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 static TRY_ESCAPE_SAVE: RefCell<Vec<(i32, i32, i32, i32)>> =
113 const { RefCell::new(Vec::new()) };
114 /// Re-entry guard for BUILTIN_DEBUG_TRAP. While the DEBUG trap
115 /// body is running, the per-statement DEBUG_TRAP dispatch in the
116 /// trap body must NOT re-fire (otherwise infinite recursion +
117 /// stack overflow). zsh's in_trap counter at Src/signals.c
118 /// serves the same purpose.
119 static DEBUG_TRAP_REENTRY: Cell<bool> = const { Cell::new(false) };
120 /// Stack of (saved_stdout, saved_stderr) tuples pushed by
121 /// `cmd_subst` around its nested-VM run. RUST-ONLY: zsh forks
122 /// each cmdsub so trap output during the cmdsub naturally
123 /// lands on the PARENT's stdout. zshrs's in-process cmdsub
124 /// dups fd 1 → pipe, so a trap firing during cmdsub would
125 /// emit into the captured value. Traps consult this stack
126 /// to route their body output to the topmost saved_stdout
127 /// instead of the cmdsub's fd 1. Bug #56 in docs/BUGS.md.
128 pub static CMDSUBST_OUTER_FDS: RefCell<Vec<(i32, i32)>> =
129 const { RefCell::new(Vec::new()) };
130 /// c:Src/exec.c:5025 getproc (PATH_DEV_FD branch) — the parent
131 /// keeps the `>(cmd)` pipe WRITE end open under `/dev/fd/N`,
132 /// parks it in the job's filelist (`fdtable[fd] =
133 /// FDT_PROC_SUBST; addfilelist(NULL, fd)`), and deletefilelist
134 /// closes it when the consuming job finishes — that close is
135 /// what lets the `>(cmd)` child's reader see EOF. zshrs runs
136 /// commands in-process, so the equivalent is: record
137 /// `(scope_depth, fd)` here and drain after the consuming
138 /// command (external exec or builtin dispatch) completes.
139 static PSUB_PENDING_FDS: RefCell<Vec<(usize, i32)>> = const { RefCell::new(Vec::new()) };
140 /// Scope depth for PSUB_PENDING_FDS tagging. Incremented around
141 /// nested execution contexts (cmd-subst bodies, shell-function
142 /// bodies) so a command running INSIDE the nested context only
143 /// drains its own `>(cmd)` fds, never the enclosing command's
144 /// (e.g. `tee >(wc) $(print x)` — print must not close tee's
145 /// fd). Mirrors C's per-job filelist ownership.
146 static PSUB_SCOPE_DEPTH: Cell<usize> = const { Cell::new(0) };
147 /// Forked `<(cmd)`/`>(cmd)` child pids awaiting reap. Drained
148 /// non-blockingly (WNOHANG) by note_psub_child so proc-sub children
149 /// don't accumulate as zombies across a shell session.
150 static PSUB_CHILDREN: RefCell<Vec<i32>> = const { RefCell::new(Vec::new()) };
151}
152
153/// Record a proc-sub child pid and best-effort reap any already-exited
154/// proc-sub children (WNOHANG). Non-blocking: still-running children
155/// stay parked and get reaped on a later call.
156pub(crate) fn note_psub_child(pid: i32) {
157 if pid <= 0 {
158 return;
159 }
160 PSUB_CHILDREN.with(|v| {
161 let mut v = v.borrow_mut();
162 v.push(pid);
163 v.retain(|&p| {
164 let mut status = 0;
165 // WNOHANG: reap if exited, else keep parked.
166 let r = unsafe { libc::waitpid(p, &mut status, libc::WNOHANG) };
167 // r == p → reaped; r == 0 → still running (keep); r < 0 →
168 // already gone/not ours (drop).
169 r == 0
170 });
171 });
172}
173
174/// Port of `fputs(s, xtrerr)` / `fprintf(xtrerr, "%s", s)` (Src/exec.c):
175/// append `s` to the buffered xtrace line. Nothing reaches stderr until
176/// [`xtrerr_flush`] (the port of `fflush(xtrerr)`) writes the line whole.
177pub(crate) fn xtrerr_fputs(s: &str) {
178 XTRERR.with(|b| b.borrow_mut().push_str(s));
179}
180
181/// Port of `fflush(xtrerr)` (Src/exec.c:1373/2123/2596): write the
182/// buffered xtrace line to stderr in ONE `write` and clear the buffer, so
183/// the line lands on the shared fd atomically (no interleaving across
184/// concurrent pipeline stages).
185pub(crate) fn xtrerr_flush() {
186 XTRERR.with(|b| {
187 let mut buf = b.borrow_mut();
188 if !buf.is_empty() {
189 use std::io::Write;
190 let _ = std::io::stderr().write_all(buf.as_bytes());
191 buf.clear();
192 }
193 });
194}
195
196/// RAII guard bumping the psub scope depth — see PSUB_SCOPE_DEPTH.
197pub(crate) struct PsubScope;
198
199impl PsubScope {
200 pub(crate) fn enter() -> Self {
201 PSUB_SCOPE_DEPTH.with(|d| d.set(d.get() + 1));
202 PsubScope
203 }
204}
205
206impl Drop for PsubScope {
207 fn drop(&mut self) {
208 PSUB_SCOPE_DEPTH.with(|d| d.set(d.get().saturating_sub(1)));
209 }
210}
211
212/// RAII guard bumping `$ZSH_SUBSHELL` for the duration of an
213/// in-process command substitution.
214///
215/// c:Src/exec.c:1161 — entersubsh() does `zsh_subshell++;` and zsh's
216/// cmdsub FORKS, so the increment dies with the child and the parent's
217/// value is untouched. zshrs runs cmdsubs in-process on a nested VM,
218/// so the visible param must be bumped on entry and restored on exit.
219/// Writes paramtab u_val directly because ZSH_SUBSHELL is PM_READONLY
220/// (same bypass pattern as the subshell-builtin bump below); also
221/// mirrors into ported::exec::zsh_subshell so exec.c:4376-style
222/// `forked | zsh_subshell` reads agree.
223pub(crate) struct CmdSubstSubshellBump {
224 saved_val: i64,
225 saved_str: Option<String>,
226}
227
228impl CmdSubstSubshellBump {
229 pub(crate) fn enter() -> Self {
230 let mut saved_val = 0i64;
231 let mut saved_str = None;
232 if let Ok(mut tab) = crate::ported::params::paramtab().write() {
233 if let Some(pm) = tab.get_mut("ZSH_SUBSHELL") {
234 saved_val = pm.u_val;
235 saved_str = pm.u_str.clone();
236 pm.u_val = saved_val + 1;
237 pm.u_str = Some((saved_val + 1).to_string());
238 pm.node.flags &= !(crate::ported::zsh_h::PM_UNSET as i32);
239 }
240 }
241 crate::ported::exec::zsh_subshell.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
242 CmdSubstSubshellBump {
243 saved_val,
244 saved_str,
245 }
246 }
247}
248
249impl Drop for CmdSubstSubshellBump {
250 fn drop(&mut self) {
251 if let Ok(mut tab) = crate::ported::params::paramtab().write() {
252 if let Some(pm) = tab.get_mut("ZSH_SUBSHELL") {
253 pm.u_val = self.saved_val;
254 pm.u_str = self.saved_str.take();
255 }
256 }
257 crate::ported::exec::zsh_subshell.fetch_sub(1, std::sync::atomic::Ordering::Relaxed);
258 }
259}
260
261/// Port of deletefilelist() from Src/jobs.c (the `>(cmd)` fd arm):
262/// closes every pending proc-subst write end created at or inside
263/// the current scope depth, exactly when C deletes the consuming
264/// job's filelist (getproc parks the fd there via
265/// `addfilelist(NULL, fd)`, Src/exec.c:5025+).
266fn close_pending_psub_fds() {
267 let depth = PSUB_SCOPE_DEPTH.with(|d| d.get());
268 PSUB_PENDING_FDS.with(|v| {
269 v.borrow_mut().retain(|&(d, fd)| {
270 if d >= depth {
271 unsafe { libc::close(fd) };
272 false
273 } else {
274 true
275 }
276 });
277 });
278}
279
280/// RAII drain guard — instantiated at the top of the consuming-
281/// command paths (dispatch_builtin, ZshrsHost::exec) so the pending
282/// `>(cmd)` write ends close on every exit path once the command
283/// finished, exactly when C's job filelist would be deleted.
284struct PsubFdGuard;
285
286impl Drop for PsubFdGuard {
287 fn drop(&mut self) {
288 close_pending_psub_fds();
289 }
290}
291
292/// Peek the outermost cmdsub-saved (stdout, stderr) fds, if any.
293/// Returns None when no cmdsub is currently capturing. Used by the
294/// trap dispatcher in `src/ported/signals.rs::dotrap` to route trap
295/// body output to the parent's real stdout (matching zsh's forked
296/// cmdsub behaviour) instead of the cmdsub's pipe-bound fd 1.
297/// Bug #56 in docs/BUGS.md.
298pub fn cmdsubst_outer_stdout() -> Option<i32> {
299 CMDSUBST_OUTER_FDS.with(|s| s.borrow().last().map(|(o, _)| *o))
300}
301
302// Thread-local pointer to the current ShellExecutor.
303// Set before VM execution, cleared after. Used by builtin handlers.
304thread_local! {
305 static CURRENT_EXECUTOR: RefCell<Option<*mut ShellExecutor>> = const { RefCell::new(None) };
306 /// Set by subshell_end after a deferred subshell `exit N` lands.
307 /// Read + cleared by the next GET_VAR sync_status path so the
308 /// vm.last_status → LASTVAL sync doesn't clobber the deferred
309 /// exit status. RUST-ONLY: needed because zshrs runs subshells
310 /// in-process (no fork) so vm.last_status doesn't track the
311 /// subshell's exit; C zsh's subshell forks and the child's
312 /// process::exit(N) becomes $? in the parent automatically.
313 static SUBSHELL_EXIT_STATUS_PENDING: std::cell::Cell<bool> = const { std::cell::Cell::new(false) };
314 /// The installed session executor, registered by
315 /// `exec::install_session_executor`. Lets [`with_session_context`]
316 /// establish a VM execution context for STARTUP work that runs
317 /// before the loop's first `execode` enters one (rc-file sourcing in
318 /// `run_init_scripts`, c:1914). Mirrors exec.rs's own
319 /// `SESSION_EXECUTOR`; kept here so the context helper lives next to
320 /// `ExecutorContext`/`CURRENT_EXECUTOR`.
321 static SESSION_EXECUTOR_PTR: std::cell::Cell<Option<*mut ShellExecutor>> = const { std::cell::Cell::new(None) };
322 /// GLOB_ASSIGN eligibility carrier. Set true by BUILTIN_MARK_GLOB_ELIGIBLE
323 /// (emitted by the compiler ONLY when a scalar-assignment RHS carries an
324 /// UNQUOTED glob token — Star/Quest/Inbrack), read+cleared by the next
325 /// BUILTIN_SET_VAR. Matches C zsh's `GLOB_ASSIGN` (Src/exec.c:2554): only
326 /// a literal unquoted glob pattern in the wordcode is globbed; values from
327 /// `$param` / `$(cmd)` / quoted strings are NOT (verified against zsh).
328 /// The runtime SET_VAR value arrives untokenized (the compiler DQ-wraps to
329 /// suppress compile-time globbing), so quoting can no longer be recovered
330 /// from the value bytes — this flag carries the compile-time decision.
331 static SET_VAR_GLOB_ELIGIBLE: std::cell::Cell<bool> = const { std::cell::Cell::new(false) };
332}
333
334/// Register the session executor pointer (called from
335/// `install_session_executor`). See [`with_session_context`].
336pub fn register_session_executor(exec: &mut ShellExecutor) {
337 SESSION_EXECUTOR_PTR.with(|c| c.set(Some(exec as *mut ShellExecutor)));
338}
339
340/// Run `f` with the registered session executor established as
341/// `CURRENT_EXECUTOR`, so code reaching the live executor via
342/// `try_with_executor` works even when no per-command `execode` context
343/// is active yet.
344///
345/// Sole caller: `zsh_main`'s `run_init_scripts()` (c:1914), which
346/// sources `.zshenv`/`.zshrc`/`.zlogin` via `source()` BEFORE the loop's
347/// first `execode`. Without an active context those sourced bodies
348/// `try_with_executor` → `None` → no-op, so the shell silently ignored
349/// the user's dotfiles. The scope is entered once around the startup
350/// sourcing window and dropped before the loop begins — deliberately NOT
351/// a global fallback inside `execute_script_zsh_pipeline`, which would
352/// re-enter the executor on nested command substitution and block on
353/// input.
354pub fn with_session_context<R>(f: impl FnOnce() -> R) -> R {
355 let ptr = SESSION_EXECUTOR_PTR.with(|c| c.get());
356 match ptr {
357 // SAFETY: set by install_session_executor to an executor that
358 // outlives the single-threaded interactive session.
359 Some(ptr) => {
360 let _ctx = ExecutorContext::enter(unsafe { &mut *ptr });
361 f()
362 }
363 None => f(),
364 }
365}
366
367/// RAII guard that sets/clears the thread-local executor pointer.
368///
369/// Idempotent: calling `enter` when a context is already active is a no-op
370/// for the entry side, and the guard's drop only clears the thread-local if
371/// *this* call was the one that set it. Nested `execute_command` invocations
372/// (e.g. from inside a builtin handler) reuse the outer pointer instead of
373/// stomping it.
374pub(crate) struct ExecutorContext {
375 we_set_it: bool,
376}
377
378impl ExecutorContext {
379 pub(crate) fn enter(executor: &mut ShellExecutor) -> Self {
380 let we_set_it = CURRENT_EXECUTOR.with(|cell| {
381 let mut slot = cell.borrow_mut();
382 if slot.is_some() {
383 false
384 } else {
385 *slot = Some(executor as *mut ShellExecutor);
386 true
387 }
388 });
389 ExecutorContext { we_set_it }
390 }
391}
392
393impl Drop for ExecutorContext {
394 fn drop(&mut self) {
395 if self.we_set_it {
396 CURRENT_EXECUTOR.with(|cell| {
397 *cell.borrow_mut() = None;
398 });
399 }
400 }
401}
402
403/// Access the current executor from a builtin handler.
404/// # Safety
405/// Only call this from within a VM execution context (after ExecutorContext::enter).
406#[inline]
407pub(crate) fn with_executor<F, R>(f: F) -> R
408where
409 F: FnOnce(&mut ShellExecutor) -> R,
410{
411 CURRENT_EXECUTOR.with(|cell| {
412 let ptr = cell
413 .borrow()
414 .expect("with_executor called outside VM context");
415 // SAFETY: The pointer is valid for the duration of VM execution,
416 // and we're single-threaded within the executor.
417 let executor = unsafe { &mut *ptr };
418 f(executor)
419 })
420}
421
422/// Non-panicking variant of [`with_executor`]: runs `f` against the
423/// current executor and returns `Some(result)`, or `None` when no
424/// executor is in scope (`CURRENT_EXECUTOR` unset — e.g. unit tests /
425/// compsys contexts with no fusevm bridge running).
426///
427/// This is the primitive the `crate::ported::exec` accessor wrappers
428/// (array/assoc/dispatch_function_call/execute_script/...) use to
429/// reach the live executor while preserving the exact "no executor →
430/// fall back to the direct param table / default value" semantics that
431/// the deleted `exec_hooks` OnceLock layer encoded via its
432/// "is-the-hook-installed?" check. `CURRENT_EXECUTOR` being set is the
433/// faithful equivalent of "the bridge installed the hooks".
434#[inline]
435pub(crate) fn try_with_executor<F, R>(f: F) -> Option<R>
436where
437 F: FnOnce(&mut ShellExecutor) -> R,
438{
439 CURRENT_EXECUTOR.with(|cell| {
440 let ptr = (*cell.borrow())?;
441 // SAFETY: same contract as with_executor — the pointer is valid
442 // for the duration of VM execution and access is single-threaded.
443 let executor = unsafe { &mut *ptr };
444 Some(f(executor))
445 })
446}
447
448/// Look up a canonical builtin by name in `BUILTINS` and dispatch
449/// via `execbuiltin` (Src/builtin.c:250). NO shadow check — calls the
450/// builtin even if a user function with the same name exists. Used by
451/// the `builtin foo` prefix opcode (which explicitly bypasses function
452/// lookup per zsh semantics) and by internal call sites where shadowing
453/// is unwanted. For zsh's normal name-resolution order (function shadows
454/// builtin), use `dispatch_builtin` instead.
455/// Shell-identifier prefix for diagnostic lines. Reads the canonical
456/// scriptname (`zsh` in `--zsh` parity mode, `zshrs` otherwise) so a
457/// single helper replaces hardcoded `"zshrs:"` literals across the
458/// file's eprintln paths.
459fn shname() -> String {
460 crate::ported::utils::scriptname_get().unwrap_or_else(|| "zshrs".to_string())
461}
462
463/// c:Src/subst.c:505-507 + Src/exec.c:3378-3380 — per-command
464/// CSH_NULL_GLOB outcome check. During this command's word expansion
465/// `expand_glob` accumulated `badcshglob |= 1` per failed glob and
466/// `|= 2` per successful one (Src/glob.c:1871-1875). Exactly 1 —
467/// failures and no successes — is the csh-style error: `no match`,
468/// command skipped, status 1. Any other value (0 = no globs, 2/3 =
469/// at least one matched) is silent. Always resets the counter for
470/// the next command (C resets at prefork entry, subst.rs:1307).
471/// Returns true when the error fired; callers mirror their
472/// glob_failed handling (builtins leave ERRFLAG_ERROR set so the
473/// script aborts, externals clear it so the next sublist runs —
474/// verified against zsh 5.9.1).
475/// Restore the user's GLOB_SUBST after a `${~spec}` carrier flip
476/// (see subst::TILDE_GLOBSUBST_CARRIER). Runs at the same
477/// command-dispatch boundaries that consume glob_failed /
478/// badcshglob — by then every glob op of the current word pipeline
479/// has read the carrier.
480pub(crate) fn consume_tilde_globsubst_carrier() {
481 crate::ported::subst::TILDE_GLOBSUBST_CARRIER.with(|c| {
482 if let Some(saved) = c.take() {
483 crate::ported::options::opt_state_set("globsubst", saved);
484 }
485 });
486}
487
488/// Pop `argc` stack slots for a whole-array assignment: the LAST popped
489/// (deepest pushed) is the param name, the rest are the values in stack
490/// order, with any `Value::Array` flattened to its elements. Mirrors the
491/// Flatten an array-assignment RHS value into scalar strings, descending
492/// through nested `Value::Array`s. zsh arrays are always flat, so recursion
493/// only collapses the wrapper layers the compiler introduces — in particular
494/// the single `Value::Array` built by `Op::MakeArray` for `arr=(...)` literals
495/// (used to dodge `CallBuiltin`'s u8 argc cap), whose own elements may
496/// themselves be arrays from an unquoted `$other_array` expansion. A
497/// one-level flatten would stringify those inner arrays into a single element.
498fn flatten_array_value(v: Value, out: &mut Vec<String>) {
499 match v {
500 Value::Array(items) => {
501 for it in items {
502 flatten_array_value(it, out);
503 }
504 }
505 other => out.push(other.to_str()),
506 }
507}
508
509/// pop/flatten prologue of BUILTIN_SET_ARRAY / BUILTIN_APPEND_ARRAY.
