `u_val` shadow agree with the atomic
6250 // the special-var getter reads. (setsparam → intsetfn's
6251 // TRY_BLOCK_ERROR arm re-stores the same value.)
6252 exec.set_scalar("TRY_BLOCK_ERROR".to_string(), try_err.to_string());
6253 exec.set_scalar("TRY_BLOCK_INTERRUPT".to_string(), try_int.to_string());
6254 });
6255 // c:Src/loop.c:768 — `errflag = 0;` ("We need to reset all
6256 // errors to allow the block to execute"). C clears the WHOLE
6257 // word, not just ERRFLAG_ERROR.
6258 crate::ported::utils::errflag.store(0, Ordering::Relaxed); // c:768
6259 Value::Status(0)
6260 });
6261
6262 // BUILTIN_BEGIN_INLINE_ENV / END_INLINE_ENV — wrap an
6263 // inline-assignment-prefixed command (`X=foo Y=bar cmd`):
6264 // BEGIN pushes a save frame; SET_VAR fires for each assign and
6265 // ALSO env::set_var's the value (visible to cmd's child); the
6266 // command runs; END pops the frame and restores both shell-var
6267 // and process-env state. Direct port of zsh's addvars() →
6268 // execute_simple → restore-after-exec contract.
6269 vm.register_builtin(BUILTIN_BEGIN_INLINE_ENV, |_vm, _argc| {
6270 with_executor(|exec| {
6271 exec.inline_env_stack.push(Vec::new());
6272 });
6273 Value::Status(0)
6274 });
6275 vm.register_builtin(BUILTIN_END_INLINE_ENV, |_vm, _argc| {
6276 with_executor(|exec| {
6277 if let Some(frame) = exec.inline_env_stack.pop() {
6278 for (name, prev_var, prev_env) in frame.into_iter().rev() {
6279 match prev_var {
6280 Some(v) => {
6281 exec.set_scalar(name.clone(), v);
6282 }
6283 None => {
6284 exec.unset_scalar(&name);
6285 }
6286 }
6287 match prev_env {
6288 Some(v) => env::set_var(&name, &v),
6289 None => env::remove_var(&name),
6290 }
6291 }
6292 }
6293 });
6294 Value::Status(0)
6295 });
6296 // c:Src/exec.c:3969-3976 — bare-exec assignment epilogue: see the
6297 // const's doc block. POSIX_BUILTINS → assignments persist (pop the
6298 // frame, discard the saved state); otherwise → restore_params
6299 // (same walk as END_INLINE_ENV).
6300 vm.register_builtin(BUILTIN_EXEC_INLINE_ENV_DONE, |_vm, _argc| {
6301 let persist = isset(crate::ported::zsh_h::POSIXBUILTINS);
6302 with_executor(|exec| {
6303 if let Some(frame) = exec.inline_env_stack.pop() {
6304 if persist {
6305 return; // c:3971 — no save/restore under POSIX_BUILTINS
6306 }
6307 for (name, prev_var, prev_env) in frame.into_iter().rev() {
6308 match prev_var {
6309 Some(v) => {
6310 exec.set_scalar(name.clone(), v);
6311 }
6312 None => {
6313 exec.unset_scalar(&name);
6314 }
6315 }
6316 match prev_env {
6317 Some(v) => env::set_var(&name, &v),
6318 None => env::remove_var(&name),
6319 }
6320 }
6321 }
6322 });
6323 Value::Status(0)
6324 });
6325
6326 // BUILTIN_RESTORE_TRY_BLOCK_STATUS — emitted at the end of an
6327 // `always` arm. Per zshmisc, the exit status of the entire
6328 // `{ try } always { finally }` construct is the try-list's
6329 // status, regardless of what happens in the always-list (the
6330 // exception is `return`/`exit` inside always, which short-
6331 // circuits and the cleanup is the only thing that runs). So
6332 // restore TRY_BLOCK_ERROR unconditionally — the always-list's
6333 // exit status is discarded for the construct.
6334 vm.register_builtin(BUILTIN_RESTORE_TRY_BLOCK_STATUS, |_vm, _argc| {
6335 use std::sync::atomic::Ordering;
6336 // c:Src/loop.c:801 — `return endval;`. The construct's exit
6337 // status is the try-list's (captured at c:755 by
6338 // SET_TRY_BLOCK_ERROR), never the always-list's.
6339 let saved = with_executor(|exec| {
6340 exec.scalar("__zshrs_try_block_saved_status")
6341 .and_then(|s| s.parse::<i32>().ok())
6342 .unwrap_or(0)
6343 });
6344 // c:Src/exec.c:1375 — `lastval = lv;` on the exectry return.
6345 // The always-list's own commands left their status in LASTVAL
6346 // (`always { : }` → 0); without this store the errflag re-raise
6347 // below aborts the shell with the always-list's 0 instead of
6348 // the try-list's failure status.
6349 crate::ported::builtin::LASTVAL.store(saved, Ordering::Relaxed); // c:1375
6350 // c:Src/loop.c:774-777 — the error RE-RAISE. This is the
6351 // whole point of TRY_BLOCK_ERROR being writable:
6352 //
6353 // if (try_errflag) errflag |= ERRFLAG_ERROR;
6354 // else errflag &= ~ERRFLAG_ERROR;
6355 // if (try_interrupt) errflag |= ERRFLAG_INT;
6356 // else errflag &= ~ERRFLAG_INT;
6357 //
6358 // SET_TRY_BLOCK_ERROR cleared errflag (c:768) so the always-arm
6359 // could run; the try-block's error is PARKED in `try_errflag`
6360 // and re-raised HERE unless the always-arm zeroed it
6361 // (`TRY_BLOCK_ERROR=0`, the documented swallow idiom — routed
6362 // to the atomic by intsetfn's IPDEF6 arm, params.rs).
6363 //
6364 // zshrs used to just drop the parked error, so
6365 // `f() { { typeset -r ro=1; ro=2 } always { … }; print reached }`
6366 // kept running and exited 0, where zsh aborts f with status 1.
6367 let te = crate::ported::r#loop::try_errflag.load(Ordering::Relaxed); // c:774
6368 let ti = crate::ported::r#loop::try_interrupt.load(Ordering::Relaxed); // c:776
6369 if te != 0 {
6370 crate::ported::utils::errflag
6371 .fetch_or(crate::ported::zsh_h::ERRFLAG_ERROR, Ordering::Relaxed); // c:775
6372 } else {
6373 crate::ported::utils::errflag
6374 .fetch_and(!crate::ported::zsh_h::ERRFLAG_ERROR, Ordering::Relaxed); // c:777
6375 }
6376 if ti != 0 {
6377 crate::ported::utils::errflag
6378 .fetch_or(crate::ported::zsh_h::ERRFLAG_INT, Ordering::Relaxed); // c:779
6379 } else {
6380 crate::ported::utils::errflag
6381 .fetch_and(!crate::ported::zsh_h::ERRFLAG_INT, Ordering::Relaxed); // c:781
6382 }
6383 // Re-apply the escape flags captured by SET_TRY_BLOCK_ERROR.
6384 // If the always-arm itself fired return/break/continue/exit,
6385 // its handler already overwrote the canonical atomics; let
6386 // those win — the always-arm's own escape always takes
6387 // priority over the try-block's deferred one.
6388 if let Some((ret, brk, cont, exit_p, try_err_save, try_int_save)) =
6389 TRY_ESCAPE_SAVE.with(|s| s.borrow_mut().pop())
6390 {
6391 // c:Src/loop.c:782-783 — `try_errflag = save_try_errflag;
6392 // try_interrupt = save_try_interrupt;`
6393 crate::ported::r#loop::try_errflag.store(try_err_save, Ordering::Relaxed); // c:782
6394 crate::ported::r#loop::try_interrupt.store(try_int_save, Ordering::Relaxed);
6395 // c:783
6396 if crate::ported::builtin::RETFLAG.load(Ordering::Relaxed) == 0 {
6397 crate::ported::builtin::RETFLAG.store(ret, Ordering::Relaxed);
6398 }
6399 if crate::ported::builtin::BREAKS.load(Ordering::Relaxed) == 0 {
6400 crate::ported::builtin::BREAKS.store(brk, Ordering::Relaxed);
6401 }
6402 if crate::ported::builtin::CONTFLAG.load(Ordering::Relaxed) == 0 {
6403 crate::ported::builtin::CONTFLAG.store(cont, Ordering::Relaxed);
6404 }
6405 if crate::ported::builtin::EXIT_PENDING.load(Ordering::Relaxed) == 0 {
6406 crate::ported::builtin::EXIT_PENDING.store(exit_p, Ordering::Relaxed);
6407 }
6408 }
6409 Value::Status(saved)
6410 });
6411
6412 // `[[ -r/-w/-x file ]]` — the cond path must use access(2) (the
6413 // C-faithful doaccess), NOT fusevm's generic Op::TestFile which only
6414 // checks existence for -r/-w (so a `chmod 000` file read as readable;
6415 // C02cond.ztst:13). Stack: [path, mode]; mode is the access(2) bit
6416 // (R_OK=4, W_OK=2, X_OK=1). Mirrors cond.rs:232/238/267 `doaccess`.
6417 vm.register_builtin(BUILTIN_COND_ACCESS, |vm, _argc| {
6418 let mode = vm.pop().to_int() as i32;
6419 let path = vm.pop().to_str();
6420 Value::Bool(crate::ported::cond::doaccess(&path, mode) != 0)
6421 });
6422
6423 // `[[ -prefix PAT ]]` / `-suffix` / `-after` / `-between` module condition.
6424 // Stack (pushed by the ModCond compile arm): arg0 … argN-1, then the
6425 // operator word last. argc = N+1.
6426 vm.register_builtin(BUILTIN_COND_MOD, |vm, argc| {
6427 use crate::ported::zle::complete::{
6428 cond_psfix, cond_range, CVT_PREPAT, CVT_SUFPAT,
6429 };
6430 let op = vm.pop().to_str(); // operator word (pushed last → popped first)
6431 let n = (argc as usize).saturating_sub(1);
6432 let mut args: Vec<String> = Vec::with_capacity(n);
6433 for _ in 0..n {
6434 args.push(vm.pop().to_str());
6435 }
6436 args.reverse(); // restore arg0 … argN-1 order
6437 // Dispatch the module/completion condition (C evalcond COND_MOD path:
6438 // condtab lookup + arity check, cond.c:149-185, over the four cotab[]
6439 // entries at complete.c:1697-1702). Handlers return 1=match/true.
6440 let name: String = op
6441 .trim_start_matches(|c: char| c == '-' || c == '\u{9b}')
6442 .to_string();
6443 let (min, max): (usize, usize) = match name.as_str() {
6444 "prefix" | "suffix" => (1, 2),
6445 "after" => (1, 1),
6446 "between" => (2, 2),
6447 _ => {
6448 crate::ported::utils::zerr(&format!(
6449 "unknown condition: {}",
6450 op.replace('\u{9b}', "-")
6451 ));
6452 return Value::Bool(false);
6453 }
6454 };
6455 if args.len() < min || args.len() > max {
6456 crate::ported::utils::zerr(&format!(
6457 "unknown condition: {}",
6458 op.replace('\u{9b}', "-")
6459 ));
6460 return Value::Bool(false);
6461 }
6462 let r = match name.as_str() {
6463 "prefix" => cond_psfix(&args, CVT_PREPAT),
6464 "suffix" => cond_psfix(&args, CVT_SUFPAT),
6465 "after" => cond_range(&args, 0),
6466 "between" => cond_range(&args, 1),
6467 _ => 0,
6468 };
6469 Value::Bool(r == 1)
6470 });
6471
6472 vm.register_builtin(BUILTIN_IS_TTY, |vm, _argc| {
6473 let fd_str = vm.pop().to_str();
6474 let fd: i32 = fd_str.trim().parse().unwrap_or(-1);
6475 let is_tty = if fd < 0 {
6476 false
6477 } else {
6478 unsafe { libc::isatty(fd) != 0 }
6479 };
6480 Value::Bool(is_tty)
6481 });
6482
6483 // c:Src/exec.c:4918/5040/5069 — a process substitution used inside a
6484 // `[[ … ]]` cond operand errors "process substitution %s cannot be
6485 // used here" (getoutputfile/getproc run with thisjob == -1). Emitted
6486 // by the compiler in place of ProcessSubIn/Out when in_cond_operand.
6487 vm.register_builtin(BUILTIN_PROCSUB_COND_ERROR, |_vm, _argc| {
6488 let cmd = _vm.pop().to_str();
6489 crate::ported::utils::zerr(&format!(
6490 "process substitution {} cannot be used here",
6491 cmd
6492 ));
6493 // c:getoutputfile returns NULL with errflag set → the enclosing
6494 // statement aborts (empty stdout, exit 1), rather than the cond
6495 // merely evaluating false.
6496 crate::ported::utils::errflag.fetch_or(
6497 crate::ported::zsh_h::ERRFLAG_ERROR,
6498 std::sync::atomic::Ordering::Relaxed,
6499 );
6500 with_executor(|exec| exec.set_last_status(1));
6501 _vm.last_status = 1;
6502 Value::str("")
6503 });
6504
6505 // Set $LINENO before executing the next statement. Direct
6506 // port of zsh's `lineno` global tracking from Src/input.c
6507 // (`if ((inbufflags & INP_LINENO) || !strin) && c == '\n')
6508 // lineno++;`). The compiler emits one of these before each
6509 // top-level pipe in `compile_sublist`, carrying the line
6510 // number captured by the parser at `ZshPipe.lineno`. Pops
6511 // [n], updates `$LINENO` in the variable table.
6512 vm.register_builtin(BUILTIN_SET_LINENO, |vm, _argc| {
6513 let n = vm.pop().to_int();
6514 // c:Src/exec.c:lineno = N — direct write to the param's
6515 // u_val. Cannot go through setsparam because LINENO carries
6516 // PM_READONLY (so `(t)LINENO` reads `integer-readonly-special`
6517 // per zsh); setsparam → assignstrvalue's PM_READONLY guard
6518 // would reject the internal write. C zsh handles this via the
6519 // PM_SPECIAL GSU vtable's setfn callback which bypasses the
6520 // generic readonly check; the Rust port writes the canonical
6521 // field directly instead.
6522 if let Ok(mut tab) = crate::ported::params::paramtab().write() {
6523 if let Some(pm) = tab.get_mut("LINENO") {
6524 pm.u_val = n;
6525 pm.node.flags &= !(crate::ported::zsh_h::PM_UNSET as i32);
6526 }
6527 }
6528 // Mirror to the file-static `lineno` (utils.c:121) that
6529 // zerrmsg reads at utils.c:301 for the `:N: msg` prefix.
6530 crate::ported::utils::set_lineno(n as i32);
6531 // Also drive lex::LEX_LINENO — zerrmsg (utils.rs:376) reads
6532 // THAT counter for the `name:N:` prefix. C zsh interleaves
6533 // parse and execute per top-level list, so its single
6534 // `lineno` global serves both; zshrs compiles the whole
6535 // script before running, leaving LEX_LINENO parked at EOF.
6536 // Without this write, every runtime zwarn/zerr reported the
6537 // script's LAST line instead of the failing statement's.
6538 crate::ported::lex::set_lineno(n as u64);
6539 // DAP hook — checks breakpoints / step mode / pause-request
6540 // for the line we just landed on. O(1) no-op when DAP is off
6541 // (single atomic load on a OnceLock). Inside `--dap` mode
6542 // this is the call that blocks the executor on a Condvar
6543 // until the IDE sends `continue`. Mirrors strykelang's
6544 // `debugger.should_stop(line) → debugger.prompt(...)` flow.
6545 crate::extensions::dap::check_line(n as u32);
6546 Value::Status(0)
6547 });
6548
6549 // Direct port of Src/prompt.c:1623 cmdpush. Token is a `CS_*`
6550 // value (zsh.h:2775-2806) emitted by compile_zsh around each
6551 // compound command (if/while/[[…]]/((…))/$(…)) and consumed by
6552 // `%_` in PS4 / prompt expansion.
6553 vm.register_builtin(BUILTIN_CMD_PUSH, |vm, _argc| {
6554 let token = vm.pop().to_int() as u8;
6555 // Route through canonical cmdpush (Src/prompt.c:1623). The
6556 // prompt expander reads from the file-static `CMDSTACK` at
6557 // `prompt.rs:2006`, not `exec.cmd_stack` — without this,
6558 // `%_` in PS4 saw an empty stack during xtrace.
6559 if (token as i32) < crate::ported::zsh_h::CS_COUNT {
6560 crate::ported::prompt::cmdpush(token);
6561 }
6562 Value::Status(0)
6563 });
6564
6565 // Direct port of Src/prompt.c:1631 cmdpop.
6566 vm.register_builtin(BUILTIN_CMD_POP, |_vm, _argc| {
6567 crate::ported::prompt::cmdpop();
6568 Value::Status(0)
6569 });
6570
6571 vm.register_builtin(BUILTIN_OPTION_SET, |vm, _argc| {
6572 let name = vm.pop().to_str();
6573 // Direct port of `optison(char *name, char *s)` at Src/cond.c:502 — `[[ -o NAME ]]`
6574 // reads through the same `opts[]` array that `setopt NAME`
6575 // writes via `dosetopt`. Earlier code read a duplicate Executor
6576 // HashMap which never saw `bin_setopt`'s writes (those land in
6577 // `OPTS_LIVE` via `opt_state_set`). Routing through the canonical
6578 // C port restores the single-store invariant: one `opts[]`,
6579 // shared between setopt/unsetopt and `[[ -o ]]`.
6580 let r = crate::ported::cond::optison(None, &name); // c:cond.c:502 (fromtest=NULL for [[ -o ]])
6581 match r {
6582 0 => Value::Bool(true), // c:cond.c:520 set
6583 1 => Value::Bool(false), // c:cond.c:518/520 unset
6584 _ => {
6585 // c:cond.c:514 — unknown option. optison already emitted the
6586 // diagnostic (via zwarn, now that fromtest=NULL for
6587 // `[[ -o ]]`); re-printing here double-emitted for
6588 // `[[ ! -o bad ]]` / `[[ -o a || -o b ]]`.
6589 Value::Bool(false)
6590 }
6591 }
6592 });
6593 // Tri-state `-o` for compile_cond's direct status path. Returns
6594 // 0 / 1 / 3 as a Value::Int that compile_cond consumes via
6595 // Op::SetStatus. Mirrors zsh's `[[ -o invalid ]]` returning $?=3.
6596 vm.register_builtin(BUILTIN_OPTION_CHECK_TRISTATE, |vm, _argc| {
6597 let name = vm.pop().to_str();
6598 let r = crate::ported::cond::optison(None, &name); // c:cond.c:502 (fromtest=NULL for [[ -o ]])
6599 // optison itself prints the diagnostic via zwarnnam when r=3
6600 // and POSIXBUILTINS is unset (the canonical path). Don't
6601 // double-emit here. r is already 0/1/3.
6602 Value::Int(r as i64)
6603 });
6604
6605 // BUILTIN_PARAM_FILTER — `${var:#pat}` / `${var:|name}` etc.
6606 // PURE PASSTHRU: rebuild `${name:#pat}` and route to paramsubst.
6607 vm.register_builtin(BUILTIN_PARAM_FILTER, |vm, _argc| {
6608 let pattern = vm.pop().to_str();
6609 let name = vm.pop().to_str();
6610 let body = format!("${{{}:#{}}}", name, pattern);
6611 paramsubst_to_value(&body)
6612 });
6613
6614 // `a[i]=(elements)` / `a[i,j]=(elements)` / `a[i]=()`
6615 // — subscripted-array assign with array RHS. Stack pushed by
6616 // compile_assign as: [elem0, elem1, …, elemN-1, name, key].
6617 vm.register_builtin(BUILTIN_SET_SUBSCRIPT_RANGE, |vm, argc| {
6618 let n = argc as usize;
6619 let mut popped: Vec<Value> = Vec::with_capacity(n);
6620 for _ in 0..n {
6621 popped.push(vm.pop());
6622 }
6623 popped.reverse();
6624 if popped.len() < 3 {
6625 return Value::Status(1);
6626 }
6627 // c:Src/params.c:3511-3526 — trailing append flag. The ARRAY
6628 // subscript path (`a[N]+=(v)` / `a[lo,hi]+=(v)`) sets it so the
6629 // AUGMENT transform below collapses the range to an empty range
6630 // after the slice end and inserts ONLY the new value. The scalar
6631 // path pre-concats the old slice (ARRAY_INDEX+Concat) and passes
6632 // 0, so it keeps plain-replace semantics.
6633 let append = popped.pop().map_or(false, |v| v.to_str() == "1");
6634 let key = popped.pop().unwrap().to_str();
6635 let name = popped.pop().unwrap().to_str();
6636 let mut values: Vec<String> = Vec::new();
6637 for v in popped {
6638 match v {
6639 Value::Array(items) => {
6640 for it in items {
6641 values.push(it.to_str());
6642 }
6643 }
6644 other => values.push(other.to_str()),
6645 }
6646 }
6647 // Bash sparse-array tracking: a single-index `a[i]=v` that pads the
6648 // dense Vec past its old end leaves indices old_len..i as HOLES (not
6649 // real elements), so `${#a[@]}`/`${!a[@]}` skip them like bash. Only
6650 // in bash mode, only for a plain non-append single index (0-based
6651 // under ksharrays). Captured before the assign; applied after.
6652 let sparse_track: Option<(String, usize, usize)> = if crate::dash_mode::bash_mode()
6653 && !append
6654 && !key.contains(',')
6655 {
6656 key.trim().parse::<usize>().ok().map(|i| {
6657 let old_len = with_executor(|exec| exec.array(&name).map(|a| a.len()).unwrap_or(0));
6658 (name.clone(), old_len, i)
6659 })
6660 } else {
6661 None
6662 };
6663 // c:Src/params.c:3383-3389 — a subscripted ARRAY assignment to an
6664 // associative array is an error, whatever the subscript looks like:
6665 // if (v && PM_TYPE(v->pm->node.flags) == PM_HASHED) {
6666 // unqueue_signals();
6667 // zerr("%s: attempt to set slice of associative array",
6668 // v->pm->node.nam);
6669 // freearray(val);
6670 // errflag |= ERRFLAG_ERROR;
6671 // return NULL;
6672 // }
6673 // assignaparam (params.rs) ports this, but a single-key `h[k]=(1 2)`
6674 // never reaches it: the VM lowers subscripted assignment to this
6675 // builtin instead. The comma form `h[a,b]=(1 2)` DID error — it takes a
6676 // different route — so the gap looked like a subscript-parsing quirk
6677 // when it was really "the check lives on a path this form doesn't
6678 // take". Untreated, the assignment was SILENTLY DISCARDED: rc=0 and
6679 // `${h[k]}` still read its old value.
6680 //
6681 // Only array-valued assignment is rejected; `h[k]=x` is a scalar
6682 // element store and stays legal.
6683 {
6684 let is_hashed = crate::ported::params::paramtab()
6685 .read()
6686 .ok()
6687 .and_then(|t| {
6688 t.get(&name)
6689 .map(|pm| crate::ported::zsh_h::PM_TYPE(pm.node.flags as u32) == crate::ported::zsh_h::PM_HASHED)
6690 })
6691 .unwrap_or(false);
6692 if is_hashed {
6693 crate::ported::utils::zerr(&format!(
6694 "{name}: attempt to set slice of associative array" // c:3385
6695 ));
6696 crate::ported::utils::errflag.fetch_or(
6697 crate::ported::zsh_h::ERRFLAG_ERROR,
6698 std::sync::atomic::Ordering::Relaxed,
6699 ); // c:3387
6700 return Value::Status(1); // c:3388
6701 }
6702 }
6703
6704 with_executor(|exec| {
6705 // Parse subscript: slice `lo,hi` or single index `i`.
6706 // setarrvalue (Src/params.c:2895) expects 1-based start/
6707 // end inclusive where start==end means replace one
6708 // element. Negative bounds translate to len+n+1 (1-based).
6709 //
6710 // c:Src/params.c — the END side accepts 0 as a valid value
6711 // that signals "insert BEFORE start position" (the canonical
6712 // `a[N,N-1]=val` prepend / mid-insert idiom). Bug #275 in
6713 // docs/BUGS.md: the previous Rust port clamped end up to 1,
6714 // collapsing `a[1,0]=(X Y)` into `a[1,1]=(X Y)` which
6715 // OVERWRITES position 1 instead of prepending. Provide two
6716 // translators — start_translate clamps to 1 (1-based);
6717 // end_translate keeps 0 intact so the splice in
6718 // setarrvalue (start_idx=0..end_idx=0) inserts at the front.
6719 // Bug #589: for scalars (no array), use the scalar's char
6720 // count as `len` so negative-index translation (`a[2,-1]`)
6721 // computes against the actual string length, not 0.
6722 let len = exec
6723 .array(&name)
6724 .map(|a| a.len() as i64)
6725 .or_else(|| {
6726 crate::ported::params::paramtab().read().ok().and_then(|t| {
6727 t.get(&name).and_then(|pm| {
6728 if crate::ported::zsh_h::PM_TYPE(pm.node.flags as u32)
6729 == crate::ported::zsh_h::PM_SCALAR
6730 {
6731 pm.u_str.as_ref().map(|s| s.chars().count() as i64)
6732 } else {
6733 None
6734 }
6735 })
6736 })
6737 })
6738 .unwrap_or(0);
6739 let start_translate = |raw: i64| -> i32 {
6740 if raw < 0 {
6741 (len + raw + 1).max(1) as i32
6742 } else {
6743 raw.max(1) as i32
6744 }
6745 };
6746 let end_translate = |raw: i64| -> i32 {
6747 if raw < 0 {
6748 (len + raw + 1).max(0) as i32
6749 } else {
6750 raw.max(0) as i32
6751 }
6752 };
6753 // c:Src/params.c — KSH_ARRAYS option flips array subscripts
6754 // from 1-based to 0-based. setarrvalue expects 1-based
6755 // inclusive bounds, so under KSH_ARRAYS we shift positive
6756 // inputs by +1 before translation. Negative bounds left
6757 // alone (count from end). Sibling of #610/#611/#612.
6758 // Bug #613.
6759 let ksh_arrays = crate::ported::zsh_h::isset(crate::ported::zsh_h::KSHARRAYS);
6760 let ksh_shift = |raw: i64| -> i64 {
6761 if ksh_arrays && raw >= 0 {
6762 raw + 1
6763 } else {
6764 raw
6765 }
6766 };
6767 // c:Src/params.c getindex — subscript bounds are MATH
6768 // expressions, not bare integers: `a[(( ${#a}+1 ))]=(x)`,
6769 // `a[n+1]=(x)`. Plain `parse::<i64>()` returned 0 on any
6770 // arithmetic subscript, which the `i == 0` guard turned into
6771 // a silent no-op (computed-index append never landed). Parse
6772 // the literal fast-path first, then fall back to mathevali
6773 // (which handles `(( ))` grouping, var refs, and operators).
6774 let eval_bound = |s: &str| -> i64 {
6775 let t = s.trim();
6776 t.parse::<i64>()
6777 .ok()
6778 .or_else(|| crate::ported::math::mathevali(t).ok())
6779 .unwrap_or(0)
6780 };
6781 let (start, end) = if let Some((s_str, e_str)) = key.split_once(',') {
6782 let s = ksh_shift(eval_bound(s_str));
6783 let e = ksh_shift(eval_bound(e_str));
6784 (start_translate(s), end_translate(e))
6785 } else {
6786 let i = ksh_shift(eval_bound(&key));
6787 if i == 0 {
6788 return;
6789 }
6790 let n = start_translate(i);
6791 (n, n)
6792 };
6793 // c:Src/params.c:3518-3520 (assignaparam ASSPM_AUGMENT) — a
6794 // subscripted `+=` to an array does NOT prepend the old slice;
6795 // it collapses the range to an EMPTY range positioned right
6796 // AFTER the slice end (`v->start = v->end--`) and splices in
6797 // ONLY the new value: `a[2]+=(d)` on (a b c) → (a b d c);
6798 // `a[2,3]+=(x)` on (1 2 3 4) → (1 2 3 x 4). In setarrvalue's
6799 // 1-based convention here that means start = end+1 (so
6800 // start_idx == end_idx == end → splice arr[end..end]).
6801 let (start, end) = if append && end > 0 {
6802 (end + 1, end)
6803 } else {
6804 (start, end)
6805 };
6806 // c:Src/params.c:392-430 IPDEF9("argv"/"@"/"*", &pparams) —
6807 // the positional parameters live in the `pparams` vector, NOT
6808 // paramtab, so a subscript splice (`argv[2]=(X Y Z)`,
6809 // `2=(X Y Z)`) must read/write pparams. Splice a synthetic
6810 // array param holding the current positionals via the
6811 // canonical setarrvalue, then store the result back to
6812 // pparams — mirroring assignaparam's argv/@/* special-case
6813 // (params.rs:6937) for the whole-array form.
6814 if name == "argv" || name == "@" || name == "*" {
6815 let mut pm = {
6816 crate::ported::params::createparam(
6817 &name,
6818 crate::ported::zsh_h::PM_ARRAY as i32,
6819 );
6820 crate::ported::params::paramtab()
6821 .write()
6822 .ok()
6823 .and_then(|mut t| t.remove(&name))
6824 };
6825 if let Some(ref mut p) = pm {
6826 p.u_arr = Some(exec.pparams());
6827 }
6828 let mut v = crate::ported::zsh_h::value {
6829 pm,
6830 arr: Vec::new(),
6831 scanflags: 0,
6832 valflags: 0,
6833 start,
6834 end,
6835 };
6836 crate::ported::params::setarrvalue(&mut v, values);
6837 let result = v.pm.and_then(|p| p.u_arr).unwrap_or_default();
6838 exec.set_pparams(result);
6839 return;
6840 }
6841 // Route through canonical setarrvalue (Src/params.c:2895).
6842 // It handles PM_READONLY rejection, PM_HASHED slice-error,
6843 // PM_ARRAY splice + bounds clamp + padding (c:2980+).
6844 let taken = match crate::ported::params::paramtab().write() {
6845 Ok(mut tab) => tab.remove(&name),
6846 Err(_) => None,
6847 };
6848 // c:Src/exec.c:2640 / getvalue(…, 1) — a subscript assignment to
6849 // a NONEXISTENT parameter auto-creates it. getindex/fetchvalue
6850 // with the create flag calls createparam(name, PM_ARRAY) so the
6851 // splice has an array to write into; `unset u; u[1,2]=(a z)`
6852 // then yields the array (a z). Without this, setarrvalue saw
6853 // v.pm == None and silently stored nothing. The single-index
6854 // scalar-value path (SET_ASSOC/SET_ARRAY_AT) already vivifies;
6855 // this brings the range/array-value path to parity.
6856 let taken = taken.or_else(|| {
6857 crate::ported::params::createparam(&name, crate::ported::zsh_h::PM_ARRAY as i32);
6858 crate::ported::params::paramtab()
6859 .write()
6860 .ok()
6861 .and_then(|mut t| t.remove(&name))
6862 });
6863 // c:Src/params.c:2748+ — PM_SCALAR with subscript range
6864 // SPLICES the value into the scalar's char string. Bug
6865 // #589: zshrs's slice handler always called setarrvalue,
6866 // erroring "attempt to assign array value to non-array"
6867 // for `a=hello; a[2,3]=XYZ`. Detect PM_SCALAR and route
6868 // through assignstrvalue (which does scalar splice via
6869 // the PM_SCALAR arm at params.rs:3709-3789).
6870 let is_scalar = taken.as_ref().map_or(false, |pm| {
6871 crate::ported::zsh_h::PM_TYPE(pm.node.flags as u32)
6872 == crate::ported::zsh_h::PM_SCALAR
6873 });
6874 let mut v = crate::ported::zsh_h::value {
6875 pm: taken,
6876 arr: Vec::new(),
6877 scanflags: 0,
6878 valflags: 0,
6879 start,
6880 end,
6881 };
6882 if is_scalar {
6883 // Scalar splice — concat values, route through
6884 // assignstrvalue which dispatches by PM_TYPE.
6885 // start_translate returns 1-based positions; assignstrvalue's
6886 // PM_SCALAR arm at params.rs:3735+ expects 0-based start
6887 // (chars before start are kept) and 0-based end-exclusive
6888 // (chars from end are kept). Convert: start-=1.
6889 if v.start > 0 {
6890 v.start -= 1;
6891 }
6892 let val: String = values.join("");
6893 crate::ported::params::assignstrvalue(Some(&mut v), Some(val), 0);
6894 } else {
6895 crate::ported::params::setarrvalue(&mut v, values);
6896 }
6897 // Write the mutated Param back to paramtab — setarrvalue
6898 // mutated v.pm in-place; the prior `tab.remove(&name)` at
6899 // the top of this handler took ownership, so we re-insert
6900 // here. setarrvalue + this re-insert IS the canonical
6901 // store (Src/params.c:2895). No further mirror needed.
6902 if let Some(pm) = v.pm {
6903 if let Ok(mut tab) = crate::ported::params::paramtab().write() {
6904 tab.insert(name, pm);
6905 }
6906 }
6907 });
6908 if let Some((nm, old_len, i)) = sparse_track {
6909 crate::bash_arrays::note_subscript_set(&nm, old_len, i);
6910 }
6911 Value::Status(0)
6912 });
6913
6914 // BUILTIN_CONCAT_SPLICE — word-segment concat for an expansion whose
6915 // ARRAY shape survives into the word (`${arr[@]}`, `$@`, `${(@)a}`,
6916 // `${=v}`, slices). c:Src/subst.c:4245 `if (isarr)` gates the two
6917 // emit shapes and c:1663 `int plan9 = isset(RCEXPANDPARAM);` picks
6918 // between them at RUNTIME — so the option, not the compile-time
6919 // segment shape, decides splice-vs-cross-product here.
6920 vm.register_builtin(BUILTIN_CONCAT_SPLICE, |vm, _argc| {
6921 let rhs = vm.pop();
6922 let lhs = vm.pop();
6923 if plan9_active() {
6924 return concat_plan9(lhs, rhs);
6925 }
6926 concat_splice(lhs, rhs)
6927 });
6928
6929 // BUILTIN_CONCAT_DISTRIBUTE — word-segment concat. With
6930 // rcexpandparam (zsh option), distributes element-wise (cartesian
6931 // product). Default mode: joins arrays with IFS first char to a
6932 // single scalar before concat, matching zsh's default unquoted
6933 // and DQ semantics. Direct port of Src/subst.c sepjoin path
6934 // (line ~1813) which gates element-vs-join on the rc_expand_param
6935 // option, defaulting to join.
6936 // BUILTIN_CONCAT_DISTRIBUTE_FORCED — same shape as
6937 // CONCAT_DISTRIBUTE, but always cartesian-distributes when one
6938 // side is Array. Used for compile-time-detected explicit
6939 // distribution forms (`${^arr}` etc.) where the source flag
6940 // overrides the rcexpandparam option default.
6941 // `${^arr}` — RC_EXPAND_PARAM forced on by the flag. concat_plan9 carries
6942 // both halves of C's plan9 block: the c:4316-4350 cartesian emit AND the
6943 // c:4362 `uremnode` word deletion for an empty array. The DISTRIBUTE_FORCED
6944 // handler below cannot be reused: it is shared with `${(@)a}` / `${(f)v}` /
6945 // `${a[@]}`, which KEEP the word on empty (`x${(@)a}y` → `xy`).
6946 vm.register_builtin(BUILTIN_CONCAT_PLAN9, |vm, _argc| {
6947 let rhs = vm.pop();
6948 let lhs = vm.pop();
6949 concat_plan9(lhs, rhs)
6950 });
6951
6952 // `${^^arr}` — RC_EXPAND_PARAM forced OFF (c:2553-2555 `plan9 = 0`). Every
6953 // other concat builtin re-checks plan9_active(), which is the OPTION, so
6954 // under `setopt rcexpandparam` they cross-product regardless of the flag.
6955 // Go straight to concat_splice — C's non-plan9 path (c:4366-4437).
6956 vm.register_builtin(BUILTIN_CONCAT_SPLICE_NOPLAN9, |vm, _argc| {
6957 let rhs = vm.pop();
6958 let lhs = vm.pop();
6959 concat_splice(lhs, rhs)
6960 });
6961
6962 vm.register_builtin(BUILTIN_CONCAT_DISTRIBUTE_FORCED, |vm, _argc| {
6963 let rhs = vm.pop();
6964 let lhs = vm.pop();
6965 match (lhs, rhs) {
6966 (Value::Array(la), Value::Array(ra)) => {
6967 if ra.is_empty() {
6968 return Value::Array(la);
6969 }
6970 if la.is_empty() {
6971 return Value::Array(ra);
6972 }
6973 let mut out = Vec::with_capacity(la.len() * ra.len());
6974 for a in &la {
6975 let a_s = a.as_str_cow();
6976 for b in &ra {
6977 let b_s = b.as_str_cow();
6978 let mut s = String::with_capacity(a_s.len() + b_s.len());
6979 s.push_str(&a_s);
6980 s.push_str(&b_s);
6981 out.push(Value::str(s));
6982 }
6983 }
6984 Value::Array(out)
6985 }
6986 (Value::Array(la), rhs_scalar) => {
6987 // An EMPTY array contributes nothing to a concatenated
6988 // word — the surrounding scalar text survives. zsh:
6989 // `x${^a}y` (a=()) / `x${(P)scalar-empty}y` → "xy", NOT
6990 // a dropped word. Without this, a `(P)` indirect to an
6991 // unset/empty scalar (which nodes_to_value collapses to
6992 // Value::Array([]) for standalone-removal semantics)
6993 // cartesian-dropped the whole word — p10k's
6994 // `typeset -g _$2=${(P)2}` then arrived as a bare
6995 // `typeset` and dumped every parameter (~217× → 19 MB
6996 // terminal flood → startup hang).
6997 if la.is_empty() {
6998 return rhs_scalar;
6999 }
7000 let r = rhs_scalar.as_str_cow();
7001 let out: Vec<Value> = la
7002 .into_iter()
7003 .map(|a| {
7004 let a_s = a.as_str_cow();
7005 let mut s = String::with_capacity(a_s.len() + r.len());
7006 s.push_str(&a_s);
7007 s.push_str(&r);
7008 Value::str(s)
7009 })
7010 .collect();
7011 Value::Array(out)
7012 }
7013 (lhs_scalar, Value::Array(ra)) => {
7014 // Symmetric empty-array-contributes-nothing rule; see
7015 // the (Array, scalar) arm above.
7016 if ra.is_empty() {
7017 return lhs_scalar;
7018 }
7019 let l = lhs_scalar.as_str_cow();
7020 let out: Vec<Value> = ra
7021 .into_iter()
7022 .map(|b| {
7023 let b_s = b.as_str_cow();
7024 let mut s = String::with_capacity(l.len() + b_s.len());
7025 s.push_str(&l);
7026 s.push_str(&b_s);
7027 Value::str(s)
7028 })
7029 .collect();
7030 Value::Array(out)
7031 }
7032 (lhs_s, rhs_s) => {
7033 let l = lhs_s.as_str_cow();
7034 let r = rhs_s.as_str_cow();
7035 let mut s = String::with_capacity(l.len() + r.len());
7036 s.push_str(&l);
7037 s.push_str(&r);
7038 Value::str(s)
7039 }
7040 }
7041 });
7042
7043 vm.register_builtin(BUILTIN_CONCAT_DISTRIBUTE, |vm, argc| {
7044 let rhs = vm.pop();
7045 let lhs = vm.pop();
7046 // c:Src/subst.c:4245 `if (isarr)` — an unquoted array embedded
7047 // in a word ALWAYS emits one word per element, never a scalar
7048 // join. The shape (splice vs plan9 cross-product) is chosen at
7049 // RUNTIME by c:1663 `int plan9 = isset(RCEXPANDPARAM);`, exactly
7050 // as BUILTIN_CONCAT_SPLICE does. The only extra case DISTRIBUTE
7051 // handles is the DQ context: the compiler emits
7052 // CallBuiltin(BUILTIN_CONCAT_DISTRIBUTE, 1) when the parent word
7053 // is DQ-wrapped (compile_zsh.rs parent_is_dq), and inside DQ
7054 // `"pre${arr}post"` joins via $IFS[0] to a single scalar
7055 // regardless of the option (c:Src/subst.c:1650-1656 isarr
7056 // comment). The default UNQUOTED path emits argc=2 (lhs + rhs).
7057 // Bug #246 in docs/BUGS.md.
7058 if argc == 1 {
7059 // DQ context: join any Array side to scalar via sepjoin's
7060 // IFS default. c:Src/utils.c:3936-3945 — set-but-empty IFS
7061 // joins with "" (`IFS=""; echo "x$*y"` → `xabcy`); only
7062 // unset / space-leading IFS yields " ".
7063 let join_arr = |arr: Vec<Value>| -> String {
7064 let strs: Vec<String> = arr.iter().map(|v| v.as_str_cow().into_owned()).collect();
7065 crate::ported::utils::sepjoin(&strs, None)
7066 };
7067 let l = match lhs {
7068 Value::Array(a) => join_arr(a),
7069 other => other.as_str_cow().into_owned(),
7070 };
7071 let r = match rhs {
7072 Value::Array(a) => join_arr(a),
7073 other => other.as_str_cow().into_owned(),
7074 };
7075 let mut s = String::with_capacity(l.len() + r.len());
7076 s.push_str(&l);
7077 s.push_str(&r);
7078 return Value::str(s);
7079 }
7080 // Unquoted plain `${arr}`: same runtime dispatch as
7081 // BUILTIN_CONCAT_SPLICE — c:4245 `if (isarr)` always distributes
7082 // one word per element; c:1663 picks splice (default, first/last
7083 // sticking, c:4366-4437) vs plan9 cross-product (c:4316-4365).
7084 // concat_splice / concat_plan9 both honor EMPTY_EXPANSION_IS_SCALAR
7085 // so the p10k `${(P)2}` empty-array word-removal semantics survive.
7086 if plan9_active() {
7087 return concat_plan9(lhs, rhs);
7088 }
7089 concat_splice(lhs, rhs)
7090 });
7091
7092 // `[[ a -ef b ]]` — same-inode test. Resolves both paths via fs::metadata
7093 // (follows symlinks the way zsh's -ef does) and compares (dev, inode).
7094 // Returns false on any I/O error (path missing, permission denied, etc.).
7095 vm.register_builtin(BUILTIN_SAME_FILE, |vm, _argc| {
7096 let b = vm.pop().to_str();
7097 let a = vm.pop().to_str();
7098 let same = match (fs::metadata(&a), fs::metadata(&b)) {
7099 (Ok(ma), Ok(mb)) => ma.dev() == mb.dev() && ma.ino() == mb.ino(),
7100 _ => false,
7101 };
7102 Value::Bool(same)
7103 });
7104
7105 // `[[ -c path ]]` — character device.
7106 vm.register_builtin(BUILTIN_IS_CHARDEV, |vm, _argc| {
7107 let path = vm.pop().to_str();
7108 let result = fs::metadata(&path)
7109 .map(|m| m.file_type().is_char_device())
7110 .unwrap_or(false);
7111 Value::Bool(result)
7112 });
7113 // `[[ -b path ]]` — block device.
7114 vm.register_builtin(BUILTIN_IS_BLOCKDEV, |vm, _argc| {
7115 let path = vm.pop().to_str();
7116 let result = fs::metadata(&path)
7117 .map(|m| m.file_type().is_block_device())
7118 .unwrap_or(false);
7119 Value::Bool(result)
7120 });
7121 // `[[ -p path ]]` — FIFO (named pipe).
7122 vm.register_builtin(BUILTIN_IS_FIFO, |vm, _argc| {
7123 let path = vm.pop().to_str();
7124 let result = fs::metadata(&path)
7125 .map(|m| m.file_type().is_fifo())
7126 .unwrap_or(false);
7127 Value::Bool(result)
7128 });
7129 // `[[ -S path ]]` — socket.
7130 vm.register_builtin(BUILTIN_IS_SOCKET, |vm, _argc| {
7131 let path = vm.pop().to_str();
7132 let result = fs::symlink_metadata(&path)
7133 .map(|m| m.file_type().is_socket())
7134 .unwrap_or(false);
7135 Value::Bool(result)
7136 });
7137
7138 // `[[ -k path ]]` / `-u` / `-g` — sticky / setuid / setgid bit.
7139 vm.register_builtin(BUILTIN_HAS_STICKY, |vm, _argc| {
7140 let path = vm.pop().to_str();
7141 let result = fs::metadata(&path)
7142 .map(|m| m.permissions().mode() & libc::S_ISVTX as u32 != 0)
7143 .unwrap_or(false);
7144 Value::Bool(result)
7145 });
7146 vm.register_builtin(BUILTIN_HAS_SETUID, |vm, _argc| {
7147 let path = vm.pop().to_str();
7148 let result = fs::metadata(&path)
7149 .map(|m| m.permissions().mode() & libc::S_ISUID as u32 != 0)
7150 .unwrap_or(false);
7151 Value::Bool(result)
7152 });
7153 vm.register_builtin(BUILTIN_HAS_SETGID, |vm, _argc| {
7154 let path = vm.pop().to_str();
7155 let result = fs::metadata(&path)
7156 .map(|m| m.permissions().mode() & libc::S_ISGID as u32 != 0)
7157 .unwrap_or(false);
7158 Value::Bool(result)
7159 });
7160 vm.register_builtin(BUILTIN_OWNED_BY_USER, |vm, _argc| {
7161 let path = vm.pop().to_str();
7162 let euid = unsafe { libc::geteuid() };
7163 let result = fs::metadata(&path)
7164 .map(|m| m.uid() == euid)
7165 .unwrap_or(false);
7166 Value::Bool(result)
7167 });
7168 vm.register_builtin(BUILTIN_OWNED_BY_GROUP, |vm, _argc| {
7169 let path = vm.pop().to_str();
7170 let egid = unsafe { libc::getegid() };
7171 let result = fs::metadata(&path)
7172 .map(|m| m.gid() == egid)
7173 .unwrap_or(false);
7174 Value::Bool(result)
7175 });
7176
7177 // `[[ -N path ]]` — file's access time is NOT newer than its
7178 // modification time (zsh man: "true if file exists and its
7179 // access time is not newer than its modification time"). Used
7180 // by zsh's mailbox-watching code. The semantic is `atime <=
7181 // mtime` (equivalent to `mtime >= atime`) — equal counts as
7182 // true, which a strict `mtime > atime` check missed for newly
7183 // created files where both stamps are identical.
7184 vm.register_builtin(BUILTIN_FILE_MODIFIED_SINCE_ACCESS, |vm, _argc| {
7185 let path = vm.pop().to_str();
7186 let result = fs::metadata(&path)
7187 .map(|m| m.atime() <= m.mtime())
7188 .unwrap_or(false);
7189 Value::Bool(result)
7190 });
7191
7192 // `[[ a -nt b ]]` — true if `a`'s mtime is strictly later than `b`'s.
7193 // BOTH files must exist; if either is missing the result is false.
7194 // (Earlier behavior was bash's "missing == infinitely-old"; zsh
7195 // strictly requires both files to exist.)
7196 vm.register_builtin(BUILTIN_FILE_NEWER, |vm, _argc| {
7197 let b = vm.pop().to_str();
7198 let a = vm.pop().to_str();
7199 // Use SystemTime modified() for nanosecond precision —
7200 // MetadataExt::mtime() returns seconds only, so two files
7201 // touched within the same second compared equal even when
7202 // 500ms apart. zsh tracks ns and uses `>=` for ties (touching
7203 // a then b in quick succession should still report b newer).
7204 let ta = fs::metadata(&a).and_then(|m| m.modified()).ok();
7205 let tb = fs::metadata(&b).and_then(|m| m.modified()).ok();
7206 let result = match (ta, tb) {
7207 (Some(ta), Some(tb)) => ta > tb,
7208 _ => false,
7209 };
7210 Value::Bool(result)
7211 });
7212
7213 // `[[ a -ot b ]]` — mirror of -nt. Same both-must-exist contract.
7214 vm.register_builtin(BUILTIN_FILE_OLDER, |vm, _argc| {
7215 let b = vm.pop().to_str();
7216 let a = vm.pop().to_str();
7217 let ta = fs::metadata(&a).and_then(|m| m.modified()).ok();
7218 let tb = fs::metadata(&b).and_then(|m| m.modified()).ok();
7219 let result = match (ta, tb) {
7220 (Some(ta), Some(tb)) => ta < tb,
7221 _ => false,
7222 };
7223 Value::Bool(result)
7224 });
7225
7226 // `set -e` / `setopt errexit` post-command check. Compiler emits
7227 // this after each top-level command's SetStatus (skipped inside
7228 // conditionals/pipelines/&&||/`!`). If errexit is on AND the last
7229 // command exited non-zero AND it's not a `return` from a function,
7230 // exit the shell with that status.
7231 // `set -x` / `setopt xtrace` — print each command before it runs.
7232 // The compiler emits this BEFORE the actual builtin/external call
7233 // with the command's literal text as a single string arg. We
7234 // print to stderr if xtrace is on. Honors `$PS4` (default `+ `).
7235 //
7236 // ── XTRACE flow control ────────────────────────────────────────
7237 // Mirror of C zsh's `doneps4` flag in execcmd_exec (Src/exec.c).
7238 // When an assignment trace fires (XTRACE_ASSIGN), it emits PS4
7239 // and sets this flag so the subsequent XTRACE_ARGS skips its own
7240 // PS4 emission — the assignment + command end up on the SAME
7241 // line: `<PS4>a=1 echo hello\n`. XTRACE_ARGS / XTRACE_NEWLINE
7242 // reset the flag after emitting the trailing `\n`.
7243 vm.register_builtin(BUILTIN_XTRACE_IS_ON, |_vm, _argc| {
7244 // Push live xtrace state. Caller pairs this with JumpIfFalse
7245 // to skip the trace-string-building block when xtrace is off,
7246 // avoiding side-effectful operand re-evaluation. Bug #159 in
7247 // docs/BUGS.md.
7248 let on = with_executor(|_| opt_state_get("xtrace").unwrap_or(false));
7249 Value::Int(if on { 1 } else { 0 })
7250 });
7251
7252 vm.register_builtin(BUILTIN_XTRACE_LINE, |vm, _argc| {
7253 let cmd_text = vm.pop().to_str();
7254 // Sync exec.last_status with the live vm.last_status BEFORE
7255 // the next command runs. Direct port of the zsh exec.c
7256 // contract — `$?` reads the exit status of the *most recent*
7257 // command. XTRACE_LINE is emitted by the compiler BEFORE
7258 // every simple command, so it's the natural sync point.