510fn pop_array_args_with_name(vm: &mut fusevm::VM, argc: u8) -> (String, Vec<String>) {
511 let n = argc as usize;
512 let mut popped: Vec<Value> = Vec::with_capacity(n);
513 for _ in 0..n {
514 popped.push(vm.pop());
515 }
516 popped.reverse();
517 let name = popped.pop().map(|v| v.to_str()).unwrap_or_default();
518 let mut values: Vec<String> = Vec::new();
519 for v in popped {
520 flatten_array_value(v, &mut values);
521 }
522 (name, values)
523}
524
525fn consume_badcshglob() -> bool {
526 let v = crate::ported::glob::BADCSHGLOB.swap(0, std::sync::atomic::Ordering::Relaxed);
527 if v == 1 {
528 crate::ported::utils::zerr("no match"); // c:Src/subst.c:507
529 true
530 } else {
531 false
532 }
533}
534
535/// Map a builtin name to the zsh module that owns it, IFF zsh does
536/// not auto-load that builtin on first use. Used by
537/// `dispatch_builtin_raw` to gate `--zsh` mode dispatch behind
538/// `zmodload`, mirroring `zsh -fc <name>` returning 127 for these
539/// names without an explicit module load.
540///
541/// Returns `Some(module_name)` if `name` belongs to a non-auto-load
542/// module per the per-module `Src/Modules/<x>.c` `bintab[]` plus
543/// the auto-load flag set at module-build time. `None` for core
544/// builtins and for auto-loaded module builtins (sched, log, echotc,
545/// echoti, zformat, zparseopts, zregexparse, zstyle, strftime,
546/// private, vared, zle, bindkey, comp*) which work without zmodload.
547fn module_bound_builtin_module(name: &str) -> Option<&'static str> {
548 match name {
549 "zftp" => Some("zsh/zftp"),
550 "zsocket" => Some("zsh/net/socket"),
551 "ztcp" => Some("zsh/net/tcp"),
552 "zstat" => Some("zsh/stat"),
553 "zselect" => Some("zsh/zselect"),
554 "zpty" => Some("zsh/zpty"),
555 "zprof" => Some("zsh/zprof"),
556 "zsystem" | "syserror" => Some("zsh/system"),
557 "clone" => Some("zsh/clone"),
558 "zcurses" => Some("zsh/curses"),
559 "ztie" | "zuntie" | "zgdbmpath" => Some("zsh/db/gdbm"),
560 "pcre_compile" | "pcre_match" | "pcre_study" => Some("zsh/pcre"),
561 "example" => Some("zsh/example"),
562 "cap" | "getcap" | "setcap" => Some("zsh/cap"),
563 "zgetattr" | "zsetattr" | "zdelattr" | "zlistattr" => Some("zsh/attr"),
564 // c:Src/Modules/datetime.c — `strftime` is registered via
565 // partab[] when zsh/datetime loads. Verified by
566 // `zsh -fc 'strftime -s s %Y 0'` → 127 "command not found".
567 "strftime" => Some("zsh/datetime"),
568 _ => None,
569 }
570}
571
572pub(crate) fn dispatch_builtin_raw(name: &str, args: Vec<String>) -> i32 {
573 // c:Src/exec.c:2700-2717 — `private` is an autoloaded builtin in
574 // zsh (autofeature b:private of zsh/param/private): first use
575 // runs ensurefeature → require_module → load_module → boot_,
576 // marking the module MOD_INIT_B (what `zmodload -e` reads) and
577 // installing the wrap_private FuncWrap (param_private.c:712).
578 // doshfunc gates the wrapper dispatch on that load state, so
579 // this require_module is what activates private scoping. The
580 // raw dispatcher is the chokepoint every builtin route funnels
581 // through; require_module is idempotent after the first call
582 // (needs_load checks MOD_INIT_B).
583 if name == "private" {
584 if let Ok(mut tab) = crate::ported::module::MODULESTAB.lock() {
585 let _ = crate::ported::module::require_module(&mut tab, "zsh/param/private", None, 0);
586 // c:2710 ensurefeature
587 }
588 }
589 // c:Src/Modules/param_private.c:682-685 setup_ — loading
590 // zsh/param/private REPLACES the `local` builtintab node's
591 // handlerfunc + optstr with bin_private's ("Even more horrible
592 // hack"), so once the module is loaded `local` IS bin_private: it
593 // accepts the -P/-Pa private-scope flags, and without -P delegates
594 // to bin_typeset, which already treats `local` and `private`
595 // identically (is_locallike, builtin.rs:3666). Replicate the swap by
596 // routing `local` through the `private` node only after the module
597 // is loaded — before then, `local -P` still errors "bad option: -P"
598 // exactly like stock zsh. The `private` node carries the augmented
599 // optstr (with P) that the `local` node lacks.
600 if name == "local"
601 && crate::ported::module::MODULESTAB
602 .lock()
603 .map(|t| t.is_bound("zsh/param/private"))
604 .unwrap_or(false)
605 {
606 return dispatch_builtin_raw("private", args);
607 }
608 // c:Bugs #475/#504/#555 — bash-only builtins (`mapfile`,
609 // `readarray`, `compopt`) should emit "command not found" in
610 // `--zsh` mode matching zsh's external-command-lookup miss.
611 // The per-opcode closures for caller/help/complete/compgen
612 // already gate via IS_ZSH_MODE at their registration sites;
613 // names without dedicated opcodes (compopt/mapfile/readarray)
614 // route through this generic builtintab lookup and need the
615 // gate here.
616 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed)
617 && matches!(name, "compopt" | "mapfile" | "readarray")
618 {
619 eprintln!("zsh:1: command not found: {}", name);
620 let _ = args;
621 return 127;
622 }
623 // c:Src/exec.c:2700-2724 resolvebuiltin — autoloaded-builtin stub
624 // (registered by `zmodload -ab MOD NAME`, Src/module.c:426
625 // add_autobin) fires on first use: ensurefeature loads the owning
626 // module, then dispatch proceeds against the real builtin. Must
627 // run BEFORE the module-bound 127 gate below — `zmodload -ab
628 // zsh/zselect zselect; zselect` previously died there with
629 // `command not found` because the gate only checked is_loaded,
630 // never the autoload ledger.
631 if let Some(rc) = crate::ported::module::resolvebuiltin(name) {
632 if rc != 0 {
633 // Load failed or feature undefined — diagnostics already
634 // printed (load_module zwarn / resolvebuiltin zerr).
635 // C's execbuiltin head returns 1 (Src/builtin.c:264-267).
636 return 1;
637 }
638 // Module loaded — fall through; the is_loaded gates below now
639 // pass and the normal dispatch chain runs the real builtin.
640 }
641 // c:Src/Modules/<mod>.c boot_/setup_ chain — module-bound builtins
642 // (zftp, zsocket, ztcp, zstat, etc.) are only registered into
643 // `builtintab` when their module is loaded via `zmodload`. In
644 // zsh `-fc` (the parity test harness's invocation), the modules
645 // are NOT pre-loaded, so each name reports "command not found"
646 // with exit 127. zshrs intentionally pre-loads all module bintabs
647 // in `createbuiltintable` (builtin.rs:131-152) for the default
648 // mode so users can call these without `zmodload`; that auto-load
649 // diverges from zsh's gate behavior. Match zsh's stance only when
650 // the user explicitly asked for parity via `--zsh`.
651 //
652 // The list is the union of builtins from modules that zsh does
653 // NOT auto-load (verified via `zsh -fc <name>` returning 127):
654 // zsh/zftp → zftp
655 // zsh/net/socket → zsocket
656 // zsh/net/tcp → ztcp
657 // zsh/stat → zstat (NOT `stat`; that name resolves to
658 // /bin/stat on PATH per zsh's setup)
659 // zsh/zselect → zselect
660 // zsh/zpty → zpty
661 // zsh/zprof → zprof
662 // zsh/system → zsystem, syserror
663 // zsh/clone → clone
664 // zsh/curses → zcurses
665 // zsh/db/gdbm → ztie, zuntie, zgdbmpath
666 // zsh/pcre → pcre_compile, pcre_match, pcre_study
667 // zsh/example → example
668 // zsh/cap → cap, getcap, setcap
669 // zsh/attr → zgetattr, zsetattr, zdelattr, zlistattr
670 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed)
671 && module_bound_builtin_module(name)
672 .map(|m| {
673 !crate::ported::module::MODULESTAB
674 .lock()
675 .map(|t| t.is_loaded(m))
676 .unwrap_or(false)
677 })
678 .unwrap_or(false)
679 {
680 eprintln!("zsh:1: command not found: {}", name);
681 let _ = args;
682 return 127;
683 }
684 // c:Src/Modules/files.c:806-824 — zsh/files registers `chmod`,
685 // `chown`, `chgrp`, `ln`, `mkdir`, `mv`, `rm`, `rmdir`, `sync`
686 // (plus their `zf_*` aliases) into builtintab on module load.
687 // Without an explicit `zmodload zsh/files`, zsh resolves the
688 // names through PATH lookup — `zsh -fc 'chmod +x f'` runs
689 // `/bin/chmod`, whose argv-parser accepts symbolic modes like
690 // `+x` that bin_chmod's octal-only parser rejects with
691 // "invalid mode `+x'". The shadow-aware wrapper at
692 // `dispatch_builtin` (line 438) already has this gate, but the
693 // direct `dispatch_builtin_raw` path used by fusevm's
694 // CallBuiltin opcode bypasses it. Mirror the gate here so the
695 // low-level dispatch matches C's PATH-fall-through behavior.
696 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed)
697 && module_gated_files_builtin(name)
698 && !crate::ported::module::MODULESTAB
699 .lock()
700 .map(|t| t.is_loaded("zsh/files"))
701 .unwrap_or(false)
702 {
703 // PATH lookup uses the LITERAL name: bare `mkdir` finds
704 // /bin/mkdir; a `zf_*` alias finds nothing and exits 127 —
705 // matching zsh -fc `zf_mkdir d` → "command not found:
706 // zf_mkdir" (the aliases exist ONLY in the loaded module's
707 // builtintab, Src/Modules/files.c:816-824; PATH has no
708 // /bin/zf_rm). The previous zf_-strip silently ran the
709 // system binary instead.
710 let status = with_executor(|exec| exec.execute_external(name, &args, &[])).unwrap_or(127);
711 return status;
712 }
713 // c:Src/Modules/stat.c:637-638 — zsh/stat registers BOTH `stat`
714 // and `zstat`. `zstat` is in the module_bound 127-gate above (no
715 // /usr/bin/zstat exists), but the bare `stat` name must FALL
716 // THROUGH to PATH when zsh/stat isn't loaded — zsh -fc
717 // 'stat -f %Lp f' runs /usr/bin/stat, while bin_stat's parser
718 // rejects stat(1) flags ("bad option: -c"). Same fall-through
719 // shape as the zsh/files gate above.
720 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed)
721 && name == "stat"
722 && !crate::ported::module::MODULESTAB
723 .lock()
724 .map(|t| t.is_loaded("zsh/stat"))
725 .unwrap_or(false)
726 {
727 let status = with_executor(|exec| exec.execute_external(name, &args, &[])).unwrap_or(127);
728 return status;
729 }
730 // c:Src/exec.c:3050-3068 — builtin lookup hits `builtintab` (the
731 // merged table containing module-provided builtins). The previous
732 // port walked only the core `BUILTINS` slice, so per-module
733 // entries like `log` (Src/Modules/watch.c:693 `BUILTIN("log", …,
734 // bin_log, …)`) were registered into builtintab via
735 // createbuiltintable but never reached at dispatch — `log` fell
736 // through to PATH and ran `/usr/bin/log`. Bug #72 in docs/BUGS.md.
737 let tab = crate::ported::builtin::createbuiltintable();
738 if let Some(bn_static) = tab.get(name) {
739 let bn_ptr = *bn_static as *const _ as *mut _;
740 return crate::ported::builtin::execbuiltin(args, Vec::new(), bn_ptr);
741 }
742 1
743}
744
745/// Shadow-aware dispatch matching zsh's name-resolution order:
746/// alias → reserved word → **function (shadows builtin)** → builtin →
747/// external. All `BUILTIN_X` opcode handlers route through here so a
748/// user-defined `cd () { … }` (or `r`, `fc`, `which`, … anything in
749/// fusevm's name→opcode map) takes precedence over the C builtin —
750/// matching `Src/exec.c:execcmd_exec`'s dispatch at c:3050-3068.
751/// Without this, compile-time builtin resolution silently ignored
752/// user wrappers (e.g. ZPWR's `cd () { builtin cd "$@"; … }`).
753/// True for builtins that are bound by zsh/files's boot_/setup_
754/// chain (Src/Modules/files.c:806-824). These are the bare-name
755/// `mkdir`/`rm`/`mv`/`ln`/`chmod`/`chown`/`chgrp`/`sync`/`rmdir`
756/// AND their `zf_*` aliases at c:816-824. Without explicit
757/// `zmodload zsh/files`, the names fall through to PATH lookup
758/// (zsh's `type rm` reports `/bin/rm`). Bug #28.
759fn module_gated_files_builtin(name: &str) -> bool {
760 matches!(
761 name,
762 "mkdir"
763 | "rmdir"
764 | "rm"
765 | "mv"
766 | "ln"
767 | "chmod"
768 | "chown"
769 | "chgrp"
770 | "sync"
771 | "zf_mkdir"
772 | "zf_rmdir"
773 | "zf_rm"
774 | "zf_mv"
775 | "zf_ln"
776 | "zf_chmod"
777 | "zf_chown"
778 | "zf_chgrp"
779 | "zf_sync"
780 )
781}
782
783pub(crate) fn dispatch_builtin(name: &str, args: Vec<String>) -> i32 {
784 // c:Src/exec.c getproc + Src/jobs.c deletefilelist — close any
785 // `>(cmd)` write ends owned by this command once it finishes
786 // (drops on every return path below).
787 let _psub_fds = PsubFdGuard;
788 // c:Src/exec.c — when any redirect in the current scope failed
789 // (e.g. noclobber blocked a `>` overwrite), zsh refuses to
790 // execute the command and exits with status 1. The Rust port
791 // still applied the command (writing to the /dev/null sink
792 // installed by host_apply_redirect's noclobber arm) so the
793 // success status overwrote the intended 1. Short-circuit here
794 // for builtins (the external-exec equivalent lives in
795 // ZshrsHost::exec).
796 let redir_failed = with_executor(|exec| {
797 let f = exec.redirect_failed;
798 exec.redirect_failed = false;
799 f
800 });
801 if redir_failed {
802 // c:Src/exec.c:4367-4386 — POSIX special-builtin escalation:
803 // a failed redirect on a PSPECIAL builtin (set, readonly,
804 // typeset, ...) under POSIX_BUILTINS is FATAL in a
805 // non-interactive shell (`exit(1)` at c:4383). The `command`
806 // prefix resets this (BINF_COMMAND, c:4369) — that path
807 // dispatches through bin_command, not here.
808 if crate::ported::zsh_h::isset(crate::ported::zsh_h::POSIXBUILTINS)
809 && !crate::ported::zsh_h::isset(crate::ported::zsh_h::INTERACTIVE)
810 && builtin_is_pspecial(name)
811 {
812 use std::sync::atomic::Ordering;
813 crate::ported::builtin::EXIT_VAL.store(1, Ordering::Relaxed);
814 crate::ported::builtin::EXIT_PENDING.store(1, Ordering::Relaxed);
815 }
816 return 1;
817 }
818 // c:Src/glob.c:1876-1880 NOMATCH path — when expand_glob() failed
819 // on a no-match glob, zsh aborts the simple command after zerr()
820 // printed "no matches found". In C, this works because zerr()
821 // sets ERRFLAG_ERROR (Src/utils.c) and execcmd_exec()
822 // (Src/exec.c:3050+) checks errflag before invoking the builtin
823 // table. Rust's builtin dispatch doesn't sit on the same errflag
824 // gate, so we explicitly consume the per-command glob-fail cell
825 // and short-circuit with status 1. Mirrors the external-path
826 // guard at host_exec_external (line 5167). Without this:
827 // `echo /never/*` would print empty (silently rolled back to ""
828 // by the empty glob expansion). Parity bug #13.
829 consume_tilde_globsubst_carrier();
830 let glob_failed = with_executor(|exec| {
831 let f = exec.current_command_glob_failed.get();
832 exec.current_command_glob_failed.set(false); // c:1879 cleanup
833 f
834 });
835 if glob_failed {
836 // c:Src/glob.c:1876-1880 + Src/exec.c — NOMATCH zerr sets
837 // ERRFLAG_ERROR (via utils.c:184). For a BUILTIN command the
838 // expansion runs IN the shell process, so errflag stays set
839 // and the rest of the input aborts (zsh -fc 'echo /nope_*;
840 // echo after' prints nothing after the error — verified
841 // against zsh 5.9). The continue-after-nomatch behaviour
842 // belongs ONLY to externals: C forks BEFORE expansion there,
843 // so the child's zerr can't touch the parent's errflag (zsh
844 // -fc 'ls /nope_*; echo after' prints `after`) — that path's
845 // clear lives in fn exec / execute_external. Leave
846 // ERRFLAG_ERROR set here; BUILTIN_ERREXIT_CHECK trigger 4
847 // aborts the remaining script at the next command boundary.
848 return 1; // c:1880 — command aborted, status 1
849 }
850 // c:Src/subst.c:505-507 — CSH_NULL_GLOB sibling of the NOMATCH
851 // gate above: all of this command's globs failed silently (words
852 // dropped, badcshglob accumulated 1s and no 2s) → `no match`,
853 // skip the builtin, status 1. Like the NOMATCH path, ERRFLAG
854 // from zerr stays set for builtins so the rest of the script
855 // aborts (zsh -fc 'setopt cshnullglob; print *nope* x; print
856 // after' prints only the error — verified zsh 5.9.1).
857 if consume_badcshglob() {
858 // c:Src/exec.c:3380 — `lastval = 1;` so the shell's final
859 // exit status reflects the aborted command.
860 with_executor(|exec| exec.set_last_status(1));
861 return 1;
862 }
863 // c:Src/exec.c:4162-4295 — assignment-builtin (BINF_ASSIGN family:
864 // typeset / declare / local / export / readonly / integer / float /
865 // private) whose `name=value` postassign arg raised errflag while
866 // its RHS was preforked (PREFORK_ASSIGN, c:4239-4245) — the classic
867 // case is a math error in `typeset -F fv=$((1/0))`. The postassign
868 // loop `break`s on errflag (c:4243) and then `if (!errflag)
869 // execbuiltin(...)` (c:4287) SKIPS the builtin entirely, so `lastval`
870 // is left UNCHANGED from before the command (0 fresh, 1 after
871 // `false`). This differs from a PLAIN assignment `x=$((1/0))`, which
872 // goes through execsimple c:1375 `lv = errflag ? errflag : cmdoutval`
873 // → 1, and from a NON-assign builtin `print $((1/0))`, whose main
874 // args-prefork errflag lands on c:3760 `lastval = 1`. Only the
875 // assignment-BUILTIN postassign path preserves the prior status.
876 // Mirror it here: the fusevm reg_passthru dispatch still calls us
877 // with errflag set (unlike C's pre-invoke gate), so consume that
878 // state and return the prior LASTVAL instead of running the builtin.
879 {
880 use std::sync::atomic::Ordering;
881 let live = crate::ported::utils::errflag.load(Ordering::Relaxed);
882 let ef = live & crate::ported::zsh_h::ERRFLAG_ERROR;
883 let hard = live & crate::ported::zsh_h::ERRFLAG_HARD;
884 // Only the SOFT recoverable error (math failure like `$((1/0))`,
885 // ERRFLAG_ERROR without ERRFLAG_HARD) preserves the prior status
886 // per c:4287. A HARD error (`${var?msg}`, which c:Src/subst.c
887 // OR's ERRFLAG_HARD onto errflag) is a script-abort that yields
888 // status 1 regardless of the prior status — leave that to the
889 // normal dispatch/abort path below (which returns 1 and keeps
890 // ERRFLAG_HARD set for the downstream errexit gate).
891 if ef != 0 && hard == 0 && builtin_is_assign_family(name) {
892 // c:4287 — execbuiltin skipped; lastval unchanged.
893 return crate::ported::builtin::LASTVAL.load(Ordering::Relaxed);
894 }
895 }
896 if let Some(status) = try_user_fn_override(name, &args) {
897 // c:Src/jobs.c:1748 waitonejob — canonical single-command
898 // pipestats update via the no-procs else-branch.
899 crate::ported::builtin::LASTVAL.store(status, std::sync::atomic::Ordering::Relaxed);
900 let mut synth = crate::ported::zsh_h::job::default();
901 crate::ported::jobs::waitonejob(&mut synth);
902 return status;
903 }
904 // c:Src/builtin.c:587 + Src/exec.c:3056 — a builtin disabled via
905 // `disable <name>` has its `DISABLED` flag set in `builtintab`;
906 // `builtintab->getnode` (the DISABLED-filtering accessor) returns
907 // NULL for it at lookup time, so execcmd_exec falls through to
908 // PATH lookup and runs the external. The Rust port stores the
909 // disabled set in `BUILTINS_DISABLED`; the previous dispatcher
910 // only checked the immutable `createbuiltintable` HashMap which
911 // never reflects disablement — so `disable echo; echo hi` kept
912 // running the bin_echo builtin. Bug #106 in docs/BUGS.md.
913 //
914 // dispatch_builtin (the high-level wrapper used by the BUILTIN_*
915 // opcode handlers and reg_passthru! callsites) is the correct
916 // gate: `dispatch_builtin_raw` is the low-level entry point
917 // used by `bin_builtin` itself which MUST bypass the disabled
918 // set (man zshbuiltins: `builtin name` runs the builtin
919 // regardless of disable state). Place the check here so the
920 // bypass path stays clean.
921 let disabled = crate::ported::builtin::BUILTINS_DISABLED
922 .lock()
923 .map(|s| s.contains(name))
924 .unwrap_or(false);
925 if disabled {
926 let status = with_executor(|exec| exec.execute_external(name, &args, &[])).unwrap_or(127);
927 crate::ported::builtin::LASTVAL.store(status, std::sync::atomic::Ordering::Relaxed);
928 let mut synth = crate::ported::zsh_h::job::default();
929 crate::ported::jobs::waitonejob(&mut synth);
930 return status;
931 }
932 // c:Src/Modules/files.c:806-814 — `mkdir`, `rm`, `mv`, `ln`, `chmod`,
933 // `chown`, `chgrp`, `sync`, `rmdir` are bound by the `zsh/files`
934 // module's boot_/setup_ chain. Without explicit `zmodload zsh/files`,
935 // these bare names fall through to PATH (`/bin/rm`, `/usr/bin/chmod`,
936 // etc.) in zsh; `type rm` reports `rm is /bin/rm`. The `zf_*`
937 // aliases (`zf_rm`, `zf_chmod`, …) are bound by the same module
938 // and gated the same way. Bug #28 in docs/BUGS.md.
939 if module_gated_files_builtin(name) {
940 if !crate::ported::module::MODULESTAB
941 .lock()
942 .unwrap()
943 .is_loaded("zsh/files")
944 {
945 // PATH lookup uses the literal name. In --zsh parity mode
946 // `zf_rm` must 127 like zsh -fc (no /bin/zf_rm); default
947 // zshrs mode keeps the convenience zf_-strip so the alias
948 // still reaches the system binary.
949 let path_name = if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed) {
950 name
951 } else {
952 name.strip_prefix("zf_").unwrap_or(name)
953 };
954 let status =
955 with_executor(|exec| exec.execute_external(path_name, &args, &[])).unwrap_or(127);
956 crate::ported::builtin::LASTVAL.store(status, std::sync::atomic::Ordering::Relaxed);
957 let mut synth = crate::ported::zsh_h::job::default();
958 crate::ported::jobs::waitonejob(&mut synth);
959 return status;
960 }
961 }
962 let status = dispatch_builtin_raw(name, args);
963 // c:Src/jobs.c:1748 waitonejob — canonical single-command pipestats update.
964 crate::ported::builtin::LASTVAL.store(status, std::sync::atomic::Ordering::Relaxed);
965 let mut synth = crate::ported::zsh_h::job::default();
966 crate::ported::jobs::waitonejob(&mut synth);
967 // c:Src/exec.c:4367-4386 — done: tail. A PSPECIAL builtin that
968 // raised errflag under POSIX_BUILTINS exits the non-interactive
969 // shell with status 1 ("hard error in POSIX" — e.g. bin_dot's
970 // zerrnam at Src/builtin.c:6133). Arm the deferred-exit pair so
971 // the next ERREXIT_CHECK unwinds; EXIT_VAL=1 matches C's
972 // hardcoded exit(1), NOT the builtin's own status (dot returns
973 // 127 but POSIX exits 1).
974 if crate::ported::zsh_h::isset(crate::ported::zsh_h::POSIXBUILTINS)
975 && !crate::ported::zsh_h::isset(crate::ported::zsh_h::INTERACTIVE)
976 && builtin_is_pspecial(name)
977 && (crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed)
978 & crate::ported::zsh_h::ERRFLAG_ERROR)
979 != 0
980 {
981 use std::sync::atomic::Ordering;
982 crate::ported::builtin::EXIT_VAL.store(1, Ordering::Relaxed);
983 crate::ported::builtin::EXIT_PENDING.store(1, Ordering::Relaxed);
984 }
985 status
986}
987
988/// c:Src/zsh.h:1467 BINF_PSPECIAL — true when `name` is a POSIX
989/// special builtin per the canonical builtin table flags
990/// (Src/builtin.c:48-129: `.`, `:`, break, continue, declare, eval,
991/// exit, export, float, integer, local, readonly, return, set,
992/// shift, source, times, trap, typeset, unset).
993fn builtin_is_pspecial(name: &str) -> bool {
994 crate::ported::builtin::createbuiltintable()
995 .get(name)
996 .map(|b| (b.node.flags as u32 & crate::ported::zsh_h::BINF_PSPECIAL) != 0)
997 .unwrap_or(false)
998}
999
1000/// c:Src/zsh.h:1486 BINF_ASSIGN — the assignment-builtin family
1001/// (typeset / declare / local / export / readonly / integer / float /
1002/// private). Their `name=value` args are handled as postassigns
1003/// (c:Src/exec.c:4162-4295), whose errflag-abort skips execbuiltin and
1004/// preserves the prior `lastval`. Read the flag straight from the
1005/// builtin table (same pattern as `builtin_is_pspecial`).
1006fn builtin_is_assign_family(name: &str) -> bool {
1007 crate::ported::builtin::createbuiltintable()
1008 .get(name)
1009 .map(|b| (b.node.flags as u32 & crate::ported::zsh_h::BINF_ASSIGN) != 0)
1010 .unwrap_or(false)
1011}
1012
1013// The former `install_exec_hooks()` fn-pointer registry is gone. Code
1014// under `src/ported/` now reaches `ShellExecutor` operations
1015// (array/assoc storage, script eval, function dispatch, command
1016// substitution) through the `crate::ported::exec::*` accessor wrappers,
1017// which resolve the live executor via `try_with_executor`
1018// (`CURRENT_EXECUTOR`). The bridge lives in exec.rs — the sanctioned
1019// fusevm-access exception — per `feedback_no_exec_script_from_ported` /
1020// `feedback_no_shellexecutor_in_ported`.
1021
1022/// Register all zsh builtins with the VM.
1023pub(crate) fn register_builtins(vm: &mut fusevm::VM) {
1024 // src/ported/ reaches the live executor (param store, function
1025 // dispatch, nested script/cmdsubst exec) through the
1026 // `crate::ported::exec::*` accessor wrappers, which read
1027 // `CURRENT_EXECUTOR` via `try_with_executor`. No install step is
1028 // needed: the executor is in scope for the duration of any VM run
1029 // (set by `ExecutorContext::enter`), so the wrappers resolve it
1030 // directly. (Replaces the former `exec_hooks` OnceLock fn-ptr
1031 // registry, now deleted.)
1032 // Engage fusevm's tiered JIT (block + tracing) so hot, fully-eligible
1033 // numeric chunks run in native code and — with the `jit-disk-cache`
1034 // feature (on by default) — persist that native code to
1035 // `~/.cache/fusevm-jit`, letting repeated zsh invocations skip Cranelift
1036 // codegen. fusevm gates the JIT on per-chunk eligibility and warms up by
1037 // an invocation threshold, falling back to the interpreter for any chunk
1038 // it cannot compile (e.g. host-builtin/`Extended` command dispatch), so
1039 // enabling it here never changes observable behaviour — it only caches
1040 // the numeric hot path. Idempotent: re-enabling on each VM is a no-op.
1041 vm.enable_tracing_jit();
1042 // Macro for builtins that user functions are allowed to shadow.
1043 // zsh dispatch order is alias → function → builtin; without the
1044 // try_user_fn_override probe a `cat() { ... }; cat` would silently
1045 // run the C builtin and ignore the user function.
1046 macro_rules! reg_overridable {
1047 ($vm:expr, $id:expr, $name:literal, $method:ident) => {
1048 $vm.register_builtin($id, |vm, argc| {
1049 let args = pop_args(vm, argc);
1050 // c:Src/exec.c getproc + Src/jobs.c deletefilelist —
1051 // close `>(cmd)` write ends owned by this command
1052 // once it finishes (shadows bypass dispatch_builtin
1053 // and ZshrsHost::exec, so they need their own guard:
1054 // `tee >(wc -c) </dev/null` left wc blocked).
1055 let _psub_fds = PsubFdGuard;
1056 if let Some(s) = try_user_fn_override($name, &args) {
1057 return Value::Status(s);
1058 }
1059 // c:Src/exec.c — redirect failure in the current
1060 // scope means the command must NOT run. coreutils
1061 // shadows (cat / head / tail / etc.) take a separate
1062 // dispatch path from dispatch_builtin, so they need
1063 // their own gate. Without this `cat <&3` after a
1064 // closed-fd diagnostic still ran the shadow and
1065 // overwrote $? from the forced 1.
1066 let redir_failed = with_executor(|exec| {
1067 let f = exec.redirect_failed;
1068 exec.redirect_failed = false;
1069 f
1070 });
1071 if redir_failed {
1072 return Value::Status(1);
1073 }
1074 // `[builtins].coreutils_shadows = off` in
1075 // ~/.zshrs/zshrs.toml (or `ZSHRS_NO_COREUTILS_SHADOWS=1`
1076 // env override) bypasses the in-process shadow and
1077 // fork-execs the real /bin/X. Safety valve for any
1078 // script that hits an edge-case divergence between
1079 // the zshrs shadow and system coreutils. Cached
1080 // after first call, so the hot path is one atomic
1081 // load per shadowed-builtin invocation.
1082 if !crate::daemon_presence::coreutils_shadows_enabled() {
1083 return Value::Status(exec_system_command($name, &args));
1084 }
1085 let status = with_executor(|exec| exec.$method(&args));
1086 Value::Status(status)
1087 });
1088 };
1089 }
1090
1091 // Pure-passthru builtin: pops args, routes to canonical
1092 // `dispatch_builtin(name, args)` (which goes via execbuiltin →
1093 // BUILTINS[name] → bin_X). No pre/post bridge work. Used by
1094 // ~25 handlers that were 4-line copy-paste boilerplate.
1095 macro_rules! reg_passthru {
1096 ($vm:expr, $id:expr, $name:literal) => {
1097 $vm.register_builtin($id, |vm, argc| {
1098 Value::Status(dispatch_builtin($name, pop_args(vm, argc)))
1099 });
1100 };
1101 }
1102
1103 // Core builtins
1104 vm.register_builtin(BUILTIN_CD, |vm, argc| {
1105 let args = pop_args(vm, argc);
1106 if let Some(s) = try_user_fn_override("cd", &args) {
1107 return Value::Status(s);
1108 }
1109 let status = dispatch_builtin("cd", args);
1110 // c:Src/builtin.c:1258 — `callhookfunc("chpwd", NULL, 1, NULL)`
1111 // after cd succeeds. The canonical port at
1112 // src/ported/utils.rs:1532 handles both the `chpwd` shfunc
1113 // dispatch AND the `chpwd_functions` array walk.
1114 if status == 0 {
1115 crate::ported::utils::callhookfunc("chpwd", None, 1, std::ptr::null_mut());
1116 }
1117 Value::Status(status)
1118 });
1119
1120 vm.register_builtin(BUILTIN_PWD, |vm, argc| {
1121 let args = pop_args(vm, argc);
1122 if let Some(s) = try_user_fn_override("pwd", &args) {
1123 return Value::Status(s);
1124 }
1125 // Route through the canonical execbuiltin path so the `rLP`
1126 // optstr at BUILTINS["pwd"] is parsed into `ops`.
1127 let status = dispatch_builtin("pwd", args);
1128 Value::Status(status)
1129 });
1130
1131 vm.register_builtin(BUILTIN_ECHO, |vm, argc| {
1132 let args = pop_args(vm, argc);
1133 if let Some(s) = try_user_fn_override("echo", &args) {
1134 return Value::Status(s);
1135 }
1136 // Update `$_` to the last arg before running. C zsh sets
1137 // zunderscore in execcmd_exec for every simple command,
1138 // including builtins.
1139 crate::ported::params::set_zunderscore(&args);
1140 let status = dispatch_builtin("echo", args);
1141 Value::Status(status)
1142 });
1143
1144 vm.register_builtin(BUILTIN_PRINT, |vm, argc| {
1145 let args = pop_args(vm, argc);
1146 if let Some(s) = try_user_fn_override("print", &args) {
1147 return Value::Status(s);
1148 }
1149 crate::ported::params::set_zunderscore(&args);
1150 let status = dispatch_builtin("print", args);
1151 Value::Status(status)
1152 });
1153
1154 reg_passthru!(vm, BUILTIN_PRINTF, "printf");
1155 reg_passthru!(vm, BUILTIN_EXPORT, "export");
1156 reg_passthru!(vm, BUILTIN_UNSET, "unset");
1157 // `source` (Src/builtin.c c:116) wired to bin_dot via BUILTINS.
1158 reg_passthru!(vm, BUILTIN_SOURCE, "source");
1159 reg_passthru!(vm, BUILTIN_DOT, ".");
1160 reg_passthru!(vm, BUILTIN_LOGOUT, "logout");
1161
1162 vm.register_builtin(BUILTIN_EXIT, |vm, argc| {
1163 let args = pop_args(vm, argc);
1164 let status = dispatch_builtin("exit", args);
1165 Value::Status(status)
1166 });
1167
1168 vm.register_builtin(BUILTIN_RETURN, |vm, argc| {
1169 let args = pop_args(vm, argc);
1170 // zsh: bare `return` (no arg) returns with the status of
1171 // the most recently executed command — `false; return`
1172 // returns 1, not 0. Direct port of zsh's bin_break/RETURN.
1173 // The executor's `last_status` is stale here (synced at
1174 // statement boundaries, not after each VM op), so read
1175 // the live `vm.last_status` instead.
1176 let live_status = vm.last_status;
1177 let status = {
1178 // Sync canonical LASTVAL to the VM's view BEFORE
1179 // bin_break("return") reads it for the no-arg fallback.
1180 with_executor(|exec| exec.set_last_status(live_status));
1181 dispatch_builtin("return", args)
1182 };
1183 Value::Status(status)
1184 });
1185
1186 vm.register_builtin(BUILTIN_TRUE, |vm, argc| {
1187 let args = pop_args(vm, argc);
1188 if let Some(s) = try_user_fn_override("true", &args) {
1189 return Value::Status(s);
1190 }
1191 // c:Src/exec.c:1257 — zsh sets `zunderscore` AT THE END of
1192 // each command (the `if (!noerrs)` block runs `zsfree(prev_argv0); …;
1193 // zunderscore = …`). For no-arg `true`, $_ becomes the
1194 // command name itself. Set DIRECTLY (not via pending_underscore)
1195 // so the NEXT command's argv-expansion of `$_` reads "true",
1196 // not the stale prior value — pending_underscore is consumed
1197 // by pop_args which runs AFTER argv expansion, too late.
1198 // c:Src/exec.c:1257 — `zunderscore = …` at end-of-command.
1199 // With args, $_ = args.last(). Without args, $_ = command name.
1200 // Write DIRECTLY to the canonical zunderscore static (the
1201 // underscoregetfn at params.rs:7003 reads from there); the
1202 // paramtab "_" slot is shadowed by lookup_special_var so
1203 // set_scalar on it has no effect on `$_` reads.
1204 if args.is_empty() {
1205 crate::ported::params::set_zunderscore(&["true".to_string()]);
1206 } else {
1207 crate::ported::params::set_zunderscore(&args);
1208 }
1209 // Route through canonical execbuiltin so PS4 xtrace fires
1210 // via the c:442 printprompt4 path.
1211 Value::Status(dispatch_builtin("true", args))
1212 });
1213 vm.register_builtin(BUILTIN_FALSE, |vm, argc| {
1214 let args = pop_args(vm, argc);
1215 if let Some(s) = try_user_fn_override("false", &args) {
1216 return Value::Status(s);
1217 }
1218 // Direct set; see BUILTIN_TRUE above for rationale.
1219 if args.is_empty() {
1220 crate::ported::params::set_zunderscore(&["false".to_string()]);
1221 } else {
1222 crate::ported::params::set_zunderscore(&args);
1223 }
1224 // Route through canonical execbuiltin — see BUILTIN_TRUE
1225 // above for the same rationale (xtrace + fast-path removal).
1226 let status = dispatch_builtin("false", args);
1227 Value::Status(status)
1228 });
1229 vm.register_builtin(BUILTIN_COLON, |vm, argc| {
1230 let args = pop_args(vm, argc);
1231 // Direct set; see BUILTIN_TRUE above for rationale.
1232 if args.is_empty() {
1233 crate::ported::params::set_zunderscore(&[":".to_string()]);
1234 } else {
1235 crate::ported::params::set_zunderscore(&args);
1236 }
1237 let status = dispatch_builtin(":", args);
1238 Value::Status(status)
1239 });
1240
1241 vm.register_builtin(BUILTIN_TEST, |vm, argc| {
1242 let args = pop_args(vm, argc);
1243 // Distinguish `[ … ]` from `test …` by sniffing the trailing
1244 // `]` — `[` requires it (c:Src/builtin.c:7241), `test` rejects
1245 // it. The compile path emits BUILTIN_TEST for both, so the
1246 // dispatch name carries the `[` vs `test` semantic for
1247 // execbuiltin's funcid (BIN_BRACKET=21 vs BIN_TEST=20). Without
1248 // this, bin_test's `if func == BIN_BRACKET` arm (which pops
1249 // the trailing `]`) never fired for `[` calls, so the `]`
1250 // leaked into evalcond as a positional and silently changed
1251 // the result. Bug surfaced via test_test_dashdash_unknown_condition.