7259 let live = vm.last_status;
7260 with_executor(|exec| {
7261 exec.set_last_status(live);
7262 });
7263 // C zsh emits xtrace for `(( … ))` / `[[ … ]]` / `case` /
7264 // `if/while/until/for/repeat` head expressions via
7265 // `printprompt4(); fprintf(xtrerr, "%s\n", expr)` at
7266 // Src/exec.c:5240 (math), c:5286 (cond), c:4117 (for), etc.
7267 // The compiler emits BUILTIN_XTRACE_LINE only at those
7268 // construct boundaries (compile_arith / compile_cond /
7269 // compile_if / compile_while / compile_for / compile_case);
7270 // simple commands route to BUILTIN_XTRACE_ARGS instead. So
7271 // this handler always emits when xtrace is on — no prefix-
7272 // string heuristic.
7273 let on = with_executor(|exec| opt_state_get("xtrace").unwrap_or(false));
7274 if on {
7275 let already = XTRACE_DONE_PS4.with(|f| f.get());
7276 if !already {
7277 printprompt4();
7278 }
7279 // c:exec.c:5240/5286 — `fprintf(xtrerr, "%s\n", expr)`. Buffer
7280 // the line + newline, flush once (single write).
7281 xtrerr_fputs(&cmd_text);
7282 xtrerr_fputs("\n");
7283 xtrerr_flush();
7284 XTRACE_DONE_PS4.with(|f| f.set(false));
7285 }
7286 Value::Status(0)
7287 });
7288
7289 // BUILTIN_XTRACE_ARRAY_LINE — xtrace line for an `arr=(...)` / `arr+=(...)`
7290 // assignment. Stack on entry: [array, prefix] (argc = 2); pops prefix
7291 // ("name=( " / "name+=( "), then the whole assembled Value::Array. Direct
7292 // port of c:Src/exec.c::addvars:2624-2632, guarded on the live xtrace
7293 // state like C's `if (xtr)`: prints `prefix qz(e0) qz(e1) … ) ` with each
7294 // element shell-quoted (quotedzputs). Replaces the former one-VM-slot-per-
7295 // element trace, which overflowed next_slot (u16) on large literals.
7296 vm.register_builtin(BUILTIN_XTRACE_ARRAY_LINE, |vm, _argc| {
7297 let prefix = vm.pop().to_str();
7298 let arr = vm.pop();
7299 let live = vm.last_status;
7300 with_executor(|exec| {
7301 exec.set_last_status(live);
7302 });
7303 let on = with_executor(|_exec| opt_state_get("xtrace").unwrap_or(false));
7304 if on {
7305 let already = XTRACE_DONE_PS4.with(|f| f.get());
7306 if !already {
7307 printprompt4();
7308 }
7309 let mut line = String::with_capacity(prefix.len() + 16);
7310 line.push_str(&prefix);
7311 if let Value::Array(items) = arr {
7312 for it in items {
7313 line.push_str(&crate::ported::utils::quotedzputs(&it.to_str()));
7314 line.push(' ');
7315 }
7316 }
7317 line.push_str(") ");
7318 line.push('\n');
7319 xtrerr_fputs(&line);
7320 xtrerr_flush();
7321 XTRACE_DONE_PS4.with(|f| f.set(false));
7322 }
7323 Value::Status(0)
7324 });
7325
7326 // BUILTIN_MAKE_ARRAY_COUNTED — pop a count Int (top), then pop that many
7327 // values below it, and push them as one Value::Array (bottom-of-group
7328 // first). Same result as Op::MakeArray(N) but N comes from the stack as an
7329 // i64, dodging MakeArray's u16 operand cap. The compiler emits this only
7330 // when a literal `arr=(...)` has more than u16::MAX elements.
7331 vm.register_builtin(BUILTIN_MAKE_ARRAY_COUNTED, |vm, _argc| {
7332 let count = vm.pop().to_int().max(0) as usize;
7333 let mut items: Vec<Value> = Vec::with_capacity(count);
7334 for _ in 0..count {
7335 items.push(vm.pop());
7336 }
7337 items.reverse();
7338 Value::Array(items)
7339 });
7340
7341 // Like XTRACE_LINE but reads the top `argc - 1` values from the
7342 // VM stack WITHOUT consuming them (peek), then pops a prefix
7343 // string at the top. Joins prefix + peeked args with spaces using
7344 // zsh's quotedzputs-equivalent quoting. Direct port of
7345 // Src/exec.c:2055-2066 — emit AFTER expansion, with each arg
7346 // shell-quoted, so `for i in a b; echo for $i` traces as
7347 // `echo for a` / `echo for b`, not `echo for $i`.
7348 //
7349 // Stack contract on entry: [arg1, arg2, ..., argN, prefix].
7350 // Pops prefix; peeks argN..arg1 below. argc = N + 1.
7351 vm.register_builtin(BUILTIN_XTRACE_ARGS, |vm, argc| {
7352 let prefix = vm.pop().to_str();
7353 let live = vm.last_status;
7354 with_executor(|exec| {
7355 exec.set_last_status(live);
7356 });
7357 let on = with_executor(|exec| opt_state_get("xtrace").unwrap_or(false));
7358 if on {
7359 let n_args = argc.saturating_sub(1) as usize;
7360 let len = vm.stack.len();
7361 // c:Src/exec.c:2055 — argv is the POST-expansion word
7362 // list, so an arg that expanded to multiple words splats
7363 // into multiple trace tokens AND an arg that expanded to
7364 // zero words (empty unquoted `${UNSET}`) emits nothing.
7365 // pop_args (line 6243) already does this splat for the
7366 // real handler; mirror the same Array → splat / empty →
7367 // drop logic here so xtrace renders `echo ${UNSET}` as
7368 // `echo` (zsh) instead of `echo ''` (the previous
7369 // single-arg stringify path returned "" and then
7370 // quotedzputs wrapped it in `''`).
7371 let arg_strs: Vec<String> = if n_args > 0 && len >= n_args {
7372 let mut out = Vec::new();
7373 for v in &vm.stack[len - n_args..] {
7374 match v {
7375 Value::Array(items) => {
7376 for item in items {
7377 out.push(quotedzputs(&item.to_str()));
7378 }
7379 }
7380 other => out.push(quotedzputs(&other.to_str())),
7381 }
7382 }
7383 out
7384 } else {
7385 Vec::new()
7386 };
7387 // Builtins dispatch through `execbuiltin` (Src/builtin.c:442)
7388 // which emits its own PS4 + name + args xtrace. To avoid
7389 // double-emission, skip our emission here when the first
7390 // arg is a known builtin with a registered HandlerFunc —
7391 // those go through execbuiltin and will trace themselves.
7392 // Externals + builtins-not-yet-routed-through-execbuiltin
7393 // keep our emission as a stand-in.
7394 let goes_through_execbuiltin = crate::ported::builtin::BUILTINS
7395 .iter()
7396 .any(|b| b.node.nam == prefix && b.handlerfunc.is_some());
7397 if !goes_through_execbuiltin {
7398 let line = if arg_strs.is_empty() {
7399 prefix
7400 } else {
7401 format!("{} {}", prefix, arg_strs.join(" "))
7402 };
7403 // Mirrors Src/exec.c:2055 xtrace emission. C does:
7404 // if (!doneps4) printprompt4();
7405 // ... emit args + spaces ...
7406 // fputc('\n', xtrerr); fflush(xtrerr);
7407 // printprompt4 + the args + `\n` all land in the xtrerr
7408 // buffer; the single fflush below writes the whole line in
7409 // one syscall so concurrent pipeline stages never
7410 // interleave (c:makecline:2122-2123).
7411 let already_ps4 = XTRACE_DONE_PS4.with(|f| f.get());
7412 if !already_ps4 {
7413 printprompt4();
7414 }
7415 xtrerr_fputs(&line);
7416 xtrerr_fputs("\n"); // c:2122 fputc('\n', xtrerr)
7417 xtrerr_flush(); // c:2123 fflush(xtrerr)
7418 }
7419 XTRACE_DONE_PS4.with(|f| f.set(false));
7420 }
7421 Value::Status(0)
7422 });
7423
7424 // BUILTIN_XTRACE_ASSIGN — direct port of the per-assignment
7425 // trace block at Src/exec.c:2517-2582. C body excerpt:
7426 // xtr = isset(XTRACE);
7427 // if (xtr) { printprompt4(); doneps4 = 1; }
7428 // while (assign) {
7429 // if (xtr) fprintf(xtrerr, "%s+=" or "%s=", name);
7430 // ... eval value into `val` ...
7431 // if (xtr) { quotedzputs(val, xtrerr); fputc(' ', xtrerr); }
7432 // ...
7433 // }
7434 //
7435 // Stack on entry: [..., name, value]. PEEKS both (they're left
7436 // on stack for SET_VAR to pop). Emits `name=<quoted-val> ` with
7437 // no newline; trailing `\n` comes from XTRACE_ARGS (cmd path)
7438 // or XTRACE_NEWLINE (assignment-only path).
7439 vm.register_builtin(BUILTIN_XTRACE_ASSIGN, |vm, _argc| {
7440 let on = with_executor(|exec| opt_state_get("xtrace").unwrap_or(false));
7441 if on {
7442 // PEEK [..., name, value] — argc==2 by contract.
7443 let len = vm.stack.len();
7444 if len >= 2 {
7445 let name = vm.stack[len - 2].to_str();
7446 let value = vm.stack[len - 1].to_str();
7447 let already_ps4 = XTRACE_DONE_PS4.with(|f| f.get());
7448 if !already_ps4 {
7449 printprompt4();
7450 XTRACE_DONE_PS4.with(|f| f.set(true));
7451 }
7452 // C: `fprintf(xtrerr, "%s=", name)` then `quotedzputs
7453 // (val); fputc(' ', xtrerr);`. Append to the xtrerr buffer
7454 // (no newline / no flush — the line continues with the
7455 // command via XTRACE_ARGS, or ends at XTRACE_NEWLINE).
7456 xtrerr_fputs(&format!("{}={} ", name, quotedzputs(&value)));
7457 }
7458 }
7459 Value::Status(0)
7460 });
7461
7462 // BUILTIN_XTRACE_NEWLINE — emit trailing `\n` + flush iff a
7463 // prior XTRACE_ASSIGN this line already emitted PS4. Mirrors
7464 // C's `fputc('\n', xtrerr); fflush(xtrerr);` at exec.c:3398
7465 // (the assignment-only path through execcmd_exec).
7466 vm.register_builtin(BUILTIN_XTRACE_NEWLINE, |_vm, _argc| {
7467 let on = with_executor(|exec| opt_state_get("xtrace").unwrap_or(false));
7468 if on {
7469 let already_ps4 = XTRACE_DONE_PS4.with(|f| f.get());
7470 if already_ps4 {
7471 xtrerr_fputs("\n"); // c:3398 fputc('\n', xtrerr)
7472 xtrerr_flush(); // c:3398 fflush(xtrerr)
7473 XTRACE_DONE_PS4.with(|f| f.set(false));
7474 }
7475 }
7476 Value::Status(0)
7477 });
7478
7479 // c:Src/exec.c WC_TRYBLOCK — post-always re-jump probes. Each
7480 // returns 1 + consumes the atomic when the corresponding
7481 // escape flag is set; the try-block compile pairs each with
7482 // a JumpIfFalse + Jump → outer scope's return / break /
7483 // continue patches.
7484 vm.register_builtin(BUILTIN_RETFLAG_CHECK, |_vm, _argc| {
7485 use std::sync::atomic::Ordering;
7486 let r = crate::ported::builtin::RETFLAG.load(Ordering::Relaxed);
7487 if r != 0 {
7488 // Don't clear here — doshfunc owns the clear at c:6047
7489 // when the function unwinds. Leaving it set propagates
7490 // through nested `eval`/`source` callers correctly.
7491 Value::Int(1)
7492 } else {
7493 Value::Int(0)
7494 }
7495 });
7496 vm.register_builtin(BUILTIN_BREAKS_CHECK, |_vm, _argc| {
7497 use std::sync::atomic::Ordering;
7498 let b = crate::ported::builtin::BREAKS.load(Ordering::Relaxed);
7499 let c = crate::ported::builtin::CONTFLAG.load(Ordering::Relaxed);
7500 // `break` sets BREAKS but NOT CONTFLAG; `continue` sets both.
7501 // Filter out the continue path here so the two checks are
7502 // mutually exclusive.
7503 if b != 0 && c == 0 {
7504 // Consume BREAKS so the outer loop's break_patches
7505 // landing doesn't double-decrement.
7506 crate::ported::builtin::BREAKS.store(0, Ordering::Relaxed);
7507 Value::Int(1)
7508 } else {
7509 Value::Int(0)
7510 }
7511 });
7512 vm.register_builtin(BUILTIN_CONTFLAG_CHECK, |_vm, _argc| {
7513 use std::sync::atomic::Ordering;
7514 let c = crate::ported::builtin::CONTFLAG.load(Ordering::Relaxed);
7515 if c != 0 {
7516 crate::ported::builtin::CONTFLAG.store(0, Ordering::Relaxed);
7517 crate::ported::builtin::BREAKS.store(0, Ordering::Relaxed);
7518 Value::Int(1)
7519 } else {
7520 Value::Int(0)
7521 }
7522 });
7523 vm.register_builtin(BUILTIN_NOEXEC_CHECK, |_vm, _argc| {
7524 // c:Src/exec.c:1390 — `set -n` / `noexec` option: parse but
7525 // don't execute. Returns Int(1) when noexec is set so the
7526 // emit-side JumpIfTrue skips the statement body.
7527 if opt_state_get("noexec").unwrap_or(false) {
7528 return Value::Int(1);
7529 }
7530 // c:Src/exec.c:1390 — execlist's list-loop gate:
7531 // `while (wc_code(code) == WC_LIST && !breaks && !retflag
7532 // && !errflag)`
7533 // — once errflag is set, the NEXT sublist never starts, so
7534 // lastval survives untouched to the shell exit. Without this
7535 // prologue gate the follow-up statement RAN, its dispatch
7536 // saw errflag, returned 1, and SetStatus clobbered lastval —
7537 // `[[ x == [a- ]]; print rc=$?` exited 1 instead of zsh's 2
7538 // (the cond syntax error set lastval=2 per exec.c:5216-5221).
7539 if (crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed)
7540 & crate::ported::zsh_h::ERRFLAG_ERROR)
7541 != 0
7542 {
7543 return Value::Int(1);
7544 }
7545 Value::Int(0)
7546 });
7547 vm.register_builtin(BUILTIN_DONETRAP_RESET, |_vm, _argc| {
7548 // c:Src/exec.c:1455 — `donetrap = 0;` at sublist start.
7549 // Reset before each top-level statement so the next
7550 // sublist's ERREXIT_CHECK fires the ZERR trap on its FIRST
7551 // non-zero command. Carries the "already fired" state
7552 // across function-call returns within the SAME outer
7553 // sublist (per C semantics — donetrap is process-global).
7554 // Bug #303 in docs/BUGS.md.
7555 crate::ported::exec::DONETRAP.store(0, std::sync::atomic::Ordering::Relaxed);
7556 Value::Status(0)
7557 });
7558
7559 // `[[ -z X ]]` / `[[ -n X ]]` — pop one Value, route through
7560 // canonical `src/ported/cond.rs::evalcond` so the actual
7561 // empty/non-empty test reuses the C-port at `cond.rs:270-271`
7562 // (`'n' => !arg.is_empty()`, `'z' => arg.is_empty()`).
7563 //
7564 // The Array→args conversion lives at the bridge because cond.rs
7565 // expects `&[&str]` (C `cond_str` signature equivalent). For
7566 // `"${arr[@]}"` in DQ context the splice yields `Value::Array`
7567 // — an empty array still expands to one implicit empty word
7568 // (per zsh's "${arr[@]}" splat preserving at least one slot
7569 // in cond context), so:
7570 // - Array(0) → ["-z", ""] → evalcond → 0 (true)
7571 // - Array(1) → ["-z", word] → evalcond → 0/1
7572 // - Array(2+) → ["-z", w1, w2, ...] → evalcond → 2 (parse
7573 // error: too many ops)
7574 // → coerced to false
7575 // - Str(s) → ["-z", s] → evalcond → 0/1
7576 //
7577 // Bug #185 in docs/BUGS.md.
7578 fn run_cond_str_empty(v: Value, op: &str) -> Value {
7579 let words: Vec<String> = match v {
7580 Value::Array(arr) => arr.into_iter().map(|x| x.to_str()).collect(),
7581 Value::Str(s) => vec![s.to_string()],
7582 other => vec![other.to_str()],
7583 };
7584 let mut args: Vec<&str> = vec![op];
7585 if words.is_empty() {
7586 args.push("");
7587 } else {
7588 args.extend(words.iter().map(|s| s.as_str()));
7589 }
7590 let opts: std::collections::HashMap<String, bool> = std::collections::HashMap::new();
7591 let vars: std::collections::HashMap<String, String> = std::collections::HashMap::new();
7592 // c:Src/cond.c:62-66 — `evalcond` returns 0=true, 1=false,
7593 // 2=syntax-error. Coerce error to false (observable behavior
7594 // in zsh: `[[ -z a b ]]` errors and the test as a whole
7595 // returns non-zero).
7596 // `[[ ]]` dispatch — C's `evalcond(state, NULL)` calling convention.
7597 // `None` for from_test → mathevali integer-compare coercion path.
7598 let ret = crate::ported::cond::evalcond(&args, &opts, &vars, false, None);
7599 Value::Int(if ret == 0 { 1 } else { 0 })
7600 }
7601 vm.register_builtin(BUILTIN_COND_STR_EMPTY, |vm, _argc| {
7602 let v = vm.pop();
7603 run_cond_str_empty(v, "-z")
7604 });
7605 vm.register_builtin(BUILTIN_COND_STR_NONEMPTY, |vm, _argc| {
7606 let v = vm.pop();
7607 run_cond_str_empty(v, "-n")
7608 });
7609
7610 // `exec N<<<"str"` — herestring redirect to explicit fd, applied
7611 // permanently. Direct port of `Src/exec.c:4655 getherestr` +
7612 // `addfd(forked, save, mfds, fn->fd1, fil, 0, ...)` at c:3766-
7613 // 3780 for the nullexec=1 bare-exec-redir path. Bug #205 in
7614 // docs/BUGS.md.
7615 vm.register_builtin(BUILTIN_EXEC_HERESTR_FD, |vm, _argc| {
7616 let fd = vm.pop().to_int() as i32;
7617 let content = vm.pop().to_str();
7618 // c:4671-4672 — append `\n` for "real" herestrings (not
7619 // heredoc-derived). zshrs's bare-exec path only fires for
7620 // the `<<<` syntax (REDIR_HERESTR), so always append.
7621 let body = format!("{}\n", content);
7622 // c:4673-4679 — gettempfile → write_loop → close → reopen
7623 // read-only → unlink. Rust equivalent via tempfile crate or
7624 // explicit O_TMPFILE; use mkstemp + unlink-immediately to
7625 // mirror C exactly.
7626 use std::ffi::CString;
7627 let mut tmpl: Vec<u8> = b"/tmp/zshrs_hs_XXXXXX\0".to_vec();
7628 let write_fd = unsafe { libc::mkstemp(tmpl.as_mut_ptr() as *mut libc::c_char) };
7629 if write_fd < 0 {
7630 crate::ported::utils::zwarn(&format!(
7631 "can't create temp file for here document: {}",
7632 std::io::Error::last_os_error()
7633 ));
7634 return Value::Status(1);
7635 }
7636 // c:4675 — write_loop(fd, t, len)
7637 let bytes = body.as_bytes();
7638 let mut off = 0;
7639 while off < bytes.len() {
7640 let n = unsafe {
7641 libc::write(
7642 write_fd,
7643 bytes[off..].as_ptr() as *const libc::c_void,
7644 bytes.len() - off,
7645 )
7646 };
7647 if n <= 0 {
7648 unsafe { libc::close(write_fd) };
7649 return Value::Status(1);
7650 }
7651 off += n as usize;
7652 }
7653 unsafe { libc::close(write_fd) }; // c:4676
7654 // Path null-terminated by mkstemp; reopen for reading.
7655 let read_fd = unsafe { libc::open(tmpl.as_ptr() as *const libc::c_char, libc::O_RDONLY) };
7656 // c:4678 — unlink immediately so the file disappears on
7657 // close, leaving only the fd reference.
7658 unsafe { libc::unlink(tmpl.as_ptr() as *const libc::c_char) };
7659 if read_fd < 0 {
7660 return Value::Status(1);
7661 }
7662 // c:3779 addfd → dup2 to target fd, close intermediate.
7663 let r = unsafe { libc::dup2(read_fd, fd) };
7664 unsafe { libc::close(read_fd) };
7665 if r < 0 {
7666 return Value::Status(1);
7667 }
7668 Value::Status(0)
7669 });
7670 // c:Src/exec.c:2418 + addfd splice — MULTIOS fan-out. Stack
7671 // layout pushed by compile_zsh's coalescing pass:
7672 // [target_1, op_byte_1, target_2, op_byte_2, …, target_N,
7673 // op_byte_N, fd]
7674 // argc = 2N + 1. Pops, opens every target, sets up a pipe +
7675 // splitter thread that reads pipe → writes every chunk to
7676 // every opened target, dup2's pipe-write-end onto fd. The
7677 // splitter is closed + joined by host_redirect_scope_end.
7678 // Bug #36 in docs/BUGS.md.
7679 vm.register_builtin(BUILTIN_MULTIOS_REDIRECT, |vm, argc| {
7680 if argc < 3 || argc % 2 == 0 {
7681 // Bad shape — bail.
7682 return Value::Status(1);
7683 }
7684 // Pop fd first (top of stack).
7685 let fd = vm.pop().to_int() as i32;
7686 // Then pop (op, target) pairs in reverse compile order. Keep
7687 // targets as Values — a glob-bearing target arrives as a
7688 // Value::Array of matches.
7689 let n_targets = ((argc - 1) / 2) as usize;
7690 let mut pairs: Vec<(u8, Value)> = Vec::with_capacity(n_targets);
7691 for _ in 0..n_targets {
7692 let op_byte = vm.pop().to_int() as u8;
7693 let target = vm.pop();
7694 pairs.push((op_byte, target));
7695 }
7696 // Restore compile order (target_1 first).
7697 pairs.reverse();
7698
7699 // c:Src/glob.c:2195-2203 xpandredir — "Loop over matches,
7700 // duplicating the redirection for each file found": a glob
7701 // target with N matches becomes N members of the same multio
7702 // (`echo hi > *.txt` with two matches writes both files).
7703 let mut entries: Vec<(u8, String)> = Vec::with_capacity(pairs.len());
7704 for (op_byte, target) in pairs {
7705 match target {
7706 Value::Array(items) => {
7707 for item in items {
7708 entries.push((op_byte, item.to_str()));
7709 }
7710 }
7711 other => entries.push((op_byte, other.to_str())),
7712 }
7713 }
7714 if entries.is_empty() {
7715 return Value::Status(1);
7716 }
7717
7718 // c:Src/exec.c:2418 — `else if (!mfds[fd1] || unset(MULTIOS))`:
7719 // with MULTIOS unset every redirect takes the REPLACE path in
7720 // script order — each target is still opened (created /
7721 // truncated) and dup2'd over the fd, so the LAST one wins and
7722 // earlier files end up empty (`unsetopt multios; print x > a
7723 // > b` leaves `a` empty, `x` in `b`). host_apply_redirect is
7724 // exactly one replace step, noclobber gate included.
7725 let multios_on = opt_state_get("multios").unwrap_or(true);
7726 if !multios_on {
7727 with_executor(|exec| {
7728 for (op_byte, target) in &entries {
7729 exec.host_apply_redirect(fd as u8, *op_byte, target);
7730 if exec.redirect_failed {
7731 // c:Src/exec.c execerr — abort the remaining
7732 // redirect list on failure.
7733 break;
7734 }
7735 }
7736 });
7737 return Value::Status(0);
7738 }
7739
7740 if entries.len() == 1 {
7741 // Single member after splicing — a plain replace
7742 // (c:2418 new-multio arm). Route through
7743 // host_apply_redirect so the noclobber gate, the
7744 // pipeline-output split partial, and error handling all
7745 // apply exactly as for an un-bagged redirect.
7746 let (op_byte, target) = &entries[0];
7747 with_executor(|exec| {
7748 exec.host_apply_redirect(fd as u8, *op_byte, target);
7749 });
7750 return Value::Status(0);
7751 }
7752
7753 // c:Src/exec.c:3722-3724 — when this command's stdout IS the
7754 // pipeline output, C seeds mfds[1] with the pipe BEFORE
7755 // walking the redirect list, so the pipe is the multio's
7756 // first member (`print x >&1 > f | cat` sends `x` down the
7757 // pipe TWICE: once for the seed, once for the `>&1` dup).
7758 let pipe_seed = fd == 1
7759 && with_executor(|exec| {
7760 exec.pipe_output_scope
7761 .is_some_and(|d| d + 1 == exec.redirect_scope_stack.len())
7762 });
7763
7764 // Save current fd state for scope-end restoration — BEFORE
7765 // the first member's replace dup2 below.
7766 // c:Src/exec.c:2425 — `int fdN = movefd(fd1); save[fd1] = fdN;`. A SAVED
7767 // descriptor is shell state and must live above the script's fd range:
7768 // plain dup() returns the LOWEST free fd, which parked the saved stdout
7769 // on fd 3, so `print -u 3 -r -- X 2>/dev/null` wrote into the shell's own
7770 // saved descriptor and reported success where zsh says `bad file number`.
7771 // F_DUPFD with a floor of 10 is exactly what movefd does.
7772 let saved = unsafe { libc::fcntl(fd, libc::F_DUPFD, 10) };
7773 if saved >= 0 {
7774 with_executor(|exec| {
7775 if let Some(top) = exec.redirect_scope_stack.last_mut() {
7776 top.push((fd, saved));
7777 } else {
7778 unsafe { libc::close(saved) };
7779 }
7780 });
7781 }
7782
7783 // Accumulate member fds in redirect order. c:Src/exec.c:
7784 // 2447-2480 addfd — the FIRST member REPLACES the fd
7785 // (c:2448-2450 `mfds[fd1]->ct=1; mfds[fd1]->fds[0]=fd1;`), so
7786 // a later numeric `>&N` self-dup resolves against the fd's
7787 // value at that point in the sequence: `print x > f >&1`
7788 // writes f TWICE; `print x >&1 > f` writes the ORIGINAL
7789 // stdout + f.
7790 let mut target_fds: Vec<i32> = Vec::with_capacity(entries.len() + 1);
7791 if pipe_seed {
7792 let p = unsafe { libc::fcntl(fd, libc::F_DUPFD, 10) };
7793 if p >= 0 {
7794 target_fds.push(p);
7795 }
7796 }
7797 let noclobber = opt_state_get("noclobber").unwrap_or(false)
7798 || !opt_state_get("clobber").unwrap_or(true);
7799 for (i, (op_byte, target)) in entries.iter().enumerate() {
7800 let open_result: std::io::Result<i32> = match *op_byte {
7801 r::DUP_WRITE | r::DUP_READ => {
7802 // Numeric `>&N` — dup the LIVE fd N (after any
7803 // earlier member's replace).
7804 match target.trim_start_matches('&').parse::<i32>() {
7805 Ok(src) => {
7806 let d = unsafe { libc::fcntl(src, libc::F_DUPFD, 10) };
7807 if d >= 0 {
7808 Ok(d)
7809 } else {
7810 Err(std::io::Error::last_os_error())
7811 }
7812 }
7813 Err(_) => Err(std::io::Error::from_raw_os_error(libc::EBADF)),
7814 }
7815 }
7816 r::WRITE => {
7817 // c:Src/exec.c clobber_open — noclobber applies
7818 // to multio file targets too; failure aborts the
7819 // remaining redirect list (execerr), so `setopt
7820 // noclobber; touch a; print x > a > b` errors on
7821 // `a` and never creates `b`.
7822 let target_meta = std::fs::metadata(target).ok();
7823 let target_is_regular_file = target_meta
7824 .as_ref()
7825 .map(|m| m.file_type().is_file())
7826 .unwrap_or(false);
7827 // c:Src/exec.c:2313 clobber_open — CLOBBER_EMPTY re-uses
7828 // an empty regular file under noclobber (same allowance
7829 // as the single-redirect path).
7830 let clobber_empty_ok = opt_state_get("clobberempty").unwrap_or(false)
7831 && target_meta.as_ref().map(|m| m.len() == 0).unwrap_or(false);
7832 if noclobber && target_is_regular_file && !clobber_empty_ok {
7833 eprintln!(
7834 "{}:{}: file exists: {}",
7835 shname(),
7836 crate::ported::lex::lineno(),
7837 target
7838 );
7839 for prev in &target_fds {
7840 unsafe {
7841 libc::close(*prev);
7842 }
7843 }
7844 with_executor(|exec| {
7845 exec.redirect_failed = true;
7846 });
7847 // Sink the upcoming command's output (mirrors
7848 // the single-redirect noclobber arm in
7849 // host_apply_redirect).
7850 if let Ok(file) = fs::OpenOptions::new().write(true).open("/dev/null") {
7851 let new_fd = file.into_raw_fd();
7852 unsafe {
7853 libc::dup2(new_fd, fd);
7854 libc::close(new_fd);
7855 }
7856 }
7857 return Value::Status(1);
7858 }
7859 fs::OpenOptions::new()
7860 .write(true)
7861 .create(true)
7862 .truncate(true)
7863 .open(target)
7864 .map(|f| f.into_raw_fd())
7865 }
7866 r::APPEND => fs::OpenOptions::new()
7867 .write(true)
7868 .create(true)
7869 .append(true)
7870 .open(target)
7871 .map(|f| f.into_raw_fd()),
7872 _ => fs::OpenOptions::new()
7873 .write(true)
7874 .create(true)
7875 .truncate(true)
7876 .open(target)
7877 .map(|f| f.into_raw_fd()),
7878 };
7879 match open_result {
7880 Ok(tfd) => {
7881 if i == 0 && !pipe_seed {
7882 // c:2448-2450 — first member replaces the fd.
7883 unsafe {
7884 libc::dup2(tfd, fd);
7885 }
7886 }
7887 target_fds.push(tfd);
7888 }
7889 Err(e) => {
7890 // c:Src/exec.c:3741 — `zwarn("%e: %s", errno, fname)`:
7891 // zwarning supplies the `name:LINE:` prefix with the
7892 // REAL current lineno; redir_errno_msg builds the `%e`
7893 // errno message (was a hardcoded ErrorKind match that
7894 // showed generic "redirect failed" for EROFS/etc.).
7895 let msg = redir_errno_msg(&e);
7896 crate::ported::utils::zwarn(&format!("{}: {}", msg, target));
7897 // Close already-opened fds to avoid leaks.
7898 for prev in &target_fds {
7899 unsafe {
7900 libc::close(*prev);
7901 }
7902 }
7903 with_executor(|exec| {
7904 exec.redirect_failed = true;
7905 });
7906 return Value::Status(1);
7907 }
7908 }
7909 }
7910
7911 // Create the splitter pipe.
7912 let (read_end, write_end) = match os_pipe::pipe() {
7913 Ok(p) => p,
7914 Err(_) => {
7915 for f in &target_fds {
7916 unsafe {
7917 libc::close(*f);
7918 }
7919 }
7920 return Value::Status(1);
7921 }
7922 };
7923 let pipe_write_raw = AsRawFd::as_raw_fd(&write_end);
7924 // Spawn the splitter thread: read pipe → write every chunk
7925 // to every target fd. Each write inside the thread uses
7926 // libc::write directly on the raw fd (no Rust File ownership
7927 // so the splitter can close after EOF without racing main).
7928 let target_fds_for_thread = target_fds.clone();
7929 let handle = std::thread::spawn(move || {
7930 let mut r = read_end;
7931 let mut buf = [0u8; 8192];
7932 loop {
7933 match std::io::Read::read(&mut r, &mut buf) {
7934 Ok(0) => break,
7935 Ok(n) => {
7936 for &tfd in &target_fds_for_thread {
7937 let mut off = 0;
7938 while off < n {
7939 let w = unsafe {
7940 libc::write(
7941 tfd,
7942 buf[off..n].as_ptr() as *const libc::c_void,
7943 n - off,
7944 )
7945 };
7946 if w <= 0 {
7947 break;
7948 }
7949 off += w as usize;
7950 }
7951 }
7952 }
7953 Err(_) => break,
7954 }
7955 }
7956 // Close every target so file contents flush.
7957 for tfd in target_fds_for_thread {
7958 unsafe {
7959 libc::close(tfd);
7960 }
7961 }
7962 });
7963
7964 // Dup the pipe write-end onto the target fd; close the
7965 // original write_end so EOF arrives when host_redirect_scope_end
7966 // closes our tracked pipe_write_fd.
7967 let write_dup = unsafe { libc::fcntl(pipe_write_raw, libc::F_DUPFD, 10) };
7968 drop(write_end);
7969 if write_dup < 0 {
7970 return Value::Status(1);
7971 }
7972 unsafe {
7973 libc::dup2(write_dup, fd);
7974 libc::close(write_dup);
7975 }
7976 // Track the running splitter so scope-end can drain + join.
7977 // The "write_fd" we store is the user-visible fd (e.g. 1).
7978 // Closing that fd at scope-end isn't quite right; we need a
7979 // way to send EOF. Solution: track the write_dup we just
7980 // closed; instead keep a second dup for the close-on-end.
7981 // Shell-internal bookkeeping fd — above the script's range (movefd).
7982 let close_on_end = unsafe { libc::fcntl(fd, libc::F_DUPFD, 10) };
7983 with_executor(|exec| {
7984 if let Some(top) = exec.multios_scope_stack.last_mut() {
7985 top.push((close_on_end, handle));
7986 } else {
7987 // No scope — leak the dup; thread will keep running
7988 // until process exit. Should not happen because
7989 // host_redirect_scope_begin pushed a frame.
7990 unsafe { libc::close(close_on_end) };
7991 }
7992 });
7993 Value::Status(0)
7994 });
7995 // c:Src/exec.c:2418 input-arm — MULTIOS read fan-in. Stack
7996 // layout pushed by compile_zsh (mirrors the write side):
7997 // [source_1, op_1, source_2, op_2, …, source_N, op_N, fd]
7998 // argc = 2N + 1; op distinguishes file opens (READ) from numeric
7999 // `<&N` dups (DUP_READ); a glob source arrives as Value::Array
8000 // and splices into one member per match (c:Src/glob.c:2195-2203).
8001 // Opens every source, sets up a pipe + producer thread that
8002 // reads each source in order and writes to the pipe write-end,
8003 // then closes its write-end so the consumer gets EOF. dup2 the
8004 // pipe read-end onto fd. Bug #36 input side in docs/BUGS.md.
8005 vm.register_builtin(BUILTIN_MULTIOS_READ, |vm, argc| {
8006 if argc < 3 || argc % 2 == 0 {
8007 return Value::Status(1);
8008 }
8009 let fd = vm.pop().to_int() as i32;
8010 let n_sources = ((argc - 1) / 2) as usize;
8011 let mut pairs: Vec<(u8, Value)> = Vec::with_capacity(n_sources);
8012 for _ in 0..n_sources {
8013 let op_byte = vm.pop().to_int() as u8;
8014 let source = vm.pop();
8015 pairs.push((op_byte, source));
8016 }
8017 pairs.reverse();
8018
8019 // Splice glob match arrays (c:Src/glob.c:2195-2203).
8020 let mut entries: Vec<(u8, String)> = Vec::with_capacity(pairs.len());
8021 for (op_byte, source) in pairs {
8022 match source {
8023 Value::Array(items) => {
8024 for item in items {
8025 entries.push((op_byte, item.to_str()));
8026 }
8027 }
8028 other => entries.push((op_byte, other.to_str())),
8029 }
8030 }
8031 if entries.is_empty() {
8032 return Value::Status(1);
8033 }
8034
8035 // c:Src/exec.c:2418 — `unset(MULTIOS)`: sequential replace,
8036 // last source wins (`unsetopt multios; cat < a < b` reads
8037 // only b; a is still opened — and errors still surface).
8038 let multios_on = opt_state_get("multios").unwrap_or(true);
8039 if !multios_on {
8040 with_executor(|exec| {
8041 for (op_byte, source) in &entries {
8042 exec.host_apply_redirect(fd as u8, *op_byte, source);
8043 if exec.redirect_failed {
8044 break;
8045 }
8046 }
8047 });
8048 return Value::Status(0);
8049 }
8050
8051 if entries.len() == 1 {
8052 // Single member after splicing — plain replace.
8053 let (op_byte, source) = &entries[0];
8054 with_executor(|exec| {
8055 exec.host_apply_redirect(fd as u8, *op_byte, source);
8056 });
8057 return Value::Status(0);
8058 }
8059
8060 // Save current fd state for scope-end restoration — BEFORE
8061 // the first member's replace dup2 below.
8062 // c:Src/exec.c:2425 — `int fdN = movefd(fd1); save[fd1] = fdN;`. A SAVED
8063 // descriptor is shell state and must live above the script's fd range:
8064 // plain dup() returns the LOWEST free fd, which parked the saved stdout
8065 // on fd 3, so `print -u 3 -r -- X 2>/dev/null` wrote into the shell's own
8066 // saved descriptor and reported success where zsh says `bad file number`.
8067 // F_DUPFD with a floor of 10 is exactly what movefd does.
8068 let saved = unsafe { libc::fcntl(fd, libc::F_DUPFD, 10) };
8069 if saved >= 0 {
8070 with_executor(|exec| {
8071 if let Some(top) = exec.redirect_scope_stack.last_mut() {
8072 top.push((fd, saved));
8073 } else {
8074 unsafe { libc::close(saved) };
8075 }
8076 });
8077 }
8078
8079 // Open every source in redirect order; numeric `<&N` dups
8080 // resolve against the LIVE fd table. First member replaces
8081 // the fd (c:2448-2450) so later self-dups see it.
8082 let mut source_fds: Vec<i32> = Vec::with_capacity(entries.len());
8083 for (i, (op_byte, source)) in entries.iter().enumerate() {
8084 let open_result: std::io::Result<i32> = match *op_byte {
8085 r::DUP_READ | r::DUP_WRITE => match source.trim_start_matches('&').parse::<i32>() {
8086 Ok(src) => {
8087 let d = unsafe { libc::fcntl(src, libc::F_DUPFD, 10) };
8088 if d >= 0 {
8089 Ok(d)
8090 } else {
8091 Err(std::io::Error::last_os_error())
8092 }
8093 }
8094 Err(_) => Err(std::io::Error::from_raw_os_error(libc::EBADF)),
8095 },
8096 _ => fs::File::open(source).map(|f| f.into_raw_fd()),
8097 };
8098 match open_result {
8099 Ok(tfd) => {
8100 if i == 0 {
8101 unsafe {
8102 libc::dup2(tfd, fd);
8103 }
8104 }
8105 source_fds.push(tfd);
8106 }
8107 Err(e) => {
8108 let msg = match e.kind() {
8109 std::io::ErrorKind::PermissionDenied => "permission denied",
8110 std::io::ErrorKind::NotFound => "no such file or directory",
8111 _ => "open failed",
8112 };
8113 // c:Src/exec.c:3741 — zwarn with real lineno prefix.
8114 crate::ported::utils::zwarn(&format!("{}: {}", msg, source));
8115 for prev in &source_fds {
8116 unsafe {
8117 libc::close(*prev);
8118 }
8119 }
8120 with_executor(|exec| {
8121 exec.redirect_failed = true;
8122 });
8123 return Value::Status(1);
8124 }
8125 }
8126 }
8127
8128 // Create the concatenator pipe.
8129 let (read_end, write_end) = match os_pipe::pipe() {
8130 Ok(p) => p,
8131 Err(_) => {
8132 for f in &source_fds {
8133 unsafe {
8134 libc::close(*f);
8135 }
8136 }
8137 return Value::Status(1);
8138 }
8139 };
8140 // dup the pipe read-end onto fd before spawning the
8141 // producer; close the original read_end so the consumer
8142 // (reading via fd) is the sole reference until scope-end.
8143 let read_dup = unsafe { libc::dup(AsRawFd::as_raw_fd(&read_end)) };
8144 drop(read_end);
8145 if read_dup < 0 {
8146 for f in &source_fds {
8147 unsafe {
8148 libc::close(*f);
8149 }
8150 }
8151 return Value::Status(1);
8152 }
8153 unsafe {
8154 libc::dup2(read_dup, fd);
8155 libc::close(read_dup);
8156 }
8157 // Spawn the producer.
8158 let source_fds_for_thread = source_fds.clone();
8159 let handle = std::thread::spawn(move || {
8160 let mut w = write_end;
8161 let mut buf = [0u8; 8192];
8162 for sfd in source_fds_for_thread {
8163 loop {
8164 let n = unsafe {
8165 libc::read(sfd, buf.as_mut_ptr() as *mut libc::c_void, buf.len())
8166 };
8167 if n <= 0 {
8168 break;
8169 }
8170 let n = n as usize;
8171 if std::io::Write::write_all(&mut w, &buf[..n]).is_err() {
8172 break;
8173 }
8174 }
8175 unsafe {
8176 libc::close(sfd);
8177 }
8178 }
8179 // Closing w (the write_end) at scope drop signals EOF
8180 // to the consumer.
8181 });
8182 with_executor(|exec| {
8183 // Track using a closed-write sentinel — the producer
8184 // owns write_end so we just need to join. Use -1 fd
8185 // marker meaning "no fd to close".
8186 if let Some(top) = exec.multios_scope_stack.last_mut() {
8187 top.push((-1, handle));
8188 } else {
8189 let _ = handle.join();
8190 }
8191 });
8192 Value::Status(0)
8193 });
8194 // c:Src/exec.c:3978-3986 — nullexec==1 marker. See the const's
8195 // doc block. Arg: 1 = entering a bare-exec redirect, 0 = leaving.
8196 vm.register_builtin(BUILTIN_EXEC_PERM_REDIRS, |vm, _argc| {
8197 let on = vm.pop().to_int() != 0;
8198 with_executor(|exec| exec.exec_redirs_permanent = on);
8199 Value::Status(0)
8200 });
8201 // Bare-exec redirect epilogue — see the const's doc block.
8202 // c:Src/exec.c:252-259 (execerr) + c:4367-4386 (done: POSIX gate).
8203 vm.register_builtin(BUILTIN_EXEC_REDIR_DONE, |vm, _argc| {
8204 use std::sync::atomic::Ordering;
8205 let failed = with_executor(|exec| {
8206 let f = exec.redirect_failed;
8207 exec.redirect_failed = false;
8208 f
8209 });
8210 if !failed {
8211 return Value::Status(0);
8212 }
8213 // c:255 — `redir_err = lastval = 1`.
8214 vm.last_status = 1;
8215 if isset(crate::ported::zsh_h::POSIXBUILTINS) && !isset(crate::ported::zsh_h::INTERACTIVE) {
8216 // c:4379-4383 — non-interactive POSIX fatal: exit(1).
8217 // In-process equivalent: arm EXIT_PENDING/EXIT_VAL so the
8218 // next BUILTIN_ERREXIT_CHECK (trigger 2) unwinds the
8219 // script with status 1 — same deferred-exit shape the
8220 // `exit` builtin uses inside subshell contexts.
8221 crate::ported::builtin::EXIT_VAL.store(1, Ordering::Relaxed);
8222 crate::ported::builtin::EXIT_PENDING.store(1, Ordering::Relaxed);
8223 }
8224 Value::Status(1)
8225 });
8226 // c:Src/exec.c:3722-3724 — see the const's doc block. No args.
8227 vm.register_builtin(BUILTIN_PIPE_OUTPUT_MARK, |_vm, _argc| {
8228 with_executor(|exec| exec.pipe_output_pending = true);
8229 Value::Status(0)
8230 });
8231 // c:Src/exec.c:3710-3724 — install this pipeline stage's fds.
8232 // /* Make a copy of stderr for xtrace output before redirecting */
8233 // fflush(xtrerr);
8234 // ...
8235 // /* Add pipeline input/output to mnodes */
8236 // if (input) addfd(forked, save, mfds, 0, input, 0, NULL);
8237 // if (output) addfd(forked, save, mfds, 1, output, 1, NULL);
8238 // Emitted into the stage chunk by compile_zsh.rs (after the arg
8239 // words' expansion ops, before the redirect scope), and fed by
8240 // BUILTIN_RUN_PIPELINE via `stage_fds_park`. Doing the dup2 HERE
8241 // rather than before the chunk runs is what makes a `$(...)` in a
8242 // stage's arguments see the shell's fd 0 instead of the pipe.
8243 vm.register_builtin(BUILTIN_PIPE_FDS_INSTALL, |vm, argc| {
8244 // Arg: `|&` merge-stderr flag (compile_zsh always passes it).
8245 let merge_stderr = pop_args(vm, argc)
8246 .first()
8247 .map(|s| s != "0" && !s.is_empty())
8248 .unwrap_or(false);
8249 let (in_fd, out_fd) = stage_fds_take();
8250 if in_fd < 0 && out_fd < 0 {
8251 return Value::Status(0);
8252 }
8253 // c:3711 `fflush(xtrerr)` — flush before the fds move, so
8254 // anything buffered from the expansion phase lands on the
8255 // ORIGINAL fd, not on the pipe.
8256 let _ = std::io::stdout().flush();
8257 let _ = std::io::stderr().flush();
8258 unsafe {
8259 if in_fd >= 0 {
8260 libc::dup2(in_fd, libc::STDIN_FILENO);
8261 if in_fd != libc::STDIN_FILENO {
8262 libc::close(in_fd);
8263 }
8264 }
8265 if out_fd >= 0 {
8266 libc::dup2(out_fd, libc::STDOUT_FILENO);
8267 if out_fd != libc::STDOUT_FILENO {
8268 libc::close(out_fd);
8269 }
8270 // `cmd |& next`: the `2>&1` C appends to cmd's redirect
8271 // list (walked at c:3730+, i.e. after this addfd), so
8272 // stderr follows the pipe, not the shell's stdout.
8273 if merge_stderr {
8274 libc::dup2(libc::STDOUT_FILENO, libc::STDERR_FILENO);
8275 }
8276 }
8277 }
8278 Value::Status(0)
8279 });
8280 // c:Src/exec.c — block-level redirect-failure gate. When a
8281 // compound command (`{ … } < file`, `( … ) > file`, etc.) has a
8282 // failing redirect (e.g. `< /nonexistent`), zsh skips the entire
8283 // body AND sets lastval to 1. The simple-command path's
8284 // redirect_failed check (line 215-221 above) only catches the
8285 // failure when a builtin dispatches and is consumed by that
8286 // single builtin call — so a multi-statement block kept running
8287 // its remaining statements after the redir error. Emit-side at
8288 // compile_zsh.rs::compile_command's Redirected arm pairs this
8289 // with a JumpIfTrue → WithRedirectsEnd to abandon the body.
8290 vm.register_builtin(BUILTIN_REDIRECT_FAILED_CHECK, |vm, _argc| {
8291 let failed = with_executor(|exec| {
8292 let f = exec.redirect_failed;
8293 exec.redirect_failed = false;
8294 f
8295 });
8296 if failed {
8297 vm.last_status = 1;
8298 Value::Int(1)
8299 } else {
8300 Value::Int(0)
8301 }
8302 });
8303 // c:Src/exec.c — drop-in replacement for fusevm's Op::Exec used by
8304 // the dynamic-first-word path (`$cmd`, `$(cmd)`, glob-named cmds).
8305 // fusevm's Op::Exec returns Value::Status(0) when post-expansion
8306 // argv is empty (vm.rs:1722) — that clobbers \$? for the
8307 // `\$(exit 1); echo \$?` case where the cmd-subst left
8308 // last_status = 1 but the empty expansion gets exec'd to 0.
8309 // Mirror C zsh: when the word list is empty after expansion,
8310 // \$? becomes whatever the inner cmd-subst's last_status is
8311 // (preserved here by returning Value::Status(last_status)).
8312 // c:Src/cond.c:308-316 — `if (!(pprog = patcompile(right, ...)))
8313 // { zwarnnam(fromtest, "bad pattern: %s", right); return 2; }`.
8314 // The cond path must NOT use str_match/glob_match_static: the
8315 // case-statement consumer of those follows Src/loop.c:667 zerr
8316 // semantics (errflag abort), while cond is a zwarn + status-2
8317 // soft failure. COND_BAD_PATTERN carries the 2 across the
8318 // Bool-shaped stack contract (so `!=`'s LogNot can't lose it).
8319 thread_local! {
8320 static COND_BAD_PATTERN: std::cell::Cell<bool> =
8321 const { std::cell::Cell::new(false) };
8322 }
8323 vm.register_builtin(BUILTIN_COND_STRMATCH, |vm, _argc| {
8324 let pat = vm.pop().to_str();
8325 let s = vm.pop().to_str();
8326 // bash `shopt -s nocasematch` → case-insensitive `[[ == ]]` / `[[ != ]]`.
8327 // Lowercase BOTH sides for the match decision (glob metacharacters are
8328 // not letters, so the pattern's `*`/`?`/`[…]` structure is preserved).
8329 // No-op unless the bash shopt is active. --zsh unaffected.
8330 let (s, pat) = if crate::dash_mode::nocasematch() {
8331 (s.to_lowercase(), pat.to_lowercase())
8332 } else {
8333 (s, pat)
8334 };
8335 let mut pat_tok = pat.clone();
8336 crate::ported::glob::tokenize(&mut pat_tok);
8337 if crate::ported::pattern::patcompile(
8338 &pat_tok,
8339 crate::ported::zsh_h::PAT_STATIC as i32,
8340 None,
8341 )
8342 .is_none()
8343 {
8344 // c:314 — zwarnnam(fromtest, "bad pattern: %s", right).
8345 crate::ported::utils::zwarn(&format!("bad pattern: {}", pat));
8346 COND_BAD_PATTERN.with(|c| c.set(true));
8347 return Value::Bool(false);
8348 }
8349 // Match via the shared engine so `(#b)`/`(#m)` backref and
8350 // MATCH-variable population stays in one place.