1252 let name = if args.last().map(|s| s.as_str()) == Some("]") {
1253 "["
1254 } else {
1255 "test"
1256 };
1257 let status = dispatch_builtin(name, args);
1258 Value::Status(status)
1259 });
1260
1261 // Variable declaration. `local` (Src/builtin.c bin_local) handles
1262 // the scope chain (`pm->old = oldpm` at Src/params.c:1137 inside
1263 // createparam, `pm->level = locallevel` at Src/builtin.c:2576).
1264 // `typeset` / `declare` are aliases — fusevm maps both to
1265 // BUILTIN_TYPESET; compile_zsh special-cases `declare` to keep
1266 // the `declare:` error prefix.
1267 reg_passthru!(vm, BUILTIN_LOCAL, "local");
1268 reg_passthru!(vm, BUILTIN_TYPESET, "typeset");
1269
1270 reg_passthru!(vm, BUILTIN_DECLARE, "declare");
1271 reg_passthru!(vm, BUILTIN_READONLY, "readonly");
1272 reg_passthru!(vm, BUILTIN_INTEGER, "integer");
1273 reg_passthru!(vm, BUILTIN_FLOAT, "float");
1274 reg_passthru!(vm, BUILTIN_READ, "read");
1275 // c:Bug #504 — fusevm reserves BUILTIN_MAPFILE for the bash
1276 // mapfile/readarray builtins. Neither exists in zsh; in --zsh
1277 // parity mode the dispatch must emit "command not found" + rc=127
1278 // matching zsh's external-command-lookup miss. The previous wiring
1279 // left BUILTIN_MAPFILE unregistered, so fusevm's VM treated the op
1280 // as a no-op rc=0 — `mapfile` (and `readarray`) silently succeeded
1281 // in --zsh mode. The host gate in `dispatch_builtin_raw` never
1282 // fired because the compile path emitted `Op::CallBuiltin(31, ..)`
1283 // directly. Register the slot so the gate runs (or a future
1284 // non-zsh mode can wire in a real impl).
1285 vm.register_builtin(fusevm::shell_builtins::BUILTIN_MAPFILE, |vm, argc| {
1286 let args = pop_args(vm, argc);
1287 // The fusevm name→id map collapses both `mapfile` and
1288 // `readarray` to the same opcode; pick the right diagnostic
1289 // by sniffing the user's actual invocation. The xtrace ARGS
1290 // push earlier records the cmd-prefix as the bottom of the
1291 // popped argv, but `args` here excludes the prefix — so we
1292 // can't recover the user-typed name from the stack. Default
1293 // to `mapfile` (the more-common spelling); both produce
1294 // identical diagnostics in any case.
1295 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed) {
1296 eprintln!("zsh:1: command not found: mapfile");
1297 let _ = args;
1298 return Value::Status(127);
1299 }
1300 // Non-zsh modes (bash drop-in) get a passthru to the canonical
1301 // dispatcher in case a future port adds a real mapfile builtin
1302 // — until then the rc=1 unknown-command default applies.
1303 Value::Status(dispatch_builtin("mapfile", args))
1304 });
1305 reg_passthru!(vm, BUILTIN_BREAK, "break");
1306 reg_passthru!(vm, BUILTIN_CONTINUE, "continue");
1307 reg_passthru!(vm, BUILTIN_SHIFT, "shift");
1308
1309 vm.register_builtin(BUILTIN_EVAL, |vm, argc| {
1310 // Direct port of `bin_eval(UNUSED(char *nam), char **argv, UNUSED(Options ops), UNUSED(int func))` body from Src/builtin.c:6151:
1311 // `if (!*argv) return 0;`
1312 // `prog = parse_string(zjoin(argv, ' ', 1), 1);`
1313 // `execode(prog, 1, 0, "eval");`
1314 // The execode invocation lives here (not in the canonical
1315 // free-fn) because it must run through the bytecode VM's
1316 // current executor — the same VM that's mid-dispatch.
1317 let mut args = pop_args(vm, argc);
1318 // c:Src/builtin.c:407-411 — generic `--` end-of-options
1319 // strip applied by `execbuiltin` for builtins that have
1320 // NULL optstr AND no BINF_HANDLES_OPTS. `eval` qualifies
1321 // (Src/builtin.c:65 `BUILTIN("eval", BINF_PSPECIAL, ...,
1322 // NULL, NULL)`). The BUILTIN_EVAL fast-path bypasses
1323 // execbuiltin, so we mirror the strip inline. Bug #319.
1324 if args.first().is_some_and(|s| s == "--") {
1325 args.remove(0);
1326 }
1327 if args.is_empty() {
1328 return Value::Status(0); // c:6160
1329 }
1330 let src = args.join(" "); // c:6166
1331 // c:Src/builtin.c:6164-6165 — `if (!ineval) scriptname =
1332 // "(eval)";`. Diagnostics emitted while the eval body runs
1333 // (command-not-found, parse errors, etc.) use scriptname as
1334 // the source-context prefix. Without setting it here the
1335 // BUILTIN_EVAL fast-path leaked the outer "zsh" prefix
1336 // through, breaking the `(eval):N:` convention zsh uses
1337 // for in-eval errors. Bug #420.
1338 let oscriptname = crate::ported::utils::scriptname_get();
1339 crate::ported::utils::set_scriptname(Some("(eval)".to_string()));
1340 let mut status = with_executor(|exec| {
1341 // c:6175 execode
1342 exec.execute_script(&src).unwrap_or(1)
1343 });
1344 // c:Src/builtin.c:6211-6212 — `if (errflag && !lastval)
1345 // lastval = errflag;`
1346 // c:Src/builtin.c:6221 — `errflag &= ~ERRFLAG_ERROR;`
1347 // eval is a CONTAINMENT boundary: an error inside the eval
1348 // body (readonly reassign, bad assoc set, ${unset?msg}, …)
1349 // breaks the eval body's lists via errflag, then eval clears
1350 // the flag and returns lastval, and the CALLER's next list
1351 // runs. zsh 5.9: `eval 'assoc=(odd)'; echo "after $?"`
1352 // prints `after 1` in -c, script, and stdin contexts.
1353 {
1354 use std::sync::atomic::Ordering;
1355 let ef = crate::ported::utils::errflag.load(Ordering::Relaxed)
1356 & crate::ported::zsh_h::ERRFLAG_ERROR;
1357 if ef != 0 && status == 0 {
1358 status = ef; // c:6212 lastval = errflag
1359 }
1360 crate::ported::utils::errflag
1361 .fetch_and(!crate::ported::zsh_h::ERRFLAG_ERROR, Ordering::Relaxed);
1362 }
1363 crate::ported::utils::set_scriptname(oscriptname);
1364 Value::Status(status)
1365 });
1366
1367 // `builtin foo args…`: precmd-modifier that forces builtin dispatch,
1368 // bypassing alias AND function lookup. Without this, `builtin cd /`
1369 // inside a user `cd () { … }` wrapper recurses (real-world ZPWR pattern).
1370 // Handler pops argc args from the stack, treats args[0] as the builtin
1371 // name, and dispatches the rest via `dispatch_builtin` → `execbuiltin`
1372 // → `bin_*` directly. No function/alias lookup happens.
1373 vm.register_builtin(BUILTIN_BUILTIN, |vm, argc| {
1374 let args = pop_args(vm, argc);
1375 let Some((name, rest)) = args.split_first() else {
1376 // `builtin` with no args → list builtins (zsh emits nothing,
1377 // exit 0). Match that behavior; the BIN_BUILTIN bin_* in C
1378 // does the same default-list-nothing.
1379 return Value::Status(0);
1380 };
1381 // c:Src/exec.c:3435-3436 — `builtin NAME` with NAME not in
1382 // builtintab emits `zwarn("no such builtin: %s", cmdarg)`
1383 // and returns 1. zshrs's dispatch_builtin_raw bare-returned 1
1384 // silently. Probe the table here so the diagnostic fires
1385 // before dispatch.
1386 let tab = crate::ported::builtin::createbuiltintable();
1387 if !tab.contains_key(name.as_str()) {
1388 eprintln!("zshrs:1: no such builtin: {}", name);
1389 return Value::Status(1);
1390 }
1391 // `builtin foo` MUST bypass function shadow — that's the whole
1392 // point of the prefix. Use the _raw helper, not the shadow-aware
1393 // one. Without this, `cd () { builtin cd "$@"; }` recurses.
1394 Value::Status(dispatch_builtin_raw(name, rest.to_vec()))
1395 });
1396
1397 // `command foo args…` — BINF_COMMAND prefix (Src/builtin.c:44). Zsh
1398 // semantic: bypass alias+function lookup, search builtin then $PATH.
1399 // Without this, `cd () { command cd "$@" }` would re-invoke the user
1400 // wrapper (same root cause as the `builtin` bug). Flags `-p`/`-v`/`-V`
1401 // route to bin_whence with BIN_COMMAND funcid; bare `command foo`
1402 // dispatches builtin if present, else external (no fork — direct
1403 // spawn via execute_external since zshrs is non-forking).
1404 // BUILTIN_COMMAND — `command [-p] [-v|-V] cmd args…` BIN_PREFIX
1405 // (Src/builtin.c:45). PURE PASSTHRU: prepend "command" and hand
1406 // to `exec::execcmd_compile_head` (the fusevm-bytecode-time head
1407 // resolver mirroring `Src/exec.c::execcmd_exec` precommand-modifier
1408 // walk at c:3104-3187). That helper already does the -p / -v / -V
1409 // option parsing, surfaces `has_command_vv` for the whence
1410 // redirect, and reports the dispatch shape (is_builtin vs external).
1411 vm.register_builtin(BUILTIN_COMMAND, |vm, argc| {
1412 let args = pop_args(vm, argc);
1413 let mut full = Vec::with_capacity(args.len() + 1);
1414 full.push("command".to_string());
1415 full.extend(args.clone());
1416 let dispatch =
1417 crate::ported::exec::execcmd_compile_head(&full, crate::ported::zsh_h::WC_SIMPLE);
1418 let post = &full[dispatch.precmd_skip..];
1419 // c:Src/builtin.c:4500 — `command -p` resets PATH for the
1420 // exec to the POSIX-defined default (`getconf PATH`), so
1421 // standard utilities resolve even when the caller has
1422 // emptied $PATH. zsh restores the original PATH after the
1423 // command returns. Mirror via a scoped env::set_var.
1424 //
1425 // command's OWN options end at the first non-flag arg —
1426 // everything after the command name belongs to IT. The
1427 // previous `.any()` scan over ALL args stole `-p` from
1428 // `command mkdir -p DIR` (zconvey.plugin.zsh:44), stripping
1429 // the flag before /bin/mkdir ran → "File exists" errors on
1430 // every re-source.
1431 let mut lead = 0usize;
1432 let mut dash_p = false;
1433 let mut kept_flags: Vec<String> = Vec::new();
1434 for a in post.iter() {
1435 let s = a.as_str();
1436 if s == "--" {
1437 lead += 1;
1438 break;
1439 }
1440 if s.starts_with('-')
1441 && s.len() >= 2
1442 && s[1..].chars().all(|c| c == 'p' || c == 'v' || c == 'V')
1443 {
1444 if s.contains('p') {
1445 dash_p = true;
1446 }
1447 // -v / -V drive the whence-style lookup downstream —
1448 // keep them in post (only the PATH-reset `p` is
1449 // consumed here).
1450 let rest: String = s[1..].chars().filter(|c| *c != 'p').collect();
1451 if !rest.is_empty() {
1452 kept_flags.push(format!("-{}", rest));
1453 }
1454 lead += 1;
1455 continue;
1456 }
1457 break;
1458 }
1459 let mut post: Vec<String> = {
1460 let mut v = kept_flags;
1461 v.extend(post[lead..].iter().cloned());
1462 v
1463 };
1464 // c:Src/exec.c:3176-3177 — `BINF_COMMAND` arm strips a single
1465 // leading `--` end-of-options marker.
1466 // `execcmd_compile_head` (src/ported/exec.rs:1042) performs
1467 // this removal on its LOCAL `preargs` Vec but doesn't surface
1468 // the modified args; the caller still sees `--` in `full` and
1469 // tried to dispatch it as the command name. Bug #251. Mirror
1470 // the C strip here so `command -- echo hi` and
1471 // `command -p -- echo hi` route correctly.
1472 if let Some(first) = post.first() {
1473 if first == "--" {
1474 post.remove(0);
1475 }
1476 }
1477 let post = post.as_slice();
1478 let _path_guard = if dash_p {
1479 let saved = env::var("PATH").ok();
1480 let default_path = std::process::Command::new("getconf")
1481 .arg("PATH")
1482 .output()
1483 .ok()
1484 .and_then(|o| String::from_utf8(o.stdout).ok())
1485 .map(|s| s.trim().to_string())
1486 .filter(|s| !s.is_empty())
1487 .unwrap_or_else(|| "/usr/bin:/bin:/usr/sbin:/sbin".to_string());
1488 env::set_var("PATH", &default_path);
1489 crate::ported::params::setsparam("PATH", &default_path);
1490 Some(saved)
1491 } else {
1492 None
1493 };
1494 struct PathGuard {
1495 saved: Option<String>,
1496 active: bool,
1497 }
1498 impl Drop for PathGuard {
1499 fn drop(&mut self) {
1500 if !self.active {
1501 return;
1502 }
1503 match self.saved.take() {
1504 Some(p) => {
1505 env::set_var("PATH", &p);
1506 crate::ported::params::setsparam("PATH", &p);
1507 }
1508 None => {
1509 env::remove_var("PATH");
1510 crate::ported::params::setsparam("PATH", "");
1511 }
1512 }
1513 }
1514 }
1515 let _restore = PathGuard {
1516 saved: _path_guard.unwrap_or(None),
1517 active: dash_p,
1518 };
1519 if dispatch.has_command_vv {
1520 // `-v` / `-V` → bin_whence with BIN_COMMAND funcid.
1521 let mut ops = options {
1522 ind: [0u8; MAX_OPS],
1523 args: Vec::new(),
1524 argscount: 0,
1525 argsalloc: 0,
1526 };
1527 let mut name_pos = 0usize;
1528 let mut flag_byte = b'v';
1529 for (i, a) in post.iter().enumerate() {
1530 if a.starts_with('-') && a.len() >= 2 {
1531 let body = &a.as_bytes()[1..];
1532 if body.contains(&b'V') {
1533 flag_byte = b'V';
1534 }
1535 name_pos = i + 1;
1536 } else {
1537 name_pos = i;
1538 break;
1539 }
1540 }
1541 ops.ind[flag_byte as usize] = 1;
1542 let whence_args: Vec<String> = post[name_pos..].to_vec();
1543 return Value::Status(crate::ported::builtin::bin_whence(
1544 "command",
1545 &whence_args,
1546 &ops,
1547 crate::ported::hashtable_h::BIN_COMMAND,
1548 ));
1549 }
1550 if dispatch.is_empty_command {
1551 return Value::Status(0);
1552 }
1553 let Some((name, rest)) = post.split_first() else {
1554 return Value::Status(0);
1555 };
1556 // c:Src/exec.c:3275-3278 — `execcmd_compile_head` cleared
1557 // hn for the BINF_COMMAND + !POSIXBUILTINS case, surfacing
1558 // is_builtin=false. Run as external. Under POSIXBUILTINS
1559 // dispatch.is_builtin would be true; honour it.
1560 let n = name.clone();
1561 let r = rest.to_vec();
1562 if dispatch.is_builtin
1563 && crate::ported::builtin::BUILTINS
1564 .iter()
1565 .any(|b| b.node.nam == n.as_str())
1566 {
1567 return Value::Status(dispatch_builtin_raw(&n, r));
1568 }
1569 Value::Status(with_executor(|exec| exec.execute_external(&n, &r, &[])).unwrap_or(127))
1570 });
1571
1572 // `exec cmd args…` — BINF_EXEC prefix (Src/builtin.c:45). Zsh
1573 // semantic: replace the current shell process with `cmd`. On Unix
1574 // this is `execvp(2)`; the call only returns on error. zshrs is
1575 // non-forking, so the shell process IS the calling process —
1576 // execvp here directly replaces it. Options `-a name` (override
1577 // argv[0]), `-c` (clean env), `-l` (login shell — prepend `-`)
1578 // ported minimally; advanced redirect-only `exec >file` is handled
1579 // upstream by compile_zsh and never reaches this handler.
1580 vm.register_builtin(BUILTIN_EXEC, |vm, argc| {
1581 let mut args = pop_args(vm, argc);
1582 let mut argv0_override: Option<String> = None;
1583 let mut clean_env = false;
1584 let mut login = false;
1585 let mut i = 0;
1586 // c:Src/builtin.c:1075-1080 — track if any flag was consumed.
1587 // `exec -c`, `exec -l`, `exec -a NAME` without a following
1588 // command emit "exec requires a command to execute" rc=1.
1589 // Bare `exec` (no args at all) is the silent-redirect-apply
1590 // form per POSIX.
1591 let mut saw_flag = false;
1592 while i < args.len() {
1593 let a = &args[i];
1594 if a == "--" {
1595 args.remove(i);
1596 break;
1597 }
1598 // c:Src/builtin.c:42 `BIN_PREFIX("-", BINF_DASH)`. A bare
1599 // `-` is its own BINF_PREFIX builtin (BINF_DASH flag —
1600 // "login shell, prepend `-` to argv[0]"). In the canonical
1601 // precmd-walk at Src/exec.c:3056-3091 a bare `-` after
1602 // `exec` is recognized AS a builtin and stripped from
1603 // preargs (precmd_skip++), accumulating BINF_DASH into
1604 // cflags. The fast-path here bypasses execcmd_compile_head,
1605 // so we mirror the strip locally: bare `-` → set login,
1606 // remove, continue. Without this `exec -` (with no command
1607 // following) tried to exec `-` as a literal command and
1608 // exited the shell. Bug #252.
1609 if a == "-" {
1610 saw_flag = true;
1611 login = true;
1612 args.remove(i);
1613 continue;
1614 }
1615 if !a.starts_with('-') || a.len() < 2 {
1616 break;
1617 }
1618 match a.as_str() {
1619 "-a" => {
1620 saw_flag = true;
1621 args.remove(i);
1622 if i < args.len() {
1623 argv0_override = Some(args.remove(i));
1624 }
1625 }
1626 "-c" => {
1627 saw_flag = true;
1628 clean_env = true;
1629 args.remove(i);
1630 }
1631 "-l" => {
1632 saw_flag = true;
1633 login = true;
1634 args.remove(i);
1635 }
1636 _ => {
1637 // c:Src/exec.c:3196-3208 — when an unrecognized
1638 // `-X`-style arg has NO following arg, the lexer's
1639 // IS_DASH walk hits the "no next node" branch at
1640 // c:3199 before the unknown-flag-letter switch at
1641 // c:3249, so the canonical message is "exec
1642 // requires a command to execute" rc=1 (verified vs
1643 // `/opt/homebrew/bin/zsh -fc 'exec --bad'`).
1644 // Consume the lone flag so the post-loop check
1645 // fires. When a following arg exists, leave the
1646 // unknown-flag arg in place — that arg becomes
1647 // the command name and execution proceeds.
1648 if args.len() == 1 {
1649 saw_flag = true;
1650 args.remove(i);
1651 continue;
1652 }
1653 break;
1654 }
1655 }
1656 }
1657 let Some(cmd) = args.first().cloned() else {
1658 if saw_flag {
1659 // c:Src/builtin.c:1078-1080 — flags consumed but no
1660 // command follows → "exec requires a command to
1661 // execute" rc=1.
1662 eprintln!("zshrs:1: exec requires a command to execute");
1663 return Value::Status(1);
1664 }
1665 // `exec` with no command + no redirects = no-op success.
1666 return Value::Status(0);
1667 };
1668 let rest: Vec<String> = args[1..].to_vec();
1669 let display_argv0 = match argv0_override {
1670 Some(a) => a,
1671 None => {
1672 if login {
1673 format!("-{}", cmd)
1674 } else {
1675 cmd.clone()
1676 }
1677 }
1678 };
1679 // c:Src/exec.c::execcmd — `exec funcname` runs the function
1680 // in-process as the shell's last act, then exits with the
1681 // function's status. zsh's dispatcher falls through from the
1682 // BINF_EXEC prefix into the normal Builtin/External/Function
1683 // resolution and only execvp's if the target ISN'T a
1684 // function. Bug #101 in docs/BUGS.md: zshrs's exec went
1685 // straight to execvp and errored `not found` for shell
1686 // functions.
1687 //
1688 // For both subshell and top-level contexts: dispatch through
1689 // the function/builtin lookup first; only fall through to
1690 // execvp/spawn if the name isn't shell-resolvable.
1691 let has_user_fn = with_executor(|exec| exec.functions_compiled.contains_key(&cmd));
1692 if has_user_fn {
1693 let status =
1694 with_executor(|exec| exec.dispatch_function_call(&cmd, &rest).unwrap_or(127));
1695 // Top-level `exec funcname` — exit the shell with the
1696 // function's status (mirrors C's "exec replaces shell as
1697 // last act"). Subshell `(exec funcname)` — return through
1698 // the EXIT_PENDING path so the subshell body aborts and
1699 // the parent resumes via subshell_end.
1700 let in_subshell_now = with_executor(|exec| !exec.subshell_snapshots.is_empty());
1701 if in_subshell_now {
1702 crate::ported::builtin::EXIT_VAL
1703 .store(status, std::sync::atomic::Ordering::Relaxed);
1704 crate::ported::builtin::EXIT_PENDING.store(1, std::sync::atomic::Ordering::Relaxed);
1705 return Value::Status(status);
1706 }
1707 std::process::exit(status);
1708 }
1709 // c:Src/exec.c — builtin path: `exec builtin` runs the
1710 // builtin in-process and exits.
1711 let bn_in_tab = crate::ported::builtin::createbuiltintable().contains_key(&cmd);
1712 if bn_in_tab {
1713 let status = dispatch_builtin_raw(&cmd, rest.clone());
1714 let in_subshell_now = with_executor(|exec| !exec.subshell_snapshots.is_empty());
1715 if in_subshell_now {
1716 crate::ported::builtin::EXIT_VAL
1717 .store(status, std::sync::atomic::Ordering::Relaxed);
1718 crate::ported::builtin::EXIT_PENDING.store(1, std::sync::atomic::Ordering::Relaxed);
1719 return Value::Status(status);
1720 }
1721 std::process::exit(status);
1722 }
1723 // c:Src/exec.c — `exec` inside a subshell (`(exec cmd)`)
1724 // replaces ONLY the subshell child process; the parent shell
1725 // continues. C zsh always forks for `(...)`, so the actual
1726 // execvp lands in the forked child. zshrs runs subshells via
1727 // a snapshot/restore pattern in the SAME process — calling
1728 // execvp here would replace the parent too. Bug #94 in
1729 // docs/BUGS.md.
1730 //
1731 // Detect subshell context via the non-empty
1732 // `subshell_snapshots` stack. When in a subshell: spawn the
1733 // command as a child, wait for it, then signal the subshell
1734 // body to abort (return Status(N) and the caller's
1735 // subshell_end will pop the snapshot and resume the parent).
1736 let in_subshell = with_executor(|exec| !exec.subshell_snapshots.is_empty());
1737 if in_subshell {
1738 let mut command = std::process::Command::new(&cmd);
1739 command.arg0(&display_argv0);
1740 command.args(&rest);
1741 if clean_env {
1742 command.env_clear();
1743 }
1744 let status = match command.spawn() {
1745 Ok(mut child) => match child.wait() {
1746 Ok(s) => s.code().unwrap_or(127),
1747 Err(_) => 127,
1748 },
1749 Err(e) => {
1750 // c:Src/exec.c:797 — `zerr("%e: %s", lerrno, arg0)`
1751 // when arg0 contains `/`.
1752 // c:872-876 — when arg0 has no `/` (PATH search
1753 // path), C tracks the "good" errno
1754 // via `isgooderr`; if all PATH entries
1755 // were ENOENT-not-good, eno stays 0
1756 // and C emits `command not found: %s`
1757 // instead of strerror.
1758 // %e expands to strerror(errno) with the first
1759 // letter lowercased (unless errno == EIO; see
1760 // Src/utils.c:362-368). `zerr` prepends the
1761 // scriptname:lineno: prefix — matching zsh's
1762 // canonical `zsh:N: <errmsg>: <cmd>` pattern.
1763 // Previously emitted `zshrs: exec: {}: not found`
1764 // (wrong prefix, hardcoded message, missing
1765 // lineno). Bug #140 in docs/BUGS.md.
1766 let errno = e.raw_os_error().unwrap_or(libc::ENOENT);
1767 let has_slash = cmd.contains('/');
1768 if !has_slash && errno == libc::ENOENT {
1769 // c:876 — PATH search exhausted with no good
1770 // errno → `command not found: arg0`.
1771 crate::ported::utils::zerr(&format!("command not found: {}", cmd));
1772 } else {
1773 let mut errmsg = crate::ported::compat::strerror(errno);
1774 if errno != libc::EIO {
1775 if let Some(c) = errmsg.chars().next() {
1776 errmsg = format!(
1777 "{}{}",
1778 c.to_ascii_lowercase(),
1779 &errmsg[c.len_utf8()..]
1780 );
1781 }
1782 }
1783 crate::ported::utils::zerr(&format!("{}: {}", errmsg, cmd));
1784 }
1785 // c:881 — `_exit((eno == EACCES || eno == ENOEXEC) ? 126 : 127);`
1786 if errno == libc::EACCES || errno == libc::ENOEXEC {
1787 126
1788 } else {
1789 127
1790 }
1791 }
1792 };
1793 // Mark the subshell as exec-replaced so subsequent body
1794 // commands skip — mirrors the post-execvp "child process
1795 // is gone" reality in C. EXIT_PENDING + EXIT_VAL drive
1796 // the next ERREXIT_CHECK to unwind to the subshell-end
1797 // patch.
1798 crate::ported::builtin::EXIT_VAL.store(status, std::sync::atomic::Ordering::Relaxed);
1799 crate::ported::builtin::EXIT_PENDING.store(1, std::sync::atomic::Ordering::Relaxed);
1800 return Value::Status(status);
1801 }
1802 let mut command = std::process::Command::new(&cmd);
1803 command.arg0(&display_argv0);
1804 command.args(&rest);
1805 if clean_env {
1806 command.env_clear();
1807 }
1808 use std::os::unix::process::CommandExt;
1809 // `exec` returns the OS error iff exec(2) failed; on success
1810 // it never returns. Match zsh: print the error to stderr with
1811 // the `exec` prefix and exit 127 (cmd not found) or 126 (not
1812 // executable).
1813 let err = command.exec();
1814 // c:Src/exec.c:797 / c:872-876 — same format as in-subshell
1815 // branch. arg0-has-/ → `<strerror>: <cmd>`; arg0-no-/ +
1816 // ENOENT → `command not found: <cmd>`. Lowercase strerror
1817 // first letter unless EIO. Bug #140 in docs/BUGS.md.
1818 let errno = err.raw_os_error().unwrap_or(libc::ENOENT);
1819 let has_slash = cmd.contains('/');
1820 if !has_slash && errno == libc::ENOENT {
1821 crate::ported::utils::zerr(&format!("command not found: {}", cmd));
1822 } else {
1823 let mut errmsg = crate::ported::compat::strerror(errno);
1824 if errno != libc::EIO {
1825 if let Some(c) = errmsg.chars().next() {
1826 errmsg = format!("{}{}", c.to_ascii_lowercase(), &errmsg[c.len_utf8()..]);
1827 }
1828 }
1829 crate::ported::utils::zerr(&format!("{}: {}", errmsg, cmd));
1830 }
1831 // c:881 — `_exit((eno == EACCES || eno == ENOEXEC) ? 126 : 127);`
1832 let code = if errno == libc::EACCES || errno == libc::ENOEXEC {
1833 126
1834 } else {
1835 127
1836 };
1837 std::process::exit(code);
1838 });
1839
1840 reg_passthru!(vm, BUILTIN_LET, "let");
1841
1842 // Job control
1843 reg_passthru!(vm, BUILTIN_JOBS, "jobs");
1844 reg_passthru!(vm, BUILTIN_FG, "fg");
1845 reg_passthru!(vm, BUILTIN_BG, "bg");
1846 reg_passthru!(vm, BUILTIN_KILL, "kill");
1847 reg_passthru!(vm, BUILTIN_DISOWN, "disown");
1848 reg_passthru!(vm, BUILTIN_WAIT, "wait");
1849 reg_passthru!(vm, BUILTIN_SUSPEND, "suspend");
1850
1851 // History — `fc` / `history` / `r` all route to `bin_fc` (zsh
1852 // registers them as aliases of the same builtin per Src/builtin.c).
1853 reg_passthru!(vm, BUILTIN_FC, "fc");
1854 reg_passthru!(vm, BUILTIN_HISTORY, "history");
1855 reg_passthru!(vm, BUILTIN_R, "r");
1856
1857 // Aliases — alias is `BINF_MAGICEQUALS` per Src/builtin.c:50.
1858 // c:Src/exec.c:3298-3304 — when a builtin has BINF_MAGICEQUALS,
1859 // execcmd_exec sets esprefork = PREFORK_TYPESET and calls
1860 // `prefork(args, esprefork, NULL)` on the argv. prefork (subst.c:
1861 // 100) drives `filesub` on each word (c:133), which (c:677-686)
1862 // looks for the assignment Equals and runs `filesubstr` on the
1863 // VALUE side. That's how `alias bad===` triggers equalsubstr's
1864 // "= not found" via the inner Equals after the first `=`.
1865 //
1866 // The fusevm dispatch path doesn't go through execcmd_exec, so
1867 // BUILTIN_ALIAS previously passed args straight to bin_alias with
1868 // no expansion — `alias x=~/foo` stored literal `~/foo` (no tilde
1869 // expand), `alias bad===` stored a broken entry without firing
1870 // the "= not found" diagnostic. The prefork(PREFORK_TYPESET) runs
1871 // per arg word via BUILTIN_MAGIC_EQUALS_PREFORK ops that
1872 // compile_simple emits BEFORE the redirect scope opens (matching
1873 // c:3304 prefork-before-addfd order), so the dispatch here is a
1874 // plain passthrough — re-running prefork would double-fire the
1875 // "= not found" diagnostic.
1876 reg_passthru!(vm, BUILTIN_ALIAS, "alias");
1877 // c:Src/exec.c:3298-3304 — per-word magic-equals prefork; see the
1878 // const doc at BUILTIN_MAGIC_EQUALS_PREFORK. prefork's filesub
1879 // trigger (subst.c:678 `strchr(*namptr+1, Equals)`) looks for the
1880 // EQUALS TOKEN, not literal `=`. The fusevm path delivers args
1881 // already-untokenized, so re-tokenize each element via
1882 // `shtokenize` (the same call C's lexer makes implicitly when
1883 // assembling the word) so prefork sees Equals tokens at `=`
1884 // boundaries and Tilde tokens at `~` starts. After prefork
1885 // expands, untokenize for storage.
1886 vm.register_builtin(BUILTIN_MAGIC_EQUALS_PREFORK, |vm, _argc| {
1887 let raw = vm.pop();
1888 let inputs: Vec<String> = match raw {
1889 Value::Array(items) => items.into_iter().map(|v| v.to_str()).collect(),
1890 other => vec![other.to_str()],
1891 };
1892 let mut as_linklist: crate::ported::linklist::LinkList<String> = Default::default();
1893 for s in &inputs {
1894 let mut tokd = s.clone();
1895 crate::ported::glob::shtokenize(&mut tokd);
1896 as_linklist.push_back(tokd);
1897 }
1898 let mut rf = 0i32;
1899 crate::ported::subst::prefork(
1900 &mut as_linklist,
1901 crate::ported::zsh_h::PREFORK_TYPESET,
1902 &mut rf,
1903 );
1904 let mut expanded: Vec<String> = Vec::with_capacity(inputs.len());
1905 while let Some(s) = as_linklist.pop_front() {
1906 expanded.push(crate::ported::lex::untokenize(&s).to_string());
1907 }
1908 if expanded.len() == 1 {
1909 Value::str(expanded.into_iter().next().unwrap())
1910 } else {
1911 Value::Array(expanded.into_iter().map(Value::str).collect())
1912 }
1913 });
1914
1915 // Options. `setopt` (BIN_SETOPT=0) / `unsetopt` (BIN_UNSETOPT=1)
1916 // share bin_setopt (options.c:580) — funcid bit discriminates
1917 // the polarity via BUILTINS table entries.
1918 reg_passthru!(vm, BUILTIN_SET, "set");
1919 reg_passthru!(vm, BUILTIN_SETOPT, "setopt");
1920 reg_passthru!(vm, BUILTIN_UNSETOPT, "unsetopt");
1921
1922 vm.register_builtin(BUILTIN_SHOPT, |vm, argc| {
1923 let args = pop_args(vm, argc);
1924 let status = crate::extensions::ext_builtins::shopt(&args);
1925 Value::Status(status)
1926 });
1927
1928 reg_passthru!(vm, BUILTIN_EMULATE, "emulate");
1929 reg_passthru!(vm, BUILTIN_GETOPTS, "getopts");
1930 reg_passthru!(vm, BUILTIN_AUTOLOAD, "autoload");
1931 reg_passthru!(vm, BUILTIN_FUNCTIONS, "functions");
1932 reg_passthru!(vm, BUILTIN_TRAP, "trap");
1933 reg_passthru!(vm, BUILTIN_DIRS, "dirs");
1934 // pushd / popd dispatch through canonical bin_cd via execbuiltin
1935 // — the BUILTINS table at src/ported/builtin.rs:9298 wires
1936 // `pushd` to bin_cd with funcid=BIN_PUSHD, and `popd` similarly
1937 // with BIN_POPD. Without these reg_passthru lines the fusevm
1938 // BUILTIN_PUSHD/POPD opcodes had no handler installed, so the
1939 // emitted CallBuiltin(110, …) silently returned a no-op and the
1940 // dirstack/$dirstack/pwd all stayed unchanged.
1941 reg_passthru!(vm, BUILTIN_PUSHD, "pushd");
1942 reg_passthru!(vm, BUILTIN_POPD, "popd");
1943 // type / whence / where / which all route through `bin_whence`
1944 // (canonical port at `src/ported/builtin.rs:3734` of
1945 // `Src/builtin.c:3975`). Each gets its own opcode so funcid +
1946 // defopts come from the BUILTINS table entry — execbuiltin
1947 // applies them correctly via the module-level dispatch_builtin.
1948 reg_passthru!(vm, BUILTIN_WHENCE, "whence");
1949 reg_passthru!(vm, BUILTIN_TYPE, "type");
1950 reg_passthru!(vm, BUILTIN_WHICH, "which");
1951 reg_passthru!(vm, BUILTIN_WHERE, "where");
1952 reg_passthru!(vm, BUILTIN_HASH, "hash");
1953 reg_passthru!(vm, BUILTIN_REHASH, "rehash");
1954
1955 // `unhash`/`unalias`/`unfunction` share `bin_unhash` (Src/builtin.c
1956 // c:4350) but each carries its own funcid (BIN_UNHASH /
1957 // BIN_UNALIAS / BIN_UNFUNCTION) in the BUILTINS table.
1958 reg_passthru!(vm, BUILTIN_UNHASH, "unhash");
1959 vm.register_builtin(BUILTIN_UNALIAS, |vm, argc| {
1960 let args = pop_args(vm, argc);
1961 Value::Status(dispatch_builtin("unalias", args))
1962 });
1963 vm.register_builtin(BUILTIN_UNFUNCTION, |vm, argc| {
1964 let args = pop_args(vm, argc);
1965 Value::Status(dispatch_builtin("unfunction", args))
1966 });
1967
1968 // Completion
1969 vm.register_builtin(BUILTIN_COMPGEN, |vm, argc| {
1970 let args = pop_args(vm, argc);
1971 // c:Bug #475/#555 — `compgen` is a bash-only builtin. In
1972 // `--zsh` mode emit "command not found" matching zsh's
1973 // external-command lookup miss.
1974 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed) {
1975 eprintln!("zsh:1: command not found: compgen");
1976 let _ = args;
1977 return Value::Status(127);
1978 }
1979 let status = with_executor(|exec| exec.builtin_compgen(&args));
1980 Value::Status(status)
1981 });
1982
1983 vm.register_builtin(BUILTIN_COMPLETE, |vm, argc| {
1984 let args = pop_args(vm, argc);
1985 // c:Bug #475 — `complete` is a bash-only builtin. Same gate
1986 // as BUILTIN_COMPGEN above.
1987 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed) {
1988 eprintln!("zsh:1: command not found: complete");
1989 let _ = args;
1990 return Value::Status(127);
1991 }
1992 let status = with_executor(|exec| exec.builtin_complete(&args));
1993 Value::Status(status)
1994 });
1995
1996 reg_passthru!(vm, BUILTIN_COMPADD, "compadd");
1997 reg_passthru!(vm, BUILTIN_COMPSET, "compset");
1998
1999 // See the const's doc comment for the contract. Stack (bottom→top):
2000 // base, e1, …, eN — argc = N + 1.
2001 vm.register_builtin(BUILTIN_TYPESET_PAREN_PACK, |vm, argc| {
2002 let mut vals: Vec<Value> = Vec::with_capacity(argc as usize);
2003 for _ in 0..argc {
2004 vals.push(vm.pop());
2005 }
2006 vals.reverse();
2007 let mut it = vals.into_iter();
2008 let mut out = it.next().map(|v| v.to_str()).unwrap_or_default();
2009 for v in it {
2010 match v {
2011 // Array → splice items as separate elements (splat);
2012 // empty array contributes nothing (empty elision).
2013 Value::Array(items) => {
2014 for item in items {
2015 out.push('\u{1f}');
2016 out.push_str(&item.to_str());
2017 }
2018 }
2019 other => {
2020 out.push('\u{1f}');
2021 out.push_str(&other.to_str());
2022 }
2023 }
2024 }
2025 Value::str(out)
2026 });
2027
2028 vm.register_builtin(BUILTIN_TYPESET_PAREN_CLOSE, |vm, _argc| {
2029 let base = vm.pop().to_str();
2030 Value::str(format!("{}\u{1f})", base))
2031 });
2032
2033 vm.register_builtin(BUILTIN_COMPDEF, |vm, argc| {
2034 let args = pop_args(vm, argc);
2035 // ACTUALLY A ZSH FUNCTION: compdef is defined by `compinit`, it is
2036 // never a builtin. Without the completion system set up it is
2037 // command-not-found (127) in every mode — `zsh -f; compdef` prints
2038 // "command not found: compdef". A user/compsys `compdef` FUNCTION
2039 // (autoload compinit → compinit defines compdef) wins and runs the
2040 // fast native impl; otherwise it's command-not-found. Previously the
2041 // extension builtin ran in native mode (bare `compdef` → "I need
2042 // arguments"), diverging from zsh.
2043 // compinit installs a `compdef` function stub (see
2044 // NATIVE_COMPDEF_MARKER) purely so `${+functions[compdef]}` is
2045 // true; route that exact body to the fast native impl instead of
2046 // dispatching the stub. A genuine user/compsys compdef function
2047 // (any other body) still wins via try_user_fn_override below.