8351 Value::Bool(crate::vm_helper::glob_match_static(&s, &pat))
8352 });
8353 vm.register_builtin(BUILTIN_COND_UNKNOWN, |vm, _argc| {
8354 // c:Src/cond.c:150-188 — `zwarnnam(fromtest, "unknown condition: %s",
8355 // name)` for a `-X` op with no matching cond module. Like a cond
8356 // syntax error it yields status 2 and aborts: arm COND_BAD_PATTERN so
8357 // the downstream BUILTIN_COND_STATUS_FROM_BOOL carries the 2 across the
8358 // Bool-shaped stack and runs the shared errflag+set_last_status(2)+abort
8359 // path (c:Src/exec.c:5216-5221). Returns Bool(false) as the operand.
8360 let op = vm.pop().to_str();
8361 crate::ported::utils::zerr(&format!("unknown condition: {}", op));
8362 COND_BAD_PATTERN.with(|c| c.set(true));
8363 Value::Bool(false)
8364 });
8365 vm.register_builtin(BUILTIN_COND_STATUS_FROM_BOOL, |vm, _argc| {
8366 // `${~pat}` / `${(P)~pat}` inside a `[[ … ]]` operand flips
8367 // GLOB_SUBST on via the tilde carrier so the pattern match sees
8368 // active metacharacters. In C that flag is prefork-scoped and
8369 // gone once the operand is consumed; zshrs restores it at the
8370 // next command-dispatch boundary, but a bare `[[ … ]]` has no
8371 // trailing assignment to trigger that — so globsubst leaked ON
8372 // into the NEXT command's word expansion, filename-generating a
8373 // scalar value it should not (p10k `_p9k_set_prompt`: line 45
8374 // `[[ … != ${(P)~disabled} ]]` leaked into line 46's
8375 // `local val=$arr[idx]`, whose glob-char-laden value then hit
8376 // "no matches found" and aborted the whole prompt build →
8377 // garbled 25-line prompt / interactive hang). Consume the
8378 // carrier here: this builtin ends EVERY `[[ … ]]`, and runs
8379 // after the operands (and their pattern match) are done.
8380 consume_tilde_globsubst_carrier();
8381 let ok = vm.pop().to_int() != 0;
8382 let bad = COND_BAD_PATTERN.with(|c| {
8383 let b = c.get();
8384 c.set(false);
8385 b
8386 });
8387 if bad {
8388 // c:Src/exec.c:5216-5221 — `stat = evalcond(...);
8389 // /* 2 indicates a syntax error. For compatibility,
8390 // turn this into a shell error. */
8391 // if (stat == 2) errflag |= ERRFLAG_ERROR;`
8392 // The errflag abort exits the script with lastval (2),
8393 // matching `zsh -fc '[[ x == [a- ]]; print rc=$?'`
8394 // printing nothing after the diagnostic and exiting 2.
8395 crate::ported::utils::errflag.fetch_or(
8396 crate::ported::zsh_h::ERRFLAG_ERROR,
8397 std::sync::atomic::Ordering::Relaxed,
8398 );
8399 with_executor(|exec| exec.set_last_status(2));
8400 return Value::Int(2); // c:Src/cond.c:316 `return 2;`
8401 }
8402 let status: i32 = if ok { 0 } else { 1 };
8403 // c:Src/exec.c:5216 — `lastval = evalcond(...)`: the conditional's
8404 // result IS the command's lastval, and c:Src/cond.c's evalcond
8405 // never inspects errflag while evaluating. So when a `[[ … ]]`
8406 // operand raised errflag (e.g. a nounset "parameter not set" zerr
8407 // on `${arr[99]}` under NO_UNSET), zsh STILL completes the test and
8408 // exits with the cond result; the errflag only aborts the FOLLOWING
8409 // commands. Sync the result to the executor's live lastval HERE —
8410 // the nounset site left it at a transient 1, and the next
8411 // BUILTIN_ERREXIT_CHECK reads the executor (not vm.last_status), so
8412 // without this sync `setopt NO_UNSET; [[ -z ${arr[99]} ]]` exited 1
8413 // instead of 0. The Op::SetStatus that follows sets vm.last_status;
8414 // this keeps the executor coherent with it before the abort check.
8415 with_executor(|exec| exec.set_last_status(status));
8416 Value::Int(status as i64)
8417 });
8418 vm.register_builtin(BUILTIN_USE_CMDOUTVAL_RESET, |_vm, _argc| {
8419 crate::ported::exec::use_cmdoutval.store(0, std::sync::atomic::Ordering::Relaxed);
8420 Value::Status(0)
8421 });
8422
8423 vm.register_builtin(BUILTIN_EXEC_DYNAMIC, |vm, argc| {
8424 let raw = pop_args(vm, argc);
8425 // Flatten Array entries into argv slots (matches fusevm
8426 // Op::Exec's flatten at vm.rs:1660-1665) so `${arr[@]}` /
8427 // splice expansions produce one argv slot per element.
8428 let args: Vec<String> = raw.into_iter().collect();
8429 // c:Src/subst.c paramsubst — when `${var:?msg}` or
8430 // `${var?msg}` set errflag, the expansion may produce empty
8431 // argv[0] which would fall into the EACCES/permission-denied
8432 // path below, masking the real paramsubst diagnostic with a
8433 // spurious "permission denied:" line and rc=126. Honour
8434 // errflag so the simple command ends with the paramsubst
8435 // error as the sole diagnostic, rc=1. Bug #86.
8436 if (crate::ported::utils::errflag.load(std::sync::atomic::Ordering::SeqCst)
8437 & crate::ported::zsh_h::ERRFLAG_ERROR)
8438 != 0
8439 {
8440 return Value::Status(1);
8441 }
8442 if args.is_empty() {
8443 // c:Src/exec.c:3442 — a command whose words expand to ZERO
8444 // words is a NULL command: `cmdoutval = use_cmdoutval ?
8445 // lastval : 0`. `use_cmdoutval` is set (below, in
8446 // BUILTIN_CMD_SUBST_TEXT) only when a command substitution
8447 // ran during this command's word expansion, so:
8448 // `false; $(exit 5)` → keep the subst status (5)
8449 // `false; $nonexistent` → reset to 0 (null command).
8450 // The previous port unconditionally kept `$?`, so
8451 // `false; $unset` wrongly stayed 1 (A01grammar.ztst:5).
8452 let keep =
8453 crate::ported::exec::use_cmdoutval.load(std::sync::atomic::Ordering::Relaxed) != 0;
8454 let status = if keep { vm.last_status } else { 0 };
8455 crate::ported::exec::use_cmdoutval.store(0, std::sync::atomic::Ordering::Relaxed);
8456 return Value::Status(status);
8457 }
8458 if args[0].is_empty() {
8459 // Explicit empty command word — exec returns EACCES.
8460 let script_name =
8461 crate::ported::utils::scriptname_get().unwrap_or_else(|| "zshrs".to_string());
8462 let lineno: u64 = with_executor(|exec| {
8463 exec.scalar("LINENO")
8464 .and_then(|s| s.parse::<u64>().ok())
8465 .unwrap_or(1)
8466 });
8467 eprintln!("{}:{}: permission denied: ", script_name, lineno);
8468 return Value::Status(126);
8469 }
8470 // AOP intercepts (zshrs extension, no C counterpart) — same
8471 // gate as host_exec_external (the static-head path): dynamic
8472 // command names (`cmd=/bin/echo; $cmd payload`) must consult
8473 // registered intercepts before dispatch, else `intercept
8474 // before /bin/echo ...` fires for the literal spelling but
8475 // not the variable one. run_intercepts runs before-advice
8476 // in-place and returns None to continue; Some(status) means
8477 // an around/after advice fully handled the command.
8478 let intercepted = with_executor(|exec| {
8479 if exec.intercepts.is_empty() {
8480 return None;
8481 }
8482 let full_cmd = if args.len() == 1 {
8483 args[0].clone()
8484 } else {
8485 args.join(" ")
8486 };
8487 let rest: Vec<String> = args[1..].to_vec();
8488 exec.run_intercepts(&args[0], &full_cmd, &rest)
8489 });
8490 if let Some(result) = intercepted {
8491 return Value::Status(result.unwrap_or(127));
8492 }
8493 // zshrs-original opcode builtins (async, doctor, peach, …) reached
8494 // via a run-time-resolved head (`$var`): they are absent from the
8495 // static BUILTINS port table / builtintab, so execcmd_exec below would
8496 // treat the head as external and report "command not found" — even
8497 // though `whence` calls it a builtin and a literal head runs it via
8498 // CallBuiltin. Dispatch by name here, but ONLY when the head is neither
8499 // a user function nor a ported builtin, so the shell's
8500 // function -> builtin -> external order is preserved.
8501 if let Some(head) = args.first() {
8502 let is_fn = with_executor(|e| e.function_exists(head));
8503 let is_ported =
8504 crate::ported::builtin::createbuiltintable().contains_key(head.as_str());
8505 if !is_fn && !is_ported {
8506 if let Some(status) = try_run_registered_builtin(head, &args[1..]) {
8507 crate::ported::builtin::LASTVAL
8508 .store(status, std::sync::atomic::Ordering::Relaxed);
8509 return Value::Status(status);
8510 }
8511 }
8512 }
8513 // c:Src/exec.c:2900 execcmd_exec — canonical simple-command
8514 // dispatcher. Runs precmd-modifier walk (c:3013-3091), then
8515 // dispatches to execbuiltin (c:4233) / runshfunc (c:3431+) /
8516 // execute (c:4314) per the resolved head. zshrs's bytecode VM
8517 // expanded the args before reaching here; we feed them in via
8518 // eparams.args and let execcmd_exec do the rest exactly as C
8519 // does for static heads. Without this, `c=builtin; $c source X`
8520 // skipped the precmd walk and emitted "command not found:
8521 // builtin".
8522 let mut state = crate::ported::zsh_h::estate {
8523 prog: Box::<crate::ported::zsh_h::eprog>::default(),
8524 pc: 0,
8525 strs: None,
8526 strs_offset: 0,
8527 };
8528 let mut eparams = crate::ported::zsh_h::execcmd_params {
8529 args: Some(args),
8530 redir: None,
8531 beg: 0,
8532 varspc: None,
8533 assignspc: None,
8534 typ: crate::ported::zsh_h::WC_SIMPLE as i32,
8535 postassigns: 0,
8536 htok: 0,
8537 };
8538 // input/output=0 → no pipe redirection (use shell stdio
8539 // directly); `output != 0` at c:2988 forks immediately. last1=2
8540 // (c:Src/exec.c:2014 `last1 ? 1 : 2`): terminal pipe stage but
8541 // the shell IS needed afterward — the VM keeps executing
8542 // bytecode after this op. last1=1 would arm the fake-exec
8543 // optimization (c:3646-3651, gate at c:3662 `last1 != 1`),
8544 // making `execute()` execve THIS process for external heads:
8545 // `p=/bin/echo; $p hi; echo after` replaced the shell and
8546 // `after` never ran (D04parameter chunk 11 shell-killer).
8547 // c:Src/exec.c:1690-1700 — execpline's job frame: save thisjob
8548 // (`pj = thisjob`) and allocate the jobtab slot that
8549 // execcmd_fork's addproc (c:2853) hangs the child pid off.
8550 // Without a live thisjob, the fork at c:3662 (last1 != 1 →
8551 // external must fork) registers no proc, nothing waits, and
8552 // the child races the rest of the script.
8553 let pj = {
8554 use crate::ported::jobs;
8555 *jobs::THISJOB
8556 .get_or_init(|| std::sync::Mutex::new(-1))
8557 .lock()
8558 .unwrap_or_else(|e| e.into_inner())
8559 };
8560 let newjob = {
8561 use crate::ported::jobs;
8562 let table = jobs::JOBTAB.get_or_init(|| std::sync::Mutex::new(Vec::new()));
8563 let mut tab = table.lock().unwrap_or_else(|e| e.into_inner());
8564 jobs::initjob(&mut tab) // c:1700 `thisjob = newjob = initjob()`
8565 };
8566 {
8567 use crate::ported::jobs;
8568 *jobs::THISJOB
8569 .get_or_init(|| std::sync::Mutex::new(-1))
8570 .lock()
8571 .unwrap_or_else(|e| e.into_inner()) = newjob as i32;
8572 }
8573 crate::ported::exec::execcmd_exec(
8574 &mut state,
8575 &mut eparams,
8576 0, // input (c:2989)
8577 0, // output (c:2988)
8578 crate::ported::zsh_h::Z_SYNC as i32, // how
8579 2, // last1=2 — shell continues (c:2014)
8580 -1, // close_if_forked
8581 );
8582 // c:Src/exec.c:1828-1835 — execpline's Z_SYNC tail: waitjobs()
8583 // reaps the forked external. c:Src/jobs.c:487-495 + 551-552 —
8584 // the job's LAST proc sets lastval (0200|sig when signalled,
8585 // else WEXITSTATUS). Builtin/shfunc heads never forked (job
8586 // has no procs) — LASTVAL was already set by execbuiltin /
8587 // doshfunc inside execcmd_exec; skip the wait.
8588 {
8589 use crate::ported::jobs;
8590 let table = jobs::JOBTAB.get_or_init(|| std::sync::Mutex::new(Vec::new()));
8591 let mut tab = table.lock().unwrap_or_else(|e| e.into_inner());
8592 if jobs::hasprocs(&tab, newjob) {
8593 jobs::waitjobs(&mut tab, newjob); // c:1835
8594 if let Some(p) = tab[newjob].procs.last() {
8595 let val = if p.is_signaled() {
8596 0o200 | p.term_sig() // c:Src/jobs.c:489-490
8597 } else {
8598 p.exit_status() // c:Src/jobs.c:494
8599 };
8600 crate::ported::builtin::LASTVAL
8601 .store(val, std::sync::atomic::Ordering::Relaxed);
8602 }
8603 }
8604 // c:1977-1979 — `deletejob(jn, 0)` once done; c:1981
8605 // `thisjob = pj` restores the caller's job.
8606 if newjob < tab.len() {
8607 jobs::deletejob(&mut tab[newjob], false);
8608 }
8609 *jobs::THISJOB
8610 .get_or_init(|| std::sync::Mutex::new(-1))
8611 .lock()
8612 .unwrap_or_else(|e| e.into_inner()) = pj;
8613 }
8614 let status = crate::ported::builtin::LASTVAL.load(std::sync::atomic::Ordering::Relaxed);
8615 let mut synth = crate::ported::zsh_h::job::default();
8616 crate::ported::jobs::waitonejob(&mut synth);
8617 Value::Status(status)
8618 });
8619 // c:Src/exec.c:3340-3364 — `< file` / `> file` with no command
8620 // word. Resolves NULLCMD/READNULLCMD at runtime then routes
8621 // through host_exec_external. Redirects are already applied by
8622 // the surrounding WithRedirectsBegin scope.
8623 vm.register_builtin(BUILTIN_NULLCMD_EXEC, |vm, argc| {
8624 let args = pop_args(vm, argc);
8625 let is_single_read = args
8626 .first()
8627 .map(|s| s != "0" && !s.is_empty())
8628 .unwrap_or(false);
8629 // c:Src/exec.c — when the surrounding redir-open failed
8630 // (e.g. `< /nonexistent`), zerr already printed the diag
8631 // and set redirect_failed. Don't invoke NULLCMD — return
8632 // status 1 like the wordcode path does.
8633 let redir_failed = with_executor(|exec| {
8634 let f = exec.redirect_failed;
8635 exec.redirect_failed = false;
8636 f
8637 });
8638 if redir_failed {
8639 crate::ported::builtin::LASTVAL.store(1, std::sync::atomic::Ordering::Relaxed);
8640 return Value::Status(1);
8641 }
8642 let nullcmd = crate::ported::params::getsparam("NULLCMD");
8643 let nc_str = nullcmd.as_deref().unwrap_or("");
8644 let nc_empty = nc_str.is_empty();
8645 // c:3340-3344 — CSHNULLCMD or no NULLCMD set → diagnostic.
8646 if nc_empty || crate::ported::zsh_h::isset(crate::ported::zsh_h::CSHNULLCMD) {
8647 let script_name =
8648 crate::ported::utils::scriptname_get().unwrap_or_else(|| "zshrs".to_string());
8649 let lineno: u64 = with_executor(|exec| {
8650 exec.scalar("LINENO")
8651 .and_then(|s| s.parse::<u64>().ok())
8652 .unwrap_or(1)
8653 });
8654 eprintln!("{}:{}: redirection with no command", script_name, lineno);
8655 return Value::Status(1);
8656 }
8657 // c:3350 — SHNULLCMD → run `:`.
8658 let cmd: String = if crate::ported::zsh_h::isset(crate::ported::zsh_h::SHNULLCMD) {
8659 ":".to_string()
8660 } else if is_single_read {
8661 // c:3354-3359 — single REDIR_READ + READNULLCMD set → readnullcmd.
8662 let rnc = crate::ported::params::getsparam("READNULLCMD");
8663 let rnc_str = rnc.as_deref().unwrap_or("");
8664 if !rnc_str.is_empty() {
8665 rnc_str.to_string()
8666 } else {
8667 nc_str.to_string() // c:3360-3363 fallback
8668 }
8669 } else {
8670 nc_str.to_string() // c:3360-3363
8671 };
8672 let status = with_executor(|exec| exec.host_exec_external(&[cmd]));
8673 crate::ported::builtin::LASTVAL.store(status, std::sync::atomic::Ordering::Relaxed);
8674 Value::Status(status)
8675 });
8676 // c:Src/exec.c:3342 — `zerr("redirection with no command")`.
8677 // Bare prefix-keyword (`builtin`, `command`, `exec`, `noglob`,
8678 // `nocorrect`) with a redirect but no command word. Emits the
8679 // canonical diagnostic via zerr (which sets errflag) and
8680 // returns Status(1). Bug #534.
8681 vm.register_builtin(BUILTIN_REDIR_NO_CMD, |_vm, _argc| {
8682 crate::ported::utils::zerr("redirection with no command");
8683 Value::Status(1)
8684 });
8685 vm.register_builtin(BUILTIN_DEBUG_TRAP, |vm, _argc| {
8686 // c:Src/signals.c:1245 dotrap(SIGDEBUG) — fires the DEBUG
8687 // trap body once per statement. The body sees the parent
8688 // shell's $? (LASTVAL). Guard against re-entry: commands
8689 // inside the DEBUG trap body would otherwise trigger
8690 // DEBUG_TRAP recursively → stack overflow. zsh guards via
8691 // its in_trap counter; we mirror with a thread-local Cell.
8692 //
8693 // c:Src/exec.c::trapcmd — before dotrap, the C source sets
8694 // `ZSH_DEBUG_CMD` to the about-to-run command text via
8695 // `dupstring(text)`. The trap body reads the parameter;
8696 // C unsets it after the trap returns. compile_list emits
8697 // the rendered statement text as the single arg here so the
8698 // shell-visible parameter reflects the command. Bug #263 in
8699 // docs/BUGS.md.
8700 let cmd_text = vm.pop().to_str();
8701 DEBUG_TRAP_REENTRY.with(|c| {
8702 if c.get() {
8703 return Value::Status(0);
8704 }
8705 // c:Src/exec.c:1423 — `if (sigtrapped[SIGDEBUG] &&
8706 // isset(DEBUGBEFORECMD) && !intrap)`. Bug #573: without
8707 // this gate, every sublist boundary called
8708 // setsparam("ZSH_DEBUG_CMD", ...) even when no DEBUG trap
8709 // was set, polluting the param table and (under
8710 // WARN_CREATE_GLOBAL) emitting a spurious
8711 // `scalar parameter ZSH_DEBUG_CMD created globally`
8712 // warning at every function call.
8713 //
8714 // Two trap registries exist (per signals.rs:1481-1511 dotrap):
8715 // - settrap path → sigtrapped[SIGDEBUG] bits set
8716 // - bin_trap path → traps_table["DEBUG"] populated, sigtrapped untouched
8717 // Mirror the dotrap dispatch decision: skip only when BOTH
8718 // are absent.
8719 let sig_debug = crate::ported::signals_h::SIGDEBUG as usize;
8720 let debug_trapped = crate::ported::signals::sigtrapped
8721 .lock()
8722 .map(|v| v.get(sig_debug).copied().unwrap_or(0))
8723 .unwrap_or(0);
8724 let debug_in_table = crate::ported::builtin::traps_table()
8725 .lock()
8726 .map(|t| t.contains_key("DEBUG"))
8727 .unwrap_or(false);
8728 if debug_trapped == 0 && !debug_in_table {
8729 return Value::Status(0);
8730 }
8731 c.set(true);
8732 // c:Src/exec.c — set ZSH_DEBUG_CMD scalar (PM_READONLY
8733 // is NOT set on ZSH_DEBUG_CMD, so the canonical
8734 // setsparam path is fine here — no direct paramtab
8735 // mutation needed).
8736 crate::ported::params::setsparam("ZSH_DEBUG_CMD", &cmd_text);
8737 let _ = crate::ported::signals::dotrap(crate::ported::signals_h::SIGDEBUG);
8738 // c:Src/exec.c::trapcmd — `unsetparam("ZSH_DEBUG_CMD")`
8739 // after the trap returns. Mirror that.
8740 crate::ported::params::unsetparam("ZSH_DEBUG_CMD");
8741 c.set(false);
8742 Value::Status(0)
8743 })
8744 });
8745
8746 // Fatal-only abort check emitted between the pipes of an `&&` / `||`
8747 // chain, where the full errexit check is suppressed. Mirrors ONLY the
8748 // errflag arm of BUILTIN_ERREXIT_CHECK below: an errflag abandons the
8749 // list in zsh, and no connector can consume it.
8750 vm.register_builtin(BUILTIN_FATAL_ABORT_CHECK, |vm, _argc| {
8751 use std::sync::atomic::Ordering;
8752 let errflag_set = (crate::ported::utils::errflag.load(Ordering::Relaxed)
8753 & crate::ported::zsh_h::ERRFLAG_ERROR)
8754 != 0;
8755 if !errflag_set || isset(crate::ported::zsh_h::INTERACTIVE) {
8756 return Value::Int(0);
8757 }
8758 // CONTINUE_ON_ERROR: clear and keep going, as the full check does.
8759 if isset(crate::ported::zsh_h::CONTINUEONERROR) {
8760 crate::ported::utils::errflag
8761 .fetch_and(!crate::ported::zsh_h::ERRFLAG_ERROR, Ordering::Relaxed);
8762 return Value::Int(0);
8763 }
8764 // Abort the chain with the failing command's own status intact —
8765 // a cond syntax error left lastval=2 (c:Src/exec.c:5216-5221), and
8766 // that 2 is what zsh exits with. Reading the executor's live
8767 // lastval (not forcing 1) is the same rule the full check uses.
8768 vm.last_status = with_executor(|exec| exec.last_status());
8769 Value::Int(1)
8770 });
8771 vm.register_builtin(BUILTIN_ERREXIT_CHECK, |vm, _argc| {
8772 // Returns Value::Int(1) when the caller should jump to the
8773 // current scope's return-patch landing (subshell-end / func-
8774 // end / chunk-end). Returns Value::Int(0) otherwise. Emit
8775 // side at `emit_errexit_check` pairs this with a JumpIfTrue
8776 // → return_patches pattern so the caller can short-circuit.
8777 //
8778 // Four triggers:
8779 // 1. RETFLAG set by a nested `return` / `exit` (eval,
8780 // sourced file, called function). Unwind THIS scope so
8781 // the flag propagates outward until something clears it.
8782 // 2. EXIT_PENDING set (mostly subshell-context exits). Same
8783 // propagation logic.
8784 // 3. `set -e` + nonzero status — the classic errexit path.
8785 // 4. errflag set in non-interactive mode — readonly
8786 // reassign, bad redirect, parse error mid-expansion etc.
8787 // Aborts the script (c:Src/init.c loop()).
8788 use std::sync::atomic::Ordering;
8789 let retflag = crate::ported::builtin::RETFLAG.load(Ordering::Relaxed);
8790 let exit_pending = crate::ported::builtin::EXIT_PENDING.load(Ordering::Relaxed);
8791 // c:Src/exec.c:1571-1603 — `sublist_done:` runs the ZERR trap for
8792 // the sublist that just failed. It is NOT gated on retflag: C only
8793 // consults retflag at the TOP of the list loop (c:1370 `while
8794 // (wc_code(code) == WC_LIST && !breaks && !retflag && !errflag)`),
8795 // which stops the NEXT sublist — the current one still completes
8796 // its sublist_done. So `return 5` fires the ERR trap on its way out.
8797 //
8798 // zshrs's escape short-circuit below returns before ever reaching
8799 // the ZERR fire, so a `return N` inside a try-list skipped the trap:
8800 // f() { { return 5 } always { print fin } }; f
8801 // printed `fin / err=5` where zsh prints `err=5 / fin / err=5`.
8802 // (Plain `f() { return 5 }` matched by luck — the inner fire was
8803 // missing but the OUTER sublist fired instead, since doshfunc had
8804 // cleared retflag by then and DONETRAP was still 0.)
8805 //
8806 // `exit` is deliberately excluded: C's `exit` goes zexit() →
8807 // realexit(), leaving the process without ever reaching
8808 // sublist_done. Verified: `zsh -fc 'trap "print err" ERR; f(){ exit
8809 // 5 }; f'` prints nothing.
8810 if retflag != 0 && exit_pending == 0 {
8811 let last = vm.last_status;
8812 // c:1598-1603 — same DONETRAP gate as the non-escape path below.
8813 if last != 0 && crate::ported::exec::DONETRAP.load(Ordering::Relaxed) == 0 {
8814 // c:Src/signals.c:1085-1087 — `int obreaks = breaks; int
8815 // oretflag = retflag; int olastval = lastval;` and c:1220-1222
8816 // — `breaks += obreaks; retflag = oretflag;`. dotrapargs
8817 // brackets EVERY trap dispatch with this save/restore because
8818 // the trap body runs as a normal list and would otherwise
8819 // consume the caller's control-flow flags. That matters
8820 // exactly here: we are firing ZERR while retflag is SET, and a
8821 // FUNCTION-form trap (`TRAPZERR() { … }`) goes through
8822 // doshfunc, whose epilogue eats retflag outright
8823 // (c:Src/exec.c:6047-6052 `if (retflag) { retflag = 0; breaks
8824 // = funcsave->breaks; }`). Without the bracket the pending
8825 // `return 5` was swallowed by its own ERR trap and the
8826 // function ran on:
8827 // TRAPZERR() { print z }; f() { { return 2 } always { : }
8828 // print after }; f
8829 // printed `after`, where zsh returns from f.
8830 //
8831 // zshrs's `dotrap` inlines the dispatch and does not carry
8832 // dotrapargs' save/restore, so the bracket lives at this call
8833 // site. lastval is restored too (c:1087 / c:1213 `lastval =
8834 // olastval`) — the trap body's own commands must not become
8835 // the caller's `$?`.
8836 let obreaks = crate::ported::builtin::BREAKS.load(Ordering::Relaxed); // c:1085
8837 let oretflag = crate::ported::builtin::RETFLAG.load(Ordering::Relaxed); // c:1086
8838 let olastval = crate::ported::builtin::LASTVAL.load(Ordering::Relaxed); // c:1087
8839 let _ = crate::ported::signals::dotrap(crate::ported::signals_h::SIGZERR); // c:1601
8840 crate::ported::exec::DONETRAP.store(1, Ordering::Relaxed); // c:1602
8841 crate::ported::builtin::BREAKS.store(obreaks, Ordering::Relaxed); // c:1220
8842 crate::ported::builtin::RETFLAG.store(oretflag, Ordering::Relaxed); // c:1222
8843 crate::ported::builtin::LASTVAL.store(olastval, Ordering::Relaxed); // c:1213
8844 }
8845 }
8846 if retflag != 0 || exit_pending != 0 {
8847 if exit_pending != 0 {
8848 // c:Src/builtin.c zexit — the deferred exit carries its
8849 // status in EXIT_VAL; sync it into the VM counter so
8850 // the top-level unwind reports it as the script's exit
8851 // (run_chunk returns vm.last_status). Without this, a
8852 // POSIX-fatal `.` failure exited 127 (bin_dot's return)
8853 // instead of C's exit(1) at Src/exec.c:4383.
8854 vm.last_status = crate::ported::builtin::EXIT_VAL.load(Ordering::Relaxed) & 0xFF;
8855 }
8856 return Value::Int(1);
8857 }
8858 let errflag_set = (crate::ported::utils::errflag.load(Ordering::Relaxed)
8859 & crate::ported::zsh_h::ERRFLAG_ERROR)
8860 != 0;
8861 // c:Src/init.c:1931 — `if (errflag && !interact &&
8862 // !isset(CONTINUEONERROR)) { errexit = 1; break; }` — with
8863 // CONTINUE_ON_ERROR set, the top-level do-while re-enters
8864 // loop() and the NEXT list runs instead of the shell exiting.
8865 // Clear the flag so the next statement starts clean (the
8866 // failed statement's lastval is already in place).
8867 if errflag_set
8868 && !crate::ported::zsh_h::isset(crate::ported::zsh_h::INTERACTIVE)
8869 && crate::ported::zsh_h::isset(crate::ported::zsh_h::CONTINUEONERROR)
8870 {
8871 crate::ported::utils::errflag
8872 .fetch_and(!crate::ported::zsh_h::ERRFLAG_ERROR, Ordering::Relaxed);
8873 return Value::Int(0);
8874 }
8875 if errflag_set && !crate::ported::zsh_h::isset(crate::ported::zsh_h::INTERACTIVE) {
8876 // c:Src/exec.c execlist — every enclosing list loop runs
8877 // `while (... && !errflag)`, so a set errflag breaks the
8878 // CURRENT scope and the check in the enclosing scope
8879 // breaks THAT one, all the way out. Leave errflag SET —
8880 // do NOT convert it to EXIT_PENDING: a process-exit
8881 // signal tunnels through the containment boundaries C
8882 // has, namely eval (Src/builtin.c:6221 `errflag &=
8883 // ~ERRFLAG_ERROR`), source (Src/init.c:1663 same), fork
8884 // boundaries (subshell/cmdsubst — child's errflag dies
8885 // with the child), and the interactive toplevel
8886 // (Src/init.c:139). Those boundaries clear errflag
8887 // themselves and execution continues past them; with
8888 // EXIT_PENDING armed here, `eval 'assoc=(odd)'; echo
8889 // after` aborted the whole script where zsh 5.9 prints
8890 // `after` (eval status 1). Bug #74's function case
8891 // (`f() { local -r x=5; x=10; }; f; echo after`) still
8892 // aborts: the function scope unwinds on THIS check, and
8893 // the caller's next ERREXIT_CHECK sees the still-set
8894 // errflag and unwinds too — exactly C's propagation.
8895 //
8896 // c:Src/init.c:234 — loop() BREAKS on errflag and
8897 // zsh_main exits with the UNTOUCHED lastval, NOT a
8898 // forced 1: `typeset -i x=3#8` (math error during the
8899 // assignment, before typeset sets a status) exits 0 in
8900 // zsh; a cond syntax error set lastval=2 (exec.c:5216-
8901 // 5221) and zsh exits 2; the readonly-reassign case
8902 // exits 1 because ITS lastval is 1. Sync the VM counter
8903 // from the executor's live lastval instead of
8904 // overwriting.
8905 vm.last_status = with_executor(|exec| exec.last_status());
8906 // c:Src/exec.c:1598-1603 — `sublist_done:` runs the ZERR trap
8907 // for the failed sublist BEFORE the enclosing list loop breaks
8908 // on errflag (`while (... && !errflag)` at c:1370). So an
8909 // errflag-setting command (readonly reassign, bad redirect)
8910 // must fire ZERR on its way out, exactly like the retflag
8911 // escape above and the non-escape fall-through below. Without
8912 // this the errflag early-return pre-empted the ZERR block
8913 // further down, so `TRAPZERR() { … }; typeset -r ro=1; ro=2`
8914 // aborted the script (correct) but never fired the trap. Same
8915 // DONETRAP gate + dotrapargs save/restore bracket
8916 // (c:signals.c:1085-1087 / 1213-1222) as the retflag branch:
8917 // a function-form TRAPZERR runs through doshfunc and would
8918 // otherwise consume the caller's breaks/retflag/lastval.
8919 let last = with_executor(|exec| exec.last_status());
8920 if last != 0 && crate::ported::exec::DONETRAP.load(Ordering::Relaxed) == 0 {
8921 let obreaks = crate::ported::builtin::BREAKS.load(Ordering::Relaxed); // c:1085
8922 let oretflag = crate::ported::builtin::RETFLAG.load(Ordering::Relaxed); // c:1086
8923 let olastval = crate::ported::builtin::LASTVAL.load(Ordering::Relaxed); // c:1087
8924 // c:Src/signals.c:1101 — dotrapargs returns early if errflag
8925 // is set, and c:1174/1205-1218 brackets the dispatch with
8926 // `traperr = errflag` … restore. The failing assignment left
8927 // errflag SET, so the trap body (`print zerr`) would itself
8928 // bail on the first op. Clear errflag across the dispatch so
8929 // the body runs, then restore it so the script still aborts.
8930 let oerrflag = crate::ported::utils::errflag.load(Ordering::Relaxed); // c:1174
8931 crate::ported::utils::errflag.store(0, Ordering::Relaxed);
8932 let _ = crate::ported::signals::dotrap(crate::ported::signals_h::SIGZERR); // c:1601
8933 crate::ported::utils::errflag.store(oerrflag, Ordering::Relaxed); // c:1216
8934 crate::ported::exec::DONETRAP.store(1, Ordering::Relaxed); // c:1602
8935 crate::ported::builtin::BREAKS.store(obreaks, Ordering::Relaxed); // c:1220
8936 crate::ported::builtin::RETFLAG.store(oretflag, Ordering::Relaxed); // c:1222
8937 crate::ported::builtin::LASTVAL.store(olastval, Ordering::Relaxed); // c:1213
8938 }
8939 return Value::Int(1);
8940 }
8941 let last = vm.last_status;
8942 if last == 0 {
8943 return Value::Int(0);
8944 }
8945 // c:Src/exec.c:1598 `if (!this_noerrexit && !donetrap &&
8946 // !this_donetrap)` — gate the ZERR trap fire on DONETRAP so
8947 // an inner sublist (e.g. `false` inside a function) that
8948 // already fired ZERR doesn't fire it AGAIN at the outer
8949 // sublist's post-command check (after the function
8950 // returned non-zero). Bug #303 in docs/BUGS.md. DONETRAP
8951 // is reset at top-level statement boundaries via
8952 // BUILTIN_DONETRAP_RESET (compile_list emit at
8953 // compile_zsh.rs).
8954 let already_done = crate::ported::exec::DONETRAP.load(Ordering::Relaxed) != 0;
8955 if !already_done {
8956 // c:Src/signals.c:1245 dotrap(SIGZERR) — canonical ZERR
8957 // trap dispatch. Fires whenever a command exits
8958 // non-zero.
8959 let _ = crate::ported::signals::dotrap(crate::ported::signals_h::SIGZERR);
8960 // c:1602 — `donetrap = 1;` after firing.
8961 crate::ported::exec::DONETRAP.store(1, Ordering::Relaxed);
8962 }
8963 // c:Src/exec.c:1605-1610 — compute errreturn / errexit.
8964 // errreturn = ERRRETURN && (INTERACTIVE || locallevel || sourcelevel)
8965 // && !(noerrexit & NOERREXIT_RETURN)
8966 // errexit = (ERREXIT || (ERRRETURN && !errreturn))
8967 // && !(noerrexit & NOERREXIT_EXIT)
8968 let no_err = crate::ported::exec::noerrexit.load(Ordering::Relaxed);
8969 let locallvl = crate::ported::params::locallevel.load(Ordering::Relaxed);
8970 let sourcelvl = crate::ported::init::sourcelevel.load(Ordering::Relaxed);
8971 let errreturn_opt = isset(crate::ported::zsh_h::ERRRETURN);
8972 let in_unwindable_scope =
8973 isset(crate::ported::zsh_h::INTERACTIVE) || locallvl != 0 || sourcelvl != 0;
8974 let errreturn = errreturn_opt
8975 && in_unwindable_scope
8976 && (no_err & crate::ported::zsh_h::NOERREXIT_RETURN) == 0;
8977 if errreturn {
8978 // c:1620-1623 — `retflag = 1; breaks = loops;` — unwind to
8979 // function boundary without exiting the shell.
8980 crate::ported::builtin::RETFLAG.store(1, Ordering::Relaxed);
8981 let loops = crate::ported::builtin::LOOPS.load(Ordering::Relaxed);
8982 crate::ported::builtin::BREAKS.store(loops, Ordering::Relaxed);
8983 return Value::Int(1);
8984 }
8985 let (errexit_on, in_subshell) = with_executor(|exec| {
8986 let on_canonical = isset(ERREXIT) || (errreturn_opt && !errreturn); // c:1608-1609
8987 let on_legacy = opt_state_get("errexit").unwrap_or(false);
8988 (
8989 (on_canonical || on_legacy) && (no_err & crate::ported::zsh_h::NOERREXIT_EXIT) == 0,
8990 !exec.subshell_snapshots.is_empty(),
8991 )
8992 });
8993 if !errexit_on {
8994 return Value::Int(0);
8995 }
8996 // c:Src/exec.c:1611-1618 — under ERR_EXIT a failing command exits the
8997 // whole shell via realexit() FROM THE POINT OF FAILURE, before any
8998 // enclosing `always` arm can run. zsh 5.9.2 (the reference) has no
8999 // `this_noerrexit` deferral, so at top-level / function scope the
9000 // faithful behavior is to process-exit here (zexit fires the SIGEXIT
9001 // trap and exits). This bypasses the always arm, fixing
9002 // `setopt errexit; { false } always { print A }` which wrongly ran the
9003 // always body: the deferred EXIT_PENDING routed the unwind through
9004 // always_entry (compile_zsh.rs re-points it there) and
9005 // SET_TRY_BLOCK_ERROR then cleared the pending exit so the body ran.
9006 if crate::ported::builtin::SUBSHELL_DEPTH.load(Ordering::Relaxed) == 0 {
9007 crate::ported::builtin::zexit(last, crate::ported::zsh_h::ZEXIT_NORMAL); // c:1618 realexit
9008 }
9009 // Subshell: zshrs runs subshells in-process, so it cannot process-exit
9010 // the whole shell here — defer to the subshell-end unwind.
9011 crate::ported::builtin::EXIT_VAL.store(last, Ordering::Relaxed);
9012 crate::ported::builtin::EXIT_PENDING.store(1, Ordering::Relaxed);
9013 let _ = in_subshell;
9014 Value::Int(1)
9015 });
9016
9017 // BUILTIN_ASSIGN_ONLY_STATUS — status of an assignment-only
9018 // simple command. c:Src/exec.c:3393-3396 (execcmd_exec, no
9019 // command word + varspc): `if (errflag) lastval = 1; else
9020 // lastval = cmdoutval;`; same shape at c:1322 (execsimple
9021 // WC_ASSIGN: `lv = (errflag ? errflag : cmdoutval)`) and
9022 // c:3977 (nullexec=2 redir variant). cmdoutval is the exit of
9023 // a `$()` that ran in an RHS (already in vm.last_status via
9024 // compile_assign's per-assign SetStatus), 0 otherwise. The
9025 // store goes to the canonical LASTVAL too — that IS C's single
9026 // `lastval` global; without it the errflag-abort path
9027 // (BUILTIN_ERREXIT_CHECK trigger 4) syncs vm.last_status from
9028 // a stale LASTVAL and `readonly r=1; r=2` exited 0, not 1.
9029 vm.register_builtin(BUILTIN_ASSIGN_ONLY_STATUS, |vm, _argc| {
9030 use std::sync::atomic::Ordering;
9031 let had_cmd_subst = vm.pop().to_int() != 0;
9032 let errflag_set = (crate::ported::utils::errflag.load(Ordering::Relaxed)
9033 & crate::ported::zsh_h::ERRFLAG_ERROR)
9034 != 0;
9035 // c:Src/exec.c addvars — `if (!pm) { lastval = 1; if
9036 // (!cmdoutval) cmdoutval = 1; }` (assignment-failed cheat).
9037 let assign_failed = ASSIGN_FAILED_FLAG.swap(false, std::sync::atomic::Ordering::Relaxed);
9038 let status = if errflag_set || assign_failed {
9039 1 // c:Src/exec.c:3394 `lastval = 1` / addvars cmdoutval=1
9040 } else if had_cmd_subst {
9041 vm.last_status // c:3396 `lastval = cmdoutval` (subst exit)
9042 } else {
9043 0 // c:3396 `lastval = cmdoutval` (cmdoutval = 0)
9044 };
9045 with_executor(|exec| exec.set_last_status(status));
9046 // c:Src/jobs.c deletefilelist — a `=(cmd)` temp file is bound to the
9047 // JOB of the command that created it and unlinked when that command
9048 // completes (Src/exec.c:5588 for the shfunc case; the simple-command
9049 // job's filelist likewise). An assignment-only command like
9050 // `f==(cmd)` has no consuming builtin/exec, so the PsubFdGuard that
9051 // cleans consuming commands never fires — the temp leaked and a later
9052 // `$(<$f)` / `[[ -f $f ]]` still saw it, where zsh deletes it at the
9053 // end of the assignment (verified: even `f==(x) && cat $f` fails).
9054 // Clean here so the assignment command is the temp's job boundary.
9055 close_pending_psub_fds();
9056 Value::Status(status)
9057 });
9058
9059 // `${var:-default}` / `${var:=default}` / `${var:?error}` / `${var:+alt}`
9060 // Pops [name, op_byte, rhs] (rhs popped first). Returns the modified
9061 // value as Value::Str. Handles unset/empty distinction (`:-` etc.
9062 // treat empty same as unset, matching POSIX).
9063 // BUILTIN_PARAM_DEFAULT_FAMILY — `${var-x}` / `${var:-x}` / `${var=x}` /
9064 // `${var:=x}` / `${var?x}` / `${var:?x}` / `${var+x}` / `${var:+x}`.
9065 // PURE PASSTHRU: pop name + op + rhs, reconstruct the canonical
9066 // brace expression, hand to `subst::paramsubst` (C port of
9067 // `Src/subst.c::paramsubst`). All "missing vs empty" gating,
9068 // nounset suppression, default-evaluation, and elide-empty-words
9069 // semantics live inside paramsubst.
9070 vm.register_builtin(BUILTIN_PARAM_DEFAULT_FAMILY, |vm, _argc| {
9071 let rhs = vm.pop().to_str();
9072 let op = vm.pop().to_int() as u8;
9073 let name = vm.pop().to_str();
9074 // op=8 is the `${+name}` set-test prefix form (distinct from the
9075 // `${name+rhs}` substitute-if-set suffix form which is op=7).
9076 // Per compile_zsh.rs::parse_param_modifier: the `+` is emitted as
9077 // a leading sigil and `rhs` is empty.
9078 let body = if op == 8 {
9079 format!("${{+{}}}", name)
9080 } else {
9081 let op_str = match op {
9082 0 => ":-",
9083 1 => ":=",
9084 2 => ":?",
9085 3 => ":+",
9086 4 => "-",
9087 5 => "=",
9088 6 => "?",
9089 7 => "+",
9090 _ => "-",
9091 };
9092 format!("${{{}{}{}}}", name, op_str, rhs)
9093 };
9094 paramsubst_to_value(&body)
9095 });
9096
9097 // `${var:offset[:length]}` — substring. Pops [name, offset, length].
9098 // length == -1 means "rest of string". Negative offset counts from end.
9099 // BUILTIN_PARAM_SUBSTRING — `${var:offset:length}` literal-int form.
9100 // PURE PASSTHRU: reconstruct `${name:offset:length}` and route
9101 // through `subst::paramsubst`. Length sentinel `i64::MIN` =
9102 // "no length given" (omit the `:length` portion).
9103 //
9104 // c:Src/subst.c:1571,3781 — `${name:-N}` is the colon-default
9105 // operator, NOT a substring with negative offset. zsh's lexical
9106 // rule disambiguates via a literal space: `${name: -N}` (space
9107 // before `-`) is the substring form. The reconstructed body MUST
9108 // preserve that space when offset < 0; otherwise paramsubst's
9109 // `:-` dispatch fires on the synthesized `${name:-N}` body and
9110 // returns N as the unset-default instead of slicing the last N
9111 // chars. Length-form `${name:-N:M}` has the same trap.
9112 vm.register_builtin(BUILTIN_PARAM_SUBSTRING, |vm, _argc| {
9113 let length = vm.pop().to_int();
9114 let offset = vm.pop().to_int();
9115 let name = vm.pop().to_str();
9116 // !!! DASH-STRICT GATE !!! dash/ash have no `${var:offset:length}`
9117 // substring expansion (it is a "Bad substitution"); bash/ksh/sh do.
9118 if crate::dash_mode::dash_strict() {
9119 crate::ported::utils::zerr("bad substitution");
9120 crate::ported::utils::errflag.fetch_or(
9121 crate::ported::zsh_h::ERRFLAG_ERROR,
9122 std::sync::atomic::Ordering::Relaxed,
9123 );
9124 with_executor(|exec| exec.set_last_status(1));
9125 return Value::str("");
9126 }
9127 let off_sep = if offset < 0 { " " } else { "" };
9128 let body = if length == i64::MIN {
9129 format!("${{{}:{}{}}}", name, off_sep, offset)
9130 } else {
9131 format!("${{{}:{}{}:{}}}", name, off_sep, offset, length)
9132 };
9133 paramsubst_to_value(&body)
9134 });
9135
9136 // BUILTIN_PARAM_SUBSTRING_EXPR — `${var:offset_expr[:length_expr]}` form.
9137 // PURE PASSTHRU: rebuild `${name:offset:length}` using the
9138 // expression text verbatim (paramsubst's offset/length
9139 // parser evaluates arith / param refs itself).
9140 //
9141 // c:Src/subst.c:1571,3781 — same `:-` disambiguation trap as
9142 // BUILTIN_PARAM_SUBSTRING. The expression text may itself start
9143 // with `-` (e.g. `${VAR:$((-1))}` arith resolves at the body-
9144 // assembly layer in some upstream paths, leaving `-1` in
9145 // off_expr). Insert a leading space when off_expr starts with
9146 // `-` so paramsubst's check_colon_subscript (subst.c:1571)
9147 // accepts the operand as a math expression instead of the
9148 // `:-` operator catching it.
9149 vm.register_builtin(BUILTIN_PARAM_SUBSTRING_EXPR, |vm, _argc| {
9150 let has_len = vm.pop().to_int() != 0;
9151 let len_expr = vm.pop().to_str();
9152 let off_expr = vm.pop().to_str();
9153 let name = vm.pop().to_str();
9154 let off_sep = if off_expr.starts_with('-') { " " } else { "" };
9155 let body = if has_len {
9156 format!("${{{}:{}{}:{}}}", name, off_sep, off_expr, len_expr)
9157 } else {
9158 format!("${{{}:{}{}}}", name, off_sep, off_expr)
9159 };
9160 paramsubst_to_value(&body)
9161 });
9162
9163 // `${var#pat}` / `${var##pat}` / `${var%pat}` / `${var%%pat}`
9164 // Pops [name, pattern, op_byte]. op: 0=`#` short-prefix, 1=`##` long,
9165 // 2=`%` short-suffix, 3=`%%` long. Glob-pattern matching via the
9166 // existing glob_match_static helper.
9167 // BUILTIN_PARAM_STRIP — `${var#pat}` / `${var##pat}` / `${var%pat}` /
9168 // `${var%%pat}`. PURE PASSTHRU: reconstruct the brace expression
9169 // and route through `subst::paramsubst`. (M)/(S) flags arrive
9170 // through SUB_FLAGS (already inside paramsubst's scope), so we
9171 // just clear the bridge-side cached read.
9172 vm.register_builtin(BUILTIN_PARAM_STRIP, |vm, _argc| {
9173 let _dq_flag = vm.pop().to_int() != 0;
9174 let op = vm.pop().to_int() as u8;
9175 let pattern = vm.pop().to_str();
9176 let name = vm.pop().to_str();
9177 let op_str = match op {
9178 0 => "#",
9179 1 => "##",
9180 2 => "%",
9181 3 => "%%",
9182 _ => "#",
9183 };
9184 let body = format!("${{{}{}{}}}", name, op_str, pattern);
9185 paramsubst_to_value(&body)
9186 });
9187
9188 // `$((expr))` — pops [expr_string], evaluates via MathEval which
9189 // honors integer-vs-float distinction (zsh-compatible). Returns
9190 // the result as Value::Str so it can be Concat'd into surrounding
9191 // word context.
9192 vm.register_builtin(BUILTIN_ARITH_EVAL, |vm, _argc| {
9193 // Pure path: evaluate expr, return string. errflag may be
9194 // set by arithsubst on math error; the caller decides
9195 // whether to clear it. For `(( ... ))` (math command) the
9196 // compile_arith path clears via BUILTIN_ARITH_CMD_FINISH;
9197 // for `$((... ))` (substitution inside another command)
9198 // errflag stays set so the surrounding command aborts —
9199 // matches c:Src/math.c "math errors propagate as errflag
9200 // through the containing word expansion".
9201 let expr = vm.pop().to_str();
9202 let result = crate::ported::subst::arithsubst(&expr, "", "");
9203 let _ = vm; // silence unused warning when no math error path mutates
9204 Value::str(result)
9205 });
9206
9207 // After-call hook used by compile_arith's `(( ... ))` path: when
9208 // arithsubst set errflag (math error), clear it and signal
9209 // status=2 in vm.last_status — matches zsh's c:exec.c arith-
9210 // failure: the math command exits 2 and the script continues.
9211 vm.register_builtin(BUILTIN_ARITH_CMD_FINISH, |vm, _argc| {
9212 use std::sync::atomic::Ordering;
9213 let live = crate::ported::utils::errflag.load(Ordering::Relaxed);
9214 let err = live & crate::ported::zsh_h::ERRFLAG_ERROR;
9215 let hard = live & crate::ported::zsh_h::ERRFLAG_HARD;
9216 if err != 0 {
9217 // c:Src/subst.c:3344 — when `${var:?msg}` fires, errflag
9218 // is OR'd with ERRFLAG_HARD to signal a script-abort
9219 // error (vs a recoverable math error like `$((1/0))`).
9220 // Clear only the ERRFLAG_ERROR bit; preserve
9221 // ERRFLAG_HARD so the next ERREXIT_CHECK aborts the
9222 // script. Bug #193 in docs/BUGS.md.
9223 if hard != 0 {
9224 // Keep ERRFLAG_HARD AND ERRFLAG_ERROR set so the
9225 // script-abort gate downstream still fires.