2048 let is_native_stub = crate::ported::hashtable::shfunctab_lock()
2049 .read()
2050 .ok()
2051 .and_then(|t| t.get("compdef").and_then(|shf| shf.body.clone()))
2052 .map(|b| b.trim() == crate::extensions::ext_builtins::NATIVE_COMPDEF_MARKER)
2053 .unwrap_or(false);
2054 if is_native_stub {
2055 return Value::Status(with_executor(|exec| exec.builtin_compdef(&args)));
2056 }
2057 if let Some(s) = try_user_fn_override("compdef", &args) {
2058 return Value::Status(s);
2059 }
2060 if with_executor(|exec| exec.function_exists("compdef")) {
2061 return Value::Status(with_executor(|exec| exec.builtin_compdef(&args)));
2062 }
2063 eprintln!("zsh:1: command not found: compdef");
2064 Value::Status(127)
2065 });
2066
2067 vm.register_builtin(BUILTIN_COMPINIT, |vm, argc| {
2068 let args = pop_args(vm, argc);
2069 // ACTUALLY A ZSH FUNCTION: compinit is a contrib FUNCTION (autoloaded
2070 // from $fpath), never a builtin. Without `autoload -Uz compinit` it is
2071 // command-not-found
2072 // (127) in every mode — `zsh -f; compinit` prints
2073 // "command not found: compinit". zshrs previously ran its builtin
2074 // unconditionally, so bare `compinit` succeeded. Gate on a compinit
2075 // function entry existing (which `autoload -Uz compinit` creates);
2076 // once the user has autoloaded/defined it, run zshrs's implementation.
2077 if !with_executor(|exec| exec.function_exists("compinit")) {
2078 eprintln!("zsh:1: command not found: compinit");
2079 let _ = args;
2080 return Value::Status(127);
2081 }
2082 Value::Status(with_executor(|exec| exec.builtin_compinit(&args)))
2083 });
2084
2085 vm.register_builtin(BUILTIN_CDREPLAY, |vm, argc| {
2086 let args = pop_args(vm, argc);
2087 Value::Status(with_executor(|exec| exec.builtin_cdreplay(&args)))
2088 });
2089
2090 // Zsh-specific
2091 reg_passthru!(vm, BUILTIN_ZSTYLE, "zstyle");
2092 reg_passthru!(vm, BUILTIN_ZMODLOAD, "zmodload");
2093 reg_passthru!(vm, BUILTIN_BINDKEY, "bindkey");
2094 reg_passthru!(vm, BUILTIN_ZLE, "zle");
2095 reg_passthru!(vm, BUILTIN_VARED, "vared");
2096 reg_passthru!(vm, BUILTIN_ZCOMPILE, "zcompile");
2097 reg_passthru!(vm, BUILTIN_ZFORMAT, "zformat");
2098 reg_passthru!(vm, BUILTIN_ZPARSEOPTS, "zparseopts");
2099 reg_passthru!(vm, BUILTIN_ZREGEXPARSE, "zregexparse");
2100
2101 // Resource limits
2102 reg_passthru!(vm, BUILTIN_ULIMIT, "ulimit");
2103 reg_passthru!(vm, BUILTIN_LIMIT, "limit");
2104 reg_passthru!(vm, BUILTIN_UNLIMIT, "unlimit");
2105 reg_passthru!(vm, BUILTIN_UMASK, "umask");
2106
2107 // Misc
2108 reg_passthru!(vm, BUILTIN_TIMES, "times");
2109
2110 vm.register_builtin(BUILTIN_CALLER, |vm, argc| {
2111 let args = pop_args(vm, argc);
2112 // c:Bug #475 — `caller` is a bash-only builtin. In `--zsh`
2113 // mode emit the canonical "command not found" diagnostic
2114 // and rc=127 matching zsh's external-command-lookup miss.
2115 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed) {
2116 eprintln!("zsh:1: command not found: caller");
2117 let _ = args;
2118 return Value::Status(127);
2119 }
2120 Value::Status(with_executor(|exec| exec.builtin_caller(&args)))
2121 });
2122
2123 vm.register_builtin(BUILTIN_HELP, |vm, argc| {
2124 let args = pop_args(vm, argc);
2125 // c:Bug #475 — `help` is a bash-only builtin. Same gate as
2126 // BUILTIN_CALLER above.
2127 if crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed) {
2128 eprintln!("zsh:1: command not found: help");
2129 let _ = args;
2130 return Value::Status(127);
2131 }
2132 Value::Status(with_executor(|exec| exec.builtin_help(&args)))
2133 });
2134
2135 reg_passthru!(vm, BUILTIN_ENABLE, "enable");
2136 reg_passthru!(vm, BUILTIN_DISABLE, "disable");
2137 reg_passthru!(vm, BUILTIN_TTYCTL, "ttyctl");
2138 reg_passthru!(vm, BUILTIN_SYNC, "sync");
2139 reg_passthru!(vm, BUILTIN_MKDIR, "mkdir");
2140 reg_passthru!(vm, BUILTIN_STRFTIME, "strftime");
2141
2142 vm.register_builtin(BUILTIN_ZSLEEP, |vm, argc| {
2143 Value::Status(crate::extensions::ext_builtins::zsleep(&pop_args(vm, argc)))
2144 });
2145
2146 reg_passthru!(vm, BUILTIN_ZSYSTEM, "zsystem");
2147
2148 // PCRE
2149 reg_passthru!(vm, BUILTIN_PCRE_COMPILE, "pcre_compile");
2150 reg_passthru!(vm, BUILTIN_PCRE_MATCH, "pcre_match");
2151 reg_passthru!(vm, BUILTIN_PCRE_STUDY, "pcre_study");
2152
2153 // Database (GDBM)
2154 reg_passthru!(vm, BUILTIN_ZTIE, "ztie");
2155 reg_passthru!(vm, BUILTIN_ZUNTIE, "zuntie");
2156 reg_passthru!(vm, BUILTIN_ZGDBMPATH, "zgdbmpath");
2157
2158 // Prompt
2159 vm.register_builtin(BUILTIN_PROMPTINIT, |vm, argc| {
2160 let args = pop_args(vm, argc);
2161 // ACTUALLY A ZSH FUNCTION: promptinit is a contrib FUNCTION
2162 // (autoloaded from $fpath), never a builtin. Command-not-found until
2163 // `autoload -Uz promptinit`; once autoloaded, run the native impl.
2164 if !with_executor(|exec| exec.function_exists("promptinit")) {
2165 eprintln!("zsh:1: command not found: promptinit");
2166 let _ = args;
2167 return Value::Status(127);
2168 }
2169 Value::Status(crate::extensions::ext_builtins::promptinit(&args))
2170 });
2171
2172 vm.register_builtin(BUILTIN_PROMPT, |vm, argc| {
2173 let args = pop_args(vm, argc);
2174 Value::Status(crate::extensions::ext_builtins::prompt(&args))
2175 });
2176
2177 // Async / Parallel (zshrs extensions)
2178 vm.register_builtin(BUILTIN_ASYNC, |vm, argc| {
2179 let args = pop_args(vm, argc);
2180 let status = with_executor(|exec| exec.builtin_async(&args));
2181 Value::Status(status)
2182 });
2183
2184 vm.register_builtin(BUILTIN_AWAIT, |vm, argc| {
2185 let args = pop_args(vm, argc);
2186 let status = with_executor(|exec| exec.builtin_await(&args));
2187 Value::Status(status)
2188 });
2189
2190 vm.register_builtin(BUILTIN_PMAP, |vm, argc| {
2191 let args = pop_args(vm, argc);
2192 let status = with_executor(|exec| exec.builtin_pmap(&args));
2193 Value::Status(status)
2194 });
2195
2196 vm.register_builtin(BUILTIN_PGREP, |vm, argc| {
2197 let args = pop_args(vm, argc);
2198 let status = with_executor(|exec| exec.builtin_pgrep(&args));
2199 Value::Status(status)
2200 });
2201
2202 vm.register_builtin(BUILTIN_PEACH, |vm, argc| {
2203 let args = pop_args(vm, argc);
2204 let status = with_executor(|exec| exec.builtin_peach(&args));
2205 Value::Status(status)
2206 });
2207
2208 vm.register_builtin(BUILTIN_BARRIER, |vm, argc| {
2209 let args = pop_args(vm, argc);
2210 let status = with_executor(|exec| exec.builtin_barrier(&args));
2211 Value::Status(status)
2212 });
2213
2214 // Intercept (AOP)
2215 vm.register_builtin(BUILTIN_INTERCEPT, |vm, argc| {
2216 let args = pop_args(vm, argc);
2217 let status = with_executor(|exec| exec.builtin_intercept(&args));
2218 Value::Status(status)
2219 });
2220
2221 vm.register_builtin(BUILTIN_INTERCEPT_PROCEED, |vm, argc| {
2222 let args = pop_args(vm, argc);
2223 let status = with_executor(|exec| exec.builtin_intercept_proceed(&args));
2224 Value::Status(status)
2225 });
2226
2227 // Debug / Profile
2228 vm.register_builtin(BUILTIN_DOCTOR, |vm, argc| {
2229 let args = pop_args(vm, argc);
2230 let status = with_executor(|exec| exec.builtin_doctor(&args));
2231 Value::Status(status)
2232 });
2233
2234 vm.register_builtin(BUILTIN_DBVIEW, |vm, argc| {
2235 let args = pop_args(vm, argc);
2236 let status = with_executor(|exec| exec.builtin_dbview(&args));
2237 Value::Status(status)
2238 });
2239
2240 vm.register_builtin(BUILTIN_PROFILE, |vm, argc| {
2241 let args = pop_args(vm, argc);
2242 let status = with_executor(|exec| exec.builtin_profile(&args));
2243 Value::Status(status)
2244 });
2245
2246 reg_passthru!(vm, BUILTIN_ZPROF, "zprof");
2247
2248 // ═══════════════════════════════════════════════════════════════════════
2249 // Coreutils builtins (anti-fork, gated by !posix_mode)
2250 //
2251 // All of these are routinely wrapped by user functions in real
2252 // dotfiles (zpwr, oh-my-zsh, etc.) — `cat() { ... }`, `ls() { ... }`,
2253 // `find() { ... }`. Each handler MUST consult try_user_fn_override
2254 // first (via reg_overridable!) so the user definition wins, matching
2255 // zsh's alias → function → builtin dispatch order.
2256 // ═══════════════════════════════════════════════════════════════════════
2257
2258 reg_overridable!(vm, BUILTIN_CAT, "cat", builtin_cat);
2259 reg_overridable!(vm, BUILTIN_HEAD, "head", builtin_head);
2260 reg_overridable!(vm, BUILTIN_TAIL, "tail", builtin_tail);
2261 reg_overridable!(vm, BUILTIN_WC, "wc", builtin_wc);
2262 reg_overridable!(vm, BUILTIN_BASENAME, "basename", builtin_basename);
2263 reg_overridable!(vm, BUILTIN_DIRNAME, "dirname", builtin_dirname);
2264 reg_overridable!(vm, BUILTIN_TOUCH, "touch", builtin_touch);
2265 reg_overridable!(vm, BUILTIN_REALPATH, "realpath", builtin_realpath);
2266 reg_overridable!(vm, BUILTIN_SORT, "sort", builtin_sort);
2267 reg_overridable!(vm, BUILTIN_FIND, "find", builtin_find);
2268 reg_overridable!(vm, BUILTIN_UNIQ, "uniq", builtin_uniq);
2269 reg_overridable!(vm, BUILTIN_CUT, "cut", builtin_cut);
2270 reg_overridable!(vm, BUILTIN_TR, "tr", builtin_tr);
2271 reg_overridable!(vm, BUILTIN_SEQ, "seq", builtin_seq);
2272 reg_overridable!(vm, BUILTIN_REV, "rev", builtin_rev);
2273 reg_overridable!(vm, BUILTIN_TEE, "tee", builtin_tee);
2274 reg_overridable!(vm, BUILTIN_SLEEP, "sleep", builtin_sleep);
2275 reg_overridable!(vm, BUILTIN_WHOAMI, "whoami", builtin_whoami);
2276 reg_overridable!(vm, BUILTIN_ID, "id", builtin_id);
2277
2278 reg_overridable!(vm, BUILTIN_HOSTNAME, "hostname", builtin_hostname);
2279 reg_overridable!(vm, BUILTIN_UNAME, "uname", builtin_uname);
2280 reg_overridable!(vm, BUILTIN_DATE, "date", builtin_date);
2281 reg_overridable!(vm, BUILTIN_MKTEMP, "mktemp", builtin_mktemp);
2282 // `cp` — zshrs extension (NOT in upstream zsh; upstream's
2283 // zsh/files module ships `ln`/`mv`/`rm`/`chmod`/`chown` but no
2284 // `cp`). In-process implementation in
2285 // `ext_builtins::cp_impl` — recursive copy with -r/-R, -f, -i,
2286 // -n, -p (chown + utimensat), -v. ID 263 is the first slot
2287 // past fusevm's built-in range (260-262) and before BUILTIN_MAX
2288 // (280).
2289 /// `BUILTIN_CP` constant.
2290 pub const BUILTIN_CP: u16 = 263;
2291 reg_overridable!(vm, BUILTIN_CP, "cp", builtin_cp);
2292
2293 // Pipeline execution — bytecode-native fork-per-stage. Pops N sub-chunk
2294 // indices, forks N children with stdin/stdout wired through N-1 pipes,
2295 // each child runs its stage's compiled bytecode and exits. Parent waits
2296 // and returns the last stage's status.
2297 //
2298 // Caveats: post-fork in a multi-threaded program, only async-signal-safe
2299 // ops are POSIX-safe. We violate this (running the bytecode VM after fork
2300 // touches mutexes like REGEX_CACHE). In practice, most pipeline stages
2301 // don't touch shared mutex state — externals fork/exec away, builtins do
2302 // pure I/O. Risks are bounded; if a stage does touch a held mutex, the
2303 // child deadlocks.
2304 vm.register_builtin(BUILTIN_RUN_PIPELINE, |vm, argc| {
2305 let n = argc as usize;
2306 if n == 0 {
2307 return Value::Status(0);
2308 }
2309
2310 // c:Src/exec.c — every pipeline stage forks from the current
2311 // shell state, so each stage observes the pre-pipeline $? until
2312 // it runs its own command. Stage sub-VMs start fresh with
2313 // last_status=0, so seed them with the parent's lastval; without
2314 // this `false; echo $? | cat` prints 0 instead of zsh's 1.
2315 let parent_status = vm.last_status;
2316
2317 // Pop N sub-chunk indices (LIFO → reverse to stage order)
2318 let mut indices: Vec<u16> = Vec::with_capacity(n);
2319 for _ in 0..n {
2320 indices.push(vm.pop().to_int() as u16);
2321 }
2322 indices.reverse();
2323
2324 // Clone each stage's sub-chunk
2325 let stages: Vec<fusevm::Chunk> = indices
2326 .iter()
2327 .filter_map(|&i| vm.chunk.sub_chunks.get(i as usize).cloned())
2328 .collect();
2329 if stages.len() != n {
2330 return Value::Status(1);
2331 }
2332
2333 // Single stage — no pipe, just run inline
2334 if n == 1 {
2335 let stage = stages.into_iter().next().unwrap();
2336 crate::fusevm_disasm::maybe_print_stdout("pipeline:single", &stage);
2337 let mut stage_vm = fusevm::VM::new(stage);
2338 stage_vm.last_status = parent_status;
2339 register_builtins(&mut stage_vm);
2340 let _ = stage_vm.run();
2341 return Value::Status(stage_vm.last_status);
2342 }
2343
2344 // Build N-1 pipes
2345 let mut pipes: Vec<(i32, i32)> = Vec::with_capacity(n - 1);
2346 for _ in 0..n - 1 {
2347 let mut fds = [0i32; 2];
2348 if unsafe { libc::pipe(fds.as_mut_ptr()) } < 0 {
2349 // Cleanup any pipes we already created
2350 for (r, w) in &pipes {
2351 unsafe {
2352 libc::close(*r);
2353 libc::close(*w);
2354 }
2355 }
2356 return Value::Status(1);
2357 }
2358 pipes.push((fds[0], fds[1]));
2359 }
2360
2361 // zsh runs the LAST stage of a pipeline in the CURRENT shell
2362 // (not a forked child) so a trailing `read x` keeps its
2363 // assignment in the parent. Other shells (bash) fork every
2364 // stage. Honor zsh by leaving stage N-1 inline. Forks the
2365 // first N-1 stages with fork(); runs the last in this process
2366 // with stdin dup2'd to the last pipe's read end and stdout
2367 // restored after.
2368 let last_idx = n - 1;
2369 let stages_vec: Vec<fusevm::Chunk> = stages.into_iter().collect();
2370
2371 let mut child_pids: Vec<libc::pid_t> = Vec::with_capacity(n - 1);
2372 for (i, chunk) in stages_vec.iter().take(last_idx).enumerate() {
2373 match unsafe { libc::fork() } {
2374 -1 => {
2375 // fork failed — kill any children we already started
2376 for pid in &child_pids {
2377 unsafe { libc::kill(*pid, libc::SIGTERM) };
2378 }
2379 for (r, w) in &pipes {
2380 unsafe {
2381 libc::close(*r);
2382 libc::close(*w);
2383 }
2384 }
2385 return Value::Status(1);
2386 }
2387 0 => {
2388 // Reset SIGPIPE to default so a broken-pipe write
2389 // kills the child cleanly instead of triggering a
2390 // Rust println! panic. The parent shell ignores
2391 // SIGPIPE so it can handle EPIPE itself, but child
2392 // pipeline stages should die quietly when their
2393 // downstream stage closes early (e.g. `seq | head -3`).
2394 unsafe {
2395 libc::signal(libc::SIGPIPE, libc::SIG_DFL);
2396 }
2397 // c:Src/exec.c — pipeline children are forked
2398 // subshells; their EXIT trap context is reset so
2399 // the parent's `trap '...' EXIT` doesn't fire when
2400 // the child exits. Mirror by dropping EXIT from
2401 // the inherited traps_table inside the child.
2402 if let Ok(mut t) = crate::ported::builtin::traps_table().lock() {
2403 t.remove("EXIT");
2404 }
2405 // c:Src/exec.c:2862 → 1219 — pipeline children run
2406 // entersubsh with ESUB_PGRP, which clears the job
2407 // table (clearjobtab, Src/jobs.c:1780). Without
2408 // this, `sleep 5 & jobs -p | wc -l` reports 1 in
2409 // the forked stage where zsh reports 0. The fork
2410 // already copy-isolates the statics, so mutating
2411 // them here can't leak to the parent.
2412 with_executor(|exec| {
2413 let monitor =
2414 crate::ported::zsh_h::isset(crate::ported::zsh_h::MONITOR) as i32;
2415 crate::ported::jobs::clearjobtab(&mut exec.jobs, monitor);
2416 });
2417 *crate::ported::jobs::THISJOB
2418 .get_or_init(|| std::sync::Mutex::new(-1))
2419 .lock()
2420 .unwrap() = -1;
2421 // Child: wire stdin from previous pipe's read end
2422 if i > 0 {
2423 unsafe {
2424 libc::dup2(pipes[i - 1].0, libc::STDIN_FILENO);
2425 }
2426 }
2427 // Wire stdout to next pipe's write end
2428 unsafe {
2429 libc::dup2(pipes[i].1, libc::STDOUT_FILENO);
2430 }
2431 // (Pipe-output MULTIOS marking — c:Src/exec.c:3724 —
2432 // is emitted INTO the stage chunk by compile_pipe
2433 // via BUILTIN_PIPE_OUTPUT_MARK, gated on the stage's
2434 // top-level command actually carrying redirects, so
2435 // a nested `{ echo a > f; } | cat` body redirect
2436 // does not wrongly join the pipe.)