9226 vm.last_status = 2;
9227 Value::Status(2)
9228 } else {
9229 crate::ported::utils::errflag
9230 .fetch_and(!crate::ported::zsh_h::ERRFLAG_ERROR, Ordering::Relaxed);
9231 vm.last_status = 2;
9232 Value::Status(2)
9233 }
9234 } else {
9235 Value::Status(vm.last_status)
9236 }
9237 });
9238
9239 // `$(cmd)` — pops [cmd_string], routes through
9240 // run_command_substitution which performs an in-process pipe-capture.
9241 // Avoids the Op::CmdSubst sub-chunk word-emit bug
9242 // (`printf "a\nb"` produced "anb" via that path). Returns trimmed
9243 // output (trailing newlines stripped per POSIX cmd-sub semantics).
9244 vm.register_builtin(BUILTIN_CMD_SUBST_TEXT, |vm, _argc| {
9245 let cmd = vm.pop().to_str();
9246 // Inherit live $? into the inner shell so cmd-subst sees the
9247 // parent's most recent exit. Same rationale as the mode-3
9248 // backtick path above.
9249 let live_status = vm.last_status;
9250 let result = with_executor(|exec| {
9251 exec.set_last_status(live_status);
9252 exec.run_command_substitution(&cmd)
9253 });
9254 // Mirror run_command_substitution's exec.last_status side
9255 // effect into the VM's live counter so a containing
9256 // assignment's BUILTIN_SET_VAR — which reads vm.last_status
9257 // — sees the cmd-subst's exit. Without this, `a=$(false);
9258 // echo $?` reads stale 0 (vm.last_status was zeroed by
9259 // compile_assign's prelude SetStatus, and run_cmd_subst only
9260 // updated exec.last_status). Pull the value back through
9261 // exec since it owns the canonical post-subst record.
9262 let cs_status = with_executor(|exec| exec.last_status());
9263 vm.last_status = cs_status;
9264 // c:Src/exec.c — a command substitution running during a
9265 // command's word expansion makes its exit the status of an
9266 // otherwise-empty command (`$(exit 5)` → 5). Flag it so
9267 // BUILTIN_EXEC_DYNAMIC's null-command branch keeps `$?` instead
9268 // of resetting to 0.
9269 crate::ported::exec::use_cmdoutval.store(1, std::sync::atomic::Ordering::Relaxed);
9270 Value::str(result)
9271 });
9272
9273 // Text-based word expansion. Pops [preserved_text, mode_byte].
9274 // mode_byte:
9275 // 0 = Default — expand_string + xpandbraces + expand_glob
9276 // 1 = DoubleQuoted — strip outer `"…"`, expand_string only
9277 // (no brace, no glob — DQ semantics)
9278 // 2 = SingleQuoted — strip outer `'…'`, no expansion
9279 // (kept for symmetry; Snull early-return covers most SQ)
9280 // 3 = AltBackquote — strip backticks, run as cmd-sub
9281 // 7 = RedirTarget — same as Default but glob gated on MULTIOS
9282 // (c:Src/glob.c:2161-2167 xpandredir)
9283 // Single result → Value::str; multi → Value::Array.
9284 vm.register_builtin(BUILTIN_EXPAND_TEXT, |vm, _argc| {
9285 let mode = vm.pop().to_int() as u8;
9286 let text = vm.pop().to_str();
9287 // Sync vm.last_status → exec.last_status so cmd-subst (mode 3)
9288 // and any nested $? reads inside singsub see the live `$?`
9289 // from the most recent VM op. Without this, cmd-subst inside
9290 // arg-eval saw a stale exec.last_status that was zeroed at
9291 // the start of the current statement. Direct port of zsh's
9292 // pre-cmdsubst lastval propagation per Src/exec.c:4770.
9293 let live_status = vm.last_status;
9294 with_executor(|exec| exec.set_last_status(live_status));
9295 let result_value = with_executor(|exec| match mode {
9296 // Mode 1 = DoubleQuoted (argument context).
9297 // Mode 5 = DoubleQuoted in scalar-assignment context.
9298 // Both share the same DQ unescape pre-processing; mode 5
9299 // additionally bumps `in_scalar_assign` so subst_port's
9300 // paramsubst sees ssub=true and suppresses split flags
9301 // `(f)` / `(s:STR:)` / `(0)` per Src/subst.c:1759 +
9302 // Src/exec.c::addvars line 2546 (the PREFORK_SINGLE bit
9303 // C zsh sets when prefork-ing the assignment RHS).
9304 1 | 5 => {
9305 // DoubleQuoted: strip outer `"…"` if present. In DQ
9306 // context, `\` escapes the DQ-special chars `$`, `` ` ``,
9307 // `"`, `\`. zsh's expand_string expects the lexer's
9308 // `\0X` literal-marker for an already-escaped char, so
9309 // we pre-process: `\$` → `\0$`, `\\` → `\0\`, etc. Then
9310 // expand_string handles the rest.
9311 let inner = if text.len() >= 2 && text.starts_with('"') && text.ends_with('"') {
9312 &text[1..text.len() - 1]
9313 } else {
9314 text.as_str()
9315 };
9316 // The lexer's dquote_parse (Src/lex.c) already tokenized
9317 // DQ contents: `$` → Qstring (\u{8c}), `\$`/`\\`/`\"`/
9318 // `` \` `` → Bnull (\u{9f}) + literal. Stringsubst /
9319 // multsub recognize these markers natively. We pass
9320 // `inner` through verbatim — no re-tokenization needed.
9321 let prepped: String = inner.to_string();
9322 // Tell parameter-flag application that we're inside
9323 // double quotes — array-only flags ((o), (O), (n),
9324 // (i), (M), (u)) must be no-ops here per zsh.
9325 exec.in_dq_context += 1;
9326 if mode == 5 {
9327 exec.in_scalar_assign += 1;
9328 }
9329 // Mode 1 = argv DQ word; mode 5 = scalar-assign RHS.
9330 // In C zsh, the corresponding prefork-on-list paths
9331 // are: argv → `prefork(argv_list, 0)` returns multi-
9332 // word LinkList (Src/exec.c::execcmd), assignment →
9333 // `prefork(rhs_list, PREFORK_SINGLE|PREFORK_ASSIGN)`
9334 // returns single-word (Src/exec.c::addvars line
9335 // 2546). zshrs's `multsub` (Src/subst.c:544) is the
9336 // multi-result variant; `singsub` (Src/subst.c:514)
9337 // asserts ≤1 node. Mode 5 keeps singsub; mode 1
9338 // switches to multsub so `"${(@)arr}"`/`"$@"`/
9339 // `"${arr[@]}"` in argv context emit multiple words
9340 // as the C path would.
9341 // c:Src/lex.c untokenize — the final argv pass C runs
9342 // on every expanded word (glob.c:1862 / exec.c) drops
9343 // the Nularg empty-word sentinel remnulargs left in
9344 // place and folds any remaining token chars. Without
9345 // it, quoted splits with empty pieces
9346 // ("${(s:|:)x}" on "|a|b|") leak U+00A1 into argv.
9347 let result_value = if mode == 5 {
9348 let out = crate::ported::subst::singsub(&prepped);
9349 Value::str(crate::ported::lex::untokenize(&out))
9350 } else {
9351 let (_first, nodes, _ms_ws, _ret) = crate::ported::subst::multsub(&prepped, 0);
9352 // c:Src/subst.c:655 — multsub returns Vec::new()
9353 // for zero-word results (quoted array splat that
9354 // resolved to empty array). Surface as
9355 // Value::Array(vec![]) so the downstream array
9356 // assignment / argv flattening sees ZERO args.
9357 // Previous Rust port returned Value::str("") which
9358 // surfaced as ONE empty arg. Bug #120 in
9359 // docs/BUGS.md.
9360 if nodes.is_empty() {
9361 Value::Array(Vec::new())
9362 } else if nodes.len() == 1 {
9363 Value::str(crate::ported::lex::untokenize(
9364 &nodes.into_iter().next().unwrap(),
9365 ))
9366 } else {
9367 Value::Array(
9368 nodes
9369 .into_iter()
9370 .map(|n| Value::str(crate::ported::lex::untokenize(&n)))
9371 .collect(),
9372 )
9373 }
9374 };
9375 if mode == 5 {
9376 exec.in_scalar_assign -= 1;
9377 }
9378 exec.in_dq_context -= 1;
9379 result_value
9380 }
9381 2 => {
9382 // SingleQuoted: pure literal, strip outer `'…'`.
9383 let inner = if text.len() >= 2 && text.starts_with('\'') && text.ends_with('\'') {
9384 &text[1..text.len() - 1]
9385 } else {
9386 text.as_str()
9387 };
9388 Value::str(inner.to_string())
9389 }
9390 3 => {
9391 // Backquote command sub: strip outer backticks.
9392 // Word-split the result on IFS when the surrounding
9393 // word is unquoted — zsh: `print -l \`echo a b c\``
9394 // emits one arg per word. The $(…) path applies the
9395 // same split via BUILTIN_WORD_SPLIT after capture; do
9396 // the equivalent here for the `…` form.
9397 let inner = if text.len() >= 2 && text.starts_with('`') && text.ends_with('`') {
9398 &text[1..text.len() - 1]
9399 } else {
9400 text.as_str()
9401 };
9402 // Apply the live VM status before running the inner
9403 // shell so the inherited $? matches zsh's lastval
9404 // propagation.
9405 exec.set_last_status(live_status);
9406 let captured = exec.run_command_substitution(inner);
9407 let trimmed = captured.trim_end_matches('\n');
9408 if exec.in_dq_context > 0 {
9409 Value::str(trimmed.to_string())
9410 } else {
9411 let ifs = exec.scalar("IFS").unwrap_or_else(|| " \t\n".to_string());
9412 let parts: Vec<Value> = trimmed
9413 .split(|c: char| ifs.contains(c))
9414 .filter(|s| !s.is_empty())
9415 .map(|s| Value::str(s.to_string()))
9416 .collect();
9417 if parts.is_empty() {
9418 Value::str(String::new())
9419 } else if parts.len() == 1 {
9420 parts.into_iter().next().unwrap()
9421 } else {
9422 Value::Array(parts)
9423 }
9424 }
9425 }
9426 4 => {
9427 // HeredocBody: expand variables / command-subst / arith
9428 // but NOT glob or brace. Heredoc lines like `[42]` must
9429 // pass through verbatim — running them through the
9430 // default pipeline triggers NOMATCH on the literal.
9431 Value::str(crate::ported::subst::singsub(&text))
9432 }
9433 _ => {
9434 // Default (unquoted): the lexer's gettokstr already
9435 // tokenized backslash-escapes (`\$` → Bnull+$, etc).
9436 // Pass `text` through verbatim — multsub/stringsubst
9437 // recognize the markers natively. No bridge-side
9438 // re-tokenization needed.
9439 //
9440 // Mode 6 = unquoted RHS in scalar-assign context.
9441 // Pass PREFORK_ASSIGN so prefork's filesub colon-walk
9442 // fires per c:Src/exec.c:2546.
9443 let prepped: String = text.clone();
9444 if std::env::var("ZSHRS_TRACE_DEFP").is_ok() {
9445 eprintln!(
9446 "[TRACE_DEFP] text={:?} prepped={:?} mode={}",
9447 text, prepped, mode
9448 );
9449 }
9450 let pf_flags = if mode == 6 {
9451 crate::ported::zsh_h::PREFORK_ASSIGN
9452 } else {
9453 0
9454 };
9455 // c:Src/subst.c:544+ — `multsub(&prepped, 0)` is the
9456 // unquoted-argv equivalent of zsh's `prefork(list,
9457 // 0, NULL)` for a single-element list. Returns the
9458 // post-expansion node list (Vec<String>) so array-
9459 // shape results (e.g. `${a:e}`, `${a[@]}`,
9460 // `${(s::)str}`) splat into multiple argv words.
9461 // singsub() collapses to one string and discards the
9462 // splat — parity bug #28 (whole-array modifier).
9463 let (_first, nodes, _ms_ws, _ret) =
9464 crate::ported::subst::multsub(&prepped, pf_flags);
9465 if std::env::var("ZSHRS_TRACE_MULTSUB").is_ok() {
9466 eprintln!("[TRACE_MULTSUB] prepped={:?} nodes={:?}", prepped, nodes);
9467 }
9468 // c:Src/subst.c:166 — xpandbraces runs AFTER prefork's
9469 // substitution pass and BEFORE untokenize/glob. Per
9470 // word, scan for Inbrace TOKEN and expand. Words that
9471 // don't contain Inbrace TOKEN pass through unchanged.
9472 // Brace expansion is done here (inside the bridge
9473 // default arm) instead of via a post-EXPAND_TEXT
9474 // BRACE_EXPAND emit because untokenize (line below)
9475 // strips TOKEN bytes, after which the strict-TOKEN
9476 // xpandbraces gate would no longer match.
9477 let brace_ccl = opt_state_get("braceccl").unwrap_or(false);
9478 // c:Src/options.c — `no_brace_expand` (negated
9479 // `braceexpand`) gates brace expansion entirely.
9480 // When off, `{a,b}` stays literal.
9481 let brace_expand = opt_state_get("braceexpand").unwrap_or(true);
9482 let pre_brace: Vec<String> = if nodes.is_empty() {
9483 vec![String::new()]
9484 } else {
9485 nodes
9486 };
9487 let brace_expanded: Vec<String> = pre_brace
9488 .into_iter()
9489 .flat_map(|w| {
9490 if brace_expand && w.contains('\u{8f}') {
9491 crate::ported::glob::xpandbraces(&w, brace_ccl)
9492 } else {
9493 vec![w]
9494 }
9495 })
9496 .collect();
9497 // zsh stores the option as `glob` (default ON);
9498 // `setopt noglob` writes `glob=false`. Honor either
9499 // form so the dispatcher behaves the same as zsh.
9500 // Mode 7 = redirect-target word: glob only under
9501 // MULTIOS (c:Src/glob.c:2161-2167 xpandredir,
9502 // "Globbing is only done for multios.").
9503 let noglob = opt_state_get("noglob").unwrap_or(false)
9504 || opt_state_get("GLOB").map(|v| !v).unwrap_or(false)
9505 || !opt_state_get("glob").unwrap_or(true)
9506 || (mode == 7 && !opt_state_get("multios").unwrap_or(true));
9507 let parts: Vec<String> = brace_expanded
9508 .into_iter()
9509 .flat_map(|s| {
9510 // The lexer leaves glob metacharacters in their
9511 // META-encoded form: `*` → `\u{87}`, `?` →
9512 // `\u{86}`, `[` → `\u{91}`, etc. expand_string
9513 // doesn't untokenize them, so the literal-char
9514 // checks below (`s.contains('*')`) would miss
9515 // every real glob and skip expand_glob — that
9516 // bug let `echo *.toml` print the literal
9517 // `*.toml` because the META `\u{87}` never
9518 // matched the literal `*`. Untokenize once so
9519 // the metacharacter checks see the canonical
9520 // form. zsh's pattern.c expects `*` etc. as
9521 // bare chars at the glob layer.
9522 // c:Src/pattern.c:4306 haswilds on the still-
9523 // TOKENIZED word (pre-untokenize), matching C's
9524 // zglob entry gate (Src/glob.c:1230) which runs
9525 // on the lexer-tokenized string. haswilds
9526 // matches ONLY token codes: source-level
9527 // `*.toml` carries Star and fires; bare literal
9528 // `[`/`*`/`?` from `$'...'` decode, `:-`
9529 // default values, or nested-substitution
9530 // results were never shtokenize'd (C
9531 // subst.c:3231 sets globsubst=0 in the `:-`
9532 // arm) and stay literal — bug #625. Plain
9533 // multibyte text (`↔`) never matches a token
9534 // codepoint — bug #627.
9535 let is_glob_pre = !noglob && crate::ported::pattern::haswilds(&s);
9536 let s = crate::lex::untokenize(&s);
9537 // Skip glob expansion for assignment-shaped
9538 // words (`NAME=value`). zsh doesn't expand the
9539 // RHS of an assignment as a path glob unless
9540 // `setopt globassign` is set, and feeding such
9541 // words through expand_glob makes NOMATCH
9542 // (default ON) fire spuriously on
9543 // `integer i=2*3+1`, `path=*.rs`, etc.
9544 let is_assignment_shape = {
9545 let bytes = s.as_bytes();
9546 let mut i = 0;
9547 if !bytes.is_empty()
9548 && (bytes[0] == b'_' || bytes[0].is_ascii_alphabetic())
9549 {
9550 i += 1;
9551 while i < bytes.len()
9552 && (bytes[i] == b'_' || bytes[i].is_ascii_alphanumeric())
9553 {
9554 i += 1;
9555 }
9556 i < bytes.len() && bytes[i] == b'='
9557 } else {
9558 false
9559 }
9560 };
9561 // Glob-trigger decision: pre-untokenize
9562 // haswilds_tokens_only result (computed above
9563 // before the untokenize that collapses META
9564 // tokens to their ASCII forms). The TOKEN-only
9565 // gate matches C `Src/pattern.c:4306-4376`
9566 // exactly — only Inbrack/Star/Quest/Inpar/Bar/
9567 // Inang/Pound/Hat token codes count as wild,
9568 // not their literal ASCII counterparts. Source-
9569 // level `*.toml` carries Star token so globs;
9570 // `$'…'`-decoded `[abc]` carries bare `[` so
9571 // stays literal. Bug #625.
9572 if is_glob_pre && !is_assignment_shape {
9573 exec.expand_glob(&s)
9574 } else if is_assignment_shape
9575 && crate::ported::zsh_h::isset(crate::ported::zsh_h::MAGICEQUALSUBST)
9576 {
9577 // c:Src/exec.c:3353 — when MAGIC_EQUAL_SUBST is set
9578 // on a non-typeset command, esprefork = PREFORK_TYPESET,
9579 // so every NAME=value arg runs through
9580 // filesub(PREFORK_TYPESET): the `~`/`=` after the
9581 // first `=` (and after each `:`) undergo filename
9582 // expansion. `print foo=~/bar` → `foo=$HOME/bar`.
9583 // filesubstr (subst.c:741) keys on the Tilde TOKEN,
9584 // not literal `~`; this `s` was already untokenized
9585 // above, so re-tokenize (as BUILTIN_MAGIC_EQUALS_PREFORK
9586 // does) before filesub, then untokenize the result.
9587 let mut tokd = s.clone();
9588 crate::ported::glob::shtokenize(&mut tokd);
9589 let exp = crate::ported::subst::filesub(
9590 &tokd,
9591 crate::ported::zsh_h::PREFORK_TYPESET,
9592 );
9593 vec![crate::lex::untokenize(&exp).to_string()]
9594 } else {
9595 vec![s]
9596 }
9597 })
9598 .collect();
9599 if parts.len() == 1 {
9600 let only = parts.into_iter().next().unwrap_or_default();
9601 // Empty unquoted expansion → drop the arg entirely
9602 // (zsh "remove empty unquoted words" rule). Returning
9603 // an empty Value::Array makes pop_args contribute zero
9604 // items. Direct port of subst.c's empty-elide pass at
9605 // the end of multsub which removes empty linknodes
9606 // from unquoted contexts. Quoted DQ/SQ paths (modes
9607 // 1/2/5) take separate arms above and always emit
9608 // Value::Str so the empty arg survives.
9609 //
9610 // c:Src/subst.c:4437 + 1650-1656 — a word CONTAINING a
9611 // quoted span never drops: `x"${v[-1]}"y` (v empty)
9612 // is the scalar "xy", and a standalone `"${v[-1]}"`
9613 // is ONE empty arg. The lexer marks DQ/SQ spans with
9614 // Dnull(\u{9e})/Snull(\u{9d})/Qstring(\u{8c})/
9615 // Bnull(\u{9f}); their presence in the SOURCE word
9616 // means qt semantics apply. Without this gate, zpwr's
9617 // global `setopt rc_expand_param` turned autopair's
9618 // `local lchar="${LBUFFER[-1]}"` (empty prompt +
9619 // backspace) into an ARGLESS `local` — the full
9620 // parameter-table dump the user saw per keystroke.
9621 if only.is_empty() {
9622 // A quote span that WRAPS the expansion keeps the
9623 // empty arg (`"${v[-1]}"` → one empty arg). But a
9624 // quote INSIDE the `${…}` braces — e.g. the alternate
9625 // of `${x:+'q'}` or `${x:-'d'}` — does NOT: when that
9626 // branch isn't taken the result is a plain unquoted
9627 // empty and must ELIDE, matching zsh (`a=(A ${x:+'q'}
9628 // C)` → 2 elements, not 3). So only count quote
9629 // markers at brace-depth 0 (outside `${…}`). Inbrace
9630 // = \u{8f}, Outbrace = \u{90}.
9631 let mut depth = 0i32;
9632 let mut word_has_quoted_span = false;
9633 for c in text.chars() {
9634 match c {
9635 '\u{8f}' => depth += 1,
9636 '\u{90}' => depth -= 1,
9637 '\u{9e}' | '\u{9d}' | '\u{8c}' | '\u{9f}' | '"' | '\''
9638 if depth <= 0 =>
9639 {
9640 word_has_quoted_span = true;
9641 break;
9642 }
9643 _ => {}
9644 }
9645 }
9646 if word_has_quoted_span {
9647 Value::str(String::new())
9648 } else {
9649 Value::Array(Vec::new())
9650 }
9651 } else {
9652 Value::str(only)
9653 }
9654 } else {
9655 Value::Array(parts.into_iter().map(Value::str).collect())
9656 }
9657 }
9658 });
9659 // Pull any inner cmd-subst (`` `cmd` `` via mode 3 or via
9660 // mode 0/6 multsub → getoutput, `$(cmd)` via the default
9661 // arm's multsub path, nested `$()`s reached through
9662 // stringsubst) back into vm.last_status so a containing
9663 // assignment's BUILTIN_SET_VAR — which reads vm.last_status —
9664 // sees the cmd-subst's exit. Without this, backtick
9665 // assignments (`a=\`false\`; echo $?`) reported 0 because the
9666 // ported LASTVAL update never reached the VM-side counter.
9667 let cs_status = with_executor(|exec| exec.last_status());
9668 vm.last_status = cs_status;
9669 result_value
9670 });
9671
9672 // `${#name}` — pops [name]. Returns the value's element count for
9673 // arrays (indexed and assoc) or character length for scalars.
9674 // BUILTIN_PARAM_LENGTH — `${#name}`. PURE PASSTHRU.
9675 vm.register_builtin(BUILTIN_PARAM_LENGTH, |vm, _argc| {
9676 let name = vm.pop().to_str();
9677 // PARAM_LENGTH's empty-result semantics differ from
9678 // paramsubst_to_value: 0 nodes → "0" (numeric length), not
9679 // empty array. paramsubst on `${#X}` always returns at least
9680 // one node in practice (the length string); the empty case
9681 // is defensive.
9682 let mut ret_flags: i32 = 0;
9683 let (_full, _pos, nodes) = crate::ported::subst::paramsubst(
9684 &format!("${{#{}}}", name),
9685 0,
9686 false,
9687 0i32,
9688 &mut ret_flags,
9689 );
9690 if crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed) != 0 {
9691 with_executor(|exec| exec.set_last_status(1));
9692 }
9693 if nodes.is_empty() {
9694 Value::str("0")
9695 } else {
9696 nodes_to_value(nodes)
9697 }
9698 });
9699
9700 // `${var/pat/repl}` / `${var//pat/repl}` / `${var/#pat/repl}` /
9701 // `${var/%pat/repl}` — Pops [name, pattern, replacement, op_byte].
9702 // op: 0=first, 1=all, 2=anchor-prefix (`/#`), 3=anchor-suffix (`/%`).
9703 // BUILTIN_PARAM_REPLACE — `${var/pat/repl}` / `${var//pat/repl}` /
9704 // `${var/#pat/repl}` / `${var/%pat/repl}`. PURE PASSTHRU.
9705 vm.register_builtin(BUILTIN_PARAM_REPLACE, |vm, _argc| {
9706 let dq_flag = vm.pop().to_int() != 0;
9707 let op = vm.pop().to_int() as u8;
9708 let repl = vm.pop().to_str();
9709 let pattern = vm.pop().to_str();
9710 let name = vm.pop().to_str();
9711 // !!! DASH-STRICT GATE !!! dash / ash have no `${var/pat/repl}`
9712 // pattern-replacement expansion — it is a "Bad substitution" error
9713 // (bash/ksh/POSIX-sh support it, so those modes fall through). Raise
9714 // the canonical zsh diagnostic + exit 1 to match /bin/dash's failure.
9715 if crate::dash_mode::dash_strict() {
9716 crate::ported::utils::zerr("bad substitution");
9717 crate::ported::utils::errflag.fetch_or(
9718 crate::ported::zsh_h::ERRFLAG_ERROR,
9719 std::sync::atomic::Ordering::Relaxed,
9720 );
9721 with_executor(|exec| exec.set_last_status(1));
9722 return Value::str("");
9723 }
9724 // DQ context: C's lexer marks every `$` inside double quotes
9725 // as the Qstring token (Src/lex.c dquote_parse) and keeps `'`
9726 // a plain char — so a DQ replacement's `$'…'` is LITERAL in
9727 // C (Src/subst.c:301 decodes only the tokenized Snull form;
9728 // `"${a/X/$'\0'}"` keeps the five chars `$'\0'`). The body
9729 // rebuilt below re-enters stringsubst as raw text, which
9730 // would mis-decode `$'…'` as ANSI-C; stamp the Qstring
9731 // marker on the repl's `$` so stringsubst sees the same DQ
9732 // signal C's tokens carry. The PATTERN side keeps decoding
9733 // (matches observed zsh: the pattern's `$'\0'` matches a
9734 // real NUL while the repl's stays literal).
9735 let repl = if dq_flag {
9736 repl.replace('$', "\u{8c}")
9737 } else {
9738 repl
9739 };
9740 // op encoding: 0 = first `/`, 1 = all `//`, 2 = anchor-prefix
9741 // `/#`, 3 = anchor-suffix `/%`. The brace form distinguishes
9742 // first-vs-all by single vs doubled slash, and anchored by
9743 // a `#` or `%` immediately after the slash(es).
9744 let body = match op {
9745 0 => format!("${{{}/{}/{}}}", name, pattern, repl),
9746 1 => format!("${{{}//{}/{}}}", name, pattern, repl),
9747 2 => format!("${{{}/#{}/{}}}", name, pattern, repl),
9748 3 => format!("${{{}/%{}/{}}}", name, pattern, repl),
9749 _ => format!("${{{}/{}/{}}}", name, pattern, repl),
9750 };
9751 // c:Src/subst.c:1625 — paramsubst's qt flag. The compiler
9752 // threads the word's DQ context onto the stack; dropping it
9753 // (the old `let _dq_flag`) ran the rebuilt body with qt=false
9754 // whenever the opcode fired outside an EXPAND_TEXT scope, so
9755 // DQ-only semantics inside the replacement (e.g. `$'` staying
9756 // literal per Src/subst.c:301 — `"${a/x/$'\t'q}"`) were lost.
9757 // Bump in_dq_context exactly like EXPAND_TEXT mode 1 so
9758 // paramsubst_to_value's qt probe sees the right context.
9759 if dq_flag {
9760 with_executor(|exec| exec.in_dq_context += 1);
9761 }
9762 let ret = paramsubst_to_value(&body);
9763 if dq_flag {
9764 with_executor(|exec| exec.in_dq_context -= 1);
9765 }
9766 ret
9767 });
9768
9769 vm.register_builtin(BUILTIN_REGISTER_COMPILED_FN, |vm, argc| {
9770 let args = pop_args(vm, argc);
9771 let mut iter = args.into_iter();
9772 let name = iter.next().unwrap_or_default();
9773 let body_b64 = iter.next().unwrap_or_default();
9774 let body_source = iter.next().unwrap_or_default();
9775 let line_base_str = iter.next().unwrap_or_default();
9776 let line_base: i64 = line_base_str.parse().unwrap_or(0);
9777 let bytes = base64_decode(&body_b64);
9778 let status = match bincode::deserialize::<fusevm::Chunk>(&bytes) {
9779 Ok(chunk) => with_executor(|exec| {
9780 // c:Src/exec.c:5383 — `shf->filename =
9781 // ztrdup(scriptfilename);` — the function's
9782 // definition-file is read from the canonical
9783 // file-scope `scriptfilename` global at compile
9784 // time, NOT from a per-executor struct field.
9785 // exec.scriptfilename is seeded once at
9786 // bins/zshrs.rs:1717 to the bin basename ("zsh")
9787 // and never updates on source/dot, so reading from
9788 // it left every user function's def_file as "zsh".
9789 // Route through scriptfilename_get() so source /
9790 // dot's set_scriptfilename calls propagate.
9791 let def_file = crate::ported::utils::scriptfilename_get()
9792 .or_else(|| exec.scriptfilename.clone());
9793 if !body_source.is_empty() {
9794 exec.function_source
9795 .insert(name.clone(), body_source.clone());
9796 }
9797 exec.function_line_base.insert(name.clone(), line_base);
9798 exec.function_def_file.insert(name.clone(), def_file);
9799 // PFA-SMR aspect: every `name() {}` / `function name { }`
9800 // funnels through here at compile time. Emit one record
9801 // with the function name + raw body source.
9802 #[cfg(feature = "recorder")]
9803 if crate::recorder::is_enabled() {
9804 let ctx = exec.recorder_ctx();
9805 let body = if body_source.is_empty() {
9806 None
9807 } else {
9808 Some(body_source.as_str())
9809 };
9810 crate::recorder::emit_function(&name, body, ctx);
9811 }
9812 // Mirror into canonical shfunctab so scanfunctions /
9813 // ${(k)functions} / functions builtin see user defs.
9814 // C: exec.c:funcdef → shfunctab->addnode(ztrdup(name),shf).
9815 if let Ok(mut tab) = crate::ported::hashtable::shfunctab_lock().write() {
9816 let mut shf = crate::ported::hashtable::shfunc_with_body(&name, &body_source);
9817 // c:Src/exec.c:5409 — `shf->lineno = lineno;`. Use
9818 // the same max(1, line_base) clamp as the synth_shf
9819 // in vm_helper::dispatch_function_call. Bug #396.
9820 shf.lineno = std::cmp::max(1, line_base);
9821 tab.add(shf);
9822 }
9823 // c:Src/exec.c:5460-5475 — `TRAP<SIG>() { ... }` is the
9824 // function-named trap install. zsh detects the `TRAP`
9825 // prefix at func-def time and calls
9826 // `settrap(signum, NULL, ZSIG_FUNC)` so the next
9827 // dispatch of that signal routes to the named shfunc.
9828 // Bug #157 in docs/BUGS.md — fusevm_bridge's funcdef
9829 // opcode skipped this dispatch entirely, so TRAPEXIT /
9830 // TRAPUSR1 / TRAPZERR / TRAPDEBUG never fired.
9831 if name.len() > 4 && name.starts_with("TRAP") {
9832 if let Some(sn) = crate::ported::jobs::getsigidx(&name[4..]) {
9833 let _ = crate::ported::signals::settrap(
9834 sn,
9835 None,
9836 crate::ported::zsh_h::ZSIG_FUNC as i32,
9837 );
9838 }
9839 }
9840 exec.functions_compiled.insert(name, chunk);
9841 0
9842 }),
9843 Err(_) => 1,
9844 };
9845 Value::Status(status)
9846 });
9847
9848 // Wire the ShellHost so direct shell ops (Op::Glob, Op::TildeExpand,
9849 // Op::ExpandParam, Op::CmdSubst, Op::CallFunction, etc.) route through
9850 // ZshrsHost back into the executor.
9851 vm.set_shell_host(Box::new(ZshrsHost));
9852}
9853
9854impl ZshrsHost {
9855 /// True iff `c` can be a `(j:…:)` / `(s:…:)` delimiter — non-alphanumeric,
9856 /// non-underscore. Restricting to punctuation avoids `(jL)` consuming `L`
9857 /// as a delim instead of as the next flag.
9858 fn is_zsh_flag_delim(c: char) -> bool {
9859 !c.is_ascii_alphanumeric() && c != '_'
9860 }
9861}
9862
9863/// Shared `${name[idx]}` subscript dispatch for BUILTIN_ARRAY_INDEX
9864/// and the KSHARRAYS-unset arm of BUILTIN_ARRAY_INDEX_UNBRACED.
9865///
9866/// c:Src/subst.c subscript parsing — when paramsubst re-parses the
9867/// synthesized `${name[idx]}` body, characters like `'` `"` `\` `$`
9868/// etc. are LEXER-active inside the `[…]` and get reinterpreted
9869/// (quote-strip, paramsubst recursion, …). For PRE-EVALUATED key
9870/// strings (the dynamic-key fast path at compile_zsh.rs:3234 already
9871/// expanded `$k` via EXPAND_TEXT), the idx is a literal string that
9872/// must match the stored key byte-for-byte — no further
9873/// reinterpretation. Direct assoc lookup bypasses the lexer for this
9874/// case, avoiding the quote-strip bug where `h[a'b]` failed to
9875/// resolve because paramsubst's subscript lexer treated the `'` as a
9876/// quote. Bug #338. Only fires for simple assoc-name + non-flag idx
9877/// (no outer-flag sentinels, no `(…)` flag prefix on idx, no splat
9878/// operator). Other paths (slice, splat, flag-based search,
9879/// magic-assoc) still flow through paramsubst.
9880fn array_index_lookup(name: &str, idx: &str) -> Value {
9881 let idx_is_simple = !idx.starts_with('(') && idx != "@" && idx != "*" && !idx.contains(',');
9882 if idx_is_simple {
9883 // assoc_key_hit: single-lock O(1) probe — exec.assoc() clones
9884 // the WHOLE map per lookup (O(n), quadratic in shell loops).
9885 // When `name` IS an assoc, exact-key semantics apply to EVERY
9886 // plain key: hit → value, miss → empty (C `${assoc[missing]}`).
9887 // Never fall through to the textual `${name[key]}` rebuild —
9888 // keys carrying `{`/`}`/`[`/`]` (zsh-autopair probes
9889 // `${AUTOPAIR_LBOUNDS[$pair]}` with pair='{') re-parse as
9890 // broken syntax there ("failed to compile regex: repetition
9891 // quantifier…" + a `}` appended per keystroke).
9892 if let Some((_, v)) = crate::vm_helper::assoc_key_hit(name, idx) {
9893 return Value::str(v.unwrap_or_default());
9894 }
9895 }
9896 // c:Src/params.c:1449-1450 getindex — a leading `(e)`/`(E)` flag
9897 // group makes the subscript LITERAL (group consumed, exact key).
9898 // The textual rebuild below re-parses a FLAT `${name[(e)KEY]}`
9899 // string, so a `]` / `}` that arrived via `$key` expansion
9900 // terminates the subscript / brace early — "bad substitution" or
9901 // spilled-junk values (zpwr expandstats iterates alias keys
9902 // containing brackets). C never re-parses: getarg scans the
9903 // TOKENIZED source where expanded data brackets are inert. Do the
9904 // exact-match lookup directly against the assoc (plain or magic
9905 // alias tables); search groups ((r)/(i)/(k)/…) and other targets
9906 // keep the textual path.
9907 if let Some(rest) = idx.strip_prefix('(') {
9908 if let Some(close) = rest.find(')') {
9909 let grp = &rest[..close];
9910 if !grp.is_empty() && grp.chars().all(|ch| ch == 'e' || ch == 'E') {
9911 let key = &rest[close + 1..];
9912 if let Some(hit) = direct_assoc_key_get(name, key) {
9913 return Value::str(hit.unwrap_or_default());
9914 }
9915 }
9916 }
9917 }
9918 // Plain assoc key that the flat rebuild would mangle (`]` closes
9919 // the subscript, `}` closes the brace): direct lookup. On a miss
9920 // return empty — the textual fallback cannot represent the key.
9921 if (idx.contains(']') || idx.contains('}')) && !idx.starts_with('(') {
9922 if let Some(hit) = direct_assoc_key_get(name, idx) {
9923 return Value::str(hit.unwrap_or_default());
9924 }
9925 }
9926 let body = format!("${{{}[{}]}}", name, idx);
9927 paramsubst_to_value(&body)
9928}
9929
9930/// Exact-key read against an assoc-like target WITHOUT the textual
9931/// `${name[key]}` reparse (see array_index_lookup — expanded `]`/`}`
9932/// in keys break the flat form). `Some(hit)` when `name` is a target
9933/// this helper understands (plain assoc, or the alias magic assocs of
9934/// zsh/parameter — Src/Modules/parameter.c getpmalias family);
9935/// `None` = not direct-capable, caller keeps the textual path.
9936fn direct_assoc_key_get(name: &str, key: &str) -> Option<Option<String>> {
9937 use crate::ported::zsh_h::{ALIAS_GLOBAL, DISABLED};
9938 // c:Src/Modules/parameter.c:1247+ getpmalias / getpmgalias /
9939 // getpmsalias — each view filters its table by flags.
9940 let alias_view = |global: bool, suffix: bool, disabled: bool| -> Option<String> {
9941 let tab = if suffix {
9942 crate::ported::hashtable::sufaliastab_lock()
9943 } else {
9944 crate::ported::hashtable::aliastab_lock()
9945 };
9946 tab.read().ok().and_then(|t| {
9947 t.iter().find_map(|(k, a)| {
9948 let f = a.node.flags as u32;
9949 if k == key
9950 && ((f & ALIAS_GLOBAL as u32 != 0) == global || suffix)
9951 && ((f & DISABLED as u32 != 0) == disabled)
9952 {
9953 Some(a.text.clone())
9954 } else {
9955 None
9956 }
9957 })
9958 })
9959 };
9960 match name {
9961 "aliases" => Some(alias_view(false, false, false)),
9962 "galiases" => Some(alias_view(true, false, false)),
9963 "saliases" => Some(alias_view(false, true, false)),
9964 "dis_aliases" => Some(alias_view(false, false, true)),
9965 "dis_galiases" => Some(alias_view(true, false, true)),
9966 "dis_saliases" => Some(alias_view(false, true, true)),
9967 _ => {
9968 // Single-lock O(1) probe (see assoc_key_hit) — the previous
9969 // double exec.assoc() cloned the whole map twice per lookup.
9970 crate::vm_helper::assoc_key_hit(name, key).map(|(_, v)| v)
9971 }
9972 }
9973}
9974
9975/// KSHARRAYS bare-`$name` expansion words for the unbraced
9976/// no-subscript form (BUILTIN_ARRAY_INDEX_UNBRACED's KSHARRAYS arm).
9977///
9978/// - `@` / `*` stay the full positional list (the c:Src/params.c:
9979/// 2293-2296 first-element collapse is gated on
9980/// `itype_end(t, IIDENT, 1) != t` — an identifier-shaped name —
9981/// which `@`/`*` are not). The literal `[idx]` then joins the LAST
9982/// word, matching zsh 5.9: `setopt ksharrays; set -- p q;
9983/// print -- $@[0]` → `zsh:1: no matches found: q[0]`.
9984/// - Identifier-named arrays collapse to the FIRST element
9985/// (c:Src/params.c:2293-2296 `v->end = 1, v->isarr = 0`).
9986/// - Assocs collapse to the first value in scan order; the `options`
9987/// magic assoc's scan order is `OPTIONTAB` bucket order (first key
9988/// `posixargzero`), matching zsh 5.9: `emulate sh -L;
9989/// print $options[posixargzero]` → `off[posixargzero]`.
9990/// - Scalars / unset names expand to their value / empty (zsh 5.9:
9991/// `setopt ksharrays; print -- $unsetvar[0]` →
9992/// `zsh:1: no matches found: [0]`).
9993fn ksharrays_bare_words(name: &str) -> Vec<String> {
9994 if name == "@" || name == "*" {
9995 return with_executor(|exec| exec.pparams());
9996 }
9997 // Magic special-parameter lookups first — mirrors the
9998 // BUILTIN_GET_VAR precedence (partab before executor tables).
9999 if let Some(vals) = crate::vm_helper::partab_array_get(name) {
10000 return vec![vals.into_iter().next().unwrap_or_default()];
10001 }
10002 if let Some(keys) = crate::vm_helper::partab_scan_keys(name) {
10003 let v = keys
10004 .first()
10005 .and_then(|k| crate::vm_helper::partab_get(name, k))
10006 .unwrap_or_default();
10007 return vec![v];
10008 }
10009 let arr_or_assoc = with_executor(|exec| {
10010 if let Some(arr) = exec.array(name) {
10011 // c:Src/params.c:2293-2296 — first element only.
10012 return Some(arr.first().cloned().unwrap_or_default());
10013 }
10014 if let Some(map) = exec.assoc(name) {
10015 // c:Src/params.c:2351-2358 — under KSH EMULATION a bare
10016 // `$assoc` is `${assoc[0]}` (KEY-"0" lookup), EMPTY unless the
10017 // hash has a key "0": `emulate -L ksh; typeset -A h=(a 1 b 2);
10018 // print $h` is empty. Every other mode (`setopt ksharrays`,
10019 // `emulate sh`) falls through to the first bucket value below.
10020 if crate::ported::zsh_h::EMULATION(crate::ported::zsh_h::EMULATE_KSH) {
10021 return Some(map.get("0").cloned().unwrap_or_default());
10022 }
10023 // Mirrors BUILTIN_GET_VAR's bare-assoc ordering
10024 // (sorted keys, first value).
10025 let mut keys: Vec<&String> = map.keys().collect();
10026 keys.sort();
10027 return Some(
10028 keys.first()
10029 .and_then(|k| map.get(*k).cloned())
10030 .unwrap_or_default(),
10031 );
10032 }
10033 None
10034 });
10035 if let Some(v) = arr_or_assoc {
10036 return vec![v];
10037 }
10038 vec![with_executor(|exec| exec.get_variable(name))]
10039}
10040
10041/// Run `body` through `crate::ported::subst::paramsubst` and convert
10042/// the resulting node list into a fusevm `Value`. Centralises the
10043/// pattern duplicated across ~10 BUILTIN_* handlers:
10044/// - build a `${...}` body string from opcode operands
10045/// - paramsubst the body
10046/// - propagate errflag to `exec.last_status`
10047/// - delegate the LinkList → Value conversion to `nodes_to_value`
10048///
10049/// **Extension** — Rust-only helper. No direct C analog because C
10050/// zsh uses LinkList everywhere; the conversion happens at the
10051/// boundary back into the VM's stack.
10052fn paramsubst_to_value(body: &str) -> Value {
10053 // c:Src/subst.c:1625 paramsubst's `qt` flag is the C signal that
10054 // the current expansion is inside `"…"`. The fast-path bridges
10055 // (BUILTIN_PARAM_*, BUILTIN_BRIDGE_BRACE_ARRAY) used to hardcode
10056 // qt=false, which silently broke DQ-only semantics inside
10057 // `${arr:^other}` / `${arr:^^other}` (Src/subst.c:3456-3520).
10058 // The executor's `in_dq_context` counter is bumped by EXPAND_TEXT
10059 // mode 1 / mode 5 before the bridge fires, so reading it here
10060 // propagates the DQ flag without changing every bridge call site.
10061 let qt = with_executor(|exec| exec.in_dq_context > 0);
10062 let mut ret_flags: i32 = 0;
10063 let (_full, _pos, nodes) = crate::ported::subst::paramsubst(body, 0, qt, 0i32, &mut ret_flags);
10064 if crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed) != 0 {
10065 with_executor(|exec| exec.set_last_status(1));
10066 }
10067 // c:Src/lex.c untokenize — the final argv pass C runs on every
10068 // expanded word (glob.c:1862 / exec.c) DROPS the Nularg
10069 // empty-word sentinel (c:2089 `if (c != Nularg)`) that
10070 // remnulargs faithfully leaves in place (glob.c:3673 re-adds it
10071 // for all-empty results). These fast-path bridges are terminal —
10072 // their output lands directly in argv slots — so apply it here.
10073 // Without it, quoted splits with empty pieces ("${(s:|:)x}" on
10074 // "|a|b|") leak U+00A1 into argv.
10075 let nodes: Vec<String> = nodes
10076 .into_iter()
10077 .map(|n| crate::ported::lex::untokenize(&n))
10078 .collect();
10079 nodes_to_value(nodes)
10080}
10081
10082/// Wrap a `Vec<String>` (e.g. paramsubst nodes, multsub parts,
10083/// xpandbraces output) into a fusevm `Value`: 0 → empty Array, 1 →
10084/// Str, >1 → Array. Same unwrap idiom every handler that calls a
10085/// canonical Vec-returning fn does.
10086/// c:Src/subst.c:1663 — `int plan9 = isset(RCEXPANDPARAM);`
10087///
10088/// zsh calls the RC_EXPAND_PARAM word shape "plan9" after the rc(1)
10089/// shell it comes from. The option is read fresh on every expansion
10090/// (`setopt` mid-script changes the very next word), so the concat
10091/// builtins must consult it at RUNTIME, not bake it in at compile time.
10092fn plan9_active() -> bool {
10093 with_executor(|_exec| opt_state_get("rcexpandparam").unwrap_or(false))
10094}
10095
10096thread_local! {
10097 /// The `isarr` bit that `Value` cannot carry.
10098 ///
10099 /// c:Src/subst.c:4245 `if (isarr)` gates the whole array emit block,
10100 /// so plan9's word-removal rule (c:4362 `uremnode`) only ever applies
10101 /// to an ARRAY-valued expansion. An empty SCALAR has `isarr == 0`,
10102 /// takes the c:4437 scalar branch, and leaves the surrounding text
10103 /// intact — `setopt rcexpandparam; v=; print -rl -- x$v y` prints `x`
10104 /// and `y`, while the same line with an empty ARRAY prints only `y`.
10105 ///
10106 /// zshrs collapses BOTH shapes to `Value::Array(vec![])`: a real empty
10107 /// array, and an unquoted empty scalar (collapsed so a standalone `$v`
10108 /// contributes zero argv words, mirroring prefork's `uremnode` at
10109 /// c:Src/subst.c:184-187). The two are indistinguishable by the time
10110 /// they reach a concat builtin, so the expansion builtins record here
10111 /// whether the empty Array they just produced came from a scalar.
10112 ///
10113 /// Sticky until the next expansion overwrites it — a word folds its
10114 /// segments left-associatively (`concat(concat(x, $e), y)`), so the
10115 /// propagating second concat must still see the first one's bit.
10116 static EMPTY_EXPANSION_IS_SCALAR: std::cell::Cell<bool> =
10117 const { std::cell::Cell::new(false) };
10118}
10119
10120/// Record whether the empty expansion just produced was a scalar (`true`)
10121/// or a genuine array (`false`). See `EMPTY_EXPANSION_IS_SCALAR`.
10122fn note_empty_is_scalar(is_scalar: bool) {
10123 EMPTY_EXPANSION_IS_SCALAR.with(|c| c.set(is_scalar));
10124}
10125
10126/// True when the empty `Value::Array` about to be concatenated stands for
10127/// an empty SCALAR (c:4437), not an empty array (c:4362).
10128fn empty_is_scalar() -> bool {
10129 EMPTY_EXPANSION_IS_SCALAR.with(|c| c.get())
10130}
10131
10132/// c:Src/subst.c:4366-4437 — the NON-plan9 arm of paramsubst's array
10133/// emit block: "simply join the first and last values."
10134///
10135/// The word prefix (`ostr..aptr`) is concatenated onto element 0
10136/// (c:4386 `strcatsub(&y, ostr, aptr, x, xlen, NULL, …)`), the interior
10137/// elements are emitted bare (c:4393-4412), and the word suffix (`fstr`)
10138/// is concatenated onto the final element (c:4414-4429). Applied left-
10139/// associatively across a word's segments this reproduces zsh's
10140/// `pre${arr}post` → `prep` / `q` / `rpost`.
10141///
10142/// An EMPTY array never reaches this arm in C: c:4261
10143/// `if ((!aval[0] || !aval[1]) && !plan9)` collapses it to the scalar ""
10144/// first, so the surrounding text survives as one word (`x$e y` → `x`).
10145/// That is what the empty-Array arms below reproduce.
10146fn concat_splice(lhs: Value, rhs: Value) -> Value {
10147 match (lhs, rhs) {
10148 (Value::Array(mut la), Value::Array(ra)) => {
10149 if la.is_empty() {
10150 return Value::Array(ra);
10151 }
10152 if ra.is_empty() {
10153 return Value::Array(la);
10154 }
10155 // Last of la merges with first of ra; rest unchanged.
10156 let last_l = la.pop().unwrap();
10157 let mut ra_iter = ra.into_iter();
10158 let first_r = ra_iter.next().unwrap();
10159 let l_s = last_l.as_str_cow();
10160 let r_s = first_r.as_str_cow();
10161 let mut merged = String::with_capacity(l_s.len() + r_s.len());
10162 merged.push_str(&l_s);
10163 merged.push_str(&r_s);
10164 la.push(Value::str(merged));
10165 la.extend(ra_iter);
10166 Value::Array(la)
10167 }
10168 (Value::Array(mut la), rhs_scalar) => {
10169 // c:4261 — empty array + empty surrounding text is zero
10170 // words, not one empty word. Bug #120 in docs/BUGS.md:
10171 // `b=("${a[@]:0:-1}")` gave len=1 instead of zsh's len=0.
10172 let rhs_s = rhs_scalar.as_str_cow();
10173 if la.is_empty() {
10174 if rhs_s.is_empty() {
10175 return Value::Array(Vec::new());
10176 }
10177 return Value::str(rhs_s.to_string());
10178 }
10179 let last = la.pop().unwrap();
10180 let l_s = last.as_str_cow();
10181 let mut s = String::with_capacity(l_s.len() + rhs_s.len());
10182 s.push_str(&l_s);
10183 s.push_str(&rhs_s);
10184 la.push(Value::str(s));
10185 Value::Array(la)
10186 }
10187 (lhs_scalar, Value::Array(mut ra)) => {
10188 let lhs_s = lhs_scalar.as_str_cow();
10189 if ra.is_empty() {
10190 // Symmetric c:4261 empty-array rule; see the arm above.
10191 if lhs_s.is_empty() {
10192 return Value::Array(Vec::new());
10193 }
10194 return Value::str(lhs_s.to_string());
10195 }
10196 let first = ra.remove(0);
10197 let r_s = first.as_str_cow();
10198 let mut s = String::with_capacity(lhs_s.len() + r_s.len());
10199 s.push_str(&lhs_s);
10200 s.push_str(&r_s);
10201 let mut out = Vec::with_capacity(ra.len() + 1);
10202 out.push(Value::str(s));
10203 out.extend(ra);
10204 Value::Array(out)
10205 }
10206 (lhs_s, rhs_s) => {
10207 let l = lhs_s.as_str_cow();
10208 let r = rhs_s.as_str_cow();
10209 let mut s = String::with_capacity(l.len() + r.len());
10210 s.push_str(&l);
10211 s.push_str(&r);
10212 Value::str(s)
10213 }
10214 }
10215}
10216
10217/// c:Src/subst.c:4316-4365 — the plan9 (RC_EXPAND_PARAM) arm of
10218/// paramsubst's array emit block.
10219///
10220/// Every element gets the FULL word prefix and suffix
10221/// (c:4341 `strcatsub(&y, ostr, aptr, x, xlen, y + 1, …)` inside the
10222/// per-element loop), giving the cross product with the surrounding
10223/// text: `pre${arr}post` → `preppost` / `preqpost` / `prerpost`.
10224///
10225/// An EMPTY array removes the WHOLE word: the c:4327
10226/// `while ((x = *aval++))` loop body never runs, so `plan9` is still
10227/// non-zero at c:4362 and the node is deleted —
10228/// `if (plan9) { uremnode(l, n); return n; }` (c:4362-4365). `e=();
10229/// setopt RC_EXPAND_PARAM; print -rl -- x$e y` prints only `y`. This is
10230/// the opposite of the non-plan9 rule at c:4261, which keeps `x`.
10231fn concat_plan9(lhs: Value, rhs: Value) -> Value {
10232 // c:4245 `if (isarr)` — an empty SCALAR never enters the array emit
10233 // block, so it contributes "" and the word survives (c:4437). Only a
10234 // real empty ARRAY reaches c:4362's `uremnode`. `Value` cannot tell
10235 // the two apart; EMPTY_EXPANSION_IS_SCALAR carries the missing bit.
10236 let scalar_empty = empty_is_scalar();
10237 match (lhs, rhs) {
10238 // c:4362-4365 — an empty array on either side deletes the word.
10239 // Propagated as an empty Array so a later concat in the same word
10240 // (`x${e[@]}y` folds twice) keeps the word deleted; pop_args
10241 // splats an empty Array into zero argv words.
10242 (Value::Array(la), rhs_v) if la.is_empty() => {
10243 if scalar_empty {
10244 // Empty scalar prefix: "" + rhs (c:4437 strcatsub).
10245 return match rhs_v {
10246 Value::Array(ra) if ra.is_empty() => Value::Array(Vec::new()),
10247 other => other,
10248 };
10249 }
10250 Value::Array(Vec::new())
10251 }
10252 (lhs_v, Value::Array(ra)) if ra.is_empty() => {
10253 if scalar_empty {
10254 // Empty scalar suffix: lhs + "" (c:4437 strcatsub).
10255 return lhs_v;
10256 }
10257 Value::Array(Vec::new())
10258 }
10259 (Value::Array(la), Value::Array(ra)) => {
10260 let mut out = Vec::with_capacity(la.len() * ra.len());
10261 for a in &la {
10262 let a_s = a.as_str_cow();
10263 for b in &ra {
10264 let b_s = b.as_str_cow();
10265 let mut s = String::with_capacity(a_s.len() + b_s.len());
10266 s.push_str(&a_s);
10267 s.push_str(&b_s);
10268 out.push(Value::str(s));
10269 }
10270 }
10271 Value::Array(out)
10272 }
10273 (Value::Array(la), rhs_scalar) => {
10274 let r = rhs_scalar.as_str_cow();
10275 let out: Vec<Value> = la
10276 .into_iter()
10277 .map(|a| {
10278 let a_s = a.as_str_cow();
10279 let mut s = String::with_capacity(a_s.len() + r.len());
10280 s.push_str(&a_s);
10281 s.push_str(&r);
10282 Value::str(s)
10283 })
10284 .collect();
10285 Value::Array(out)
10286 }
10287 (lhs_scalar, Value::Array(ra)) => {
10288 let l = lhs_scalar.as_str_cow();
10289 let out: Vec<Value> = ra
10290 .into_iter()
10291 .map(|b| {
10292 let b_s = b.as_str_cow();
10293 let mut s = String::with_capacity(l.len() + b_s.len());
10294 s.push_str(&l);
10295 s.push_str(&b_s);
10296 Value::str(s)
10297 })
10298 .collect();
10299 Value::Array(out)
10300 }
10301 (lhs_s, rhs_s) => {
10302 // Both scalar: nothing to distribute (c:4444 scalar branch).
10303 let l = lhs_s.as_str_cow();
10304 let r = rhs_s.as_str_cow();
10305 let mut s = String::with_capacity(l.len() + r.len());
10306 s.push_str(&l);
10307 s.push_str(&r);
10308 Value::str(s)
10309 }
10310 }
10311}
10312
10313fn nodes_to_value(nodes: Vec<String>) -> Value {
10314 // c:Src/glob.c:3649 remnulargs — strip the Nularg (`\u{a1}`)
10315 // sentinel and other INULL bytes that paramsubst's splat block
10316 // emits for empty array elements (so prefork's empty-node-delete
10317 // pass doesn't drop them). Downstream consumers (cond `-z`/`-n`,
10318 // command args, etc.) must see the post-remnulargs strings. Bug
10319 // #185 in docs/BUGS.md: `[[ -z "${b[@]}" ]]` for b=("") returned
10320 // false because the leftover `\u{a1}` had StringLen=1.
10321 let stripped: Vec<String> = nodes
10322 .into_iter()
10323 .map(|mut s| {
10324 crate::ported::glob::remnulargs(&mut s);
10325 s
10326 })
10327 .collect();
10328 if stripped.is_empty() {
10329 // Zero nodes = an ARRAY-shaped expansion that produced no words
10330 // (empty array splat, empty slice). c:4245 `if (isarr)` holds, so
10331 // plan9 deletes the surrounding word (c:4362).
10332 note_empty_is_scalar(false);
10333 Value::Array(Vec::new())
10334 } else if stripped.len() == 1 {
10335 let only = stripped.into_iter().next().unwrap();
10336 // c:Src/subst.c:183-186 — `else if (!(flags & PREFORK_SINGLE)
10337 // && !(*ret_flags & PREFORK_KEY_VALUE) && !keep)
10338 // uremnode(list, node);`
10339 // C zsh's prefork removes empty linknodes from the result
10340 // list when in non-SINGLE (argv-context) mode. The ported
10341 // prefork at subst.rs:388-396 honors the same delete-empty
10342 // pass, but some paramsubst paths land here with a single-
10343 // empty-string Vec instead of an empty Vec (paramsubst's
10344 // slice / substring / parameter-flag branches allocate a
10345 // result before checking emptiness). Mirror the prefork
10346 // drop at this layer: single-empty under !in_dq_context
10347 // collapses to Value::Array(empty), and pop_args (line 6243)
10348 // splats the empty Array → zero argv words. DQ context
10349 // (in_dq_context > 0) keeps the empty string so
10350 // `echo "${UNSET}"` still produces an empty arg per zsh's
10351 // quoting rules (c:Src/subst.c:1650-1656 isarr comment).
10352 if only.is_empty() {
10353 let in_dq = with_executor(|exec| exec.in_dq_context > 0);
10354 if !in_dq {
10355 // One empty node = a SCALAR-shaped empty result (c:4437),
10356 // not an empty array — see EMPTY_EXPANSION_IS_SCALAR.
10357 note_empty_is_scalar(true);
10358 return Value::Array(Vec::new());
10359 }
10360 }
10361 Value::str(only)
10362 } else {
10363 Value::Array(stripped.into_iter().map(Value::str).collect())
10364 }
10365}
10366
10367fn pop_args(vm: &mut fusevm::VM, argc: u8) -> Vec<String> {
10368 let mut popped: Vec<Value> = Vec::with_capacity(argc as usize);
10369 for _ in 0..argc {
10370 popped.push(vm.pop());
10371 }
10372 popped.reverse();
10373 let mut args: Vec<String> = Vec::with_capacity(popped.len());
10374 for v in popped {
10375 match v {
10376 Value::Array(items) => {
10377 for item in items {
10378 args.push(item.to_str());
10379 }
10380 }
10381 other => args.push(other.to_str()),
10382 }
10383 }
10384 // `expand_glob` set the glob-failed cell when a no-match glob
10385 // triggered nomatch (c:Src/glob.c:1877). Signal the failure via
10386 // last_status + the per-command glob_failed cell; the dispatcher
10387 // (`host_exec_external`) consumes + clears it and returns status 1
10388 // without running the command body.
10389 if with_executor(|exec| exec.current_command_glob_failed.get()) {
10390 with_executor(|exec| exec.set_last_status(1));
10391 }
10392 // `$_` tracks the last argument of the PREVIOUSLY executed
10393 // command (zsh / bash convention). Promote the deferred value
10394 // into `$_` BEFORE this command runs (so `echo $_` reads the
10395 // prior command's last arg) then stash THIS command's last arg
10396 // for the next dispatch.
10397 let new_last = args.last().cloned();
10398 with_executor(|exec| {
10399 if let Some(prev) = exec.pending_underscore.take() {
10400 exec.set_scalar("_".to_string(), prev);
10401 }
10402 if let Some(last) = new_last {
10403 exec.pending_underscore = Some(last);
10404 }
10405 });
10406 args
10407}
10408
10409/// zsh dispatch order is alias → function → builtin → external. The
10410/// compiler emits direct CallBuiltin ops for known builtin names for
10411/// perf, which silently skips a user function that shadows the same
10412/// name (e.g. `echo() { ... }; echo hi` would run the C builtin
10413/// without this check). Returns Some(status) when the call is routed
10414/// to the user function; the builtin handler should fall through to
10415/// its native impl when None.
10416/// Fork+exec a system binary by name. Used by `reg_overridable!` as
10417/// the fall-through path when `[builtins].coreutils_shadows = off`
10418/// (the default) — runs the canonical `/bin/X` instead of zshrs's
10419/// in-process shadow so old scripts hit zero behavioral divergence.
10420///
10421/// Inherits stdin/stdout/stderr from the parent so pipelines work
10422/// transparently. Resolves the binary via PATH; mirrors what zsh's
10423/// own external-command dispatch would do. Returns the child's exit
10424/// status (or 127 if PATH lookup fails — the standard "command not
10425/// found" code).
10426/// RAII guard that queues signals for the lifetime of a synchronous
10427/// foreground `waitpid` (via `std::process::Command::status`/`wait`).
10428///
10429/// zshrs installs a process-wide SIGCHLD handler (`zhandler` →
10430/// `wait_for_processes` → `waitpid(-1, WNOHANG)`) that reaps EVERY
10431/// exited child to drive the job table. `std::process::Command` does
10432/// its own targeted `waitpid(pid)`; when the reaper fires on any
10433/// thread between the fork and that wait, it reaps the child first and
10434/// `Command::status()` fails with ECHILD ("No child processes (os
10435/// error 10)"). This surfaced as `zshrs: hostname: No child processes
10436/// (os error 10)` when a coreutils shadow (`coreutils_shadows = off`
10437/// default) fork-execs `/usr/bin/hostname` while a background prewarm
10438/// child exits at the same instant.
10439///
10440/// Holding this guard bumps `queueing_enabled` (a global SeqCst atomic
10441/// that `zhandler` honors on every thread), so a SIGCHLD arriving
10442/// during the wait is pushed onto the deferred queue instead of being
10443/// reaped — `Command::status()` reaps its own child and reads the real
10444/// status. On drop, `unqueue_signals()` drains the queue, so any
10445/// genuine background children that exited meanwhile still get reaped
10446/// and routed to the job table. This is the same queue_signals /
10447/// unqueue_signals fencing zsh uses around its own foreground waits
10448/// (Src/exec.c). Panic-safe via `Drop`.
10449pub(crate) struct ForegroundWaitGuard;
10450
10451impl ForegroundWaitGuard {
10452 #[inline]
10453 pub(crate) fn enter() -> Self {
10454 crate::ported::signals_h::queue_signals();
10455 ForegroundWaitGuard
10456 }
10457}
10458
10459impl Drop for ForegroundWaitGuard {
10460 #[inline]
10461 fn drop(&mut self) {
10462 crate::ported::signals_h::unqueue_signals();
10463 }
10464}
10465
10466fn exec_system_command(name: &str, args: &[String]) -> i32 {
10467 // c:Src/jobs.c — count the fork so `time` reports for an
10468 // overridable coreutils shadow run as an external (`time sleep 0`,
10469 // `time cat …`). This is a distinct spawn path from
10470 // execute_external_bg; without the bump BUILTIN_TIME_SUBLIST saw no
10471 // job and stayed silent. (Builtins that don't reach a spawn never
10472 // hit this fn.)
10473 crate::vm_helper::FORK_EVENTS.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
10474 // Queue signals across the wait so the SIGCHLD reaper can't steal
10475 // this child out from under Command::status — see ForegroundWaitGuard.
10476 let status = {
10477 let _wait_guard = ForegroundWaitGuard::enter();
10478 std::process::Command::new(name)
10479 .args(args)
10480 .stdin(std::process::Stdio::inherit())
10481 .stdout(std::process::Stdio::inherit())
10482 .stderr(std::process::Stdio::inherit())
10483 .status()
10484 };
10485 match status {
10486 Ok(s) => s.code().unwrap_or(if s.success() { 0 } else { 1 }),
10487 Err(e) => {
10488 eprintln!("zshrs: {}: {}", name, e);
10489 127
10490 }
10491 }
10492}
10493
10494fn try_user_fn_override(name: &str, args: &[String]) -> Option<i32> {
10495 let has_fn = with_executor(|exec| {
10496 exec.functions_compiled.contains_key(name) || exec.function_exists(name)
10497 });
10498 if !has_fn {
10499 return None;
10500 }
10501 Some(with_executor(|exec| {
10502 exec.dispatch_function_call(name, args).unwrap_or(127)
10503 }))
10504}
10505
10506/// Builtin ID for `${name}` reads — routes through canonical
10507/// `getsparam` (Src/params.c:3076) via paramtab + env walk so nested
10508/// VMs (function calls) see the same storage.
10509pub const BUILTIN_GET_VAR: u16 = 283;
10510
10511/// Like `BUILTIN_GET_VAR` but forces double-quoted (DQ) semantics on
10512/// the read regardless of the runtime `in_dq_context`. The compiler
10513/// emits this for a QUOTED simple-var read (`"$name"`) — those compile
10514/// to a direct GET_VAR with no EXPAND_TEXT wrapper, so `in_dq_context`
10515/// is 0 and the plain GET_VAR would wrongly word-elide an array's empty
10516/// elements (`a=(1 "" 3); "$a"` must keep the empty → `1 3`, not
10517/// `1 3`). With force_dq the array joins via sepjoin keeping empties and
10518/// a scalar is returned verbatim (no empty-drop, no SH_WORD_SPLIT).
10519pub const BUILTIN_GET_VAR_DQ: u16 = 639;
10520
10521/// Builtin ID for `name=value` assignments — pops [name, value] and
10522/// routes through canonical `setsparam` (Src/params.c:3350).
10523pub const BUILTIN_SET_VAR: u16 = 284;
10524
10525/// Builtin ID that sets the thread-local [`SET_VAR_GLOB_ELIGIBLE`] flag true.
10526/// Emitted by the compiler immediately before a `BUILTIN_SET_VAR` whose scalar
10527/// RHS carried an UNQUOTED glob token, so the runtime knows the RHS is a literal
10528/// glob pattern eligible for GLOB_ASSIGN. Takes no stack args, pushes nothing.
10529pub const BUILTIN_MARK_GLOB_ELIGIBLE: u16 = 640;
10530
10531/// Builtin ID for pipeline execution. Pops N sub-chunk indices from the stack;
10532/// each index points into `vm.chunk.sub_chunks` (compiled stage bodies). Forks
10533/// N children, wires stdin/stdout between them via pipes, runs each stage's
10534/// bytecode on a fresh VM in its child, parent waits for all and pushes the
10535/// last stage's exit status. This is bytecode-native pipeline execution —
10536/// no tree-walker delegation.
10537pub const BUILTIN_RUN_PIPELINE: u16 = 285;
10538
10539/// Builtin ID for `Array → String` joining. Pops one value: if it's an Array,
10540/// joins its string-coerced elements with a single space; otherwise passes
10541/// through. Used after `Op::Glob` to convert the pattern's matched paths into
10542/// the single argv-token form the bytecode word model expects (no per-word
10543/// splitting yet — that's a future phase).
10544pub const BUILTIN_ARRAY_JOIN: u16 = 286;
10545
10546/// Builtin ID for `cmd &` background execution. IDs 287/288/289 are reserved
10547/// for the planned array work in Phase G1 (SET_ARRAY/SET_ASSOC/ARRAY_INDEX),
10548/// so this lands at 290. Pops the sub-chunk index then the job text; forks;
10549/// child detaches (`setsid`), runs the sub-chunk on a fresh VM, exits with
10550/// last_status; parent registers the job in the canonical JOBTAB
10551/// (initjob/addproc/spawnjob per c:Src/exec.c:1700-1758) so `jobs` / `wait
10552/// %N` / `kill %N` / `disown` and the zsh/parameter assocs all see it, then
10553/// returns Status(0) immediately.
10554pub const BUILTIN_RUN_BG: u16 = 290;
10555
10556/// Indexed-array assignment: `arr=(a b c)`. Compile_simple emits N element
10557/// pushes followed by name push, then `CallBuiltin(BUILTIN_SET_ARRAY, N+1)`.
10558/// The handler pops args (last popped = name in our pushing order) and stores
10559/// `Vec<String>` into `executor.arrays`. Tree-walker callers see the same
10560/// storage. Any prior scalar binding in `executor.variables` for `name` is
10561/// removed so `${name}` (scalar context) consistently reflects the array's
10562/// first element via `get_variable`.
10563pub const BUILTIN_SET_ARRAY: u16 = 287;
10564
10565/// Single-key set on an associative array: `foo[key]=val`. Stack (top-down):
10566/// [name, key, value]. Stores `value` into `executor.assoc_arrays[name][key]`,
10567/// creating the outer entry if missing. compile_simple detects `var[...]=...`
10568/// in assignments and emits this builtin.
10569pub const BUILTIN_SET_ASSOC: u16 = 288;
10570
10571/// `${arr[idx]}` — single-element array index. Pops two args:
10572/// stack: [name, idx_str]
10573/// Returns the indexed element as Value::str. Indexing semantics: zsh is
10574/// 1-based by default; bash is 0-based. We follow zsh.
10575/// Special idx values: `@` and `*` return the whole array as Value::Array
10576/// (which fuses correctly via the Op::Exec splice for argv splice).
10577pub const BUILTIN_ARRAY_INDEX: u16 = 289;
10578
10579/// `${#arr[@]}` and `${#arr}` (when arr is an array name) — array length.
10580/// Pops one arg: name. Returns Value::str of len.
10581
10582/// `${arr[@]}` — splice all elements as a Value::Array. Pops one arg: name.
10583/// The Array gets flattened by Op::Exec/ExecBg/CallFunction into argv.
10584pub const BUILTIN_ARRAY_ALL: u16 = 292;
10585
10586/// Flatten one level of Value::Array nesting. Pops N values; for each, if it's
10587/// a Value::Array, its elements are appended directly; otherwise the value is
10588/// appended as-is. Pushes a single Value::Array of the flattened result. Used
10589/// by the for-loop word-list compile path: when a word like `${arr[@]}`
10590/// produces a nested Array, this lets `for i in ${arr[@]}` iterate over the
10591/// inner elements rather than the outer single-element array.
10592pub const BUILTIN_ARRAY_FLATTEN: u16 = 293;
10593
10594/// `coproc [name] { body }` — bidirectional pipe to async child. Pops a name
10595/// (optional, "" for default) and a sub-chunk index. Creates two pipes, forks,
10596/// child redirects its fd 0/1 to the inner ends and runs the body, parent
10597/// stores [write_fd, read_fd] into the named array (default `COPROC`). Caller
10598/// closes the fds and `wait`s when done. Job-table integration deferred to
10599/// Phase G6 alongside the bg `&` work.
10600pub const BUILTIN_RUN_COPROC: u16 = 294;
10601
10602/// `arr+=(d e f)` — append N elements to an existing indexed array. Compile
10603/// emits N element pushes + name push, then `CallBuiltin(295, N+1)`. Handler
10604/// drains args (last popped = name), extends `executor.arrays[name]` (creates
10605/// the entry if missing). Mirrors zsh's `+=` semantics for indexed arrays.
10606pub const BUILTIN_APPEND_ARRAY: u16 = 295;
10607
10608/// `name[@]=(...)` / `name[*]=(...)` whole-array SET. Identical to
10609/// BUILTIN_SET_ARRAY for an indexed array / scalar (whole replace), but
10610/// rejects an associative target with "attempt to set slice of
10611/// associative array" (c:Src/params.c:3324-3327).
10612pub const BUILTIN_SET_ARRAY_AT: u16 = 633;
10613
10614/// `name[@]+=(...)` / `name[*]+=(...)` whole-array APPEND. Indexed
10615/// append (push), assoc target → same slice-of-assoc error as 633.
10616pub const BUILTIN_APPEND_ARRAY_AT: u16 = 634;
10617
10618/// `select var in words; do body; done` — interactive numbered-menu loop.
10619/// Compile emits N word pushes + var-name push + sub-chunk index push, then
10620/// `CallBuiltin(296, N+2)`. Handler prints `1) word1\n2) word2\n...` to
10621/// stderr, prints `$PROMPT3` (default `?# `) to stderr, reads a line from
10622/// stdin. On EOF returns 0. On a valid 1-based number, sets `var` to the
10623/// chosen word, runs the sub-chunk, then redisplays the menu and loops. On
10624/// invalid input redraws the menu without running the body. `break` from
10625/// inside the body exits the loop (handled by the body's own bytecode).
10626pub const BUILTIN_RUN_SELECT: u16 = 296;
10627
10628/// `m[k]+=value` — append onto an existing assoc-array value (string concat).
10629/// If the key doesn't exist, behaves like SET_ASSOC. Stack: [name, key, value].
10630
10631/// `break` from inside a body that runs on a sub-VM (select, future
10632/// loop-via-builtin constructs). Writes the canonical
10633/// `crate::ported::builtin::BREAKS` atomic (port of `Src/loop.c:46
10634/// breaks`). Outer-loop builtins drain BREAKS/CONTFLAG after each
10635/// body run, matching the loop.c:529-534 drain pattern.
10636pub const BUILTIN_SET_BREAK: u16 = 299;
10637
10638/// `continue` from inside a sub-VM body. Sets CONTFLAG=1 + bumps
10639/// BREAKS, matching `bin_break`'s WC_CONTINUE arm at Src/builtin.c
10640/// c:5836 `contflag = 1; FALLTHROUGH; breaks++;`.
10641pub const BUILTIN_SET_CONTINUE: u16 = 300;
10642
10643/// Brace expansion: `{a,b,c}` → 3 values, `{1..5}` → 5 values, `{01..05}` →
10644/// zero-padded numerics, `{a..e}` → letter range. Pops one string, returns
10645/// Value::Array of expansions (empty array → original string preserved).
10646pub const BUILTIN_BRACE_EXPAND: u16 = 301;
10647
10648/// Glob qualifier filter: `*(qualifier)` filters glob results by predicate.
10649/// Pops [pattern, qualifier_string]. Returns Value::Array of matching paths.
10650
10651/// Re-export the regex_match host method as a builtin so `[[ s =~ pat ]]`
10652/// works even when fusevm's Op::RegexMatch isn't routed (compat fallback).
10653
10654/// Word-split a string on IFS (default: whitespace). Pops one string,
10655/// returns Value::Array of fields. Used in array-literal context where
10656/// `arr=($(cmd))` should expand cmd's stdout into multiple elements.
10657pub const BUILTIN_WORD_SPLIT: u16 = 304;
10658
10659/// `${=name}` / SH_WORD_SPLIT forced IFS split — c:Src/subst.c:3920-3928
10660/// `aval = sepsplit(val, spsep, 0, 1);`.
10661///
10662/// Unlike BUILTIN_WORD_SPLIT (which routes through `multsub`'s
10663/// PREFORK_SPLIT walker — the c:553-620 loop that COLLAPSES runs of
10664/// separators and never emits an empty field), this is the *other* zsh
10665/// splitter: `sepsplit` → `Src/utils.c:3711 spacesplit(s, allownull=0)`,
10666/// which distinguishes the two IFS classes and preserves empty fields.
10667///
10668/// Stack: \[value\]. argc selects the empty-field rule:
10669/// * argc == 0 — the expansion is a bare unquoted word (`${=v}`): the
10670/// leading/trailing `""` fields spacesplit emits for skipped
10671/// IFS-WHITESPACE are empty argv words and prefork deletes them
10672/// (c:Src/subst.c:184-187 `uremnode`).
10673/// * argc == 1 — the expansion is quoted (`"${=v}"`) or has adjacent
10674/// word segments (`x${=v}y`): those fields survive, because in C the
10675/// word's `Dnull` quote markers / literal prefix+suffix attach to the
10676/// first and last elements (c:4386 / c:4429 strcatsub) and the node is
10677/// no longer empty. `v=$' a:b '` → `""`, `a:b`, `""` quoted; `x`,
10678/// `a:b`, `y` with surrounding text.
10679///
10680/// Empty fields that come from an IFS-NON-whitespace separator are the
10681/// `nulstring` (`Nularg`, c:Src/subst.c:36) and survive in BOTH cases —
10682/// `IFS=x; v=xaxbx` splits to `""`, `a`, `b`, `""` quoted or not.
10683pub const BUILTIN_FORCE_SPLIT: u16 = 643;
10684
10685/// Register a pre-compiled fusevm chunk as a function. Stack: [name,
10686/// base64-bincode-of-Chunk]. Used by compile_zsh's compile_funcdef to
10687/// register functions parsed via parse_init+parse without going through the
10688/// ShellCommand JSON serialization path.
10689pub const BUILTIN_REGISTER_COMPILED_FN: u16 = 305;
10690/// `BUILTIN_VAR_EXISTS` constant.
10691pub const BUILTIN_VAR_EXISTS: u16 = 306;
10692/// Native param-modifier builtins. Each takes a fixed argv shape and
10693/// returns the modified value as Value::Str.
10694///
10695/// `${var:-default}` / `${var:=default}` / `${var:?error}` / `${var:+alt}`
10696/// — pop [name, op_byte, rhs]. op_byte: 0=`:-`, 1=`:=`, 2=`:?`, 3=`:+`.
10697pub const BUILTIN_PARAM_DEFAULT_FAMILY: u16 = 307;
10698/// `${var:offset[:length]}` — pop [name, offset, length] (length=-1 means
10699/// "rest of value"; negative offset counts from end).
10700pub const BUILTIN_PARAM_SUBSTRING: u16 = 308;
10701/// `${var#pat}` / `${var##pat}` / `${var%pat}` / `${var%%pat}` — pop
10702/// [name, pattern, op_byte]. op_byte: 0=`#`, 1=`##`, 2=`%`, 3=`%%`.
10703pub const BUILTIN_PARAM_STRIP: u16 = 309;
10704/// `${var/pat/repl}` / `${var//pat/repl}` / `${var/#pat/repl}` /
10705/// `${var/%pat/repl}` — pop [name, pattern, replacement, op_byte].
10706/// op_byte: 0=first, 1=all, 2=anchor-prefix, 3=anchor-suffix.
10707pub const BUILTIN_PARAM_REPLACE: u16 = 310;
10708/// `${#name}` — character length of a scalar value, or element count
10709/// of an indexed/assoc array. Pops \[name\], returns count as Value::Str.
10710pub const BUILTIN_PARAM_LENGTH: u16 = 311;
10711/// `$((expr))` arithmetic substitution. Pops \[expr_string\], evaluates
10712/// via the executor's MathEval (integer-aware), returns result as
10713/// Value::Str. Bypasses ArithCompiler's float-only Op::Div path so
10714/// `$((10/3))` returns "3" not "3.333...".
10715pub const BUILTIN_ARITH_EVAL: u16 = 312;
10716/// `(( ... ))` math command post-eval status hook. Pops nothing,
10717/// pushes Value::Status. If errflag is set (math error in the
10718/// preceding BUILTIN_ARITH_EVAL call), clears it and emits status=2
10719/// matching c:Src/math.c arith-failure semantics. Otherwise emits
10720/// the current vm.last_status. Used by compile_arith's `(( ... ))`
10721/// path so the math command swallows errors without halting the
10722/// script — `$((... ))` substitutions skip this hook so their
10723/// errflag propagates up to the containing command.
10724pub const BUILTIN_ARITH_CMD_FINISH: u16 = 527;
10725/// `$(cmd)` command substitution. Pops \[cmd_string\], runs through
10726/// `run_command_substitution` which compiles via parse_init+parse + ZshCompiler
10727/// and captures stdout via an in-process pipe. Returns trimmed output
10728/// as Value::Str. Avoids the sub-chunk word-emit quoting bug in the
10729/// raw Op::CmdSubst path.
10730pub const BUILTIN_CMD_SUBST_TEXT: u16 = 313;
10731/// Text-based word expansion. Pops \[preserved_text\]: the word with
10732/// quotes preserved (Dnull→`"`, Snull→`'`, Bnull→`\`), runs
10733/// `expand_string` (variable + cmd-sub + arith) then `xpandbraces`
10734/// then `expand_glob`. Returns Value::str (single match) or
10735/// Value::Array (multi-match brace/glob).
10736pub const BUILTIN_EXPAND_TEXT: u16 = 314;
10737
10738/// `[[ a -ef b ]]` — same-inode test. Stack: [a, b]. Pushes Bool true iff
10739/// both paths resolve to the same `(dev, inode)` pair (zsh + bash semantics).
10740pub const BUILTIN_SAME_FILE: u16 = 315;
10741
10742/// `[[ a -nt b ]]` — file `a` newer than file `b` (mtime strict).
10743/// Stack: [path_a, path_b]. Pushes Bool. zsh-compatible "missing"
10744/// rules: if both exist, compare mtime; if only `a` exists → true;
10745/// otherwise false.
10746pub const BUILTIN_FILE_NEWER: u16 = 324;
10747
10748/// `[[ a -ot b ]]` — mirror of `-nt`. If both exist, compare mtime;
10749/// if only `b` exists → true; otherwise false.
10750pub const BUILTIN_FILE_OLDER: u16 = 325;
10751
10752/// `[[ -k path ]]` — sticky bit (S_ISVTX) set on path.
10753pub const BUILTIN_HAS_STICKY: u16 = 326;
10754/// `[[ -u path ]]` — setuid bit (S_ISUID).
10755pub const BUILTIN_HAS_SETUID: u16 = 327;
10756/// `[[ -g path ]]` — setgid bit (S_ISGID).
10757pub const BUILTIN_HAS_SETGID: u16 = 328;
10758/// `[[ -O path ]]` — owned by effective UID.
10759pub const BUILTIN_OWNED_BY_USER: u16 = 329;
10760/// `[[ -G path ]]` — owned by effective GID.
10761pub const BUILTIN_OWNED_BY_GROUP: u16 = 330;
10762/// `[[ -N path ]]` — file modified since last accessed (atime <= mtime).
10763pub const BUILTIN_FILE_MODIFIED_SINCE_ACCESS: u16 = 341;
10764
10765/// `name+=val` (no parens) — runtime-dispatched append.
10766/// If name is an indexed array → push val as element.
10767/// If name is an assoc array → error (zsh requires `(k v)` form).
10768/// Else → scalar concat (existing SET_VAR behavior).
10769pub const BUILTIN_APPEND_SCALAR_OR_PUSH: u16 = 331;
10770
10771/// `[[ -c path ]]` — character device.
10772pub const BUILTIN_IS_CHARDEV: u16 = 332;
10773/// `[[ -b path ]]` — block device.
10774pub const BUILTIN_IS_BLOCKDEV: u16 = 333;
10775/// `[[ -p path ]]` — FIFO / named pipe.
10776pub const BUILTIN_IS_FIFO: u16 = 334;
10777/// `[[ -S path ]]` — socket.
10778pub const BUILTIN_IS_SOCKET: u16 = 335;
10779/// `BUILTIN_ERREXIT_CHECK` constant.
10780pub const BUILTIN_ERREXIT_CHECK: u16 = 336;
10781/// Fatal-error-only abort check, for use INSIDE an `&&` / `||` chain.
10782///
10783/// A chain suppresses the errexit check (a non-zero status is "consumed"
10784/// by the connector — `false && x` must not fire ERREXIT or the ZERR
10785/// trap). But an errflag — a *fatal* error such as a `[[ ]]` bad pattern
10786/// — is not a status the connector can consume: zsh abandons the whole
10787/// list. Without this, zshrs ran the `||` right-hand side after the
10788/// error (`[[ x = [a- ]] || touch f` created `f`; zsh does not) and the
10789/// aborted builtin then overwrote the cond's status 2 with 1.
10790///
10791/// Same errflag arm as `BUILTIN_ERREXIT_CHECK`, with the errexit/ZERR
10792/// half omitted.
10793pub const BUILTIN_FATAL_ABORT_CHECK: u16 = 641;
10794/// Post-`always`-arm checks for the canonical RETFLAG / BREAKS /
10795/// CONTFLAG atomics that mark try-block escapes. Each returns
10796/// Value::Int(1) when the corresponding atomic is set (and consumes
10797/// it so the next escape doesn't re-fire) and Value::Int(0) otherwise.
10798/// Paired with JumpIfFalse + Jump to outer return_patches /
10799/// break_patches / continue_patches by compile_zsh's `Try` arm.
10800pub const BUILTIN_RETFLAG_CHECK: u16 = 600;
10801/// `BUILTIN_BREAKS_CHECK` constant.
10802pub const BUILTIN_BREAKS_CHECK: u16 = 601;
10803/// `BUILTIN_CONTFLAG_CHECK` constant.
10804pub const BUILTIN_CONTFLAG_CHECK: u16 = 602;
10805/// Fire the DEBUG trap (SIGDEBUG) before each statement.
10806/// c:Src/exec.c:1357-1500 DEBUGBEFORECMD — when a "DEBUG" entry is
10807/// installed via `trap '...' DEBUG`, run the body just before the
10808/// next command. Cheap when no DEBUG trap is set (one hashmap lookup
10809/// returns None and we early-out).
10810pub const BUILTIN_DEBUG_TRAP: u16 = 603;
10811/// `set -n` / `set -o noexec` — parse but don't execute. Returns
10812/// Value::Int(1) when the noexec option is set so the caller's
10813/// JumpIfTrue skips the statement body. c:Src/exec.c:1390 main loop
10814/// check.
10815pub const BUILTIN_NOEXEC_CHECK: u16 = 604;
10816/// Block-level redirect-failure gate. Reads exec.redirect_failed
10817/// (set by host.redirect when a redirect open fails); returns
10818/// Value::Int(1) AND clears the flag if set, else 0. Emit-side at
10819/// compile_zsh.rs::compile_command's Redirected arm pairs with a
10820/// JumpIfTrue → WithRedirectsEnd to abandon the body. Without this,
10821/// a multi-statement block after a failed redir kept running every
10822/// statement after the first (the first builtin consumed the flag,
10823/// subsequent statements ran unimpeded).
10824pub const BUILTIN_REDIRECT_FAILED_CHECK: u16 = 605;
10825/// Drop-in replacement for fusevm's Op::Exec for the dynamic-first-
10826/// word path (`$cmd`, `$(cmd)`, `~/bin/foo`). Returns
10827/// Value::Status(vm.last_status) when post-expansion argv is empty
10828/// (preserves the inner cmd-subst's exit), Value::Status(126) with
10829/// "permission denied" when `argv[0]` is empty, otherwise routes
10830/// through executor.host_exec_external like Op::Exec did.
10831pub const BUILTIN_EXEC_DYNAMIC: u16 = 606;
10832/// Reset `use_cmdoutval` to 0 at the START of a dynamic command (before
10833/// its words expand), so a command substitution from a PREVIOUS command
10834/// can't leak into this command's null-command status decision
10835/// (c:Src/exec.c:3009 `use_cmdoutval = !args`). See BUILTIN_EXEC_DYNAMIC.
10836pub const BUILTIN_USE_CMDOUTVAL_RESET: u16 = 637;
10837/// `[[ lhs == pat ]]` / `!=` glob compare — cond-specific so the
10838/// bad-pattern diagnostic follows Src/cond.c:308-316: zwarnnam
10839/// "bad pattern: %s" WITHOUT errflag (the script continues) and the
10840/// cond statement exits 2. Stack: [lhs, pat] → Bool. On compile
10841/// failure pushes Bool(false) and arms COND_BAD_PATTERN so
10842/// BUILTIN_COND_STATUS_FROM_BOOL reports 2.
10843pub const BUILTIN_COND_STRMATCH: u16 = 624;
10844/// Pops the cond result Bool → Int status per Src/cond.c: true→0,
10845/// false→1, but 2 when COND_BAD_PATTERN was armed during this cond
10846/// (covers `!=` where LogNot flips the Bool before status time).
10847pub const BUILTIN_COND_STATUS_FROM_BOOL: u16 = 625;
10848/// `[[ ]]` unknown condition. Pops \[op_name\], emits `zerr("unknown
10849/// condition: %s")` and sets ERRFLAG_ERROR so the next BUILTIN_ERREXIT_CHECK
10850/// (trigger 4) aborts the input — matching zsh's COND_MODI "unknown condition"
10851/// path (Src/cond.c:150-188) for a `-X` op with no matching loadable module.
10852/// Returns Bool(false). Replaces a compile-time `eprintln!` hack that printed
10853/// the message but never set errflag (so the line didn't abort).
10854pub const BUILTIN_COND_UNKNOWN: u16 = 632;
10855/// Bare-`exec` redirect epilogue. Consumes `exec.redirect_failed` and
10856/// applies the C `done:` tail of execcmd_exec:
10857/// - c:Src/exec.c:252-259 execerr — `redir_err = lastval = 1` (the
10858/// failed redirect makes the exec statement exit 1, NOT fatal by
10859/// itself);
10860/// - c:Src/exec.c:4367-4386 — `if (isset(POSIXBUILTINS) && (cflags
10861/// & (BINF_PSPECIAL|BINF_EXEC)) ...) { if (redir_err || errflag)
10862/// { if (!isset(INTERACTIVE)) exit(1); } }` — POSIX_BUILTINS makes
10863/// a failed exec redirect fatal in a non-interactive shell.
10864/// Returns Value::Status(0|1) for the trailing SetStatus.
10865pub const BUILTIN_EXEC_REDIR_DONE: u16 = 626;
10866/// Assignment-prefix epilogue for bare `exec` redirects
10867/// (`x=$(cmd) exec >file`). c:Src/exec.c:3969-3976 — nullexec==1
10868/// runs addvars THEN, without POSIX_BUILTINS, restores the params
10869/// (`save_params` / `restore_params`): the RHS side effects fire but
10870/// the values don't persist. With POSIX_BUILTINS the assignments
10871/// stick. Pops the BEGIN_INLINE_ENV frame either way.
10872pub const BUILTIN_EXEC_INLINE_ENV_DONE: u16 = 627;
10873
10874/// `< file` / `> file` with no command word (NULLCMD path).
10875/// Resolves NULLCMD (default "cat") / READNULLCMD (default "more")
10876/// at runtime per Src/exec.c:3340-3364 and exec's it through
10877/// host_exec_external. Argc is 1: the int (0 or 1) on the stack
10878/// indicates whether this is a single REDIR_READ redirect
10879/// (selects READNULLCMD when set + non-empty).
10880pub const BUILTIN_NULLCMD_EXEC: u16 = 607;
10881/// `.` (dot) — alias of source/bin_dot but dispatches with the
10882/// literal name "." so the diagnostic prefix matches zsh's
10883/// (`zsh:.:1: …` vs source's `zsh:source:1: …`).
10884/// c:Src/builtin.c:9308 — `BUILTIN(".", BINF_PSPECIAL, bin_dot, …)`.
10885pub const BUILTIN_DOT: u16 = 608;
10886/// `logout` — fusevm maps this to BUILTIN_EXIT alongside `exit`/`bye`,
10887/// which drops the name and dispatches with BIN_EXIT funcid. zsh's
10888/// `logout` outside a login shell must emit "not login shell" + exit 1,
10889/// which only fires when bin_break sees BIN_LOGOUT funcid. Dedicated
10890/// opcode dispatches via BUILTINS table by literal name "logout".
10891pub const BUILTIN_LOGOUT: u16 = 610;
10892/// `BUILTIN_PARAM_SUBSTRING_EXPR` constant.
10893pub const BUILTIN_PARAM_SUBSTRING_EXPR: u16 = 337;
10894/// `BUILTIN_XTRACE_LINE` constant.
10895pub const BUILTIN_XTRACE_LINE: u16 = 338;
10896/// `BUILTIN_XTRACE_ARRAY_LINE` — xtrace an `arr=(...)` assignment from the
10897/// whole assembled `Value::Array` (see compile_zsh array-literal codegen).
10898pub const BUILTIN_XTRACE_ARRAY_LINE: u16 = 649;
10899/// `BUILTIN_MAKE_ARRAY_COUNTED` — like `Op::MakeArray(u16)` but the element
10900/// count is a runtime `Int` on the stack, so it is not capped at 65535. Used
10901/// by the array-literal codegen only when the literal has > u16::MAX elements
10902/// (e.g. a .zcompdump's ~103k-element `_comps=(...)`).
10903pub const BUILTIN_MAKE_ARRAY_COUNTED: u16 = 650;
10904/// `BUILTIN_ARRAY_JOIN_STAR` constant.
10905pub const BUILTIN_ARRAY_JOIN_STAR: u16 = 339;
10906/// `BUILTIN_SET_RAW_OPT` constant.
10907pub const BUILTIN_SET_RAW_OPT: u16 = 340;
10908
10909/// `time { compound; ... }` — wall-clock-time the sub-chunk and print
10910/// elapsed seconds. Stack: [sub_chunk_idx as Int]. Runs the sub-chunk
10911/// on the current VM (so positional/local state is shared) and prints
10912/// the timing summary to stderr in zsh's format. Pushes Status.
10913pub const BUILTIN_TIME_SUBLIST: u16 = 316;
10914
10915/// `{name}>file` / `{name}<file` / `{name}>>file` — named-fd allocation.
10916/// Stack: [path, varid, op_byte]. Opens `path` per `op_byte`, gets the
10917/// new fd (≥10 in zsh; we use libc::open with O_CLOEXEC bit cleared so
10918/// the inherited fd survives Command::new spawns), stores the fd number
10919/// as a string in `$varid`. Pushes Status (0 success, 1 error).
10920pub const BUILTIN_OPEN_NAMED_FD: u16 = 317;
10921
10922/// Word-segment concat that does cartesian-product distribution over
10923/// arrays. Stack: [lhs, rhs]. Used for RC_EXPAND_PARAM `${arr}` and
10924/// explicit-distribute forms (`${^arr}`, `${(@)…}`).
10925///
10926/// - both scalar: `Value::str(a + b)` (fast path, identical to Op::Concat)
10927/// - lhs Array, rhs scalar: `Value::Array([a + rhs for a in lhs])`
10928/// - lhs scalar, rhs Array: `Value::Array([lhs + b for b in rhs])`
10929/// - both Array: cartesian product `[a + b for a in lhs for b in rhs]`
10930pub const BUILTIN_CONCAT_DISTRIBUTE: u16 = 318;
10931
10932/// Forced-distribute concat — like `BUILTIN_CONCAT_DISTRIBUTE` but
10933/// always distributes cartesian regardless of the `rcexpandparam`
10934/// option. Emitted by the segments fast-path when an
10935/// `is_distribute_expansion` segment is present (`${^arr}`,
10936/// `${(@)arr}`, `${(s.…)arr}` etc.) per zsh: the source-level
10937/// distribution flag overrides the option default.
10938/// Direct port of Src/subst.c:1875 `case Hat: nojoin = 1` and the
10939/// `rcexpandparam` test bypass for the explicit-distribute flags.
10940pub const BUILTIN_CONCAT_DISTRIBUTE_FORCED: u16 = 522;
10941
10942/// Capture current `last_status` into the `TRY_BLOCK_ERROR` variable.
10943/// Emitted between the try block and the always block of `{ … } always
10944/// { … }` so the finally arm can read $TRY_BLOCK_ERROR.
10945pub const BUILTIN_SET_TRY_BLOCK_ERROR: u16 = 320;
10946/// `BUILTIN_RESTORE_TRY_BLOCK_STATUS` constant.
10947pub const BUILTIN_RESTORE_TRY_BLOCK_STATUS: u16 = 432;
10948/// `BUILTIN_BEGIN_INLINE_ENV` constant.
10949pub const BUILTIN_BEGIN_INLINE_ENV: u16 = 433;
10950/// `BUILTIN_END_INLINE_ENV` constant.
10951pub const BUILTIN_END_INLINE_ENV: u16 = 434;
10952
10953/// `[[ -o option ]]` — shell-option-set test. Stack: \[option_name\].
10954/// Normalizes the name (strip underscores, lowercase) and reads
10955/// `exec.options`. Pushes Bool.
10956pub const BUILTIN_OPTION_SET: u16 = 321;
10957/// Tri-state `[[ -o NAME ]]` — same lookup as BUILTIN_OPTION_SET
10958/// but returns a Value::Int (0=set, 1=unset, 3=invalid-name). The
10959/// 3-state code matches zsh's `[[ -o invalid ]]` exit (Src/cond.c
10960/// :502 `optison()`). Used by compile_cond's `-o` arm to skip the
10961/// generic bool→status conversion and preserve the invalid-name
10962/// signal in `$?`.
10963pub const BUILTIN_OPTION_CHECK_TRISTATE: u16 = 609;
10964
10965/// `${var:#pattern}` — array filter: remove elements matching `pattern`.
10966/// Stack: [name, pattern]. For scalar `var`, returns empty if it matches
10967/// the pattern, else the value. For array `var`, returns Array of
10968/// non-matching elements.
10969pub const BUILTIN_PARAM_FILTER: u16 = 322;
10970
10971/// `a[i]=(elements)` / `a[i,j]=(elements)` / `a[i]=()` —
10972/// subscripted-array assign with array-literal RHS. Stack:
10973/// [...elements, name, key]. Empty elements + single-int key `a[i]=()`
10974/// removes that element. Comma-key `a[i,j]=(...)` splices.
10975pub const BUILTIN_SET_SUBSCRIPT_RANGE: u16 = 323;
10976
10977/// `[[ -X file ]]` for unknown unary test op `-X`. Stack: \[op_name\].
10978/// Emits zsh's `unknown condition: -X` diagnostic to stderr and
10979/// pushes Bool(false). Without this, unknown conditions silently
10980/// returned false matching neither zsh's error format nor the
10981/// expected status code (zsh returns 2 for parse error).
10982
10983/// `[[ -t fd ]]` — fd-is-a-tty check. Stack: \[fd_string\].
10984/// Routes through libc::isatty. Pushes Bool.
10985///
10986/// ID 644 (unique, next free above the previous max of 643). This was
10987/// 325, which COLLIDED with BUILTIN_FILE_OLDER (also 325). The VM's
10988/// builtin table is last-registration-wins, and FILE_OLDER registered
10989/// after IS_TTY, so every `[[ -t fd ]]` silently dispatched to the
10990/// file-`-ot` handler: it compared the mtime of a file NAMED by the fd
10991/// string ("0", "1", …) — which never exists — so `[[ -t 0 ]]` was
10992/// always false. That broke interactive detection (`[[ -t 0 && -t 1 ]]`)
10993/// and any config gated on it. c:Src/cond.c:390 `return !isatty(...)`.
10994pub const BUILTIN_IS_TTY: u16 = 644;
10995/// Runtime rejection of a process substitution used inside a `[[ … ]]`
10996/// cond operand. c:Src/exec.c:4918/5040/5069 — `getoutputfile`/`getproc`
10997/// error `"process substitution %s cannot be used here"` when `thisjob ==
10998/// -1`, which is the case during cond evaluation. zshrs's THISJOB never
10999/// distinguishes that context at runtime, so the compiler emits this
11000/// builtin (gated on `in_cond_operand`) instead of the ProcessSubIn/Out
11001/// opcode. Pops the substitution text, zerrs, sets errflag (aborting the
11002/// statement → empty stdout, exit 1, matching zsh), returns empty.
11003pub const BUILTIN_PROCSUB_COND_ERROR: u16 = 645;
11004/// `${^arr}` cross-product concat — RC_EXPAND_PARAM forced ON by the `^` flag.
11005///
11006/// Distinct from BUILTIN_CONCAT_DISTRIBUTE_FORCED, which the other distribute
11007/// shapes (`${(@)a}`, `${(f)v}`, `${a[@]}`) share: those keep the word when the
11008/// array is EMPTY (`x${(@)a}y` → `xy`), but plan9 DELETES it
11009/// (c:Src/subst.c:4362-4365 `if (plan9) { uremnode(l, n); return n; }`), so
11010/// `x${^a}y` with `a=()` produces no word at all. One builtin cannot serve both
11011/// — the plan9-ness is known only at compile time, from the `^` flag itself.
11012/// Routes to `concat_plan9`, which already ports both c:4362's removal and the
11013/// c:4316-4350 cartesian emit, and is what the OPTION path
11014/// (`setopt rcexpandparam`) has always used.
11015pub const BUILTIN_CONCAT_PLAN9: u16 = 646;
11016/// `${^^arr}` concat — RC_EXPAND_PARAM forced OFF by the doubled flag
11017/// (c:Src/subst.c:2553-2555 `plan9 = 0`).
11018///
11019/// The mirror of BUILTIN_CONCAT_PLAN9. Needed because every other concat
11020/// builtin consults `plan9_active()` (the runtime OPTION) and so cross-products
11021/// anyway under `setopt rcexpandparam`, while `^^` must override the option:
11022/// setopt rcexpandparam; a=(a b c); print -rl -- ${^^a}.x
11023/// # zsh: `a`, `b`, `c.x` — spliced, NOT `a.x b.x c.x`
11024/// The override is computed in paramsubst but the distribution call is made
11025/// here, so — like the `^` flag — the only place that knows is the compiler.
11026/// Routes straight to `concat_splice`, C's non-plan9 join-first-and-last path
11027/// (c:4366-4437).
11028pub const BUILTIN_CONCAT_SPLICE_NOPLAN9: u16 = 647;
11029/// `break N`/`continue N` runtime-count validator (see registration).
11030pub const BUILTIN_BREAK_COUNT_VALIDATE: u16 = 648;
11031/// `[[ -r/-w/-x file ]]` via access(2) (doaccess) — see handler.