2437 // Close all original pipe fds (keeping stdin/stdout dups)
2438 for (r, w) in &pipes {
2439 unsafe {
2440 libc::close(*r);
2441 libc::close(*w);
2442 }
2443 }
2444
2445 // Run this stage's bytecode on a fresh VM
2446 crate::fusevm_disasm::maybe_print_stdout(
2447 &format!("pipeline:child:stage:{i}"),
2448 chunk,
2449 );
2450 let mut stage_vm = fusevm::VM::new(chunk.clone());
2451 stage_vm.last_status = parent_status;
2452 register_builtins(&mut stage_vm);
2453 let _ = stage_vm.run();
2454 // Flush any buffered output before exiting
2455 let _ = std::io::stdout().flush();
2456 let _ = std::io::stderr().flush();
2457 std::process::exit(stage_vm.last_status);
2458 }
2459 pid => {
2460 child_pids.push(pid);
2461 }
2462 }
2463 }
2464
2465 // Parent runs the LAST stage inline. Save stdin, dup the last
2466 // pipe's read end onto fd 0, run the chunk, restore stdin.
2467 // Close every other pipe fd so the producer side gets EOF
2468 // when the last upstream stage exits.
2469 let saved_stdin = unsafe { libc::dup(libc::STDIN_FILENO) };
2470 if last_idx > 0 {
2471 let read_fd = pipes[last_idx - 1].0;
2472 unsafe {
2473 libc::dup2(read_fd, libc::STDIN_FILENO);
2474 }
2475 }
2476 // Close all pipe fds in the parent now that stdin is wired.
2477 // (Children already have their own copies. The dup2 above
2478 // already gave us a fresh fd 0 if needed.)
2479 for (r, w) in &pipes {
2480 unsafe {
2481 libc::close(*r);
2482 libc::close(*w);
2483 }
2484 }
2485
2486 // Run the last stage's bytecode on a sub-VM with the host
2487 // wired up. The host points back at the executor so reads
2488 // (`read x`) update the parent's variables directly.
2489 let last_stage_status = {
2490 let last_chunk = stages_vec.into_iter().last().unwrap();
2491 crate::fusevm_disasm::maybe_print_stdout("pipeline:last", &last_chunk);
2492 let mut stage_vm = fusevm::VM::new(last_chunk);
2493 stage_vm.last_status = parent_status;
2494 register_builtins(&mut stage_vm);
2495 stage_vm.set_shell_host(Box::new(ZshrsHost));
2496 let _ = stage_vm.run();
2497 let _ = std::io::stdout().flush();
2498 let _ = std::io::stderr().flush();
2499 stage_vm.last_status
2500 };
2501
2502 // Restore stdin
2503 if saved_stdin >= 0 {
2504 unsafe {
2505 libc::dup2(saved_stdin, libc::STDIN_FILENO);
2506 libc::close(saved_stdin);
2507 }
2508 }
2509
2510 // Wait for all forked stages, capture per-stage statuses for PIPESTATUS.
2511 let mut pipestatus: Vec<i32> = Vec::with_capacity(n);
2512 for pid in child_pids {
2513 let mut status: i32 = 0;
2514 unsafe {
2515 libc::waitpid(pid, &mut status, 0);
2516 }
2517 let s = if libc::WIFEXITED(status) {
2518 libc::WEXITSTATUS(status)
2519 } else if libc::WIFSIGNALED(status) {
2520 128 + libc::WTERMSIG(status)
2521 } else {
2522 1
2523 };
2524 pipestatus.push(s);
2525 }
2526 // Append the in-parent last-stage status so `pipestatus` ends
2527 // with N entries (one per stage).
2528 pipestatus.push(last_stage_status);
2529 // Pipeline exit status: by default, the LAST stage's status.
2530 // With `setopt pipefail` (or `set -o pipefail`), use the
2531 // first non-zero stage status (so failures earlier in the
2532 // pipeline propagate even if the last stage succeeded).
2533 let pipefail_on = with_executor(|exec| opt_state_get("pipefail").unwrap_or(false));
2534 let last_status = if pipefail_on {
2535 pipestatus
2536 .iter()
2537 .copied()
2538 .rfind(|&s| s != 0)
2539 .or_else(|| pipestatus.last().copied())
2540 .unwrap_or(0)
2541 } else {
2542 *pipestatus.last().unwrap_or(&0)
2543 };
2544
2545 // c:Src/params.c:265,438 — only `pipestatus` (lowercase) is the
2546 // zsh special parameter; bash's `PIPESTATUS` doesn't exist in
2547 // zsh's special-params table. Prior port also populated
2548 // `PIPESTATUS` "for portability" — but that's a real divergence
2549 // from zsh: a script doing `[[ -z $PIPESTATUS ]]` to detect
2550 // zsh-vs-bash would mis-classify. Bug #64 in docs/BUGS.md.
2551 with_executor(|exec| {
2552 let strs: Vec<String> = pipestatus.iter().map(|s| s.to_string()).collect();
2553 exec.set_array("pipestatus".to_string(), strs);
2554 });
2555
2556 Value::Status(last_status)
2557 });
2558
2559 // Array→String join. Pops one value; if it's an Array (e.g. from Op::Glob),
2560 // joins string-coerced elements with a single space. Pass-through for
2561 // non-arrays so the op is safe to chain after any String-or-Array producer.
2562 // Scalar coercion of an assembled word: pop a Value; if it's an
2563 // Array (produced by a splice segment like `"$@"` / `"${arr[@]}"`),
2564 // IFS[0]-join it to a single scalar; a scalar passes through. This
2565 // is the assignment-context coercion C zsh applies in multsub when
2566 // the expansion is the RHS of a SCALAR assignment (Src/subst.c
2567 // c:3032 sepjoin under ssub) — `v="$@"` joins the positionals with
2568 // ${IFS[1]} rather than leaving an array whose splat would lose all
2569 // but the first element. Joins via sepjoin so a custom / empty IFS
2570 // is honored (not a hardcoded space).
2571 vm.register_builtin(BUILTIN_ARRAY_JOIN, |vm, _argc| {
2572 let val = vm.pop();
2573 match val {
2574 Value::Array(items) => {
2575 let strs: Vec<String> = items.iter().map(|v| v.to_str()).collect();
2576 Value::str(crate::ported::utils::sepjoin(&strs, None))
2577 }
2578 other => other,
2579 }
2580 });
2581
2582 // `cmd &` background execution. Compile_list emits this for any item
2583 // followed by ListOp::Amp: the job text + the cmd's sub-chunk index are
2584 // pushed, then this builtin pops both, looks up the chunk, forks. The
2585 // child detaches via setsid (so SIGINT to the foreground job doesn't kill
2586 // it), runs the bytecode on a fresh VM with builtins re-registered, exits
2587 // with the last status. The parent registers the job in the canonical
2588 // JOBTAB (c:Src/exec.c::execpline Z_ASYNC arm) and returns Status(0).
2589 vm.register_builtin(BUILTIN_RUN_BG, |vm, _argc| {
2590 let sub_idx = vm.pop().to_int() as usize;
2591 let job_text = vm.pop().to_str();
2592 let chunk = match vm.chunk.sub_chunks.get(sub_idx).cloned() {
2593 Some(c) => c,
2594 None => return Value::Status(1),
2595 };
2596
2597 match unsafe { libc::fork() } {
2598 -1 => Value::Status(1),
2599 0 => {
2600 // Child: detach and run.
2601 unsafe { libc::setsid() };
2602 crate::fusevm_disasm::maybe_print_stdout("background_job", &chunk);
2603 let mut bg_vm = fusevm::VM::new(chunk);
2604 register_builtins(&mut bg_vm);
2605 let _ = bg_vm.run();
2606 let _ = std::io::stdout().flush();
2607 let _ = std::io::stderr().flush();
2608 std::process::exit(bg_vm.last_status);
2609 }
2610 pid => {
2611 // Parent: record the PID into `$!` (most recent
2612 // backgrounded job's pid). zsh exposes this for any
2613 // script that needs `wait $!`. Also register the
2614 // bare-pid job so a no-args `wait` can synchronize.
2615 // c:Src/jobs.c:73 — `lastpid = pid;` after a
2616 // background fork. zshrs's `$!` getter
2617 // (params.rs::lookup_special_var "!") reads from
2618 // the same atomic, so a single store here is the
2619 // canonical writer.
2620 crate::ported::modules::clone::lastpid
2621 .store(pid, std::sync::atomic::Ordering::Relaxed);
2622 // c:Src/exec.c:1700 — `thisjob = newjob = initjob()`:
2623 // allocate the canonical jobtab slot. c:Src/exec.c:2950
2624 // zfork path → addproc(pid, text, 0, &bgtime, ...) hangs
2625 // the proc entry (with its display text) off the job.
2626 // c:Src/exec.c:1744-1746 — `clearoldjobtab();
2627 // jobtab[thisjob].stat |= STAT_NOSTTY;` then c:1758
2628 // `spawnjob()` promotes it to curjob (top-level shell
2629 // only), marks STAT_LOCKED and resets thisjob.
2630 {
2631 use crate::ported::jobs;
2632 use std::sync::Mutex;
2633 let table = jobs::JOBTAB.get_or_init(|| Mutex::new(Vec::new()));
2634 let idx = {
2635 let mut tab = table.lock().unwrap_or_else(|e| e.into_inner());
2636 let idx = jobs::initjob(&mut tab); // c:exec.c:1700
2637 jobs::addproc(
2638 &mut tab[idx],
2639 pid,
2640 &job_text,
2641 false,
2642 Some(std::time::Instant::now()),
2643 -1,
2644 -1,
2645 ); // c:exec.c:2950 addproc
2646 tab[idx].stat |= crate::ported::zsh_h::STAT_NOSTTY; // c:exec.c:1746
2647 idx
2648 };
2649 jobs::clearoldjobtab(); // c:exec.c:1744
2650 if let Ok(mut tj) = jobs::THISJOB.get_or_init(|| Mutex::new(-1)).lock() {
2651 *tj = idx as i32;
2652 }
2653 jobs::spawnjob(); // c:exec.c:1758
2654 }
2655 with_executor(|exec| {
2656 exec.jobs
2657 .add_pid_job(pid, job_text.clone(), JobState::Running);
2658 });
2659 Value::Status(0)
2660 }
2661 }
2662 });
2663
2664 // ── Indexed-array storage ─────────────────────────────────────────────
2665 //
2666 // Stack: pushed values then name (LAST). `arr=(a b c)` → 4 args
2667 // (a, b, c, arr). `arr=($(cmd))` → 2 args (FlatArray, arr).
2668 //
2669 // PURE PASSTHRU: pop name + values, dispatch to canonical
2670 // `setaparam` / `sethparam` (C port of `Src/params.c:3595/3602`).
2671 // assignaparam already handles PM_UNIQUE dedupe, type-flag flip,
2672 // PM_NAMEREF rejection, ASSPM_AUGMENT prepend, and createparam
2673 // for fresh names.
2674 vm.register_builtin(BUILTIN_SET_ARRAY, |vm, argc| {
2675 // `${~spec}` carrier: an assignment statement is a word-
2676 // pipeline boundary too — restore the user's GLOB_SUBST
2677 // before the NEXT word expands (`Z[d]=${~Z[d]}; print
2678 // ${options[globsubst]}` must read the user value).
2679 consume_tilde_globsubst_carrier();
2680 let n = argc as usize;
2681 let mut popped: Vec<Value> = Vec::with_capacity(n);
2682 for _ in 0..n {
2683 popped.push(vm.pop());
2684 }
2685 popped.reverse();
2686 if popped.is_empty() {
2687 return Value::Status(1);
2688 }
2689 let name = popped.pop().unwrap().to_str();
2690 let mut values: Vec<String> = Vec::new();
2691 for v in popped {
2692 flatten_array_value(v, &mut values);
2693 }
2694 let blocked = with_executor(|exec| {
2695 // Assoc init `typeset -A m; m=(k v k v ...)` — route to
2696 // canonical sethparam (Src/params.c:3602) which parses the
2697 // flat (k,v) pair list internally.
2698 if exec.assoc(&name).is_some() {
2699 // `[k]=v` / `[k]+=v` elements arrive from the compiler
2700 // as Marker / key / value triples (compile_zsh's port
2701 // of keyvalpairelement, c:Src/subst.c:49-79).
2702 let marker = crate::ported::zsh_h::Marker;
2703 let values = if values.iter().any(|e| e.starts_with(marker)) {
2704 // c:Src/params.c:3544-3560 — under ASSPM_KEY_VALUE
2705 // assocs strictly enforce `[key]=value`: every
2706 // stride-of-3 element must be a Marker. Mixing
2707 // plain pairs with kv triads is an error.
2708 let mut i = 0usize;
2709 while i < values.len() {
2710 if !values[i].starts_with(marker) {
2711 crate::ported::utils::zerr(
2712 "bad [key]=value syntax for associative array",
2713 );
2714 crate::ported::utils::errflag.fetch_or(
2715 crate::ported::zsh_h::ERRFLAG_ERROR,
2716 std::sync::atomic::Ordering::Relaxed,
2717 );
2718 exec.set_last_status(1);
2719 return true;
2720 }
2721 i += 3;
2722 }
2723 if values.len() % 3 != 0 {
2724 // c:Src/params.c:4124-4131 arrhashsetfn — a
2725 // truncated triad leaves an odd non-Marker
2726 // count → "bad set of key/value pairs".
2727 crate::ported::utils::zerr(
2728 "bad set of key/value pairs for associative array",
2729 );
2730 crate::ported::utils::errflag.fetch_or(
2731 crate::ported::zsh_h::ERRFLAG_ERROR,
2732 std::sync::atomic::Ordering::Relaxed,
2733 );
2734 exec.set_last_status(1);
2735 return true;
2736 }
2737 // c:Src/params.c:4136-4168 arrhashsetfn — whole
2738 // assignment builds a FRESH table; a `Marker +`
2739 // triad (`[k]+=v`) appends to the value inserted
2740 // EARLIER IN THIS SAME LITERAL (assignstrvalue
2741 // with eltflags=ASSPM_AUGMENT against the new ht),
2742 // so `h=([k]=a [k]+=b)` yields "ab". Resolve the
2743 // appends here, then hand flat pairs to sethparam.
2744 let mut order: Vec<String> = Vec::new();
2745 let mut map: std::collections::HashMap<String, String> =
2746 std::collections::HashMap::new();
2747 for ch in values.chunks(3) {
2748 let elt_append = ch[0].chars().nth(1) == Some('+');
2749 let k = ch[1].clone();
2750 let v = ch[2].clone();
2751 let nv = if elt_append {
2752 format!("{}{}", map.get(&k).cloned().unwrap_or_default(), v)
2753 } else {
2754 v
2755 };
2756 if !map.contains_key(&k) {
2757 order.push(k.clone());
2758 }
2759 map.insert(k, nv);
2760 }
2761 order
2762 .into_iter()
2763 .flat_map(|k| {
2764 let v = map.get(&k).cloned().unwrap_or_default();
2765 [k, v]
2766 })
2767 .collect()
2768 } else {
2769 values
2770 };
2771 // Odd-count rejection lives in the canonical chain:
2772 // sethparam → setarrvalue (c:3651/c:2920) →
2773 // arrhashsetfn's zerr "bad set of key/value pairs"
2774 // (Src/params.c:4128-4131). zerr sets ERRFLAG_ERROR,
2775 // which aborts the remaining list at the next command
2776 // boundary (BUILTIN_ERREXIT_CHECK trigger 4) —
2777 // matching `zsh -fc 'typeset -A m; m=(odd); print x'`
2778 // printing nothing after the error. Like C's sethparam
2779 // (c:3652-3653 returns v->pm regardless), the Rust
2780 // port returns Some on the odd-count path — the
2781 // failure travels via errflag, so check BOTH.
2782 let pre_err = crate::ported::utils::errflag
2783 .load(std::sync::atomic::Ordering::Relaxed)
2784 & crate::ported::zsh_h::ERRFLAG_ERROR;
2785 let res = crate::ported::params::sethparam(&name, values.clone());
2786 let now_err = crate::ported::utils::errflag
2787 .load(std::sync::atomic::Ordering::Relaxed)
2788 & crate::ported::zsh_h::ERRFLAG_ERROR;
2789 if res.is_none() || (pre_err == 0 && now_err != 0) {
2790 // c:Src/exec.c:2632-2633 addvars — `if
2791 // (!assignaparam(name, arr, myflags)) lastval = 1;`
2792 // — failed assignment sets lastval so the errflag
2793 // abort exits 1 (init.c loop() breaks, zsh_main
2794 // returns lastval).
2795 exec.set_last_status(1);
2796 return true;
2797 }
2798 #[cfg(feature = "recorder")]
2799 if crate::recorder::is_enabled() && exec.local_scope_depth == 0 {
2800 let ctx = exec.recorder_ctx();
2801 let attrs = exec.recorder_attrs_for(&name);
2802 let mut pairs: Vec<(String, String)> = Vec::with_capacity(values.len() / 2);
2803 let mut iter = values.iter().cloned();
2804 while let Some(k) = iter.next() {
2805 if let Some(v) = iter.next() {
2806 pairs.push((k, v));
2807 }
2808 }
2809 crate::recorder::emit_assoc_assign(&name, pairs, attrs, false, ctx);
2810 }
2811 return false;
2812 }
2813 // Indexed-array: setaparam (Src/params.c:3766) wraps
2814 // assignaparam with ASSPM_WARN — handles PM_UNIQUE dedupe,
2815 // type-flag flip, PM_READONLY rejection.
2816 //
2817 // `[k]=v` elements arrive as Marker / key / value triples
2818 // (compile_zsh's keyvalpairelement port). Mirror
2819 // c:Src/exec.c:2552-2553 — `if (prefork_ret &
2820 // PREFORK_KEY_VALUE) myflags |= ASSPM_KEY_VALUE;` — so
2821 // assignaparam runs its kv-resolution block (sparse fill
2822 // for PM_ARRAY, c:3447-3541; strict-triad enforcement for
2823 // special PM_HASHED targets like `options`, c:3544-3560).
2824 let values = values;
2825 let has_kv = values
2826 .iter()
2827 .any(|e| e.starts_with(crate::ported::zsh_h::Marker));
2828 // The tied-array mirror to a PM_TIED scalar
2829 // (`typeset -T PATH path`) lives canonically in
2830 // setarrvalue's dispatch in C zsh; until that wires
2831 // through assignaparam, mirror here so PATH stays in sync
2832 // after `path=(/x)`.
2833 if let Some((scalar_name, sep)) = exec.tied_array_to_scalar.get(&name).cloned() {
2834 let joined = values.join(&sep);
2835 exec.set_scalar(scalar_name, joined);
2836 }
2837 // c:Src/exec.c:2632-2633 addvars — `if (!assignaparam(...))
2838 // lastval = 1;` — a failed assignment (bad subscript, bad
2839 // [key]=value syntax, readonly) exits 1 and the errflag
2840 // abort stops the remaining list. Track errflag pre/post
2841 // like the assoc branch above.