11032pub const BUILTIN_COND_ACCESS: u16 = 638;
11033
11034/// Evaluate a `[[ ]]` module/completion condition (`-prefix`/`-suffix`/
11035/// `-after`/`-between`). Stack (top-first): argc operand words, then the
11036/// operator word. Dispatches to `complete::eval_mod_cond`. Result pushed as
11037/// Bool (true = condition matched). Used by the `ZshCond::ModCond` compile arm.
11038pub const BUILTIN_COND_MOD: u16 = 651;
11039
11040/// Update `$LINENO` to track the source line of the next statement.
11041/// Stack: \[n\] (the line number from `ZshPipe.lineno`). Direct port
11042/// of zsh's `lineno` global tracking (Src/input.c:330) — the
11043/// compiler emits one of these per top-level pipe so `$LINENO`
11044/// reflects the source position at runtime. ID 342 picked because
11045/// the previous `326` collided with `BUILTIN_HAS_STICKY` (the 325
11046/// collision between IS_TTY and FILE_OLDER has since been fixed by
11047/// moving IS_TTY to 644).
11048pub const BUILTIN_SET_LINENO: u16 = 342;
11049
11050/// Pop a scalar from the VM stack, run expand_glob on it, push the
11051/// result as Value::Array. Used by the segment-concat compile path
11052/// when var refs concatenate with glob meta literals (`$D/*`,
11053/// `${prefix}*`, etc.) — those skip the bridge's pathname-expansion
11054/// pass and would otherwise leak the glob meta to argv as a literal.
11055pub const BUILTIN_GLOB_EXPAND: u16 = 343;
11056
11057/// MULTIOS-gated glob expansion for redirect-target words
11058/// (c:Src/glob.c:2161-2167 xpandredir: "Globbing is only done for
11059/// multios."). Same stack shape as BUILTIN_GLOB_EXPAND; additionally
11060/// passes the word through literally when `unsetopt multios`.
11061// 624 is BUILTIN_COND_STRMATCH — the VM's builtin table is
11062// last-registration-wins, so a duplicate id silently shadows the
11063// earlier handler.
11064pub const BUILTIN_REDIR_GLOB_EXPAND: u16 = 628;
11065
11066/// Reset the default-word glob-pending carrier at the START of a word
11067/// whose source contains a glob metachar (so the flag never leaks from a
11068/// prior word/statement). Paired with BUILTIN_DEFAULT_WORD_GLOB.
11069pub const BUILTIN_DEFAULT_WORD_GLOB_RESET: u16 = 635;
11070
11071/// Filename-generate the ASSEMBLED word ONLY when the default/alternate
11072/// paramsubst arm took a SOURCE word carrying glob metachars
11073/// (subst::DEFAULT_WORD_GLOB_PENDING). A parameter VALUE never sets the
11074/// flag, so `x='*file'; ${x:-d}` stays literal while `${x:-*file}` /
11075/// `${x:-a*}bar` glob. Reads+clears the flag. c:Src/subst.c → globlist.
11076pub const BUILTIN_DEFAULT_WORD_GLOB: u16 = 636;
11077/// `BUILTIN_SET_LOOP_VAR` constant — for-loop variable binding via
11078/// `setloopvar` (Src/params.c:6362): a PM_NAMEREF loop var REBINDS
11079/// to each word (SETREFNAME + setscope) instead of assigning
11080/// through the chain. Returns Bool(false) when zerr fired
11081/// (read-only reference / invalid self reference) so the loop
11082/// driver aborts, mirroring C execfor's errflag check.
11083pub const BUILTIN_SET_LOOP_VAR: u16 = 629;
11084
11085/// EXTEND step of typeset paren-init packing. Pops `argc` values:
11086/// [base, e1, …, eN] — base is either the opener (`name=(` /
11087/// `name+=(`) or a previous EXTEND result. Pushes base with
11088/// `\u{1f}` + element appended per element. Array values SPLICE
11089/// their items as separate elements (`typeset b=( x $arr )` splat);
11090/// an empty Array contributes nothing (unquoted-empty elision).
11091/// CallBuiltin's argc is u8, so the compiler emits one EXTEND per
11092/// ≤200-element chunk — p10k's 408-element `__p9k_colors=( … )`
11093/// overflowed a single-shot pack (argc wrapped mod 256 and the
11094/// stack spilled into the arg list: "not an identifier: 173…").
11095/// BUILTIN_TYPESET_PAREN_CLOSE appends the final `\u{1f})`,
11096/// yielding the exact REJOIN_SEP-delimited one-arg form
11097/// bin_typeset's single-arg splitter consumes (builtin.rs ~4891,
11098/// empties preserved, leading/trailing sentinel-empties trimmed
11099/// once). One arg in → one arg out: bin_typeset's multi-arg rejoin
11100/// (paren-depth scan, unsafe on EXPANDED paren-literal elements
11101/// like p10k's `')' ''`) never runs.
11102pub const BUILTIN_TYPESET_PAREN_PACK: u16 = 630;
11103
11104/// CLOSE step — pops the EXTEND chain's result, pushes it with
11105/// `\u{1f})` appended. See BUILTIN_TYPESET_PAREN_PACK.
11106pub const BUILTIN_TYPESET_PAREN_CLOSE: u16 = 631;
11107
11108/// Shared body of BUILTIN_GLOB_EXPAND / BUILTIN_REDIR_GLOB_EXPAND.
11109/// c:Src/glob.c:1872 — `zglob` runs per-word in the argv pipeline.
11110/// When the upstream EXPAND_TEXT returned an array (e.g. `${a:e}`
11111/// splat → ["txt","md"]), glob each element separately, not a
11112/// sepjoin'd scalar. `skip_glob` short-circuits to a literal
11113/// pass-through (noglob, or a redirect target under
11114/// `unsetopt multios`).
11115fn glob_expand_word_value(raw: Value, skip_glob: bool) -> Value {
11116 let patterns: Vec<String> = match raw {
11117 Value::Array(items) => items.into_iter().map(|v| v.to_str()).collect(),
11118 other => vec![other.to_str()],
11119 };
11120 if skip_glob {
11121 return if patterns.is_empty() {
11122 Value::Array(Vec::new())
11123 } else if patterns.len() == 1 {
11124 Value::str(patterns.into_iter().next().unwrap())
11125 } else {
11126 Value::Array(patterns.into_iter().map(Value::str).collect())
11127 };
11128 }
11129 let mut out: Vec<String> = Vec::with_capacity(patterns.len());
11130 for pattern in &patterns {
11131 // c:Src/subst.c — filename generation runs `filesub` (tilde/`=`
11132 // expansion) BEFORE globbing. A `~`/`=` reaching this word-glob op
11133 // comes from `${~spec}` / GLOB_SUBST marking a substituted VALUE:
11134 // literal and quoted `~` words are filesub'd (or skip glob) upstream
11135 // and never arrive here. filesubstr matches the Tilde TOKEN, so
11136 // shtokenize first (raw `~`->Tilde; already-Tilde `${~a[@]}` results
11137 // pass through), run filesub, then untokenize surviving glob metas
11138 // for expand_glob. Gated on `~`/`=` (raw or token) so ordinary
11139 // substituted words skip the roundtrip. Fixes `${~x}` x="~/foo".
11140 let filesubbed = if pattern.contains('~')
11141 || pattern.contains('=')
11142 || pattern.contains(crate::ported::zsh_h::Tilde)
11143 || pattern.contains(crate::ported::zsh_h::Equals)
11144 {
11145 let mut tok = pattern.clone();
11146 crate::ported::glob::shtokenize(&mut tok);
11147 crate::ported::lex::untokenize(&crate::ported::subst::filesub(&tok, 0))
11148 } else {
11149 pattern.clone()
11150 };
11151 let matches = with_executor(|exec| exec.expand_glob(&filesubbed));
11152 if matches.is_empty() {
11153 // c:1872 nullglob — drop this word, don't emit a hole
11154 continue;
11155 }
11156 for m in matches {
11157 out.push(m);
11158 }
11159 }
11160 if out.is_empty() {
11161 return Value::Array(Vec::new());
11162 }
11163 if patterns.len() == 1 && out.len() == 1 && out[0] == patterns[0] {
11164 // No real matches; expand_glob returned the literal. Pass
11165 // back as scalar so downstream ops don't re-flatten.
11166 return Value::str(out.into_iter().next().unwrap());
11167 }
11168 Value::Array(out.into_iter().map(Value::str).collect())
11169}
11170
11171/// Push a `CmdState` token onto the command-context stack. Direct
11172/// port of zsh's `cmdpush(int cmdtok)` (Src/prompt.c:1623). The
11173/// stack is consulted by `%_` in PS4/prompt expansion to produce
11174/// the cumulative control-flow-context labels (`if`, `then`,
11175/// `cmdand`, `cmdor`, `cmdsubst`, …) that `zsh -x` xtrace shows
11176/// in the trace prefix. Compile_zsh emits push/pop pairs around
11177/// each compound command (if/while/[[…]]/((…))/$(…) etc.).
11178/// Token is a `CmdState as u8`.
11179pub const BUILTIN_CMD_PUSH: u16 = 344;
11180
11181/// Pop the top of the command-context stack. Direct port of zsh's
11182/// `cmdpop(void)` (Src/prompt.c:1631).
11183pub const BUILTIN_CMD_POP: u16 = 345;
11184
11185/// Emit an xtrace line built from the top `argc` values on the VM
11186/// stack, peeked WITHOUT consuming. Used to trace simple commands
11187/// AFTER expansion, so `echo for $i` shows as `echo for a` / `echo
11188/// for b`. Direct port of Src/exec.c:2055-2066.
11189pub const BUILTIN_XTRACE_ARGS: u16 = 346;
11190
11191/// Trace one assignment: emits `name=<quoted-value> ` (no newline)
11192/// to xtrerr if XTRACE is on. Coalesces with subsequent
11193/// XTRACE_ASSIGN / XTRACE_ARGS calls onto the SAME line via the
11194/// `XTRACE_DONE_PS4` flag so `a=1 b=2 echo $a $b` produces:
11195/// `<PS4>a=1 b=2 echo 1 2\n`
11196/// matching C zsh's `execcmd_exec` body (Src/exec.c:2517-2582):
11197/// xtr = isset(XTRACE);
11198/// if (xtr) { printprompt4(); doneps4 = 1; }
11199/// while (assign) {
11200/// if (xtr) fprintf(xtrerr, "%s=", name);
11201/// ... eval value ...
11202/// if (xtr) { quotedzputs(val, xtrerr); fputc(' ', xtrerr); }
11203/// }
11204///
11205/// Stack contract on entry: [..., name, value]. Both peeked, NOT
11206/// consumed (the matching SET_VAR call pops them after). argc = 2.
11207pub const BUILTIN_XTRACE_ASSIGN: u16 = 525;
11208
11209/// Emit a trailing `\n` + flush iff XTRACE is on AND PS4 was
11210/// emitted by an earlier XTRACE_ASSIGN this line. Used at the end
11211/// of compile_simple's assignment-only path so the trace line gets
11212/// terminated. Mirrors C's exec.c:3397-3399 (the assign-only return
11213/// path through execcmd_exec which does `fputc('\n', xtrerr);
11214/// fflush(xtrerr)`).
11215///
11216/// Stack: untouched. argc = 0.
11217pub const BUILTIN_XTRACE_NEWLINE: u16 = 526;
11218
11219/// Push the live `xtrace` opt-state as `Value::Int(1)` (on) or
11220/// `Value::Int(0)` (off). Used by `compile_cond` to gate the
11221/// trace-string-building block on xtrace state at runtime — without
11222/// this the trace path's `compile_word_str` on each operand re-
11223/// evaluates side-effectful expressions (`$((i++))`) once for the
11224/// trace string and once for the actual condition, doubling the
11225/// effective increment. Bug #159 in docs/BUGS.md.
11226///
11227/// Stack: pushes Int(0|1). argc = 0.
11228pub const BUILTIN_XTRACE_IS_ON: u16 = 611;
11229
11230/// Reset the `DONETRAP` flag at the start of each top-level statement
11231/// (sublist boundary). Mirrors C `Src/exec.c:1455` — `donetrap = 0`.
11232/// Stack: untouched. argc = 0. Bug #303 in docs/BUGS.md.
11233pub const BUILTIN_DONETRAP_RESET: u16 = 612;
11234
11235/// `[[ -z X ]]` / `[[ -n X ]]` operand-empty test that honours zsh's
11236/// array-splice semantics. C zsh evaluates `[[ -z X ]]` per
11237/// `Src/cond.c:347` (case 'z'): `s` is the SCALAR operand passed
11238/// through `cond_str`'s singsub. For `"${arr[@]}"` zsh expands per
11239/// `Src/subst.c:multsub` which yields each element as its own word
11240/// list node; cond.c then sees the joined-or-single-element form.
11241///
11242/// The compile-side `-z` shortcut at `compile_zsh.rs:5371` used
11243/// `Op::StringLen` which calls `Value::len` — for `Value::Array`
11244/// that returns ARRAY LENGTH, not string length. `b=("")` produced
11245/// `Value::Array([""])` → `len = 1` → `-z` returned false.
11246///
11247/// This builtin pops one `Value` and pushes `1` (empty) or `0`
11248/// (non-empty) per the cond context:
11249/// - `Value::Str(s)` → s.is_empty()
11250/// - `Value::Array([])` → true (zero words → vacuous-empty)
11251/// - `Value::Array([s])` → s.is_empty() (single-word case)
11252/// - `Value::Array([_; n>=2])` → false (multiple non-empty
11253/// words; zsh would raise "unknown condition" but the
11254/// observable test result is non-empty/false)
11255///
11256/// Companion to BUILTIN_COND_STR_NONEMPTY (#185 in docs/BUGS.md).
11257pub const BUILTIN_COND_STR_EMPTY: u16 = 613;
11258
11259/// `[[ -n X ]]` operand-non-empty test (logical complement of
11260/// BUILTIN_COND_STR_EMPTY).
11261pub const BUILTIN_COND_STR_NONEMPTY: u16 = 614;
11262
11263/// `exec N<<<"str"` — herestring redirect to explicit fd, applied
11264/// permanently to the shell (no scope restoration). Pops `[content,
11265/// fd]` from the stack; creates a temp file, writes
11266/// `content + "\n"`, reopens read-only, dup2's to `fd`, unlinks the
11267/// temp path so it disappears on close. Mirrors C `Src/exec.c:4655
11268/// getherestr` + `addfd(forked, save, mfds, fn->fd1, fil, 0, ...)`
11269/// at c:3766-3780 for the bare-exec-redir code path (nullexec=1).
11270/// Bug #205 in docs/BUGS.md.
11271///
11272/// Stack: pushes `Value::Status(0)` on success, `Status(1)` on
11273/// failure. argc = 2.
11274pub const BUILTIN_EXEC_HERESTR_FD: u16 = 615;
11275
11276/// MULTIOS write/append fan-out for `cmd > a > b` / `cmd > a >> b`
11277/// style redirects (Bug #36 in docs/BUGS.md). zsh's MULTIOS option
11278/// (Src/exec.c:2418 `mfds[fd1]` check + addfd splice) creates a
11279/// pipe at fd1, spawns an internal "tee" process that copies
11280/// stdin → every collected target file. Without this, only the
11281/// LAST redirect target survives because each dup2 overwrites the
11282/// previous binding.
11283///
11284/// Stack layout (pushed by compile_zsh's compile_redirs coalescing
11285/// pass): `[target_1, op_byte_1, target_2, op_byte_2, …, target_N,
11286/// op_byte_N, fd]`. Pops 2N+1 elements; `argc = 2*N + 1`. A target
11287/// may be a Value::Array of glob matches (spliced into one member
11288/// per match, c:Src/glob.c:2195-2203); an op may be DUP_WRITE for a
11289/// numeric `>&N` member (c:Src/exec.c:3895-3917).
11290///
11291/// Runtime (MULTIOS set):
11292/// 1. Seed the member list with `dup(1)` when this command's
11293/// stdout is the pipeline output (c:Src/exec.c:3722-3724).
11294/// 2. Open/dup all targets per their op_byte in redirect order
11295/// (WRITE truncate + noclobber gate / APPEND / DUP_WRITE live
11296/// dup); the first member replaces the fd (c:2448-2450).
11297/// 3. Save `dup(fd)` onto the active redirect_scope_stack so
11298/// `host_redirect_scope_end` restores the original fd.
11299/// 4. Create a pipe; spawn a thread that reads from the pipe
11300/// read-end and writes every chunk to every opened target.
11301/// 5. dup2 the pipe write-end onto `fd` so the command's writes
11302/// go through the splitter.
11303/// 6. Track `(pipe_write_fd, JoinHandle)` so scope-end can close
11304/// the pipe (draining the thread) and join before restoring.
11305///
11306/// MULTIOS unset (c:2418 `unset(MULTIOS)` replace arm): each entry
11307/// is applied as a plain sequential replace via host_apply_redirect
11308/// — every file still opened/truncated, last one wins.
11309pub const BUILTIN_MULTIOS_REDIRECT: u16 = 617;
11310
11311/// MULTIOS input-side concatenation for `cmd < a < b` shapes
11312/// (Bug #36 input arm). C zsh's `Src/exec.c:2418` mfds dispatch
11313/// also covers the read direction — when multiple `<` redirects
11314/// target the same fd, mfds[fd] grows and addfd splices a
11315/// concatenating cat into the pipe.
11316///
11317/// Stack layout (mirrors the write side): `[source_1, op_1,
11318/// source_2, op_2, …, source_N, op_N, fd]`. Pops 2N + 1 elements
11319/// (argc = 2N + 1). op is READ for file sources, DUP_READ for
11320/// numeric `<&N` members; a source may be a Value::Array of glob
11321/// matches (spliced, c:Src/glob.c:2195-2203).
11322///
11323/// Runtime (MULTIOS set):
11324/// 1. Open/dup every source in redirect order; first member
11325/// replaces the fd (c:Src/exec.c:2448-2450).
11326/// 2. Save `dup(fd)` onto the redirect_scope_stack.
11327/// 3. Create a pipe; spawn a thread that reads each source in
11328/// order and writes every chunk to the pipe write-end. Close
11329/// write-end when done so the consumer sees EOF.
11330/// 4. dup2 the pipe read-end onto `fd`.
11331/// 5. Track the JoinHandle so scope-end joins (no fd-close needed
11332/// here — the producer thread closes its own pipe write-end
11333/// on exit).
11334///
11335/// MULTIOS unset: sequential replace via host_apply_redirect — last
11336/// source wins (c:2418).
11337pub const BUILTIN_MULTIOS_READ: u16 = 618;
11338
11339/// Toggle `ShellExecutor::exec_redirs_permanent`. Emitted by
11340/// compile_zsh's bare-`exec`-with-redirects arm tightly around each
11341/// `Op::Redirect`: `LoadInt(1); CallBuiltin; …Redirect…; LoadInt(0);
11342/// CallBuiltin`. While set, `host_apply_redirect` skips pushing the
11343/// saved fd into the enclosing redirect scope, making the fd change
11344/// permanent.
11345///
11346/// c:Src/exec.c:3978-3986 — nullexec==1 (`exec` carrying only
11347/// redirections): "If nullexec is 1 we specifically *don't* restore
11348/// the original fd's before returning" — the per-execcmd `save[]`
11349/// dups are closed unrestored. An ENCLOSING group's own saves are a
11350/// different execcmd's `save[]` and still restore (verified:
11351/// `{ exec 2>/dev/null; } 2>&1; ls /nope` prints the ls error in zsh).
11352pub const BUILTIN_EXEC_PERM_REDIRS: u16 = 619;
11353
11354/// Set `ShellExecutor::pipe_output_pending`. Emitted by compile_pipe
11355/// at the head of a NON-LAST pipeline-stage sub-chunk when that
11356/// stage's top-level command carries redirects (`Simple` with redirs
11357/// or `Redirected` compound). The forked stage child runs the chunk
11358/// with stdout already dup2'd onto the pipe; the first
11359/// `host_redirect_scope_begin` (the stage command's own redirect
11360/// list) consumes the flag into `pipe_output_scope`, enabling the
11361/// MULTIOS stream-split for fd-1 write redirects in that list.
11362///
11363/// c:Src/exec.c:3722-3724 — `addfd(forked, save, mfds, 1, output, 1,
11364/// NULL)` registers the pipe in mfds[1] in the SAME execcmd that
11365/// walks the stage command's redirect list; mfds is per-execcmd, so
11366/// nested body commands (`{ echo a > f; } | cat`) never see it.
11367pub const BUILTIN_PIPE_OUTPUT_MARK: u16 = 620;
11368
11369/// Install the pipeline stage's parked fds onto 0/1.
11370///
11371/// c:Src/exec.c:3720-3724 — `addfd(forked, save, mfds, 0, input, 0,
11372/// NULL)` / `addfd(..., 1, output, 1, NULL)`. Runs after prefork
11373/// (c:3304) and globlist (c:3702) have expanded the stage's argument
11374/// words, which is why a `$(...)` in those words reads the shell's
11375/// original stdin rather than the pipe. Emitted by
11376/// `compile_zsh.rs::emit_stage_fds_install`; the fds themselves are
11377/// parked by `BUILTIN_RUN_PIPELINE`.
11378pub const BUILTIN_PIPE_FDS_INSTALL: u16 = 642;
11379
11380/// Magic-equals prefork for a single arg word of a
11381/// `BINF_MAGICEQUALS` builtin head (`alias`). Direct port of
11382/// c:Src/exec.c:3298-3304 — `esprefork = PREFORK_TYPESET;
11383/// prefork(args, esprefork, NULL)` runs on the argv BEFORE the addfd
11384/// redirect loop at c:3720, so an expansion zerr (`alias bad===` →
11385/// equalsubstr "= not found" at Src/subst.c:726) prints to the
11386/// command's UN-redirected stderr. argc=1: pops the just-pushed
11387/// word value, runs shtokenize → prefork(PREFORK_TYPESET) →
11388/// untokenize on it (each element for Array splices), pushes the
11389/// result back. Emitted by compile_simple per arg word when the
11390/// dispatch head is `alias`; BUILTIN_ALIAS itself no longer
11391/// preforks (it would double-fire the diagnostic).
11392pub const BUILTIN_MAGIC_EQUALS_PREFORK: u16 = 621;
11393
11394/// Bare (unbraced) `$name[idx]` subscript — same dispatch as
11395/// `BUILTIN_ARRAY_INDEX` while KSHARRAYS is unset, but under
11396/// KSHARRAYS the unbraced form does NOT subscript (c:Src/subst.c:
11397/// 2800-2802 + 2867): `$name` expands bare and `[idx]` stays literal
11398/// trailing text that undergoes filename generation. Operands:
11399/// [name, idx, suffix, quoted].
11400pub const BUILTIN_ARRAY_INDEX_UNBRACED: u16 = 622;
11401
11402/// Assignment-only simple-command exit status. Direct port of
11403/// `lv = (errflag ? errflag : cmdoutval)` (c:Src/exec.c:1322,
11404/// execsimple's WC_ASSIGN arm) / `if (errflag) lastval = 1; else
11405/// lastval = cmdoutval;` (c:Src/exec.c:3393-3396, execcmd_exec's
11406/// no-command-word varspc path; redir variant at c:3977). Pops
11407/// [had_cmd_subst]; cmdoutval is the live vm.last_status when a
11408/// `$()` ran in any RHS of the chain, 0 otherwise. Writes the
11409/// canonical LASTVAL (C's single `lastval` global) so the
11410/// non-interactive errflag abort exits with this value per
11411/// Src/init.c:234. Caller pairs with SetStatus.
11412pub const BUILTIN_ASSIGN_ONLY_STATUS: u16 = 623;
11413
11414/// c:Src/exec.c addvars — `if (!pm) { lastval = 1; if (!cmdoutval)
11415/// cmdoutval = 1; }`. Set by BUILTIN_SET_VAR on assignsparam
11416/// failure, consumed by BUILTIN_ASSIGN_ONLY_STATUS so the
11417/// assignment-only command reports status 1. Process-global like
11418/// C's `cmdoutval` (function bodies may run on a different thread
11419/// than the opcode that reads the status back).
11420pub static ASSIGN_FAILED_FLAG: std::sync::atomic::AtomicBool =
11421 std::sync::atomic::AtomicBool::new(false);
11422
11423/// `redirection with no command` parse-time error for bare
11424/// `builtin 2>&1` / `command < file` / `exec >&-` precmd-keyword
11425/// shapes with a redirect but no following command. Direct port
11426/// of `Src/exec.c:3342 zerr("redirection with no command")`.
11427/// argc=0; pushes Value::Status(1).
11428pub const BUILTIN_REDIR_NO_CMD: u16 = 616;
11429
11430/// GLOB_SUBST guard for `[[ x == $pat ]]` pattern RHS coming from
11431/// parameter / command substitution. C-zsh's `[[ == ]]` semantics
11432/// (Src/options.c GLOB_SUBST default OFF + Src/cond.c:552
11433/// `cond_match` + Src/pattern.c patcompile tokenization-based
11434/// meta detection) treat chars from substitution as LITERAL
11435/// unless GLOB_SUBST is on. The Rust patcompile accepts both
11436/// tokenized and raw-ASCII meta chars, losing the distinction,
11437/// so `pat="h*"; [[ hello == $pat ]]` matched in zshrs but not
11438/// in zsh. Bug #116 in docs/BUGS.md.
11439///
11440/// Compile-time signal: emitted by `compile_cond_expr` ONLY when
11441/// the RHS contains `$` or backtick. Runtime checks the live
11442/// option state. If GLOB_SUBST is OFF, the popped string has
11443/// its glob metachars escaped with `\` so the downstream StrMatch
11444/// → patcompile treats them as literals. If GLOB_SUBST is ON,
11445/// the value passes through unchanged so `setopt glob_subst`
11446/// restores zsh's pattern-on-expansion behavior.
11447///
11448/// Stack: pops one string, pushes the (possibly escaped) result.
11449/// argc = 1.
11450pub const BUILTIN_GLOB_SUBST_GUARD: u16 = 528;
11451
11452/// Coerce a string parameter value to a math number (Int or Float)
11453/// for arithmetic-context reads, mirroring C-zsh's `getmathparam`
11454/// (Src/math.c:337). When the variable holds a string like "hello"
11455/// that isn't numeric, C falls back to recursively evaluating the
11456/// raw string as an arith expression; if that fails too, returns 0.
11457///
11458/// Used by the ArithCompiler pre-load path so `(( y = x ))` with
11459/// `x="hello"` reads `x` as integer 0, then assigns y as integer 0
11460/// — matching zsh's behaviour. The previous Rust port used
11461/// BUILTIN_GET_VAR which returned the raw string "hello"; the
11462/// ArithCompiler stored it verbatim in y's slot, and the post-sync
11463/// BUILTIN_SET_VAR wrote y="hello" as scalar instead of y=0 as
11464/// integer. Bug #118 in docs/BUGS.md.
11465///
11466/// Stack: pops `name` (string), pushes coerced numeric Value.
11467/// argc = 1.
11468pub const BUILTIN_GET_MATH_VAR: u16 = 529;
11469
11470/// GLOB_SUBST runtime gate for words containing parameter / command
11471/// substitution. C-zsh's `prefork` (Src/subst.c) runs `shtokenize`
11472/// on the substituted value when `GLOB_SUBST` is set, making the
11473/// substituted chars eligible for filename generation. With the
11474/// option off, substituted chars stay literal.
11475///
11476/// The Rust port's compile_zsh emits `compile_word_str` for words
11477/// like `/tmp/X/$pat`, which returns the post-expansion string but
11478/// never runs glob expansion (no path here triggers
11479/// BUILTIN_GLOB_EXPAND). Bug #119 in docs/BUGS.md: with `setopt
11480/// glob_subst`, `for f in /tmp/X/$pat` (pat="*.txt") never matched
11481/// `*.txt` files.
11482///
11483/// This opcode wraps the substitution result and dispatches at
11484/// runtime: when GLOB_SUBST is OFF, return unchanged; when ON,
11485/// pass the value through `expand_glob` so glob metas become
11486/// active. Emitted by `compile_for_words` (and similar sites)
11487/// after WORD_SPLIT for words with unquoted expansion.
11488///
11489/// Stack: pops a Value (Str or Array of Str), pushes the glob-
11490/// expanded result (still Str or Array depending on input shape).
11491/// argc = 1.
11492pub const BUILTIN_GLOB_SUBST_EXPAND: u16 = 530;
11493/// `BUILTIN_ASSOC_HAS_KEY` constant — `${(k)assoc[name]}` key-existence
11494/// query. Returns the key text on hit, empty string on miss. Bug #145.
11495pub const BUILTIN_ASSOC_HAS_KEY: u16 = 531;
11496/// `BUILTIN_ARRAY_DROP_EMPTY` constant — filter empty elements from
11497/// an Array on the stack. Used by `for x in $@` / `for x in $*`
11498/// unquoted forms. Bug #166.
11499pub const BUILTIN_ARRAY_DROP_EMPTY: u16 = 532;
11500/// `BUILTIN_QUOTEDZPUTS` constant — run top-of-stack value through
11501/// `crate::ported::utils::quotedzputs` and push the quoted result.
11502/// Used by the cond xtrace path so non-printable bytes (e.g.
11503/// `$'\C-[OP'` expanded ESC+OP) get re-wrapped in `$'…'` form for
11504/// the trace prefix line, matching zsh's `Src/exec.c` cond trace
11505/// which calls `quotedzputs(operand, xtrerr)` on each side. Bug
11506/// surfaced when `[[ -n $'\C-[OP' ]]` traced as `[[ -n OP ]]`
11507/// (raw bytes leaked through the terminal) vs zsh's
11508/// `[[ -n $'\C-[OP' ]]` source-form preservation.
11509pub const BUILTIN_QUOTEDZPUTS: u16 = 533;
11510/// `BUILTIN_QUOTE_TOKENIZED_OUTPUT` — port of
11511/// `crate::ported::exec::quote_tokenized_output` (Src/exec.c:2114)
11512/// applied to top-of-stack scalar. Used by cond xtrace for the RHS
11513/// of pattern-context comparisons (`=` / `==` / `!=`) where C zsh
11514/// emits the SOURCE form: untokenize lexer tokens (Star → `*`,
11515/// Inpar → `(`, …) and backslash-escape special chars, but
11516/// preserve literal ASCII unchanged. Distinct from quotedzputs
11517/// which wraps the whole string in `'…'` / `$'…'` based on
11518/// non-printability — that's wrong for `[[ x = a* ]]` which must
11519/// render as `[[ x = a* ]]`, not `'a*'`.
11520pub const BUILTIN_QUOTE_TOKENIZED_OUTPUT: u16 = 534;
11521
11522/// Bridge into subst_port::substitute_brace_array for nested forms
11523/// that need to PRESERVE array shape across the expand_string
11524/// boundary. Stack: `[content_string]`. Returns Value::Array of the
11525/// per-element words. Used by the compile path for
11526/// `${(@)<nested>...##pat}` shapes — the standard substitute_brace
11527/// returns String which collapses array→scalar; this builtin
11528/// preserves the multi-word output via paramsubst's third return
11529/// (`nodes` vec, the C source's `aval` thread).
11530pub const BUILTIN_BRIDGE_BRACE_ARRAY: u16 = 347;
11531
11532/// Word-segment concat with FIRST/LAST sticking. Stack: [lhs, rhs].
11533/// Used for default unquoted splice forms (`${arr[@]}`, `$@`, `$*`)
11534/// where prefix sticks to first element only and suffix to last only.
11535///
11536/// Distribution table:
11537/// - both scalar: `Value::str(a + b)` (fast path)
11538/// - lhs scalar, rhs Array(b₀..bₙ): `Value::Array([lhs+b₀, b₁, …, bₙ])`
11539/// - lhs Array(a₀..aₙ), rhs scalar: `Value::Array([a₀, …, aₙ₋₁, aₙ+rhs])`
11540/// - both Array: `Value::Array([a₀, …, aₙ₋₁, aₙ+b₀, b₁, …, bₙ])`
11541/// (last of lhs merges with first of rhs; the rest stay separate)
11542///
11543/// This is the default zsh semantics for `print -l X${arr[@]}Y` →
11544/// "Xa", "b", "cY" — three distinct args, surrounding text only on ends.
11545pub const BUILTIN_CONCAT_SPLICE: u16 = 319;
11546
11547/// `${(flags)name}` — zsh parameter expansion flags. Stack: [name, flags].
11548/// Flags applied left-to-right. Supported subset (high-value, used by zpwr):
11549///
11550/// `L` — lowercase the value (scalar; or each element if array)
11551/// `U` — uppercase
11552/// `j:sep:` — join array with `sep` (delim is the char after `j`)
11553/// `s:sep:` — split scalar on `sep` (returns Value::Array)
11554/// `f` — split on newlines (shorthand for `s.\n.`)
11555/// `o` — sort array ascending
11556/// `O` — sort array descending
11557/// `P` — indirect: read name's value as another var name, return that's value
11558/// `@` — keep as array (returns Value::Array — useful before `j` etc.)
11559/// `k` — keys of assoc array
11560/// `v` — values of assoc array
11561/// `#` — word count (array length as scalar)
11562///
11563/// Flags can stack: `(jL)` joins then lowercases; `(s.,.U)` splits on `,`
11564/// then uppercases each element. The long-tail flags (`q`, `qq`, `qqq` for
11565/// quoting, `A` for assoc, `%` for prompt expansion, `e`/`g` for re-eval,
11566/// `n`/`p` for numeric, `t` for type, etc.) are deferred — they hit the
11567/// runtime fallback via the catch-all expansion path.
11568pub const BUILTIN_PARAM_FLAG: u16 = 297;
11569
11570/// `ShellHost` implementation that delegates to the current `ShellExecutor`
11571/// via the `with_executor` thread-local.
11572///
11573/// Construct fresh on each VM run (it carries no state itself). The VM
11574/// dispatches host method calls during `vm.run()`, and `with_executor`
11575/// resolves to the executor pointer set by `ExecutorContext::enter`.
11576/// fusevm-host implementation tying bytecode ops to the
11577/// shell executor.
11578/// zshrs-original — no C counterpart. C zsh has no bytecode VM
11579/// to host; everything runs through `execlist()`/`execpline()`
11580/// directly (Src/exec.c lines 1349/1668).
11581pub struct ZshrsHost;
11582
11583impl fusevm::ShellHost for ZshrsHost {
11584 fn glob(&mut self, pattern: &str, _recursive: bool) -> Vec<String> {
11585 with_executor(|exec| exec.expand_glob(pattern))
11586 }
11587
11588 fn tilde_expand(&mut self, s: &str) -> String {
11589 with_executor(|exec| s.to_string())
11590 }
11591
11592 fn brace_expand(&mut self, s: &str) -> Vec<String> {
11593 // Direct call to the canonical brace expander
11594 // (Src/glob.c::xpandbraces port at glob.rs:1678). Was
11595 // routing through singsub which uses PREFORK_SINGLE — that
11596 // flag explicitly suppresses brace expansion in subst.c:166,
11597 // so `print X{1,2,3}Y` returned the literal string.
11598 //
11599 // brace_ccl: respect the BRACE_CCL option which the bracket-
11600 // class form `{a-z}` requires. Pull from executor options.
11601 let brace_ccl = with_executor(|exec| opt_state_get("braceccl").unwrap_or(false));
11602 crate::ported::glob::xpandbraces(s, brace_ccl)
11603 }
11604
11605 fn str_match(&mut self, s: &str, pattern: &str) -> bool {
11606 // Shell glob match — `*`, `?`, `[...]`, alternation. After the
11607 // cond path moved to BUILTIN_COND_STRMATCH, the consumer here
11608 // is the `case` arm dispatch, whose bad-pattern semantics are
11609 // Src/loop.c:663-667: `if (!(pprog = patcompile(pat, ...)))
11610 // zerr("bad pattern: %s", pat);` — errflag set, the arm
11611 // doesn't match, and the script aborts at the next command
11612 // boundary (matching `zsh -fc 'case x in [a-) ...'` printing
11613 // the diagnostic with exit 0 = untouched lastval).
11614 let mut pat_tok = pattern.to_string();
11615 crate::ported::glob::tokenize(&mut pat_tok);
11616 if crate::ported::pattern::patcompile(
11617 &pat_tok,
11618 crate::ported::zsh_h::PAT_STATIC as i32,
11619 None,
11620 )
11621 .is_none()
11622 {
11623 crate::ported::utils::zerr(&format!("bad pattern: {}", pattern)); // c:667
11624 return false;
11625 }
11626 glob_match_static(s, pattern)
11627 }
11628
11629 fn expand_param(&mut self, name: &str, _modifier: u8, _args: &[Value]) -> Value {
11630 // Sole funnel: route through `getsparam` matching C zsh's
11631 // `getsparam(name)` → `getvalue` → `getstrvalue` →
11632 // `Param.gsu->getfn` dispatch (Src/params.c:3076 / 2335).
11633 //
11634 // The lookup chain (GSU dispatch + variables + env + array-
11635 // join) lives in `params::getsparam`; subst.rs and this
11636 // bridge both call into it so the logic is in exactly one
11637 // place — mirroring C's "every read goes through getsparam"
11638 // architecture. fuseVM bytecode triggers this bridge when
11639 // the VM hits a PARAM opcode, equivalent to C's wordcode VM
11640 // resolving a parameter read during `exec.c` execution.
11641 //
11642 // Modifier handling: the `_modifier` / `_args` parameters
11643 // are populated by the bytecode compiler but applied by
11644 // separate VM opcodes (LENGTH/STRIP/SUBST/etc.) downstream
11645 // of this fetch — matching C's split between getsparam
11646 // (value fetch) and paramsubst's modifier-walk loop. This
11647 // bridge is the value-fetch step only.
11648 let val_str = crate::ported::params::getsparam(name).unwrap_or_default();
11649 Value::str(val_str)
11650 }
11651
11652 fn process_sub_in(&mut self, sub: &fusevm::Chunk) -> String {
11653 // c:Src/exec.c:4906 getoutputfile — `=(cmd)` (marked "equalsubst" by the
11654 // compiler) is the TEMP-FILE flavor: create a real regular file, fork a
11655 // writer whose stdout is the file, WAIT for it (so the file is complete
11656 // and seekable before the consumer runs), and return the file path. It
11657 // is unlinked at job end. This differs from `<(cmd)` below, which is a
11658 // /dev/fd pipe that is never waited on.
11659 if sub.source == "equalsubst" {
11660 let nam = crate::ported::utils::gettempname(None, true)
11661 .unwrap_or_else(|| format!("/tmp/zshrs_eqsub_{}", std::process::id()));
11662 let cpath = match std::ffi::CString::new(nam.as_str()) {
11663 Ok(c) => c,
11664 Err(_) => return String::from("/dev/null"),
11665 };
11666 // c:4945 — O_WRONLY|O_CREAT|O_EXCL|O_NOCTTY, 0600.
11667 let fd = unsafe {
11668 libc::open(
11669 cpath.as_ptr(),
11670 libc::O_WRONLY | libc::O_CREAT | libc::O_EXCL | libc::O_NOCTTY,
11671 0o600,
11672 )
11673 };
11674 if fd < 0 {
11675 return String::from("/dev/null");
11676 }
11677 let sub_for_child = sub.clone();
11678 match unsafe { libc::fork() } {
11679 -1 => {
11680 unsafe { libc::close(fd) };
11681 let _ = fs::remove_file(&nam);
11682 return String::from("/dev/null");
11683 }
11684 0 => {
11685 // c:4985 — child: stdout → the temp file, run the body, exit.
11686 // Clear the inherited pending-file list so this child never
11687 // unlinks the PARENT's =() temp files when its own commands
11688 // dispatch (fork copies the list; unlink hits the shared fs).
11689 PSUB_PENDING_FILES.with(|v| v.borrow_mut().clear());
11690 unsafe {
11691 libc::dup2(fd, libc::STDOUT_FILENO);
11692 libc::close(fd);
11693 }
11694 let mut vm = fusevm::VM::new(sub_for_child);
11695 register_builtins(&mut vm);
11696 vm.set_shell_host(Box::new(ZshrsHost));
11697 let _ = vm.run();
11698 let _ = std::io::stdout().flush();
11699 unsafe { libc::_exit(0) };
11700 }
11701 child_pid => {
11702 // c:4976-4980 — parent: close the write fd and WAIT so the
11703 // file is fully written before the consumer opens it.
11704 unsafe {
11705 libc::close(fd);
11706 let mut status: libc::c_int = 0;
11707 libc::waitpid(child_pid, &mut status, 0);
11708 }
11709 let depth = PSUB_SCOPE_DEPTH.with(|d| d.get());
11710 PSUB_PENDING_FILES.with(|v| v.borrow_mut().push((depth, nam.clone())));
11711 return nam;
11712 }
11713 }
11714 }
11715 // c:Src/exec.c::getproc — `<(cmd)` uses pipe + fork + the
11716 // `/dev/fd/N` filesystem entry (where N is the read end of
11717 // the pipe held open in the parent). Consumer opens
11718 // `/dev/fd/N`, reads the cmd's stdout through the pipe.
11719 // Both macOS and Linux expose `/dev/fd` for held-open file
11720 // descriptors. Previous Rust port captured stdout into
11721 // `/tmp/zshrs_psub_*` tempfiles synchronously — works for
11722 // `diff <(a) <(b)` style readers that scan once but diverges
11723 // from zsh's observable path string and breaks any consumer
11724 // that introspects the path or expects a non-seekable pipe.
11725 let mut fds: [libc::c_int; 2] = [-1, -1];
11726 if unsafe { libc::pipe(fds.as_mut_ptr()) } != 0 {
11727 // Pipe creation failed — fall back to tempfile so we at
11728 // least return SOMETHING.
11729 let fifo_path = format!(
11730 "/tmp/zshrs_psub_fallback_{}_{}",
11731 std::process::id(),
11732 with_executor(|e| {
11733 let n = e.process_sub_counter;
11734 e.process_sub_counter += 1;
11735 n
11736 })
11737 );
11738 let _ = fs::remove_file(&fifo_path);
11739 return fifo_path;
11740 }
11741 let (read_end, write_end) = (fds[0], fds[1]);
11742 let sub_for_child = sub.clone();
11743 match unsafe { libc::fork() } {
11744 -1 => {
11745 unsafe {
11746 libc::close(read_end);
11747 libc::close(write_end);
11748 }
11749 return String::from("/dev/null");
11750 }
11751 0 => {
11752 // Child: close read end, dup write end to stdout,
11753 // run the sub-chunk, exit. The exit closes the
11754 // write end automatically, so the parent's reader
11755 // gets EOF when the cmd finishes.
11756 PSUB_PENDING_FILES.with(|v| v.borrow_mut().clear());
11757 unsafe {
11758 libc::close(read_end);
11759 libc::dup2(write_end, libc::STDOUT_FILENO);
11760 libc::close(write_end);
11761 }
11762 // c:Src/exec.c:5101/5150 — `execode(prog, 0, 1, out ?
11763 // "outsubst" : "insubst");`. execode (c:1245-1266) APPENDS its
11764 // context for the duration of the body, so `<(cmd)` — whose child WRITES (out=1) — runs as `…:outsubst`.
11765 // zshrs's ported getproc carries these citations but is NOT the
11766 // live path (established in #1062) — the VM forks here instead,
11767 // so the push belongs in this child. No pop needed: this runs
11768 // INSIDE the forked child and dies with it. Bug #1069 (procsub
11769 // legs).
11770 if let Ok(mut ctx) = crate::ported::exec::zsh_eval_context.lock() {
11771 ctx.push("outsubst".to_string());
11772 let joined = ctx.join(":");
11773 if let Ok(mut tab) = crate::ported::params::paramtab().write() {
11774 if let Some(pm) = tab.get_mut("zsh_eval_context") {
11775 pm.u_arr = Some(ctx.clone());
11776 pm.node.flags &= !(crate::ported::zsh_h::PM_UNSET as i32);
11777 }
11778 if let Some(pm) = tab.get_mut("ZSH_EVAL_CONTEXT") {
11779 pm.u_str = Some(joined);
11780 pm.node.flags &= !(crate::ported::zsh_h::PM_UNSET as i32);
11781 }
11782 }
11783 }
11784 crate::fusevm_disasm::maybe_print_stdout("process_subst_in", &sub_for_child);
11785 let mut vm = fusevm::VM::new(sub_for_child);
11786 register_builtins(&mut vm);
11787 vm.set_shell_host(Box::new(ZshrsHost));
11788 let _ = vm.run();
11789 let _ = std::io::stdout().flush();
11790 unsafe { libc::_exit(0) };
11791 }
11792 child_pid => {
11793 // c:Src/exec.c:5092 `procsubstpid = pid;` — record the
11794 // forked child's PID so `${sysparams[procsubstpid]}`
11795 // returns it (was reading the never-updated atomic, so it
11796 // always came back 0). p10k's gitstatus daemon reads
11797 // `sysparams[procsubstpid]` right after `sysopen <(cmd)`
11798 // to track its worker PID; with 0 the daemon's self-check
11799 // failed and gitstatus fell back to re-downloading
11800 // gitstatusd — surfacing as "no prebuilt gitstatusd".
11801 crate::ported::exec::procsubstpid
11802 .store(child_pid, std::sync::atomic::Ordering::Relaxed);
11803 // Parent: close write end, keep read end open under
11804 // the same fd value so `/dev/fd/N` resolves to the
11805 // pipe's read side. NOTE: FD_CLOEXEC must STAY clear
11806 // — consumers like `cat <(cmd)` and `diff <(a) <(b)`
11807 // discover the fd via exec inheritance, so closing
11808 // on exec defeats the whole point. C zsh's getproc
11809 // (Src/exec.c:5045+) leaves the fd open across exec.
11810 unsafe {
11811 libc::close(write_end);
11812 }
11813 // Park read_end for close-after-consuming-command,
11814 // exactly like process_sub_out does for its write_end
11815 // (c:Src/exec.c addfilelist(NULL, fd) → deletefilelist).
11816 // WITHOUT this the parent's read_end stayed open for
11817 // the whole shell lifetime: p10k's async worker /
11818 // realtime clock do `exec {fd}< <(cmd)` on every prompt,
11819 // so each keystroke/redraw leaked a pipe fd until the
11820 // ~256-fd limit was hit and the shell locked up
11821 // (107 leaked pipes + 107 unreaped children observed).
11822 let depth = PSUB_SCOPE_DEPTH.with(|d| d.get());
11823 PSUB_PENDING_FDS.with(|v| v.borrow_mut().push((depth, read_end)));
11824 // Reap the forked child so it doesn't linger as a
11825 // zombie. `<(cmd)` children are fire-and-forget (their
11826 // output flows through the pipe); C reaps them via the
11827 // job machinery. A non-blocking reap here is scheduled;
11828 // do a best-effort WNOHANG now and the rest drain on
11829 // subsequent proc-subs / prompt cycles.
11830 crate::fusevm_bridge::note_psub_child(child_pid);
11831 }
11832 }
11833 format!("/dev/fd/{}", read_end)
11834 }
11835
11836 fn process_sub_out(&mut self, sub: &fusevm::Chunk) -> String {
11837 // c:Src/exec.c:5025 getproc, PATH_DEV_FD branch — `>(cmd)`
11838 // (out == 0): `mpipe(pipes)`, fork; the CHILD `redup(pipes[0],
11839 // 0)` (pipe read end onto stdin) and `closem` drops the write
11840 // end; the PARENT closes pipes[0] and hands the consumer
11841 // `/dev/fd/<pipes[1]>` (the write end). The previous Rust port
11842 // used mkfifo + a child that BLOCKED in open(FIFO, O_RDONLY)
11843 // before running cmd — with no writer the child never started,
11844 // never exited, and kept its inherited stdout (e.g. a `$()`
11845 // capture pipe) open forever: `a=$(print -r -- >(true))` hung.
11846 // With the pipe shape the child runs immediately and exits,
11847 // releasing inherited fds exactly like zsh (verified: zsh
11848 // blocks ~2s on `a=$(print -r -- >(sleep 2))`, then EOFs).
11849 let mut fds: [libc::c_int; 2] = [-1, -1];
11850 if unsafe { libc::pipe(fds.as_mut_ptr()) } != 0 {
11851 // Pipe creation failed — fall back to a plain temp file so
11852 // the consumer at least has a writable path.
11853 let fallback = format!(
11854 "/tmp/zshrs_psub_out_{}_{}",
11855 std::process::id(),
11856 with_executor(|e| {
11857 let n = e.process_sub_counter;
11858 e.process_sub_counter += 1;
11859 n
11860 })
11861 );
11862 let _ = fs::write(&fallback, "");
11863 return fallback;
11864 }
11865 let (read_end, write_end) = (fds[0], fds[1]);
11866 let sub_for_child = sub.clone();
11867 match unsafe { libc::fork() } {
11868 -1 => {
11869 unsafe {
11870 libc::close(read_end);
11871 libc::close(write_end);
11872 }
11873 String::from("/dev/null")
11874 }
11875 0 => {
11876 // Child: close the write end (c: closem after redup),
11877 // dup the read end onto stdin (c: redup(pipes[0], 0)),
11878 // run the sub-chunk, exit. Other std fds stay
11879 // inherited — zsh's child keeps the surrounding
11880 // command's stdout/stderr.
11881 unsafe {
11882 libc::close(write_end);
11883 libc::dup2(read_end, libc::STDIN_FILENO);
11884 libc::close(read_end);
11885 }
11886 // c:Src/exec.c:5101/5150 — `execode(prog, 0, 1, out ?
11887 // "outsubst" : "insubst");`. execode (c:1245-1266) APPENDS its
11888 // context for the duration of the body, so `>(cmd)` — whose child READS (out=0) — runs as `…:insubst`.
11889 // zshrs's ported getproc carries these citations but is NOT the
11890 // live path (established in #1062) — the VM forks here instead,
11891 // so the push belongs in this child. No pop needed: this runs
11892 // INSIDE the forked child and dies with it. Bug #1069 (procsub
11893 // legs).
11894 if let Ok(mut ctx) = crate::ported::exec::zsh_eval_context.lock() {
11895 ctx.push("insubst".to_string());
11896 let joined = ctx.join(":");
11897 if let Ok(mut tab) = crate::ported::params::paramtab().write() {
11898 if let Some(pm) = tab.get_mut("zsh_eval_context") {
11899 pm.u_arr = Some(ctx.clone());
11900 pm.node.flags &= !(crate::ported::zsh_h::PM_UNSET as i32);
11901 }
11902 if let Some(pm) = tab.get_mut("ZSH_EVAL_CONTEXT") {
11903 pm.u_str = Some(joined);
11904 pm.node.flags &= !(crate::ported::zsh_h::PM_UNSET as i32);
11905 }
11906 }
11907 }
11908 crate::fusevm_disasm::maybe_print_stdout("process_subst_out:child", &sub_for_child);
11909 let mut vm = fusevm::VM::new(sub_for_child);
11910 register_builtins(&mut vm);
11911 vm.set_shell_host(Box::new(ZshrsHost));
11912 let _ = vm.run();
11913 unsafe { libc::_exit(0) };
11914 }
11915 child_pid => {
11916 // c:Src/exec.c:5143 `procsubstpid = pid;` — same fix as
11917 // the `<(cmd)` in-path above: record the forked child's
11918 // PID for `${sysparams[procsubstpid]}` (was always 0).
11919 crate::ported::exec::procsubstpid
11920 .store(child_pid, std::sync::atomic::Ordering::Relaxed);
11921 // Parent: close the read end, keep the write end open
11922 // under its fd value so `/dev/fd/N` resolves to the
11923 // pipe's write side. FD_CLOEXEC must STAY clear —
11924 // consumers (`tee >(cmd)`) discover the fd via exec
11925 // inheritance, matching process_sub_in above and C's
11926 // fdtable[fd] = FDT_PROC_SUBST bookkeeping. Park the
11927 // fd for close-after-consuming-command (c: addfilelist
11928 // (NULL, fd) → deletefilelist) so the child's reader
11929 // EOFs — without this `tee >(wc -c) </dev/null` left
11930 // wc blocked until shell exit.
11931 unsafe {
11932 libc::close(read_end);
11933 }
11934 let depth = PSUB_SCOPE_DEPTH.with(|d| d.get());
11935 PSUB_PENDING_FDS.with(|v| v.borrow_mut().push((depth, write_end)));
11936 format!("/dev/fd/{}", write_end)
11937 }
11938 }
11939 }
11940
11941 fn subshell_begin(&mut self) {
11942 with_executor(|exec| {
11943 // Special parameters whose value lives in a process GLOBAL behind a
11944 // GSU (`Src/params.c`'s `ifs`, `wordchars`, `home`, `histsiz`, …)
11945 // rather than in the param table. Mirrors the getfn dispatch list at
11946 // params.rs:12548. C isolates these for free by forking `(...)`;
11947 // zshrs's in-process subshell has to snapshot them by hand, or a
11948 // subshell-local `IFS=,` rewrites the parent's word-splitting.
11949 const SUBSHELL_SPECIAL_GLOBALS: &[&str] = &[
11950 "IFS",
11951 "HOME",
11952 "TERM",
11953 "USERNAME",
11954 "WORDCHARS",
11955 "TERMINFO",
11956 "TERMINFO_DIRS",
11957 "KEYBOARD_HACK",
11958 "histchars",
11959 "HISTSIZE",
11960 "SAVEHIST",
11961 ];
11962 // An UNSET special yields None and is skipped: the paramtab snapshot
11963 // restores its PM_UNSET flag, and the getfn dispatch refuses to read
11964 // the (stale) global while PM_UNSET is set (params.rs:12552).
11965 let special_globals_snap: Vec<(String, String)> = SUBSHELL_SPECIAL_GLOBALS
11966 .iter()
11967 .filter_map(|n| {
11968 crate::ported::params::getsparam(n).map(|v| ((*n).to_string(), v))
11969 })
11970 .collect();
11971 // libc::umask returns the previous mask AND sets the new
11972 // one; call with current value to read without changing.
11973 let cur_umask = unsafe {
11974 let m = libc::umask(0o022);
11975 libc::umask(m);
11976 m as u32
11977 };
11978 // Snapshot paramtab + hashed-storage too (step 1 of the
11979 // store unification mirrors writes there; restoring only
11980 // the HashMaps leaks subshell-scoped writes to the parent
11981 // via paramtab readers like `paramsubst → vars_get`).
11982 let paramtab_snap = crate::ported::params::paramtab()
11983 .read()
11984 .ok()
11985 .map(|t| t.clone())
11986 .unwrap_or_default();
11987 let paramtab_hashed_snap = crate::ported::params::paramtab_hashed_storage()
11988 .lock()
11989 .ok()
11990 .map(|m| m.clone())
11991 .unwrap_or_default();
11992 exec.subshell_snapshots.push(SubshellSnapshot {
11993 paramtab: paramtab_snap,
11994 paramtab_hashed_storage: paramtab_hashed_snap,
11995 special_globals: special_globals_snap,
11996 positional_params: exec.pparams(),
11997 env_vars: env::vars().collect(),
11998 // Save the LOGICAL pwd ($PWD env), not `current_dir()`'s
11999 // symlink-resolved path. zsh's subshell isolation per
12000 // Src/exec.c at the `entersubsh` path treats `pwd` (the
12001 // shell-tracked logical PWD) as the carrier — see
12002 // `Src/builtin.c:1239-1242` where cd writes the logical
12003 // dest into `pwd`. Falling back to current_dir() only
12004 // when PWD is unset matches `setupvals` at
12005 // `Src/init.c:1100+`.
12006 cwd: env::var("PWD")
12007 .ok()
12008 .map(PathBuf::from)
12009 .or_else(|| env::current_dir().ok()),
12010 umask: cur_umask,
12011 // Snapshot canonical `traps_table` — bin_trap writes
12012 // there (`Src/builtin.c`).
12013 traps: crate::ported::builtin::traps_table()
12014 .lock()
12015 .map(|t| t.clone())
12016 .unwrap_or_default(),
12017 // Snapshot option store so `(set -e)` /
12018 // `(setopt extendedglob)` don't leak to parent.
12019 opts: crate::ported::options::opt_state_snapshot(),
12020 // c:Src/exec.c — fork() copies the alias table to
12021 // the subshell. `(alias x=y)` inside the subshell
12022 // dies with the child; the parent doesn't see x.
12023 // Snapshot here so subshell_end can restore.
12024 // Bug #209 in docs/BUGS.md.
12025 aliases: crate::ported::hashtable::aliastab_lock()
12026 .read()
12027 .ok()
12028 .map(|t| {
12029 t.iter()
12030 .map(|(k, v)| (k.clone(), v.text.clone(), v.node.flags))
12031 .collect()
12032 })
12033 .unwrap_or_default(),
12034 // c:Src/exec.c::entersubsh — same fork-copy
12035 // semantics for shfunctab. `(f() { ... })` defined
12036 // inside the subshell dies with the child; parent's
12037 // `type f` reports "not found". Bug #208 in
12038 // docs/BUGS.md.
12039 shfuncs: crate::ported::hashtable::shfunctab_lock()
12040 .read()
12041 .ok()
12042 .map(|t| t.snapshot())
12043 .unwrap_or_default(),
12044 functions_compiled: exec.functions_compiled.clone(),
12045 function_source: exec.function_source.clone(),
12046 // c:Src/exec.c::entersubsh — subshell forks its own
12047 // modulestab. A `(zmodload zsh/X)` inside the
12048 // subshell flips MOD_INIT_B on the CHILD's
12049 // modulestab; when the child exits the change
12050 // dies with it. zshrs's in-process subshell would
12051 // otherwise leak the load to the parent.
12052 // Bug #210 in docs/BUGS.md. Snapshot just the
12053 // (name → flags) pairs since the only mutating
12054 // field is the flags bitmask (MOD_INIT_B for
12055 // loaded, MOD_UNLOAD for unloaded).
12056 modules: crate::ported::module::MODULESTAB
12057 .lock()
12058 .ok()
12059 .map(|t| {
12060 t.modules
12061 .iter()
12062 .map(|(k, v)| (k.clone(), v.node.flags))
12063 .collect()
12064 })
12065 .unwrap_or_default(),
12066 // c:Src/exec.c::entersubsh — fork-copy semantics for
12067 // THINGYTAB (ZLE widget registry). A subshell `zle -N`
12068 // / `zle -D` mutation dies with the child in C zsh;
12069 // mirror via in-process snapshot. Bug #453.
12070 thingytab: crate::ported::zle::zle_thingy::thingytab()
12071 .lock()
12072 .ok()
12073 .map(|t| t.clone())
12074 .unwrap_or_default(),
12075 // c:Src/exec.c::entersubsh — same fork-copy for the
12076 // KEYMAPNAMTAB (named keymap registry). `bindkey -N km`
12077 // / `bindkey -D km` inside a subshell dies with the
12078 // child. Bug #454.
12079 keymapnamtab: crate::ported::zle::zle_keymap::keymapnamtab()
12080 .lock()
12081 .ok()
12082 .map(|t| t.clone())
12083 .unwrap_or_default(),
12084 // c:Src/exec.c::entersubsh fork semantics — `$!`
12085 // (clone::lastpid) set by a `&` INSIDE the subshell
12086 // dies with the child: `( : & ); echo $!` -> 0.
12087 lastpid: crate::ported::modules::clone::lastpid
12088 .load(std::sync::atomic::Ordering::Relaxed),
12089 // c:Src/exec.c::entersubsh fork semantics — the
12090 // subshell gets a COPY of the job table; its disown/
12091 // wait/`&` mutations die with it. Bug #462.
12092 jobtab: crate::ported::jobs::JOBTAB
12093 .get_or_init(|| std::sync::Mutex::new(Vec::new()))
12094 .lock()
12095 .map(|t| t.clone())
12096 .unwrap_or_default(),
12097 curjob: *crate::ported::jobs::CURJOB
12098 .get_or_init(|| std::sync::Mutex::new(-1))
12099 .lock()
12100 .unwrap(),
12101 prevjob: *crate::ported::jobs::PREVJOB
12102 .get_or_init(|| std::sync::Mutex::new(-1))
12103 .lock()
12104 .unwrap(),
12105 maxjob: *crate::ported::jobs::MAXJOB
12106 .get_or_init(|| std::sync::Mutex::new(0))
12107 .lock()
12108 .unwrap(),
12109 thisjob: *crate::ported::jobs::THISJOB
12110 .get_or_init(|| std::sync::Mutex::new(-1))
12111 .lock()
12112 .unwrap(),
12113 // c:Src/exec.c entersubsh — fork copies the fd table;
12114 // the child's `exec >file` / `exec N<&-` mutations die
12115 // with it. Dup each user-range fd to >= 10 so
12116 // subshell_end can restore the parent's exact table.
12117 saved_fds: (0..10)
12118 .map(|fd| {
12119 let dup = unsafe { libc::fcntl(fd, libc::F_DUPFD, 10) };
12120 (fd, dup)
12121 })
12122 .collect(),
12123 });
12124 // C forks for `(...)` — count the fork-equivalent so
12125 // `time (builtin)` reports like zsh (see FORK_EVENTS).
12126 crate::vm_helper::FORK_EVENTS.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
12127 // c:Src/exec.c:1088-1092 — entersubsh resets traps in the child:
12128 // if (!(flags & ESUB_KEEPTRAP))
12129 // for (sig = 0; sig <= SIGCOUNT; sig++)
12130 // if (!(sigtrapped[sig] & ZSIG_FUNC) &&
12131 // !(isset(POSIXTRAPS) && (sigtrapped[sig] & ZSIG_IGNORED)))
12132 // unsettrap(sig);
12133 //
12134 // A subshell does NOT inherit string-form traps. Two exemptions:
12135 // FUNCTION-form traps (`TRAPUSR1() { … }`, ZSIG_FUNC) survive, and
12136 // under POSIX_TRAPS an IGNORED trap (`trap '' SIG`) survives.
12137 //
12138 // The ZSIG_FUNC exemption is structural here rather than a flag
12139 // test: zshrs keeps string-form bodies in traps_table and
12140 // function-form ones in shfunctab as TRAPxxx, so filtering only
12141 // traps_table leaves the function form untouched by construction.
12142 // ZSIG_IGNORED is `trap '' SIG`, which stores an empty body.
12143 //
12144 // The LOOP BOUND is also part of the spec, not an implementation
12145 // detail: `sig <= SIGCOUNT` never reaches the PSEUDO-signals,
12146 // which zsh numbers above the real ones —
12147 // #define SIGZERR (SIGCOUNT+1)
12148 // #define SIGDEBUG (SIGCOUNT+2) (c:Src/signals.h:34-35)
12149 // so ERR/ZERR and DEBUG traps SURVIVE a subshell, while SIGEXIT
12150 // (sig 0) is inside the loop and is cleared. Verified against the
12151 // oracle: `trap 'print e' ERR; (trap)` lists the ERR trap;
12152 // `trap 'print u' USR1; (trap)` lists nothing.
12153 //
12154 // Without this a subshell kept the parent's traps: `(trap)` listed
12155 // them where zsh lists nothing, and — the part that isn't
12156 // cosmetic — an inherited trap FIRED inside the child, so
12157 // `trap 'print p' USR1; (kill -USR1 $$; print after)` printed
12158 // p before after instead of after…p (the signal is meant to reach
12159 // the parent, whose trap runs there).
12160 //
12161 // The snapshot pushed above restores the parent's table at
12162 // subshell_end, which is what makes clearing safe for zshrs's
12163 // in-process subshell.
12164 {
12165 let posixtraps = crate::ported::zsh_h::isset(crate::ported::zsh_h::POSIXTRAPS);
12166 if let Ok(mut t) = crate::ported::builtin::traps_table().lock() {
12167 t.retain(|name, body| {
12168 // Above SIGCOUNT — outside c:1088's loop entirely.
12169 if name == "ERR" || name == "ZERR" || name == "DEBUG" {
12170 return true;
12171 }
12172 // c:1090-1092 — otherwise keep ONLY (POSIXTRAPS && ignored).
12173 posixtraps && body.is_empty()
12174 });
12175 }
12176 }
12177 // c:Src/exec.c:2862 — subshell fork flags carry ESUB_PGRP,
12178 // so entersubsh runs `clearjobtab(monitor)` (c:1219): the
12179 // child gets an EMPTY job table plus the procless control
12180 // job grabbed at Src/jobs.c:1828 (`thisjob = initjob()`).
12181 // That's why zsh's `(jobs)` prints nothing and `(kill %1)`
12182 // hits the empty control job instead of the parent's job 1.
12183 // The snapshot pushed above restores the parent's table at
12184 // subshell_end. Bug #462.
12185 let monitor = crate::ported::zsh_h::isset(crate::ported::zsh_h::MONITOR) as i32;
12186 crate::ported::jobs::clearjobtab(&mut exec.jobs, monitor);
12187 // clearjobtab left THISJOB on the control job (Src/jobs.c:
12188 // 1828). In C the very next pipeline's execpline reassigns
12189 // thisjob (Src/exec.c:1700 `thisjob = newjob = initjob()`),
12190 // so by the time any builtin runs, thisjob never aliases
12191 // the control job. zshrs has no per-pipeline job slot —
12192 // model the between-pipelines state (-1) so getjob's
12193 // `jobnum != thisjob` (c:jobs.c:2107) doesn't reject %1 and
12194 // setcurjob doesn't demote an inherited curjob that
12195 // collides with the control slot.
12196 *crate::ported::jobs::THISJOB
12197 .get_or_init(|| std::sync::Mutex::new(-1))
12198 .lock()
12199 .unwrap() = -1;
12200 // Subshell starts with EXIT trap cleared so the parent's
12201 // EXIT handler doesn't fire when the subshell ends. zsh:
12202 // each subshell has its own trap context. Other signals
12203 // are inherited (well, parent's are still in place — but
12204 // a trap set INSIDE the subshell shouldn't leak out).
12205 if let Ok(mut t) = crate::ported::builtin::traps_table().lock() {
12206 t.remove("EXIT");
12207 }
12208 let level = exec
12209 .scalar("ZSH_SUBSHELL")
12210 .and_then(|s| s.parse::<i32>().ok())
12211 .unwrap_or(0);
12212 // c:Src/exec.c — ZSH_SUBSHELL carries PM_READONLY (declared
12213 // in params.rs special_params); setsparam would be rejected
12214 // by assignstrvalue's PM_READONLY guard. Write u_val
12215 // directly — same bypass pattern as BUILTIN_SET_LINENO at
12216 // line 2784. C zsh's PM_SPECIAL GSU vtable handles this
12217 // implicitly via the setfn callback.
12218 let new_level = (level + 1) as i64;
12219 if let Ok(mut tab) = crate::ported::params::paramtab().write() {
12220 if let Some(pm) = tab.get_mut("ZSH_SUBSHELL") {
12221 pm.u_val = new_level;
12222 pm.u_str = Some(new_level.to_string());
12223 pm.node.flags &= !(crate::ported::zsh_h::PM_UNSET as i32);
12224 }
12225 }
12226 });
12227 // Bump SUBSHELL_DEPTH so zexit defers process::exit (see
12228 // SUBSHELL_DEPTH declaration in src/ported/builtin.rs for
12229 // rationale).
12230 crate::ported::builtin::SUBSHELL_DEPTH.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
12231 // c:Src/exec.c::entersubsh — C zsh's subshell is a forked
12232 // child process: signals sent to the parent (via `kill $$`
12233 // inside the subshell, where `$$` is the parent's pid)
12234 // never reach the child's signal handlers. zshrs's
12235 // in-process subshell shares the process pid with the
12236 // parent, so without queueing the subshell's trap handler
12237 // fires for signals that zsh would deliver only to the
12238 // parent. Queue signals across the subshell body so the
12239 // parent's restored trap table sees them after
12240 // subshell_end's unqueue drain. Bug #450.
12241 crate::ported::signals_h::queue_signals();
12242 }
12243
12244 fn subshell_end(&mut self) -> Option<i32> {
12245 // Fire subshell's EXIT trap BEFORE restoring parent state so
12246 // the trap body sees the subshell's vars and exit status. zsh
12247 // forks for `(...)` so the trap runs in the child process,
12248 // before exit. We mirror by running it here, just before the
12249 // pop+restore. REMOVE the trap before firing so the inner
12250 // execute_script doesn't fire it again at its own end.
12251 let exit_trap_body = crate::ported::builtin::traps_table()
12252 .lock()
12253 .ok()
12254 .and_then(|mut t| t.remove("EXIT"));
12255 if let Some(body) = exit_trap_body {
12256 // Execute the trap body. Errors during trap execution
12257 // don't bubble — zsh ignores trap-body errors.
12258 with_executor(|exec| {
12259 let _ = exec.execute_script(&body);
12260 });
12261 }
12262 with_executor(|exec| {
12263 if let Some(snap) = exec.subshell_snapshots.pop() {
12264 // Restore paramtab + hashed storage so subshell-scoped
12265 // writes via setsparam/setaparam/sethparam don't leak
12266 // to the parent via paramtab readers.
12267 if let Some(tab) = crate::ported::params::paramtab()
12268 .write()
12269 .ok()
12270 .as_deref_mut()
12271 {
12272 *tab = snap.paramtab;
12273 }
12274 // Restore the global-backed specials (see
12275 // SubshellSnapshot::special_globals). MUST run after the
12276 // paramtab restore above: setsparam writes through the GSU setfn
12277 // to BOTH the process global and the param node, so the paramtab
12278 // overwrite would otherwise clobber the node half of it.
12279 for (name, val) in &snap.special_globals {
12280 crate::ported::params::setsparam(name, val);
12281 }
12282 // c:Src/exec.c::entersubsh fork semantics — restore
12283 // the parent's `$!`; a background job inside `(...)`
12284 // dies with the child in C zsh.
12285 crate::ported::modules::clone::lastpid
12286 .store(snap.lastpid, std::sync::atomic::Ordering::Relaxed);
12287 // c:Src/exec.c::entersubsh fork semantics — restore the
12288 // parent's job table + curjob/prevjob/maxjob/thisjob.
12289 // The subshell mutated only its own copy. Bug #462.
12290 if let Ok(mut t) = crate::ported::jobs::JOBTAB
12291 .get_or_init(|| std::sync::Mutex::new(Vec::new()))
12292 .lock()
12293 {
12294 *t = snap.jobtab;
12295 }
12296 *crate::ported::jobs::CURJOB
12297 .get_or_init(|| std::sync::Mutex::new(-1))
12298 .lock()
12299 .unwrap() = snap.curjob;
12300 *crate::ported::jobs::PREVJOB
12301 .get_or_init(|| std::sync::Mutex::new(-1))
12302 .lock()
12303 .unwrap() = snap.prevjob;
12304 *crate::ported::jobs::MAXJOB
12305 .get_or_init(|| std::sync::Mutex::new(0))
12306 .lock()
12307 .unwrap() = snap.maxjob;
12308 *crate::ported::jobs::THISJOB
12309 .get_or_init(|| std::sync::Mutex::new(-1))
12310 .lock()
12311 .unwrap() = snap.thisjob;
12312 if let Some(m) = crate::ported::params::paramtab_hashed_storage()
12313 .lock()
12314 .ok()
12315 .as_deref_mut()
12316 {
12317 *m = snap.paramtab_hashed_storage;
12318 }
12319 exec.set_pparams(snap.positional_params);
12320 // Restore the OS env to its pre-subshell state.
12321 // Removes any `export` writes the subshell made, and
12322 // restores any vars the subshell unset. Without this
12323 // `(export y=sub)` would leak `y` to the parent shell.
12324 let current: HashMap<String, String> = env::vars().collect();
12325 for k in current.keys() {
12326 if !snap.env_vars.contains_key(k) {
12327 env::remove_var(k);
12328 }
12329 }
12330 for (k, v) in &snap.env_vars {
12331 if current.get(k) != Some(v) {
12332 env::set_var(k, v);
12333 }
12334 }
12335 if let Some(cwd) = snap.cwd {
12336 let _ = env::set_current_dir(&cwd);
12337 // Resync $PWD env so a parent `pwd` doesn't read
12338 // the cwd the subshell `cd`'d into.
12339 env::set_var("PWD", &cwd);
12340 }
12341 // Restore umask. zsh's `(umask 077)` doesn't leak to
12342 // parent because the subshell forks; we run in-process
12343 // so we manually reset.
12344 unsafe {
12345 libc::umask(snap.umask as libc::mode_t);
12346 }
12347 // Restore parent's traps (the subshell's own traps die
12348 // with it). zsh: `(trap "X" USR1)` doesn't leak the
12349 // USR1 trap out of the subshell. Write back to the
12350 // canonical `traps_table` (bin_trap writes there).
12351 if let Ok(mut t) = crate::ported::builtin::traps_table().lock() {
12352 *t = snap.traps;
12353 }
12354 // Restore parent's option store so `(set -e)` /
12355 // `(setopt extendedglob)` don't leak. zsh forks
12356 // subshells so child option changes die with the
12357 // child; we run in-process and must restore.
12358 crate::ported::options::opt_state_restore(snap.opts);
12359 // c:Src/exec.c — fork() means alias mutations in a
12360 // subshell die with the child. Restore parent's
12361 // alias table from snapshot. Clear current entries
12362 // then re-add parent's. Bug #209 in docs/BUGS.md.
12363 if let Ok(mut tab) = crate::ported::hashtable::aliastab_lock().write() {
12364 tab.clear();
12365 for (name, text, flags) in snap.aliases {
12366 tab.add(crate::ported::zsh_h::alias {
12367 node: crate::ported::zsh_h::hashnode {
12368 next: None,
12369 nam: name,
12370 // ALIAS_GLOBAL / DISABLED must survive the
12371 // round-trip — flags:0 turned every global
12372 // alias regular on ANY subshell exit.
12373 flags,
12374 },
12375 text,
12376 inuse: 0,
12377 });
12378 }
12379 }
12380 // c:Src/exec.c::entersubsh — same fork-copy
12381 // semantics for shfunctab. Restore parent's function
12382 // table from snapshot so `(f() { ... })` definitions
12383 // inside the subshell don't leak to the parent.
12384 // Bug #208 in docs/BUGS.md.
12385 if let Ok(mut tab) = crate::ported::hashtable::shfunctab_lock().write() {
12386 tab.restore(snap.shfuncs);
12387 }
12388 // Restore the runtime dispatch tables (compiled chunks
12389 // + source). Without these, a subshell-defined
12390 // override leaves its bytecode in place even after
12391 // shfunctab is restored — `g` after the subshell would
12392 // still run the override.
12393 exec.functions_compiled = snap.functions_compiled;
12394 exec.function_source = snap.function_source;
12395 // c:Src/exec.c::entersubsh — restore parent's
12396 // modulestab so a subshell `(zmodload zsh/X)` doesn't
12397 // leak to the parent. Bug #210 in docs/BUGS.md.
12398 // Restore via per-module flag write since the
12399 // snapshot is `(name → flags)` only.
12400 if let Ok(mut t) = crate::ported::module::MODULESTAB.lock() {
12401 for (name, saved_flags) in &snap.modules {
12402 if let Some(m) = t.modules.get_mut(name) {
12403 m.node.flags = *saved_flags;
12404 }
12405 }
12406 }
12407 // c:Src/exec.c::entersubsh — restore parent's THINGYTAB
12408 // so a subshell's `zle -N w f` / `zle -D w` doesn't
12409 // affect the parent's widget registry. Bug #453.
12410 if let Ok(mut t) = crate::ported::zle::zle_thingy::thingytab().lock() {
12411 *t = snap.thingytab;
12412 }
12413 // Same for KEYMAPNAMTAB. Bug #454.
12414 if let Ok(mut t) = crate::ported::zle::zle_keymap::keymapnamtab().lock() {
12415 *t = snap.keymapnamtab;
12416 }
12417 // c:Src/exec.c entersubsh fork semantics — restore the
12418 // parent's user-range fd table. A bare `exec >file` /
12419 // `exec N>&-` inside `(...)` died with the C child;
12420 // the in-process subshell must undo it here. Flush
12421 // Rust's stdout buffer FIRST so bytes the subshell
12422 // printed drain to the SUBSHELL's fd 1, not the
12423 // restored parent fd.
12424 {
12425 use std::io::Write;
12426 let _ = std::io::stdout().flush();
12427 }
12428 for (fd, saved) in snap.saved_fds {
12429 unsafe {
12430 if saved >= 0 {
12431 libc::dup2(saved, fd);
12432 libc::close(saved);
12433 } else {
12434 // fd was closed at entry; close whatever
12435 // the subshell opened on that slot.
12436 libc::close(fd);
12437 }
12438 }
12439 }
12440 }
12441 });
12442 // Decrement SUBSHELL_DEPTH. If a deferred subshell exit
12443 // landed inside (EXIT_PENDING set with depth > 0), promote
12444 // the deferred status into the subshell's exit status now
12445 // that we're at the boundary, then clear so the parent
12446 // continues. Matches C zsh's "subshell-exit-via-fork"
12447 // boundary where the child's process::exit(N) becomes
12448 // $WAITSTATUS / $? in the parent.
12449 crate::ported::builtin::SUBSHELL_DEPTH.fetch_sub(1, std::sync::atomic::Ordering::Relaxed);
12450 // c:Src/exec.c — drain the signal queue against the now-
12451 // restored parent trap table. Pairs with the
12452 // queue_signals() call at the end of subshell_begin.
12453 // Any `kill $$` from inside the subshell is processed
12454 // here against OUTER's trap, matching C zsh's
12455 // signal-delivery-to-parent semantics. Bug #450.
12456 crate::ported::signals_h::unqueue_signals();
12457 // c:Src/exec.c — a `( … )` subshell is a FORK in C: an errflag
12458 // abort inside the child ends the child with its lastval as
12459 // the exit status, and the flag dies with the child process.
12460 // The parent's $? picks up the status and the parent's lists
12461 // keep running. zsh 5.9: `(readonly r=1; r=2); echo "after
12462 // $?"` prints `after 1`. zshrs runs the subshell in-process,
12463 // so mirror the fork isolation by clearing ERRFLAG_ERROR at
12464 // the subshell boundary — exec.last_status() already carries
12465 // the child's lastval (synced by ERREXIT_CHECK trigger 4).
12466 //
12467 // ERRFLAG_HARD must die at this boundary too: `${u:?msg}` sets
12468 // errflag |= ERRFLAG_HARD (c:Src/subst.c:3344) and then, in a
12469 // C forked subshell, `_exit(1)` (c:3353) — the parent never
12470 // sees ANY errflag bit. A leaked HARD bit here made every
12471 // subsequent zerr() take the silent arm (c:Src/utils.c:175-177
12472 // `if (errflag || noerrs) { errflag |= ERRFLAG_ERROR; return; }`),
12473 // so the next eval/source's parse silently "failed" and the
12474 // D04 harness shell wedged after chunk 10's
12475 // `(print ${unset1:?exiting1})`.
12476 crate::ported::utils::errflag.fetch_and(
12477 !(crate::ported::zsh_h::ERRFLAG_ERROR | crate::ported::zsh_h::ERRFLAG_HARD),
12478 std::sync::atomic::Ordering::Relaxed,
12479 );
12480 let exit_pending =
12481 crate::ported::builtin::EXIT_PENDING.load(std::sync::atomic::Ordering::Relaxed);
12482 if exit_pending != 0 {
12483 // c:Src/builtin.c — `exit N` masks N to 8 bits because
12484 // POSIX _exit takes the low byte as status. `(exit 256)`
12485 // and `(exit 0)` are indistinguishable to the parent;
12486 // `(exit 257)` exits with 1. Without the mask zshrs's
12487 // in-process subshell propagated the full i32 (256) into
12488 // the parent's $?, diverging from zsh.
12489 let raw = crate::ported::builtin::EXIT_VAL.load(std::sync::atomic::Ordering::Relaxed);
12490 let val = raw & 0xFF;
12491 with_executor(|exec| exec.set_last_status(val));
12492 crate::ported::builtin::EXIT_PENDING.store(0, std::sync::atomic::Ordering::Relaxed);
12493 crate::ported::builtin::RETFLAG.store(0, std::sync::atomic::Ordering::Relaxed);
12494 crate::ported::builtin::BREAKS.store(0, std::sync::atomic::Ordering::Relaxed);
12495 // Return the deferred-exit status so the VM updates its
12496 // own `last_status`. Otherwise run_chunk's post-script
12497 // `set_last_status(vm.last_status)` would clobber LASTVAL
12498 // back to the stale pre-subshell value.
12499 return Some(val);
12500 }
12501 None
12502 }
12503
12504 fn redirect(&mut self, fd: u8, op: u8, target: &str) {
12505 // Apply a redirection at the OS level for the next command/builtin.
12506 // The host tracks saved fds in a per-executor stack so a future
12507 // `with_redirects_end` can restore. For now, this is a thin wrapper
12508 // that performs the dup2; pairing with explicit save/restore is
12509 // delivered by `with_redirects_begin/end`.
12510 with_executor(|exec| exec.host_apply_redirect(fd, op, target));
12511 }
12512
12513 fn with_redirects_begin(&mut self, count: u8) {
12514 with_executor(|exec| exec.host_redirect_scope_begin(count));
12515 }
12516
12517 fn regex_match(&mut self, s: &str, regex: &str) -> bool {
12518 // c:Src/Modules/regex.c:54 `zcond_regex_match` — POSIX ERE
12519 // matching + populate `$MATCH` / `$MBEGIN` / `$MEND` /
12520 // `$match[]` / `$mbegin[]` / `$mend[]` (or `$BASH_REMATCH`
12521 // under BASHREMATCH). Direct delegation to the canonical
12522 // port at src/ported/modules/regex.rs:58.
12523 //
12524 // The bridge passthru path delivers TOKEN-form bytes here
12525 // (Inbrack \u{91}, Outbrack \u{92}, Star \u{87}, Quest
12526 // \u{86}, etc.) since the lexer tokenizes regex meta chars
12527 // inside `[[ ]]`. The host regex engine expects ASCII, so
12528 // untokenize the pattern (and subject, for safety) once at
12529 // this boundary. zsh C reaches its POSIX-ERE engine through
12530 // the same untokenize path inside zcond_regex_match.
12531 let s_clean = crate::lex::untokenize(s);
12532 let regex_clean = crate::lex::untokenize(regex);
12533 // c:Src/cond.c:113-119 — WHICH engine `=~` uses is an option:
12534 //
12535 // char *modname = isset(REMATCHPCRE) ? "zsh/pcre" : "zsh/regex";
12536 //
12537 // and the two speak different languages (POSIX ERE vs PCRE), so the
12538 // option decides whether `\d` is a digit class or a literal `d`, and
12539 // whether `(?<name>…)` compiles at all. This dispatch was missing:
12540 // `=~` always used the regex module, so `setopt rematchpcre` silently
12541 // did nothing.
12542 if crate::ported::zsh_h::isset(crate::ported::zsh_h::REMATCHPCRE) {
12543 // c:115 — "zsh/pcre" → the `-pcre-match` cond.
12544 crate::ported::modules::pcre::cond_pcre_match(
12545 &[s_clean, regex_clean],
12546 crate::ported::modules::pcre::CPCRE_PLAIN,
12547 ) != 0
12548 } else {
12549 // c:115 — "zsh/regex" → the `-regex-match` cond.
12550 crate::ported::modules::regex::zcond_regex_match(
12551 &[s_clean.as_str(), regex_clean.as_str()],
12552 crate::ported::modules::regex::ZREGEX_EXTENDED,
12553 ) != 0
12554 }
12555 }
12556
12557 fn with_redirects_end(&mut self) {
12558 with_executor(|exec| exec.host_redirect_scope_end());
12559 // c:Src/exec.c:5172 — if any redirect in this scope failed
12560 // (noclobber-blocked, ENOENT for read, etc.), the command's
12561 // exit status is forced to 1 regardless of what the (still-
12562 // executed) command's own exit was. C zsh prevents the
12563 // command from running at all when a redirect fails; the
12564 // Rust port still runs it (sinking output to /dev/null in
12565 // the noclobber arm at host_apply_redirect:5481) and then
12566 // overrides $? here. Same observable effect for the common
12567 // pattern `echo x > existing-file` under noclobber.
12568 let failed = with_executor(|exec| {
12569 let f = exec.redirect_failed;
12570 exec.redirect_failed = false;
12571 f
12572 });
12573 if failed {
12574 with_executor(|exec| exec.set_last_status(1));
12575 }
12576 }
12577
12578 fn heredoc(&mut self, content: &str) {
12579 // C `Src/exec.c:4641` — `parsestr(&buf)` runs parameter +
12580 // command substitution on the heredoc body. The lexer's
12581 // quoted-delimiter detection (`<<'EOF'`) routes through the
12582 // `Op::HereDoc` path in `compile_zsh.rs` which short-circuits
12583 // before reaching here; unquoted forms route through the
12584 // BUILTIN_EXPAND_TEXT mode-4 emit path that calls singsub.
12585 // This handler covers the verbatim/quoted case.
12586 with_executor(|exec| exec.host_set_pending_stdin(content.to_string()));
12587 }
12588
12589 fn herestring(&mut self, content: &str) {
12590 // Shell semantics: herestring appends a newline. `<<<` body
12591 // substitution (`Src/exec.c:4655 getherestr` calls
12592 // `quotesubst` + `untokenize`) lands here verbatim; the
12593 // upstream compiler routes through `Op::HereString` after
12594 // BUILTIN_EXPAND_TEXT for the substitution pass, so callers
12595 // of `host.herestring` see the already-expanded form.
12596 let mut s = content.to_string();
12597 s.push('\n');
12598 with_executor(|exec| exec.host_set_pending_stdin(s));
12599 }
12600
12601 fn exec(&mut self, args: Vec<String>) -> i32 {
12602 // c:Src/exec.c getproc + Src/jobs.c deletefilelist — close
12603 // any `>(cmd)` write ends owned by this command once it
12604 // finishes (drops on every return path below).
12605 let _psub_fds = PsubFdGuard;
12606 // c:Src/subst.c paramsubst — when `${var:?msg}` or `${var?msg}`
12607 // triggered the "parameter null or not set" error, errflag
12608 // is raised and zsh aborts the simple command without
12609 // attempting exec. The expansion may have produced empty
12610 // argv[0] which falls into the c:?/permission-denied path
12611 // below, masking the real diagnostic with a spurious
12612 // "permission denied:" line and rc=126 instead of rc=1.
12613 // Honour errflag here so the script ends with the
12614 // paramsubst error as the sole diagnostic. Bug #86.
12615 //
12616 // c:Src/exec.c — C's execlist loop clears ERRFLAG_ERROR
12617 // between sublists when the error came from a NOMATCH-style
12618 // command failure (glob no-match, etc.) so subsequent
12619 // sublists run. zshrs's vm dispatch handles this at the
12620 // post-command-boundary HERE: if THIS command has its
12621 // `current_command_glob_failed` cell set (meaning the glob
12622 // NOMATCH happened during this command's argv prep), surface
12623 // status 1 and clear BOTH the cell AND ERRFLAG_ERROR so the
12624 // NEXT exec call sees a clean state. The errflag from
12625 // genuine script-fatal errors (parse, redirect, paramsubst
12626 // `${:?msg}`) does NOT come paired with glob_failed, so
12627 // those still short-circuit + propagate.
12628 consume_tilde_globsubst_carrier();
12629 let glob_failed = with_executor(|exec| {
12630 let f = exec.current_command_glob_failed.get();
12631 exec.current_command_glob_failed.set(false);
12632 f
12633 });
12634 if glob_failed {
12635 crate::ported::utils::errflag.fetch_and(
12636 !crate::ported::zsh_h::ERRFLAG_ERROR,
12637 std::sync::atomic::Ordering::Relaxed,
12638 );
12639 with_executor(|exec| exec.set_last_status(1));
12640 return 1;
12641 }
12642 // c:Src/subst.c:505-507 — CSH_NULL_GLOB external-path
12643 // boundary: command skipped with `no match` but the NEXT
12644 // sublist runs (zsh -fc 'setopt cshnullglob; ls *nope*;
12645 // print after' prints the error then `after` — verified
12646 // zsh 5.9.1), so clear ERRFLAG like the glob_failed arm.
12647 if consume_badcshglob() {
12648 crate::ported::utils::errflag.fetch_and(
12649 !crate::ported::zsh_h::ERRFLAG_ERROR,
12650 std::sync::atomic::Ordering::Relaxed,
12651 );
12652 with_executor(|exec| exec.set_last_status(1));
12653 return 1;
12654 }
12655 if (crate::ported::utils::errflag.load(std::sync::atomic::Ordering::SeqCst)
12656 & crate::ported::zsh_h::ERRFLAG_ERROR)
12657 != 0
12658 {
12659 return 1;
12660 }
12661 // c:Src/exec.c — two distinct empty-command cases:
12662 //
12663 // 1. args=[""] — an explicit empty-string command word
12664 // (`""`, `"\$unset"`, `\$'\$x'`). zsh attempts exec(2)
12665 // on the empty path → EACCES → "permission denied", \$?
12666 // = 126.
12667 //
12668 // 2. args=[] — the WORD LIST is empty (unquoted \$(\$cmd)
12669 // that produced empty, or an unquoted unset \$var that
12670 // elided). zsh: no exec is attempted; \$? becomes the
12671 // last cmd-subst's exit status (the inner sub-VM
12672 // already set last_status), and the line completes
12673 // silently. Critically NOT 126.
12674 if args.is_empty() {
12675 // c:Src/exec.c — empty word list passes through to a
12676 // no-op; preserve whatever the inner cmd-subst's exit
12677 // is. Return last_status so the caller's SetStatus
12678 // round-trips correctly.
12679 return with_executor(|exec| exec.last_status());
12680 }
12681 if args[0].is_empty() {
12682 let script_name =
12683 crate::ported::utils::scriptname_get().unwrap_or_else(|| "zshrs".to_string());
12684 let lineno: u64 = with_executor(|exec| {
12685 exec.scalar("LINENO")
12686 .and_then(|s| s.parse::<u64>().ok())
12687 .unwrap_or(1)
12688 });
12689 eprintln!("{}:{}: permission denied: ", script_name, lineno);
12690 return 126;
12691 }
12692 // c:Src/exec.c — when any redirect in the current scope
12693 // failed (e.g. noclobber blocked a `>` overwrite), zsh
12694 // refuses to execute the command and exits with status 1.
12695 // The Rust port still applied the command (writing to the
12696 // /dev/null sink installed by host_apply_redirect's
12697 // noclobber arm), but the success status overwrote the
12698 // intended `1`. Short-circuit here so the exec returns 1
12699 // without running the body.
12700 let redir_failed = with_executor(|exec| {
12701 let f = exec.redirect_failed;
12702 exec.redirect_failed = false;
12703 f
12704 });
12705 if redir_failed {
12706 return 1;
12707 }
12708 // Track `$_` as the last argument of the last command (zsh /
12709 // bash convention). Empty arglists leave it untouched.
12710 if let Some(last) = args.last() {
12711 with_executor(|exec| {
12712 exec.set_scalar("_".to_string(), last.clone());
12713 });
12714 }
12715 // Route external command spawning through `executor.execute_external`
12716 // so intercepts (AOP before/after/around), command_hash lookups,
12717 // pre/postexec hooks, and zsh-specific fork-then-exec all apply.
12718 // Without this override, fusevm's default `host.exec` calls
12719 // `Command::new` directly, bypassing zshrs's dispatch logic.
12720 let status = with_executor(|exec| exec.host_exec_external(&args));
12721 // c:Src/jobs.c:1748 waitonejob (no-procs else-branch). zshrs's
12722 // exec model routes external commands through host_exec_external
12723 // (which already waitpid'd in-line); the canonical waitonejob
12724 // expects a Job to derive lastval, but here we already know
12725 // it. Synthesize a procs-less job so waitonejob's no-procs
12726 // branch fires the `pipestats[0]=lastval; numpipestats=1;`
12727 // update via the canonical port.
12728 crate::ported::builtin::LASTVAL.store(status, std::sync::atomic::Ordering::Relaxed);
12729 let mut synth = crate::ported::zsh_h::job::default();
12730 crate::ported::jobs::waitonejob(&mut synth);
12731 status
12732 }
12733
12734 fn cmd_subst(&mut self, sub: &fusevm::Chunk) -> String {
12735 // Run the sub-chunk on a nested VM with the same host wired up,
12736 // capturing stdout. The current executor remains active via the
12737 // thread-local — the nested VM uses CallBuiltin to dispatch shell
12738 // ops back through `with_executor`.
12739 let (read_end, write_end) = match os_pipe::pipe() {
12740 Ok(p) => p,
12741 Err(_) => return String::new(),
12742 };
12743 let saved_stdout = unsafe { libc::fcntl(libc::STDOUT_FILENO, libc::F_DUPFD, 10) };
12744 if saved_stdout < 0 {
12745 return String::new();
12746 }
12747 let saved_stderr = unsafe { libc::fcntl(libc::STDERR_FILENO, libc::F_DUPFD, 10) };
12748 let write_fd = AsRawFd::as_raw_fd(&write_end);
12749 unsafe {
12750 libc::dup2(write_fd, libc::STDOUT_FILENO);
12751 }
12752 drop(write_end);
12753
12754 // c:Bug #56 — publish the saved outer fds so a trap firing
12755 // during the nested VM run can route its body output to the
12756 // PARENT's stdout instead of the cmdsub's pipe-bound fd 1.
12757 // zsh's forked cmdsub gets this for free (trap runs in the
12758 // parent process whose fd 1 is untouched). zshrs's
12759 // in-process cmdsub needs this thread-local stack so the
12760 // trap dispatcher can find the right destination fd.
12761 CMDSUBST_OUTER_FDS.with(|s| s.borrow_mut().push((saved_stdout, saved_stderr)));
12762
12763 // Nested scope for `>(cmd)` fd ownership — commands inside
12764 // the cmdsub must not drain the enclosing command's pending
12765 // psub fds (see PSUB_SCOPE_DEPTH).
12766 let _psub_scope = PsubScope::enter();
12767
12768 // c:Src/exec.c:1161 — forked cmdsub child runs entersubsh()
12769 // which does `zsh_subshell++`; in-process equivalent.
12770 let _subshell_bump = CmdSubstSubshellBump::enter();
12771
12772 crate::fusevm_disasm::maybe_print_stdout("host.cmd_subst", sub);
12773 let mut vm = fusevm::VM::new(sub.clone());
12774 register_builtins(&mut vm);
12775 vm.set_shell_host(Box::new(ZshrsHost));
12776 let _ = vm.run();
12777 let cmd_status = vm.last_status;
12778
12779 CMDSUBST_OUTER_FDS.with(|s| {
12780 s.borrow_mut().pop();
12781 });
12782
12783 unsafe {
12784 libc::dup2(saved_stdout, libc::STDOUT_FILENO);
12785 libc::close(saved_stdout);
12786 if saved_stderr >= 0 {
12787 libc::close(saved_stderr);
12788 }
12789 }
12790
12791 // Inner cmd's status not propagated for the same reason as
12792 // run_command_substitution — see GAPS.md.
12793 let _ = cmd_status;
12794
12795 let mut buf = String::new();
12796 let mut reader = read_end;
12797 let _ = reader.read_to_string(&mut buf);
12798 // Strip trailing newlines (POSIX command substitution semantics)
12799 while buf.ends_with('\n') {
12800 buf.pop();
12801 }
12802 buf
12803 }
12804
12805 fn call_function(&mut self, name: &str, args: Vec<String>) -> Option<i32> {
12806 // c:Src/exec.c — when the command word is empty (e.g. `""`
12807 // or `"$nonexistent"`), zsh attempts the exec(2) which
12808 // returns EACCES ("permission denied") and exits 126. The
12809 // Rust port silently treated empty as a no-op (status 0).
12810 // Match zsh by emitting the diagnostic and returning 126.
12811 if name.is_empty() {
12812 let script_name =
12813 crate::ported::utils::scriptname_get().unwrap_or_else(|| "zshrs".to_string());
12814 let lineno: u64 = with_executor(|exec| {
12815 exec.scalar("LINENO")
12816 .and_then(|s| s.parse::<u64>().ok())
12817 .unwrap_or(1)
12818 });
12819 eprintln!("{}:{}: permission denied: ", script_name, lineno);
12820 with_executor(|exec| exec.set_last_status(126));
12821 return Some(126);
12822 }
12823 // c:Src/exec.c — redirect failure in this scope means the
12824 // command should NOT run. The Host::exec path already has
12825 // this gate (at fn exec above); call_function takes external
12826 // commands like `cat <&3` through a different code path, so
12827 // gate here too. Without this, bad-fd redirects produced
12828 // the diagnostic but the external command still ran, so $?
12829 // came out from the command's natural exit instead of the
12830 // forced 1.
12831 let redir_failed = with_executor(|exec| {
12832 let f = exec.redirect_failed;
12833 exec.redirect_failed = false;
12834 f
12835 });
12836 if redir_failed {
12837 with_executor(|exec| exec.set_last_status(1));
12838 return Some(1);
12839 }
12840 // ACTUALLY A ZSH FUNCTION: zmv/zcp/zln/zcalc are zsh autoload
12841 // functions, NOT builtins. zshrs ships fast native impls, but they
12842 // must behave like the zsh functions — command-not-found until
12843 // `autoload -Uz <name>` creates a function entry. When autoloaded we
12844 // run the native impl here (short-circuiting the fpath source, which
12845 // can hang zshrs's parser on zsh-specific syntax); when NOT autoloaded
12846 // we fall through (return None → resolution ends in command-not-found),
12847 // matching `zsh -f; zmv` → "command not found: zmv".
12848 if matches!(name, "zmv" | "zcp" | "zln" | "zcalc")
12849 && !with_executor(|exec| exec.function_exists(name))
12850 {
12851 return None;
12852 }
12853 match name {
12854 "zmv" => {
12855 return Some(crate::extensions::ext_builtins::zmv(&args, "mv"));
12856 }
12857 "zcp" => {
12858 return Some(crate::extensions::ext_builtins::zmv(&args, "cp"));
12859 }
12860 "zln" => {
12861 return Some(crate::extensions::ext_builtins::zmv(&args, "ln"));
12862 }
12863 "zcalc" => {
12864 return Some(crate::extensions::ext_builtins::zcalc(&args));
12865 }
12866 // znative — the plugin package manager (src/extensions/pkg/). Installs
12867 // + loads zsh script and native (Rust cdylib) plugins from a global
12868 // content-addressed store. `znative add owner/repo`, `znative load`, ...
12869 "znative" => {
12870 return Some(crate::extensions::pkg::builtin::znative(&args));
12871 }
12872 // ztest framework (src/extensions/ztest.rs — port of
12873 // ../strykelang's unit-test framework). All zassert_*/
12874 // ztest_* names route through the single try_dispatch
12875 // helper so adding/removing assertions only touches
12876 // ztest.rs.
12877 n if crate::extensions::ztest::try_dispatch_known(n) => {
12878 let status = with_executor(|exec| {
12879 crate::extensions::ztest::try_dispatch(exec, n, &args).unwrap_or(1)
12880 });
12881 return Some(status);
12882 }
12883 // Daemon-managed z* builtins — thin IPC wrappers. Short-circuit BEFORE
12884 // the function-lookup path so a missing daemon doesn't fall through to
12885 // "command not found". The name list is owned by the daemon crate
12886 // (zshrs_daemon::builtins::ZSHRS_BUILTIN_NAMES); routing through
12887 // try_dispatch keeps this site zero-touch as new z* builtins land.
12888 n if crate::daemon::builtins::is_zshrs_builtin(n) => {
12889 let argv: Vec<String> = std::iter::once(name.to_string()).chain(args).collect();
12890 return Some(crate::daemon::builtins::try_dispatch(n, &argv).unwrap_or(1));
12891 }
12892 _ => {}
12893 }
12894
12895 // c:Src/exec.c:3050-3068 — module-provided builtins (registered
12896 // via each module's `bintab` and folded into the canonical
12897 // `builtintab` by `createbuiltintable`) must dispatch BEFORE
12898 // PATH lookup. fusevm's `shell_builtins::builtin_id` doesn't
12899 // know about per-module entries like `log`
12900 // (Src/Modules/watch.c:693) — they reach call_function as
12901 // CallFunction ops. Consult the merged builtintab here so
12902 // `log` runs the canonical `bin_log` instead of falling
12903 // through to `/usr/bin/log` on macOS. Bug #72 in docs/BUGS.md.
12904 //
12905 // User-defined functions still take precedence over builtins
12906 // (zsh's `alias → function → builtin → external` resolution
12907 // order, c:Src/exec.c:3038-3068). Check `functions_compiled`
12908 // first so a user `log() { ... }` shadows the module bin_log.
12909 // c:Src/exec.c — shfunctab->getnode (the DISABLED-filtering
12910 // accessor) returns NULL for entries flipped to DISABLED via
12911 // `disable -f NAME`. functions_compiled holds the body
12912 // independently of the DISABLED flag, so check shfunctab first
12913 // and mask the lookup when the entry is disabled. Bug #221
12914 // in docs/BUGS.md.
12915 let user_fn_disabled = crate::ported::hashtable::shfunctab_lock()
12916 .read()
12917 .ok()
12918 .and_then(|t| {
12919 let entry = t.get_including_disabled(name)?;
12920 Some((entry.node.flags as u32 & crate::ported::zsh_h::DISABLED as u32) != 0)
12921 })
12922 .unwrap_or(false);
12923 let has_user_fn =
12924 !user_fn_disabled && with_executor(|exec| exec.functions_compiled.contains_key(name));
12925 if !has_user_fn {
12926 // c:Src/exec.c:3056 — `builtintab->getnode(builtintab,
12927 // cmdarg)` returns NULL for DISABLED entries, falling
12928 // execcmd through to PATH lookup. Mirror by gating the
12929 // bn_in_tab match on the BUILTINS_DISABLED set. Bug #106
12930 // in docs/BUGS.md.
12931 let disabled = crate::ported::builtin::BUILTINS_DISABLED
12932 .lock()
12933 .map(|s| s.contains(name))
12934 .unwrap_or(false);
12935 let bn_in_tab =
12936 !disabled && crate::ported::builtin::createbuiltintable().contains_key(name);
12937 if bn_in_tab {
12938 return Some(dispatch_builtin_raw(name, args));
12939 }
12940 // zshrs-original opcode builtins (async, doctor, peach, …) are not
12941 // in builtintab, so a run-time-resolved name (`$var`) never reaches
12942 // them. Dispatch by name here — after ported builtins, before
12943 // external — matching the shell's function -> builtin -> external
12944 // order (`has_user_fn` was checked above, so functions still win).
12945 if let Some(status) = try_run_registered_builtin(name, &args) {
12946 return Some(status);
12947 }
12948 }
12949
12950 // c:Src/lex.c — alias expansion is a LEXER-TIME pass, not a
12951 // run-time lookup. zsh parses the whole `-c` argument (or
12952 // script) before executing, so aliases defined in the same
12953 // parse unit don't apply to commands parsed earlier. Only at
12954 // an INTERACTIVE prompt does each line parse separately with
12955 // the latest aliastab visible.
12956 //
12957 // Gate the run-time alias-rewrite path on `interactive` so
12958 // `alias hi='echo hello'; hi` in `-c` mode falls through to
12959 // "command not found" (matching zsh) while interactive REPL
12960 // input still re-parses with the live aliastab.
12961 let interactive = crate::ported::zsh_h::isset(crate::ported::zsh_h::INTERACTIVE);
12962 let already_expanding = if interactive {
12963 crate::ported::hashtable::aliastab_lock()
12964 .read()
12965 .ok()
12966 .and_then(|tab| tab.get(name).map(|a| a.inuse != 0))
12967 .unwrap_or(false)
12968 } else {
12969 true // suppress lookup entirely in non-interactive mode
12970 };
12971 let alias_body = if already_expanding {
12972 None
12973 } else {
12974 with_executor(|exec| exec.alias(name))
12975 };
12976 if let Some(body) = alias_body {
12977 let combined = if args.is_empty() {
12978 body
12979 } else {
12980 let quoted: Vec<String> = args
12981 .iter()
12982 .map(|a| {
12983 let escaped = a.replace('\'', "'\\''");
12984 format!("'{}'", escaped)
12985 })
12986 .collect();
12987 format!("{} {}", body, quoted.join(" "))
12988 };
12989 // Bump inuse → run → clear, matching C's lexer behavior.
12990 if let Ok(mut tab) = crate::ported::hashtable::aliastab_lock().write() {
12991 if let Some(a) = tab.get_mut(name) {
12992 a.inuse += 1;
12993 }
12994 }
12995 let status = with_executor(|exec| exec.execute_script(&combined).unwrap_or(1));
12996 if let Ok(mut tab) = crate::ported::hashtable::aliastab_lock().write() {
12997 if let Some(a) = tab.get_mut(name) {
12998 a.inuse = (a.inuse - 1).max(0);
12999 }
13000 }
13001 return Some(status);
13002 }
13003
13004 // $_ pre-body bump and pending-underscore tracking are
13005 // ZshrsHost-only concerns (prompt rendering). Apply BEFORE
13006 // delegating to dispatch_function_call so the body sees the
13007 // bumped value.
13008 //
13009 // c:Src/exec.c:3491 — `setunderscore((args && nonempty(args))
13010 // ? ((char *) getdata(lastnode(args))) : "")`. C sets $_ to
13011 // the LAST node of the WHOLE args list (which includes argv[0]
13012 // == the function name). So for a no-arg `f`, $_ becomes "f"
13013 // inside the function body. The Rust port at the CallFunction
13014 // op-handler receives `args` WITHOUT the command name
13015 // (compile_zsh.rs:1571 only pushes simple.words[1..]). The
13016 // last() fallback `|| fn_name.clone()` already covers the
13017 // no-arg case, but `exec.set_scalar("_", ...)` writes paramtab
13018 // — the canonical `$_` read goes through `underscoregetfn`
13019 // (params.rs:7836) which reads the `zunderscore` Mutex.
13020 // setsparam("_") doesn't update that mutex, so the body's
13021 // `${_}` returned empty. Bug #279 in docs/BUGS.md. Mirror the
13022 // C `setunderscore` by writing via `set_zunderscore` directly.
13023 let fn_name = name.to_string();
13024 with_executor(|exec| {
13025 let dollar_underscore = args.last().cloned().unwrap_or_else(|| fn_name.clone());
13026 exec.set_scalar("_".to_string(), dollar_underscore.clone());
13027 crate::ported::params::set_zunderscore(std::slice::from_ref(&dollar_underscore));
13028 exec.pending_underscore = Some(dollar_underscore);
13029 });
13030
13031 // Delegate the actual function dispatch to the canonical
13032 // `dispatch_function_call` (which itself wraps the canonical
13033 // `doshfunc` port from `Src/exec.c:5823`). Single doshfunc
13034 // call-site keeps scope-mgmt invariants in one place.
13035 let status = with_executor(|exec| exec.dispatch_function_call(&fn_name, &args));
13036
13037 // Anonymous functions (`() { … } args`, compiled by
13038 // parse_anon_funcdef as `_zshrs_anon_N` / `_zshrs_anon_kw_N`)
13039 // execute exactly ONCE and must not persist. zsh runs the body and
13040 // frees the function, so `${functions}` / `typeset -f` never show
13041 // it. Remove every trace right after the single invocation —
13042 // AFTER `status` is captured, so the body's exit code is preserved
13043 // ($? — calling `unfunction` here would reset it to 0 instead).
13044 // Without this, real plugins that use `() { … }` (fzf-tab, zinit,
13045 // p10k, …) leaked dozens of `_zshrs_anon_N` into `$functions`,
13046 // diverging from zsh's function table on every such config.
13047 if fn_name.starts_with("_zshrs_anon_") {
13048 // `${functions}` / `typeset -f` enumerate the canonical
13049 // `shfunctab` (via scanpmfunctions); the bytecode call path
13050 // also keeps the body in the executor's compiled-fn maps. Clear
13051 // BOTH so no trace of the one-shot anon survives.
13052 crate::ported::hashtable::removeshfuncnode(&fn_name);
13053 with_executor(|exec| {
13054 exec.functions_compiled.remove(&fn_name);
13055 exec.function_source.remove(&fn_name);
13056 exec.function_line_base.remove(&fn_name);
13057 exec.function_def_file.remove(&fn_name);
13058 });
13059 }
13060
13061 // $_ post-body — last call-arg or function name. Mirrors the
13062 // C `setunderscore` invocation after the body returns.
13063 with_executor(|exec| {
13064 let last_call_arg = args.last().cloned().unwrap_or_else(|| fn_name.clone());
13065 exec.set_scalar("_".to_string(), last_call_arg.clone());
13066 exec.pending_underscore = Some(last_call_arg);
13067 });
13068
13069 status
13070 }
13071}
13072
13073// ───────────────────────────────────────────────────────────────────────────
13074/// Render a failed-redirect open error the way C's `zerrmsg` `%e` format
13075/// code does (Src/utils.c): `strerror(errno)` with the first character
13076/// lowercased, except `EIO` (kept capitalized) and `EINTR` (→ "interrupt").
13077/// C's redirect open failures call `zwarn("%e: %s", errno, fname)`
13078/// (Src/exec.c:3741); zshrs's `zwarning` takes a pre-built string, so the
13079/// `%e` part is built here. Replaces the prior hardcoded `ErrorKind` match
13080/// that fell back to a generic "redirect failed" for `EROFS`/`EACCES`/etc.
13081fn redir_errno_msg(err: &std::io::Error) -> String {
13082 let errno = match err.raw_os_error() {
13083 Some(n) if n != 0 => n,
13084 _ => return "redirect failed".to_string(),
13085 };
13086 if errno == libc::EINTR {
13087 return "interrupt".to_string(); // c:zerrmsg %e — EINTR special-case
13088 }
13089 let cptr = unsafe { libc::strerror(errno) };
13090 if cptr.is_null() {
13091 return "redirect failed".to_string();
13092 }
13093 let msg = unsafe { std::ffi::CStr::from_ptr(cptr) }.to_string_lossy();
13094 if errno == libc::EIO {
13095 return msg.into_owned(); // c:zerrmsg %e — EIO keeps capitalization
13096 }
13097 // c:zerrmsg %e — `fputc(tulower(errmsg[0])); fputs(errmsg + 1)`.
13098 let mut chars = msg.chars();
13099 match chars.next() {
13100 Some(first) => first.to_lowercase().collect::<String>() + chars.as_str(),
13101 None => "redirect failed".to_string(),
13102 }
13103}
13104
13105// Host-routed shell ops: ShellExecutor methods invoked by ZshrsHost from the
13106// fusevm VM. Not a port of Src/exec.c (see file-level docs above) — they're
13107// the bridge between fusevm opcodes and ShellExecutor state.
13108// ───────────────────────────────────────────────────────────────────────────
13109impl ShellExecutor {
13110 // ─── Host-routed shell ops (called by ZshrsHost from fusevm) ────────────
13111
13112 /// Apply a single redirection. The current scope's saved-fd vec gets a
13113 /// dup of the original fd so it can be restored by `host_redirect_scope_end`.
13114 /// `op_byte` matches `fusevm::op::redirect_op::*`.
13115 /// Apply a file-open result to a redirect fd; on error, emit
13116 /// zsh-format diagnostic, set redirect_failed, sink fd to /dev/null.
13117 /// Shared between WRITE/APPEND/READ/CLOBBER arms in
13118 /// host_apply_redirect to keep the error-handling identical.
13119 fn redir_open_or_fail(
13120 fd: i32,
13121 result: std::io::Result<fs::File>,
13122 target: &str,
13123 redirect_failed: &mut bool,
13124 ) -> bool {
13125 match result {
13126 Ok(file) => {
13127 let new_fd = file.into_raw_fd();
13128 unsafe {
13129 // When the target fd was already closed (e.g. `exec 0<&-;
13130 // cmd < file`), open() returns the lowest free fd, which is
13131 // `fd` itself. Then `dup2(fd, fd)` is a no-op: closing new_fd
13132 // would CLOSE the fd we just opened, AND — since Rust's
13133 // File::open sets O_CLOEXEC and a no-op dup2 does NOT clear
13134 // it — an exec'd child would lose the descriptor. So in the
13135 // reuse case, keep the fd and clear its close-on-exec flag;
13136 // otherwise dup2 (which clears cloexec on the copy) + close.
13137 if new_fd != fd {
13138 libc::dup2(new_fd, fd);
13139 libc::close(new_fd);
13140 } else {
13141 libc::fcntl(fd, libc::F_SETFD, 0);
13142 }
13143 }
13144 true
13145 }
13146 Err(e) => {
13147 // c:Src/exec.c:3741 — zwarn("%e: %s", errno, fname) with the
13148 // real lineno prefix; redir_errno_msg builds the `%e` errno
13149 // message for all errnos (not just the few hardcoded before).
13150 let msg = redir_errno_msg(&e);
13151 crate::ported::utils::zwarn(&format!("{}: {}", msg, target));
13152 *redirect_failed = true;
13153 // The /dev/null sink keeps a failed scoped redirect
13154 // from leaking the aborted command's output to the
13155 // wrong fd until scope-end restores it. For a bare
13156 // `exec` redirect (permanent, no scope restore) C
13157 // leaves the fd UNTOUCHED — execerr() aborts the
13158 // statement and the original fd 1 keeps flowing
13159 // (A04redirect: `exec >./nonexistent/x` then `echo
13160 // output` still prints). c:Src/exec.c:3735-3742.
13161 let permanent = with_executor(|exec| exec.exec_redirs_permanent);
13162 if !permanent {
13163 if let Ok(devnull) = fs::OpenOptions::new()
13164 .read(true)
13165 .write(true)
13166 .open("/dev/null")
13167 {
13168 let new_fd = devnull.into_raw_fd();
13169 unsafe {
13170 if new_fd != fd {
13171 libc::dup2(new_fd, fd);
13172 libc::close(new_fd);
13173 } else {
13174 libc::fcntl(fd, libc::F_SETFD, 0);
13175 }
13176 }
13177 }
13178 }
13179 false
13180 }
13181 }
13182 }
13183 /// `host_apply_redirect` — see implementation.
13184 pub fn host_apply_redirect(&mut self, fd: u8, op_byte: u8, target: &str) {
13185 // `&>` / `&>>` always target both fd 1 and fd 2 regardless of the
13186 // fd byte the parser supplied (the lexer's tokfd clamp makes the
13187 // raw value unreliable for these forms).
13188 let fd: i32 = if matches!(op_byte, r::WRITE_BOTH | r::APPEND_BOTH) {
13189 1
13190 } else {
13191 fd as i32
13192 };
13193 // c:Src/exec.c — for DUP_READ / DUP_WRITE forms (<&N / >&N),
13194 // validate the source fd is open BEFORE the save-and-dup
13195 // dance below. The save's `dup(fd)` reclaims the lowest free
13196 // fd, which on closed-fd reuse would let dup2(src=N, …)
13197 // succeed against the freshly-claimed slot — masking the
13198 // user's "bad file descriptor" error. Check src_fd first.
13199 if matches!(op_byte, r::DUP_READ | r::DUP_WRITE) {
13200 let n_check = target.trim_start_matches('&');
13201 if n_check != "-" {
13202 if let Ok(src_fd) = n_check.parse::<i32>() {
13203 if unsafe { libc::fcntl(src_fd, libc::F_GETFD) } == -1 {
13204 // c:Src/exec.c — zwarn with real lineno prefix.
13205 crate::ported::utils::zwarn(&format!("{}: bad file descriptor", src_fd));
13206 self.set_last_status(1);
13207 self.redirect_failed = true;
13208 return;
13209 }
13210 }
13211 }
13212 }
13213 // c:Src/exec.c:3978-3986 — bare `exec` redirects (nullexec==1)
13214 // skip the save entirely: "we specifically *don't* restore the
13215 // original fd's". C's save[] is per-execcmd, so exec's redirs
13216 // never enter an enclosing group's save list either; pushing
13217 // into `redirect_scope_stack.last_mut()` here (the enclosing
13218 // group's scope) made `{ exec 1>&-; … } 2>/dev/null` restore
13219 // stdout at group end — diverging from zsh, which keeps fd 1
13220 // closed for the rest of the script.
13221 if !self.exec_redirs_permanent {
13222 let saved = unsafe { libc::fcntl(fd, libc::F_DUPFD, 10) };
13223 if saved >= 0 {
13224 if let Some(top) = self.redirect_scope_stack.last_mut() {
13225 top.push((fd, saved));
13226 } else {
13227 // No scope — leave saved fd open and let the next scope
13228 // reclaim it. (Caller without a scope leaks the dup; this
13229 // matches `WithRedirects` parser construction always wrapping.)
13230 unsafe { libc::close(saved) };
13231 }
13232 }
13233 // For `&>` / `&>>` also save fd 2 so the scope restores it after
13234 // the body. Otherwise stderr stays redirected past the command.
13235 if matches!(op_byte, r::WRITE_BOTH | r::APPEND_BOTH) {
13236 let saved2 = unsafe { libc::fcntl(2, libc::F_DUPFD, 10) };
13237 if saved2 >= 0 {
13238 if let Some(top) = self.redirect_scope_stack.last_mut() {
13239 top.push((2, saved2));
13240 } else {
13241 unsafe { libc::close(saved2) };
13242 }
13243 }
13244 }
13245 }
13246 // c:Src/exec.c:3722-3724 + 2447-2480 — MULTIOS split when this
13247 // command's stdout IS the pipeline output. C registers the pipe
13248 // in mfds[1] (`addfd(forked, save, mfds, 1, output, 1, NULL)`)
13249 // BEFORE walking the explicit redirect list, so a write-side
13250 // redirect of fd 1 finds mfds[1] occupied and, with MULTIOS
13251 // set, "split[s] the stream": fd 1 becomes the write end of an
13252 // internal pipe whose reader tees every chunk to BOTH the
13253 // pipeline pipe and the new target. That is why
13254 // `{ echo a; echo b >&2; } 3>&1 1>&2 2>&3 3>&- | cat` sends
13255 // `a` to the pipe (via the tee) AND to stderr — plain dup2
13256 // replacement loses the pipe stream. The scope-depth gate
13257 // mirrors mfds being per-execcmd: only the redirect list
13258 // attached to the stage's own command joins the pipe; nested
13259 // commands inside the body (`{ echo a > f; } | cat`) get a
13260 // fresh "mfds" and replace as usual.
13261 if fd == 1
13262 && self
13263 .pipe_output_scope
13264 .is_some_and(|d| d + 1 == self.redirect_scope_stack.len())
13265 && crate::ported::options::opt_state_get("multios").unwrap_or(true)
13266 {
13267 // Resolve the new write target exactly as the plain arms
13268 // below would, but as a raw fd for the tee.
13269 let new_target_fd: i32 = match op_byte {
13270 r::DUP_WRITE => {
13271 // Numeric `>&N` only; `-` (close) and `p` (coproc)
13272 // fall through to the plain arms.
13273 target
13274 .trim_start_matches('&')
13275 .parse::<i32>()
13276 .map(|src| unsafe { libc::fcntl(src, libc::F_DUPFD, 10) })
13277 .unwrap_or(-1)
13278 }
13279 r::WRITE | r::CLOBBER => fs::File::create(target)
13280 .map(|f| f.into_raw_fd())
13281 .unwrap_or(-1),
13282 r::APPEND => fs::OpenOptions::new()
13283 .create(true)
13284 .append(true)
13285 .open(target)
13286 .map(|f| f.into_raw_fd())
13287 .unwrap_or(-1),
13288 _ => -1,
13289 };
13290 if new_target_fd >= 0 {
13291 let pipe_dup = unsafe { libc::fcntl(1, libc::F_DUPFD, 10) };
13292 match (pipe_dup >= 0).then(os_pipe::pipe) {
13293 Some(Ok((read_end, write_end))) => {
13294 // Splitter: same read-loop shape as
13295 // BUILTIN_MULTIOS_REDIRECT, with one ordering
13296 // refinement. C's tee is a forked process
13297 // (closemn → teeproc) whose wakeup latency lets
13298 // the stage's DIRECT pipe writes land first —
13299 // observed zsh output for `{ echo a; echo b >&2; }
13300 // 3>&1 1>&2 2>&3 3>&- | cat` is `b` then `a`,
13301 // 15/15 runs. A Rust thread wakes faster than
13302 // the debug-build VM dispatches the next echo,
13303 // inverting the order. Emulate the C timing
13304 // observably: stream to the NEW target (file /
13305 // stderr dup) immediately, but defer the
13306 // pipe-bound copy until EOF (or a 64KB cap so a
13307 // long-running stream still flows instead of
13308 // growing memory unboundedly).
13309 let write_now = |tfd: i32, data: &[u8]| {
13310 let mut off = 0;
13311 while off < data.len() {
13312 let w = unsafe {
13313 libc::write(
13314 tfd,
13315 data[off..].as_ptr() as *const libc::c_void,
13316 data.len() - off,
13317 )
13318 };
13319 if w <= 0 {
13320 break;
13321 }
13322 off += w as usize;
13323 }
13324 };
13325 let handle = std::thread::spawn(move || {
13326 let mut rd = read_end;
13327 let mut buf = [0u8; 8192];
13328 let mut pipe_pending: Vec<u8> = Vec::new();
13329 loop {
13330 match std::io::Read::read(&mut rd, &mut buf) {
13331 Ok(0) | Err(_) => break,
13332 Ok(n) => {
13333 write_now(new_target_fd, &buf[..n]);
13334 pipe_pending.extend_from_slice(&buf[..n]);
13335 if pipe_pending.len() >= 65536 {
13336 write_now(pipe_dup, &pipe_pending);
13337 pipe_pending.clear();
13338 }
13339 }
13340 }
13341 }
13342 write_now(pipe_dup, &pipe_pending);
13343 unsafe {
13344 libc::close(pipe_dup);
13345 libc::close(new_target_fd);
13346 }
13347 });
13348 let write_raw = AsRawFd::as_raw_fd(&write_end);
13349 unsafe { libc::dup2(write_raw, 1) };
13350 drop(write_end);
13351 // Scope-end closes this dup (the last writer once
13352 // the saved fd 1 is restored) → EOF → join.
13353 let close_on_end = unsafe { libc::fcntl(1, libc::F_DUPFD, 10) };
13354 if let Some(top) = self.multios_scope_stack.last_mut() {
13355 top.push((close_on_end, handle));
13356 } else {
13357 unsafe { libc::close(close_on_end) };
13358 let _ = handle.join();
13359 }
13360 return;
13361 }
13362 _ => unsafe {
13363 // pipe()/dup failure — fall through to plain replace.
13364 if pipe_dup >= 0 {
13365 libc::close(pipe_dup);
13366 }
13367 libc::close(new_target_fd);
13368 },
13369 }
13370 }
13371 }
13372 match op_byte {
13373 r::WRITE => {
13374 // Honor `setopt noclobber`: refuse to overwrite an
13375 // existing regular file unless `>!` / `>|` (CLOBBER).
13376 // zsh internally stores the inverted-name `clobber`
13377 // (default ON); `setopt noclobber` writes
13378 // `clobber=false`. Honor both keys.
13379 //
13380 // c:Src/exec.c:2241-2245 clobber_open recover path:
13381 // after O_EXCL fails, reopen and `if (!S_ISREG(...))
13382 // return fd;` — non-regular targets (char/block-
13383 // special, FIFO, socket) bypass the noclobber check.
13384 // Bug #30 in docs/BUGS.md: this bridge-side check did
13385 // a bare `Path::exists()` and treated `/dev/null` as
13386 // a protected file, breaking `setopt no_clobber; echo
13387 // hi > /dev/null` and every `2> /dev/null` idiom.
13388 // Add a regular-file stat gate that matches the C
13389 // semantic. The canonical clobber_open at
13390 // src/ported/exec.rs:2123 already handles this; the
13391 // bridge duplicates a stripped-down version here and
13392 // must mirror the same check.
13393 let noclobber = opt_state_get("noclobber").unwrap_or(false)
13394 || !opt_state_get("clobber").unwrap_or(true);
13395 let target_meta = std::fs::metadata(target).ok();
13396 let target_is_regular_file = target_meta
13397 .as_ref()
13398 .map(|m| m.file_type().is_file())
13399 .unwrap_or(false);
13400 // c:Src/exec.c:2313 clobber_open — CLOBBER_EMPTY permits
13401 // re-using an EMPTY regular file under noclobber: `setopt
13402 // noclobber clobberempty; : >f; echo hi >f` overwrites f.
13403 // The inline bridge check ignored this and errored.
13404 let clobber_empty_ok = opt_state_get("clobberempty").unwrap_or(false)
13405 && target_meta.as_ref().map(|m| m.len() == 0).unwrap_or(false);
13406 if noclobber && target_is_regular_file && !clobber_empty_ok {
13407 eprintln!(
13408 "{}:{}: file exists: {}",
13409 shname(),
13410 crate::ported::lex::lineno(),
13411 target
13412 );
13413 self.set_last_status(1);
13414 // c:Src/exec.c — set redirect_failed so the scope-end
13415 // hook (`with_redirects_end` in this file) forces
13416 // $? to 1 regardless of the still-running command's
13417 // own exit. Without this the next command (e.g.
13418 // `echo x` writing to /dev/null below) succeeds
13419 // and overwrites the redirect-failure status,
13420 // making noclobber unobservable from $?.
13421 self.redirect_failed = true;
13422 // Sink the upcoming command's stdout to /dev/null
13423 // so we don't leak its output to the terminal.
13424 // zsh skips the command entirely; we approximate by
13425 // discarding the output (the redirect target was
13426 // the user's chosen sink, but with noclobber the
13427 // file is protected — discarding matches the
13428 // user's intent better than printing to terminal).
13429 if let Ok(file) = fs::OpenOptions::new().write(true).open("/dev/null") {
13430 let new_fd = file.into_raw_fd();
13431 unsafe {
13432 libc::dup2(new_fd, fd);
13433 libc::close(new_fd);
13434 }
13435 }
13436 return;
13437 }
13438 if !Self::redir_open_or_fail(
13439 fd,
13440 fs::File::create(target),
13441 target,
13442 &mut self.redirect_failed,
13443 ) {
13444 self.set_last_status(1);
13445 }
13446 }
13447 r::CLOBBER => {
13448 if !Self::redir_open_or_fail(
13449 fd,
13450 fs::File::create(target),
13451 target,
13452 &mut self.redirect_failed,
13453 ) {
13454 self.set_last_status(1);
13455 }
13456 }
13457 r::APPEND => {
13458 // c:Src/exec.c:3924-3927 — `>>` honors NO_CLOBBER+!APPENDCREATE
13459 // by opening O_APPEND|O_WRONLY WITHOUT O_CREAT, so missing
13460 // files yield ENOENT. zsh source:
13461 // if (!isset(CLOBBER) && !isset(APPENDCREATE) &&
13462 // !IS_CLOBBER_REDIR(fn->type))
13463 // mode = O_WRONLY|O_APPEND|O_NOCTTY;
13464 // else mode = O_WRONLY|O_APPEND|O_CREAT|O_NOCTTY;
13465 // (IS_CLOBBER_REDIR — `>>!`/`>>|` — is currently flattened
13466 // to plain APPEND at compile time in
13467 // src/extensions/compile_zsh.rs:1654-1655, so the bang/pipe
13468 // forms can't be distinguished here yet.)
13469 let noclobber = opt_state_get("noclobber").unwrap_or(false)
13470 || !opt_state_get("clobber").unwrap_or(true);
13471 let append_create = opt_state_get("appendcreate").unwrap_or(false)
13472 || opt_state_get("append_create").unwrap_or(false);
13473 let open_result = if noclobber && !append_create {
13474 fs::OpenOptions::new().append(true).open(target) // no create
13475 } else {
13476 fs::OpenOptions::new()
13477 .create(true)
13478 .append(true)
13479 .open(target)
13480 };
13481 if !Self::redir_open_or_fail(fd, open_result, target, &mut self.redirect_failed) {
13482 self.set_last_status(1);
13483 }
13484 }
13485 r::READ => {
13486 if !Self::redir_open_or_fail(
13487 fd,
13488 fs::File::open(target),
13489 target,
13490 &mut self.redirect_failed,
13491 ) {
13492 self.set_last_status(1);
13493 }
13494 }
13495 r::READ_WRITE => {
13496 if let Ok(file) = fs::OpenOptions::new()
13497 .create(true)
13498 .truncate(false) // <> opens existing-or-new without truncating
13499 .read(true)
13500 .write(true)
13501 .open(target)
13502 {
13503 let new_fd = file.into_raw_fd();
13504 unsafe {
13505 // See redir_open_or_fail: when the opened fd IS the
13506 // destination (target fd was closed), keep it and clear
13507 // O_CLOEXEC; else dup2 + close.
13508 if new_fd != fd {
13509 libc::dup2(new_fd, fd);
13510 libc::close(new_fd);
13511 } else {
13512 libc::fcntl(fd, libc::F_SETFD, 0);
13513 }
13514 }
13515 }
13516 }
13517 r::DUP_READ | r::DUP_WRITE => {
13518 // Target is a numeric fd reference like `&3`. The parser
13519 // strips the `&` prefix before we get here in some paths,
13520 // others retain it — accept both. Also support `-` for
13521 // close-fd (`<&-` / `>&-`) per POSIX. The src_fd
13522 // validity check ran above before the save-and-dup.
13523 let n = target.trim_start_matches('&');
13524 if n == "-" {
13525 unsafe { libc::close(fd) };
13526 } else if n == "p" {
13527 // c:Src/exec.c — `<&p` / `>&p` route through the
13528 // coprocin / coprocout globals. zsh's `coproc CMD`
13529 // launch publishes those fds; the canonical
13530 // bin_print / bin_read `-p` arms already consume
13531 // them. The DUP redirect form is the third
13532 // consumer: it must dup the coproc fd onto the
13533 // target slot so the next command's stdin/stdout
13534 // is wired to the running coprocess. Bug #388.
13535 let coproc_fd = if op_byte == r::DUP_READ {
13536 crate::ported::modules::clone::coprocin
13537 .load(std::sync::atomic::Ordering::Relaxed)
13538 } else {
13539 crate::ported::modules::clone::coprocout
13540 .load(std::sync::atomic::Ordering::Relaxed)
13541 };
13542 if coproc_fd < 0 {
13543 eprintln!("{}:1: no coprocess", shname());
13544 self.set_last_status(1);
13545 self.redirect_failed = true;
13546 } else {
13547 unsafe {
13548 libc::dup2(coproc_fd, fd);
13549 }
13550 }
13551 } else if let Ok(src_fd) = n.parse::<i32>() {
13552 unsafe { libc::dup2(src_fd, fd) };
13553 } else if op_byte == r::DUP_WRITE {
13554 // c:Src/glob.c:2184-2187 xpandredir — a MERGEOUT
13555 // word that expands to a non-number becomes
13556 // REDIR_ERRWRITE: `cmd >& word` opens `word` and
13557 // routes BOTH fd 1 and fd 2 there. Reached only
13558 // for dynamic words (`>&$var`); static filenames
13559 // were converted at compile time.
13560 if let Ok(file) = fs::File::create(target) {
13561 let new_fd = file.into_raw_fd();
13562 unsafe {
13563 libc::dup2(new_fd, 1);
13564 libc::dup2(new_fd, 2);
13565 libc::close(new_fd);
13566 }
13567 }
13568 } else {
13569 // c:Src/glob.c:2185 — MERGEIN non-number:
13570 // `zerr("file number expected")`.
13571 crate::ported::utils::zerr("file number expected");
13572 self.set_last_status(1);
13573 self.redirect_failed = true;
13574 }
13575 }
13576 r::WRITE_BOTH => {
13577 if let Ok(file) = fs::File::create(target) {
13578 let new_fd = file.into_raw_fd();
13579 unsafe {
13580 libc::dup2(new_fd, 1);
13581 libc::dup2(new_fd, 2);
13582 libc::close(new_fd);
13583 }
13584 }
13585 }
13586 r::APPEND_BOTH => {
13587 if let Ok(file) = fs::OpenOptions::new()
13588 .create(true)
13589 .append(true)
13590 .open(target)
13591 {
13592 let new_fd = file.into_raw_fd();
13593 unsafe {
13594 libc::dup2(new_fd, 1);
13595 libc::dup2(new_fd, 2);
13596 libc::close(new_fd);
13597 }
13598 }
13599 }
13600 _ => {}
13601 }
13602 }
13603
13604 /// Push a fresh redirect scope. `_count` is informational — the actual
13605 /// saved fds are appended by host_apply_redirect into the top scope.
13606 pub fn host_redirect_scope_begin(&mut self, _count: u8) {
13607 // c:Src/exec.c:3722-3724 — the pipeline child set
13608 // `pipe_output_pending` right after dup2'ing its stdout onto
13609 // the pipe; the FIRST redirect scope opened in that child is
13610 // the stage command's own redirect list (same execcmd as the
13611 // pipe's addfd into mfds[1]). Capture the depth so only THAT
13612 // list's fd-1 write redirects MULTIOS-join the pipe.
13613 if self.pipe_output_pending {
13614 self.pipe_output_pending = false;
13615 self.pipe_output_scope = Some(self.redirect_scope_stack.len());
13616 }
13617 self.redirect_scope_stack.push(Vec::new());
13618 self.multios_scope_stack.push(Vec::new());
13619 }
13620
13621 /// Pop the top redirect scope, restoring saved fds.
13622 pub fn host_redirect_scope_end(&mut self) {
13623 // c:Src/exec.c — restore saved fds FIRST so the multios
13624 // pipe-write end is released from `fd`, then close our
13625 // tracked close_on_end (the last surviving writer dup), then
13626 // join the splitter thread. If we closed close_on_end before
13627 // restoring saved, `fd` would still hold a pipe writer and
13628 // the thread would block forever waiting for EOF.
13629 if let Some(saved) = self.redirect_scope_stack.pop() {
13630 for (fd, saved_fd) in saved.into_iter().rev() {
13631 unsafe {
13632 libc::dup2(saved_fd, fd);
13633 libc::close(saved_fd);
13634 }
13635 }
13636 }
13637 if let Some(scope) = self.multios_scope_stack.pop() {
13638 // Close ALL tracked writer dups BEFORE joining any
13639 // thread. When one splitter holds a dup of another's
13640 // pipe write-end (two multios in one scope where a later
13641 // one duped fd 1 while an earlier splitter owned it),
13642 // joining in push order deadlocks: splitter A's EOF
13643 // waits on splitter B's writer dup, which only closes
13644 // after B's thread exits — blocked behind A's join.
13645 let mut handles = Vec::with_capacity(scope.len());
13646 for (write_fd, handle) in scope {
13647 if write_fd >= 0 {
13648 unsafe {
13649 libc::close(write_fd);
13650 }
13651 }
13652 handles.push(handle);
13653 }
13654 for handle in handles {
13655 let _ = handle.join();
13656 }
13657 }
13658 // The scope that captured the pipeline-output marker is gone;
13659 // deeper-nested future scopes must not re-match its depth.
13660 if self.pipe_output_scope == Some(self.redirect_scope_stack.len()) {
13661 self.pipe_output_scope = None;
13662 }
13663 }
13664
13665 /// Set up `content` as stdin (fd 0) for the next command.
13666 /// Used by `Op::HereDoc(idx)` and `Op::HereString`.
13667 ///
13668 /// c:Src/exec.c:4655 getherestr — C writes the body to a TEMP
13669 /// FILE (gettempfile → write_loop → close → reopen O_RDONLY →
13670 /// unlink), NOT a pipe. The previous pipe+writer-thread shape
13671 /// SIGPIPE'd the whole shell when the consumer never read the
13672 /// body (`: <<< ${(F)x/y}` — D04parameter chunk 211, flaky
13673 /// rc=141): the redirect-scope teardown closed the read end
13674 /// while the detached thread was still in write_all, and the
13675 /// shell's SIGPIPE disposition is SIG_DFL. A temp file has no
13676 /// reader/writer coupling — matching C exactly, including
13677 /// lseek-ability of fd 0, which pipes don't give.
13678 pub fn host_set_pending_stdin(&mut self, content: String) {
13679 // c:4673 — `gettempfile(NULL, 1, &s)`.
13680 let mut tmp = std::env::temp_dir();
13681 tmp.push(format!(
13682 "zshrs-herestr-{}-{:x}",
13683 std::process::id(),
13684 std::time::SystemTime::now()
13685 .duration_since(std::time::UNIX_EPOCH)
13686 .map(|d| d.as_nanos())
13687 .unwrap_or(0)
13688 ));
13689 // c:4675 — `write_loop(fd, t, len); close(fd);`
13690 if std::fs::write(&tmp, content.as_bytes()).is_err() {
13691 return; // c:4674 — tempfile failure → no redirect
13692 }
13693 // c:Src/utils.c gettempfile → mkstemp creates the temp file mode
13694 // 0600 IGNORING the umask, so the O_RDONLY reopen below always
13695 // succeeds. `std::fs::write` honors the umask, so under `umask
13696 // 0777` the file landed mode 0000 and the reopen failed with
13697 // EACCES — `cat <<<x` then read empty stdin. Force 0600 to match
13698 // mkstemp's umask-independent permissions.
13699 let _ = std::fs::set_permissions(
13700 &tmp,
13701 <std::fs::Permissions as std::os::unix::fs::PermissionsExt>::from_mode(0o600),
13702 );
13703 // c:4677 — `fd = open(s, O_RDONLY | O_NOCTTY);`
13704 let file = match std::fs::File::open(&tmp) {
13705 Ok(f) => f,
13706 Err(_) => {
13707 let _ = std::fs::remove_file(&tmp);
13708 return;
13709 }
13710 };
13711 // c:4678 — `unlink(s);` — fd stays valid, name disappears.
13712 let _ = std::fs::remove_file(&tmp);
13713 let saved = unsafe { libc::fcntl(libc::STDIN_FILENO, libc::F_DUPFD, 10) };
13714 if saved >= 0 {
13715 if let Some(top) = self.redirect_scope_stack.last_mut() {
13716 top.push((libc::STDIN_FILENO, saved));
13717 } else {
13718 unsafe { libc::close(saved) };
13719 }
13720 }
13721 let read_fd = AsRawFd::as_raw_fd(&file);
13722 unsafe { libc::dup2(read_fd, libc::STDIN_FILENO) };
13723 drop(file);
13724 }
13725
13726 /// Spawn an external command using zshrs's full dispatch logic
13727 /// (intercepts, command_hash, redirect handling). Used by
13728 /// `ZshrsHost::exec` so the bytecode VM's `Op::Exec` and
13729 /// `Op::CallFunction` external fallback get the same semantics as
13730 /// the tree-walker's `execute_external` rather than a plain
13731 /// `Command::new` shortcut. Returns the exit status.
13732 pub fn host_exec_external(&mut self, args: &[String]) -> i32 {
13733 // Native p10k API: the `p10k(){ zshrs-p10k-api "$@" }` stub's
13734 // body lands here (the name is neither function nor builtin).
13735 // Route into the engine instead of a PATH miss.
13736 if let Some(name) = args.first() {
13737 if let Some(status) = crate::p10k::maybe_intercept_command(name, &args[1..]) {
13738 self.set_last_status(status);
13739 return status;
13740 }
13741 }
13742 // If a glob expansion in this command's argv triggered the
13743 // nomatch error path, suppress the actual exec and return
13744 // status 1 — mirrors zsh's command-aborted-on-glob-error
13745 // behaviour. The flag is reset BEFORE returning so the next
13746 // command starts clean.
13747 //
13748 // c:Src/glob.c:1876-1880 + Src/exec.c — NOMATCH sets
13749 // ERRFLAG_ERROR but C's execlist clears the bit per-sublist
13750 // so subsequent commands run. Symmetric with the builtin
13751 // dispatcher's clear at fusevm_bridge.rs:299 — clear it here
13752 // too at the external-command post-command-boundary.
13753 consume_tilde_globsubst_carrier();
13754 if self.current_command_glob_failed.get() {
13755 self.current_command_glob_failed.set(false);
13756 crate::ported::utils::errflag.fetch_and(
13757 !crate::ported::zsh_h::ERRFLAG_ERROR,
13758 std::sync::atomic::Ordering::Relaxed,
13759 );
13760 self.set_last_status(1);
13761 return 1;
13762 }
13763 // c:Src/subst.c:505-507 — CSH_NULL_GLOB sibling of the
13764 // NOMATCH gate above, same external-path semantics (skip
13765 // command, `no match`, clear ERRFLAG so the next sublist
13766 // runs).
13767 if consume_badcshglob() {
13768 crate::ported::utils::errflag.fetch_and(
13769 !crate::ported::zsh_h::ERRFLAG_ERROR,
13770 std::sync::atomic::Ordering::Relaxed,
13771 );
13772 self.set_last_status(1);
13773 return 1;
13774 }
13775 let Some((cmd, rest)) = args.split_first() else {
13776 return 0;
13777 };
13778 // Empty command name (e.g. result of an empty `$(false)`
13779 // command-sub being the only word) — zsh: no command runs,
13780 // exit status preserved from prior step. Was hitting the
13781 // "command not found: " path with empty name.
13782 if cmd.is_empty() && rest.is_empty() {
13783 return self.last_status();
13784 }
13785 let rest_vec: Vec<String> = rest.to_vec();
13786 // Update `$_` with the just-arriving argv so the next command
13787 // reads `_=<last_arg>`. Mirrors C zsh's writeback in
13788 // `execcmd_exec` (Src/exec.c). Per `args.last()` semantics,
13789 // when invoked as `cmd a b c`, `$_` becomes "c" — for a bare
13790 // command with no args, `$_` becomes the command name itself.
13791 crate::ported::params::set_zunderscore(args);
13792
13793 // Builtins not in fusevm's name→id table fall through to
13794 // host.exec. Catch them here before the OS-level exec attempts
13795 // to spawn a non-existent binary.
13796 match cmd.as_str() {
13797 "sched" => return dispatch_builtin("sched", rest_vec.clone()),
13798 "echotc" => return dispatch_builtin("echotc", rest_vec.clone()),
13799 "echoti" => return dispatch_builtin("echoti", rest_vec.clone()),
13800 "zpty" => return dispatch_builtin("zpty", rest_vec.clone()),
13801 "ztcp" => return dispatch_builtin("ztcp", rest_vec.clone()),
13802 "zsocket" => {
13803 // c:Src/Modules/socket.c:276 BUILTIN spec — BUILTINS["zsocket"]
13804 // optstr "ad:ltv" parsed by execbuiltin.
13805 return dispatch_builtin("zsocket", rest_vec.clone());
13806 }
13807 "private" => {
13808 // c:Src/Modules/param_private.c:217 — bin_private via
13809 // BUILTINS["private"]. The autoload require_module
13810 // (exec.c:2700-2717) fires inside
13811 // dispatch_builtin_raw, the chokepoint for all routes.
13812 return dispatch_builtin("private", rest_vec.clone());
13813 }
13814 "zformat" => return dispatch_builtin("zformat", rest_vec.clone()),
13815 "zregexparse" => return dispatch_builtin("zregexparse", rest_vec.clone()),
13816 // `unalias`/`unhash`/`unfunction` share `bin_unhash` but
13817 // each carries its own funcid (BIN_UNALIAS / BIN_UNHASH /
13818 // BIN_UNFUNCTION) — dispatch_builtin handles the BUILTINS
13819 // lookup + funcid propagation via execbuiltin.
13820 "unalias" | "unhash" | "unfunction" => {
13821 return dispatch_builtin(cmd.as_str(), rest_vec.clone());
13822 }
13823 // ACTUALLY A ZSH FUNCTION: zmv/zcp/zln/zcalc are zsh autoload
13824 // functions — implemented natively in Rust so `autoload -Uz zmv`
13825 // works without shipping the function source (and without the
13826 // fpath source hanging the parser). The `function_exists` guard
13827 // keeps them command-not-found until autoloaded, exactly like zsh;
13828 // an un-guarded arm ran them for bare `zmv`, diverging from
13829 // `zsh -f; zmv` → "command not found: zmv".
13830 "zmv" if self.function_exists("zmv") => {
13831 return crate::extensions::ext_builtins::zmv(&rest_vec, "mv")
13832 }
13833 "zcp" if self.function_exists("zcp") => {
13834 return crate::extensions::ext_builtins::zmv(&rest_vec, "cp")
13835 }
13836 "zln" if self.function_exists("zln") => {
13837 return crate::extensions::ext_builtins::zmv(&rest_vec, "ln")
13838 }
13839 "zcalc" if self.function_exists("zcalc") => {
13840 return crate::extensions::ext_builtins::zcalc(&rest_vec)
13841 }
13842 "zselect" => {
13843 // Route through canonical dispatch_builtin which goes
13844 // via execbuiltin → BUILTINS["zselect"] (zselect.c:272).
13845 return dispatch_builtin("zselect", rest_vec.clone());
13846 }
13847 "cap" => return dispatch_builtin("cap", rest_vec.clone()),
13848 "getcap" => return dispatch_builtin("getcap", rest_vec.clone()),
13849 "setcap" => return dispatch_builtin("setcap", rest_vec.clone()),
13850 "yes" => return self.builtin_yes(&rest_vec),
13851 "nl" => return self.builtin_nl(&rest_vec),
13852 "env" => return self.builtin_env(&rest_vec),
13853 "printenv" => return self.builtin_printenv(&rest_vec),
13854 "tty" => return self.builtin_tty(&rest_vec),
13855 // c:Src/Modules/files.c:806 — BUILTINS["chgrp"] with
13856 // BIN_CHGRP funcid + "hRs" optstr.
13857 "chgrp" => return dispatch_builtin("chgrp", rest_vec.clone()),
13858 "nproc" => return self.builtin_nproc(&rest_vec),
13859 "expr" => return self.builtin_expr(&rest_vec),
13860 "sha256sum" => return self.builtin_sha256sum(&rest_vec),
13861 "base64" => return self.builtin_base64(&rest_vec),
13862 "tac" => return self.builtin_tac(&rest_vec),
13863 "expand" => return self.builtin_expand(&rest_vec),
13864 "unexpand" => return self.builtin_unexpand(&rest_vec),
13865 "paste" => return self.builtin_paste(&rest_vec),
13866 "fold" => return self.builtin_fold(&rest_vec),
13867 "shuf" => return self.builtin_shuf(&rest_vec),
13868 "comm" => return self.builtin_comm(&rest_vec),
13869 "cksum" => return self.builtin_cksum(&rest_vec),
13870 "factor" => return self.builtin_factor(&rest_vec),
13871 "tsort" => return self.builtin_tsort(&rest_vec),
13872 "sum" => return self.builtin_sum(&rest_vec),
13873 "mkfifo" => return self.builtin_mkfifo(&rest_vec),
13874 "link" => return self.builtin_link(&rest_vec),
13875 "unlink" => return self.builtin_unlink(&rest_vec),
13876 "dircolors" => return self.builtin_dircolors(&rest_vec),
13877 "groups" => return self.builtin_groups(&rest_vec),
13878 "arch" => return self.builtin_arch(&rest_vec),
13879 "nice" => return self.builtin_nice(&rest_vec),
13880 "logname" => return self.builtin_logname(&rest_vec),
13881 "tput" => return self.builtin_tput(&rest_vec),
13882 "users" => return self.builtin_users(&rest_vec),
13883 // "sync" => return self.bin_sync(&rest_vec),
13884 "zbuild" => return self.builtin_zbuild(&rest_vec),
13885 // `zf_*` aliases from `zsh/files` (Src/Modules/files.c
13886 // BUILTIN table at line 816-824). The C source binds
13887 // both unprefixed (`chmod`) and prefixed (`zf_chmod`)
13888 // names to the SAME `bin_chmod` etc. handlers — the
13889 // prefixed forms exist so a script can portably reach
13890 // the builtin even when a function or alias has shadowed
13891 // the bare name. Each arm routes through the canonical
13892 // zf_* aliases route through canonical BUILTINS entries
13893 // (files.c:816-824) — execbuiltin parses each fn's optstr
13894 // automatically.
13895 "mkdir" | "zf_mkdir" | "zf_rm" | "zf_rmdir" | "zf_chmod" | "zf_chown" | "zf_chgrp"
13896 | "zf_ln" | "zf_mv" | "zf_sync"
13897 // `--zsh` parity gate: zsh -fc has zsh/files UNLOADED
13898 // — bare `mkdir` is /bin/mkdir (so `command mkdir -p`
13899 // honors the system flag set; zconvey.plugin.zsh:44
13900 // got "File exists" from the in-process bin_mkdir
13901 // that this arm intercepted) and `zf_*` names are
13902 // command-not-found 127 until `zmodload zsh/files`.
13903 // Fall through to the external/exec path in --zsh
13904 // mode; default zshrs mode keeps the anti-fork
13905 // intercept.
13906 if !crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed) =>
13907 {
13908 return dispatch_builtin(cmd.as_str(), rest_vec.clone());
13909 }
13910 // `zstat` — port of zsh/stat module (Src/Modules/stat.c
13911 // BUILTIN("zstat", …)). Returns file metadata as
13912 // `field value` pairs / an assoc / a plus-separated
13913 // list depending on flags. zsh ALSO registers `stat`
13914 // bound to the same handler, but that name conflicts
13915 // with the system `stat(1)` binary (every script that
13916 // calls `stat -f '%Lp' …` would break). zsh resolves
13917 // this through opt-in `zmodload`; zshrs's modules are
13918 // statically linked so we keep `stat` routing to the
13919 // external command and only intercept the unambiguous
13920 // `zstat` name.
13921 "zstat" => {
13922 // Canonical bin_stat per stat.c:638 via BUILTINS["zstat"].
13923 return dispatch_builtin("zstat", rest_vec.clone());
13924 }
13925 _ => {}
13926 }
13927
13928 // AOP intercepts: when an `intercept :before/:around/:after foo` block
13929 // is registered, dynamic-command-name dispatch must consult it before
13930 // spawning. Without this, `cmd=ls; $cmd` bypasses every intercept that
13931 // a literal `ls` would trigger. The full_cmd string mirrors what the
13932 // tree-walker era passed (cmd + args joined by space) so existing
13933 // pattern matchers continue to work.
13934 if !self.intercepts.is_empty() {
13935 let full_cmd = if rest_vec.is_empty() {
13936 cmd.clone()
13937 } else {
13938 format!("{} {}", cmd, rest_vec.join(" "))
13939 };
13940 if let Some(intercept_result) = self.run_intercepts(cmd, &full_cmd, &rest_vec) {
13941 return intercept_result.unwrap_or(127);
13942 }
13943 }
13944
13945 // User-defined function lookup before OS-level exec. zsh's
13946 // dynamic-command-name dispatch (`cmd=hook1; $cmd`) checks
13947 // the function table FIRST — without this, `$f` for a
13948 // function-name `f` was always falling through to
13949 // `execute_external` and erroring "command not found".
13950 // Plugin code uses this pattern constantly:
13951 // for f in "${precmd_functions[@]}"; do "$f"; done
13952 if self.function_exists(cmd) {
13953 if let Some(status) = self.dispatch_function_call(cmd, &rest_vec) {
13954 return status;
13955 }
13956 }
13957
13958 self.execute_external(cmd, &rest_vec, &[]).unwrap_or(127)
13959 }
13960}