2842 let kv_flag = if has_kv {
2843 crate::ported::zsh_h::ASSPM_KEY_VALUE
2844 } else {
2845 0
2846 };
2847 let pre_err = crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed)
2848 & crate::ported::zsh_h::ERRFLAG_ERROR;
2849 let res = crate::ported::params::assignaparam(
2850 &name,
2851 values.clone(),
2852 crate::ported::zsh_h::ASSPM_WARN | kv_flag,
2853 );
2854 let now_err = crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed)
2855 & crate::ported::zsh_h::ERRFLAG_ERROR;
2856 if res.is_none() && pre_err == 0 && now_err != 0 {
2857 exec.set_last_status(1);
2858 return true;
2859 }
2860 #[cfg(feature = "recorder")]
2861 if crate::recorder::is_enabled() && exec.local_scope_depth == 0 {
2862 let ctx = exec.recorder_ctx();
2863 let attrs = exec.recorder_attrs_for(&name);
2864 emit_path_or_assign(&name, &values, attrs, false, &ctx);
2865 }
2866 false
2867 });
2868 let status = if blocked { 1 } else { 0 };
2869 // c:Src/jobs.c:1748-1757 waitonejob — in C an array-assignment
2870 // simple command goes through execpline → waitjobs; with no
2871 // procs the else-branch stores `pipestats[0] = lastval;
2872 // numpipestats = 1`. Bare SCALAR assignments never create a
2873 // job (no waitjobs), so this clobber is array/assoc-assignment
2874 // specific: `false|true; x=(1 2); echo $pipestatus` → `0` in
2875 // zsh while `x=1` preserves `1 0`. Bug #373.
2876 crate::ported::builtin::LASTVAL.store(status, std::sync::atomic::Ordering::Relaxed);
2877 let mut synth = crate::ported::zsh_h::job::default();
2878 crate::ported::jobs::waitonejob(&mut synth);
2879 Value::Status(status)
2880 });
2881 // `arr+=(d e f)` — array append. Same calling conventions as SET_ARRAY.
2882 //
2883 // PURE PASSTHRU shape: pop name + values, dispatch through the
2884 // canonical assoc / array setter. assignaparam's ASSPM_AUGMENT
2885 // flag handles the C-source-equivalent "preserve prior value"
2886 // semantics; for now we read the current array, extend with
2887 // new values, write through set_array (which routes to
2888 // setaparam → assignaparam where PM_UNIQUE dedupe lands).
2889 vm.register_builtin(BUILTIN_APPEND_ARRAY, |vm, argc| {
2890 let n = argc as usize;
2891 let mut popped: Vec<Value> = Vec::with_capacity(n);
2892 for _ in 0..n {
2893 popped.push(vm.pop());
2894 }
2895 popped.reverse();
2896 if popped.is_empty() {
2897 return Value::Status(1);
2898 }
2899 let name = popped.pop().unwrap().to_str();
2900 let mut values: Vec<String> = Vec::new();
2901 for v in popped {
2902 flatten_array_value(v, &mut values);
2903 }
2904 let blocked = with_executor(|exec| -> bool {
2905 // Assoc append `m+=(k1 v1 ...)`: merge the (k,v) pairs into
2906 // the existing map and write back via canonical sethparam
2907 // (Src/params.c:3602). The canonical C path would go
2908 // assignaparam(ASSPM_AUGMENT) → arrhashsetfn(ASSPM_AUGMENT)
2909 // at Src/params.c:3850, but the zshrs port of
2910 // arrhashsetfn doesn't yet implement value-storage
2911 // (pending Param.u_hash backend wireup) — until that
2912 // lands, do the augment + write here so the storage
2913 // actually mutates.
2914 if exec.assoc(&name).is_some() {
2915 // `[k]=v` / `[k]+=v` elements arrive as Marker / key /
2916 // value triples (compile_zsh's port of
2917 // keyvalpairelement, c:Src/subst.c:49-79).
2918 let marker = crate::ported::zsh_h::Marker;
2919 let mut map = exec.assoc(&name).unwrap_or_default();
2920 if values.iter().any(|e| e.starts_with(marker)) {
2921 // c:Src/params.c:3544-3560 — strict triad rule.
2922 let mut i = 0usize;
2923 while i < values.len() {
2924 if !values[i].starts_with(marker) {
2925 crate::ported::utils::zerr(
2926 "bad [key]=value syntax for associative array",
2927 );
2928 crate::ported::utils::errflag.fetch_or(
2929 crate::ported::zsh_h::ERRFLAG_ERROR,
2930 std::sync::atomic::Ordering::Relaxed,
2931 );
2932 exec.set_last_status(1);
2933 return true;
2934 }
2935 i += 3;
2936 }
2937 if values.len() % 3 != 0 {
2938 // c:Src/params.c:4124-4131 — odd pair count.
2939 crate::ported::utils::zerr(
2940 "bad set of key/value pairs for associative array",
2941 );
2942 crate::ported::utils::errflag.fetch_or(
2943 crate::ported::zsh_h::ERRFLAG_ERROR,
2944 std::sync::atomic::Ordering::Relaxed,
2945 );
2946 exec.set_last_status(1);
2947 return true;
2948 }
2949 // c:Src/params.c:4133-4168 arrhashsetfn with
2950 // ASSPM_AUGMENT — ht = the EXISTING table, so
2951 // `[k]+=v` appends to the current value
2952 // (assignstrvalue eltflags=ASSPM_AUGMENT,
2953 // c:4144-4150) and `[k]=v` overwrites.
2954 for ch in values.chunks(3) {
2955 let elt_append = ch[0].chars().nth(1) == Some('+');
2956 let k = ch[1].clone();
2957 let v = ch[2].clone();
2958 let nv = if elt_append {
2959 format!("{}{}", map.get(&k).cloned().unwrap_or_default(), v)
2960 } else {
2961 v
2962 };
2963 map.insert(k, nv);
2964 }
2965 } else {
2966 let mut it = values.iter().cloned();
2967 while let Some(k) = it.next() {
2968 if let Some(v) = it.next() {
2969 map.insert(k, v);
2970 }
2971 }
2972 }
2973 exec.set_assoc(name, map);
2974 return false;
2975 }
2976 // Indexed-array append `arr+=(d e f)` — route directly
2977 // through canonical assignaparam with ASSPM_AUGMENT
2978 // (`Src/params.c:3570-3585` append-on-array branch).
2979 // assignaparam reads the prior array internally and
2980 // appends the new values, so the bridge no longer needs
2981 // to pre-concat manually. Marker triples from `[k]=v`
2982 // elements add ASSPM_KEY_VALUE (c:Src/exec.c:2552-2553)
2983 // so the kv sparse-fill block (c:Src/params.c:3447-3541)
2984 // resolves them against the existing elements.
2985 let kv_flag = if values
2986 .iter()
2987 .any(|e| e.starts_with(crate::ported::zsh_h::Marker))
2988 {
2989 crate::ported::zsh_h::ASSPM_KEY_VALUE
2990 } else {
2991 0
2992 };
2993 let pre_err = crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed)
2994 & crate::ported::zsh_h::ERRFLAG_ERROR;
2995 let res = crate::ported::params::assignaparam(
2996 &name,
2997 values.clone(),
2998 crate::ported::zsh_h::ASSPM_AUGMENT | kv_flag,
2999 );
3000 let now_err = crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed)
3001 & crate::ported::zsh_h::ERRFLAG_ERROR;
3002 if res.is_none() && pre_err == 0 && now_err != 0 {
3003 // c:Src/exec.c:2632-2633 — failed assignment → lastval 1.
3004 exec.set_last_status(1);
3005 return true;
3006 }
3007 #[cfg(feature = "recorder")]
3008 if crate::recorder::is_enabled() && exec.local_scope_depth == 0 {
3009 let ctx = exec.recorder_ctx();
3010 let attrs = exec.recorder_attrs_for(&name);
3011 emit_path_or_assign(&name, &values, attrs, true, &ctx);
3012 }
3013 // Tied-scalar mirror — TODO faithful: should live in
3014 // setarrvalue's gsu dispatch once boot_ paramtab wiring
3015 // lands (Task #16). Re-read the canonical post-augment
3016 // array so the joined scalar matches.
3017 let tied_scalar = exec.tied_array_to_scalar.get(&name).cloned();
3018 if let Some((scalar_name, sep)) = tied_scalar {
3019 let merged = exec.array(&name).unwrap_or_default();
3020 let joined = merged.join(&sep);
3021 exec.set_scalar(scalar_name.clone(), joined.clone());
3022 let _ = crate::ported::params::zputenv(&format!("{}={}", &scalar_name, &joined));
3023 // c:Src/params.c:5354
3024 }
3025 false
3026 });
3027 // c:Src/jobs.c:1748-1757 waitonejob — `arr+=(...)` is an
3028 // array-assignment simple command and clobbers pipestats to
3029 // `[lastval]` exactly like `arr=(...)` above. Bug #373.
3030 let status = if blocked { 1 } else { 0 };
3031 crate::ported::builtin::LASTVAL.store(status, std::sync::atomic::Ordering::Relaxed);
3032 let mut synth = crate::ported::zsh_h::job::default();
3033 crate::ported::jobs::waitonejob(&mut synth);
3034 Value::Status(status)
3035 });
3036 // `name[@]=(...)` / `name[*]=(...)` — whole-array SET with the assoc
3037 // guard (c:Src/params.c:3324-3327). Stack: [v0..vn, name].
3038 vm.register_builtin(BUILTIN_SET_ARRAY_AT, |vm, argc| {
3039 let (name, values) = pop_array_args_with_name(vm, argc);
3040 let status = with_executor(|exec| {
3041 if exec.assoc(&name).is_some() {
3042 // c:Src/params.c:3324-3327 — `[@]` (any slice) on a
3043 // PM_HASHED target is an error.
3044 crate::ported::utils::zerr(&format!(
3045 "{}: attempt to set slice of associative array",
3046 name
3047 ));
3048 crate::ported::utils::errflag.fetch_or(
3049 crate::ported::zsh_h::ERRFLAG_ERROR,
3050 std::sync::atomic::Ordering::Relaxed,
3051 );
3052 exec.set_last_status(1);
3053 return 1;
3054 }
3055 exec.set_array(name, values); // whole replace (c:3528 setarrvalue)
3056 0
3057 });
3058 Value::Status(status)
3059 });
3060 // `name[@]+=(...)` / `name[*]+=(...)` — whole-array APPEND (push) with
3061 // the same assoc guard.
3062 vm.register_builtin(BUILTIN_APPEND_ARRAY_AT, |vm, argc| {
3063 let (name, values) = pop_array_args_with_name(vm, argc);
3064 let status = with_executor(|exec| {
3065 if exec.assoc(&name).is_some() {
3066 crate::ported::utils::zerr(&format!(
3067 "{}: attempt to set slice of associative array",
3068 name
3069 ));
3070 crate::ported::utils::errflag.fetch_or(
3071 crate::ported::zsh_h::ERRFLAG_ERROR,
3072 std::sync::atomic::Ordering::Relaxed,
3073 );
3074 exec.set_last_status(1);
3075 return 1;
3076 }
3077 let mut cur = exec.array(&name).unwrap_or_default();
3078 cur.extend(values);
3079 exec.set_array(name, cur); // c:3511-3528 AUGMENT on array → push
3080 0
3081 });
3082 Value::Status(status)
3083 });
3084 vm.register_builtin(BUILTIN_RUN_SELECT, |vm, argc| {
3085 if argc < 2 {
3086 return Value::Status(1);
3087 }
3088 let n = argc as usize;
3089 let mut popped: Vec<Value> = Vec::with_capacity(n);
3090 for _ in 0..n {
3091 popped.push(vm.pop());
3092 }
3093 // popped: [sub_idx, name, word_N, ..., word_1] (popping from top)
3094 let sub_idx_val = popped.remove(0);
3095 let name_val = popped.remove(0);
3096 // c:Src/loop.c — `select` flattens Array values (from `$@`,
3097 // `${arr[@]}`, etc.) into the menu. Without per-element
3098 // splice, `select x do ... done` (bare, iterating $@)
3099 // collapsed all positionals into one joined entry.
3100 let mut words: Vec<String> = Vec::new();
3101 for v in popped.into_iter().rev() {
3102 match v {
3103 Value::Array(items) => {
3104 for item in items {
3105 words.push(item.to_str());
3106 }
3107 }
3108 other => words.push(other.to_str()),
3109 }
3110 }
3111
3112 let sub_idx = sub_idx_val.to_int() as usize;
3113 let name = name_val.to_str();
3114
3115 // c:Src/loop.c:248-252 — `if (!args || empty(args)) {
3116 // state->pc = end; ... return 0; }`. An empty option list
3117 // skips the body entirely; without this gate the prompt loop
3118 // runs indefinitely (or twice on the EOF stdin case before
3119 // exiting). Bug #401.
3120 if words.is_empty() {
3121 return Value::Status(0);
3122 }
3123
3124 let chunk = match vm.chunk.sub_chunks.get(sub_idx).cloned() {
3125 Some(c) => c,
3126 None => return Value::Status(1),
3127 };
3128
3129 let prompt =
3130 with_executor(|exec| exec.scalar("PROMPT3").unwrap_or_else(|| "?# ".to_string()));
3131
3132 let stdin = std::io::stdin();
3133 let mut reader = stdin.lock();
3134 let mut last_status: i32 = 0;
3135
3136 loop {
3137 // Direct port of zsh's selectlist from
3138 // src/zsh/Src/loop.c:347-409. Layout is column-major
3139 // ("down columns, then across") — NOT row-major. With
3140 // 6 items in 3 cols zsh produces:
3141 // 1 3 5
3142 // 2 4 6
3143 // The previous Rust impl walked row-major which
3144 // produced 1 2 3 / 4 5 6 (visually similar but wrong
3145 // for prompts that mention ordering and breaks scripts
3146 // that rely on column count == ceil(N/rows)).
3147 //
3148 // C variable mapping:
3149 // ct -> word count (n)
3150 // longest -> max item width + 1, then plus digits-of-ct
3151 // fct -> column count
3152 // fw -> per-column width
3153 // colsz -> row count = ceil(ct / fct)
3154 // t1 -> row index, walks 0..colsz
3155 // ap -> item pointer; advances by colsz to step
3156 // DOWN a column.
3157 let term_width: usize = env::var("COLUMNS")
3158 .ok()
3159 .and_then(|v| v.parse().ok())
3160 .unwrap_or(80);
3161 let ct = words.len();
3162 // loop.c:354-363 — find longest item width.
3163 let mut longest = 1usize;
3164 for w in &words {
3165 let aplen = w.chars().count();
3166 if aplen > longest {
3167 longest = aplen;
3168 }
3169 }
3170 // loop.c:365-367 — `longest++` then add digits of `ct`.
3171 longest += 1;
3172 let mut t0 = ct;
3173 while t0 > 0 {
3174 t0 /= 10;
3175 longest += 1;
3176 }
3177 // loop.c:369-373 — fct = (cols - 1) / (longest + 3); if
3178 // 0, fct = 1; else fw = (cols - 1) / fct.
3179 let raw_fct = (term_width.saturating_sub(1)) / (longest + 3);
3180 let (fct, fw) = if raw_fct == 0 {
3181 (1, longest + 3)
3182 } else {
3183 (raw_fct, (term_width.saturating_sub(1)) / raw_fct)
3184 };
3185 // loop.c:374 — colsz = (ct + fct - 1) / fct.
3186 let colsz = ct.div_ceil(fct);
3187 // loop.c:375-395 — for each row t1, walk down columns.
3188 for t1 in 0..colsz {
3189 let mut ap_idx = t1;
3190 while ap_idx < ct {
3191 let w = &words[ap_idx];
3192 let n = ap_idx + 1;
3193 let _ = write!(std::io::stderr(), "{}) {}", n, w);
3194 let mut t2 = w.chars().count() + 2;
3195 let mut t3 = n;
3196 while t3 > 0 {
3197 t2 += 1;
3198 t3 /= 10;
3199 }
3200 // Pad to fw (loop.c:389-390).
3201 while t2 < fw {
3202 let _ = write!(std::io::stderr(), " ");
3203 t2 += 1;
3204 }
3205 ap_idx += colsz;
3206 }
3207 let _ = writeln!(std::io::stderr());
3208 }
3209 let _ = write!(std::io::stderr(), "{}", prompt);
3210 let _ = std::io::stderr().flush();
3211
3212 let mut line = String::new();
3213 match reader.read_line(&mut line) {
3214 Ok(0) => {
3215 // EOF — emit the final newline that zsh prints
3216 // after the prompt-then-EOF sequence (c:Src/loop.c
3217 // selectlist falls through to fputc('\n', stderr)
3218 // at the end of the read failure path). Without
3219 // this the next process's output runs directly
3220 // after `-->>>> ` on the same line.
3221 let _ = writeln!(std::io::stderr());
3222 let _ = std::io::stderr().flush();
3223 break;
3224 }
3225 Ok(_) => {}
3226 Err(_) => break,
3227 }
3228 let trimmed = line.trim_end_matches(['\n', '\r'][..].as_ref()).to_string();
3229
3230 with_executor(|exec| {
3231 exec.set_scalar("REPLY".to_string(), trimmed.clone());
3232 });
3233
3234 if trimmed.is_empty() {
3235 // Empty input → redraw menu without running body.
3236 continue;
3237 }
3238
3239 let chosen = match trimmed.parse::<usize>() {
3240 Ok(n) if n >= 1 && n <= words.len() => words[n - 1].clone(),
3241 _ => String::new(),
3242 };
3243
3244 with_executor(|exec| {
3245 exec.set_scalar(name.clone(), chosen);
3246 });
3247
3248 // Reset canonical BREAKS/CONTFLAG before running the body
3249 // so a stale value from a sibling construct doesn't leak in.
3250 crate::ported::builtin::BREAKS.store(0, SeqCst);
3251 crate::ported::builtin::CONTFLAG.store(0, SeqCst);
3252
3253 // c:Src/loop.c — `select` increments LOOPS for the body so
3254 // `break` / `continue` inside the body see loops > 0 and
3255 // don't emit `not in while, until, select, or repeat loop`.
3256 // Mirrors execwhile/execrepeat's `LOOPS.fetch_add` pattern.
3257 // The decrement happens after the body call so a body that
3258 // explicitly returns / errors still leaves the counter
3259 // balanced for the next iteration.
3260 crate::ported::builtin::LOOPS.fetch_add(1, SeqCst);
3261
3262 crate::fusevm_disasm::maybe_print_stdout("select:body", &chunk);
3263 let mut body_vm = fusevm::VM::new(chunk.clone());
3264 register_builtins(&mut body_vm);
3265 let _ = body_vm.run();
3266 last_status = body_vm.last_status;
3267
3268 crate::ported::builtin::LOOPS.fetch_sub(1, SeqCst);
3269
3270 // Drain the canonical BREAKS/CONTFLAG counters. Mirrors
3271 // loop.c:529-534's `if (breaks) { breaks--; if (breaks ||
3272 // !contflag) break; contflag = 0; }` drain pattern.
3273 // The legacy `BREAK_SELECT=1` env-var sentinel is still
3274 // honored for backward compat.
3275 let break_legacy = with_executor(|exec| {
3276 let v = exec.scalar("BREAK_SELECT");
3277 exec.unset_scalar("BREAK_SELECT");
3278 v.map(|s| s != "0" && !s.is_empty()).unwrap_or(false)
3279 });
3280 use std::sync::atomic::Ordering::SeqCst;
3281 let breaks = crate::ported::builtin::BREAKS.load(SeqCst);
3282 if breaks > 0 {
3283 let cont = crate::ported::builtin::CONTFLAG.load(SeqCst);
3284 crate::ported::builtin::BREAKS.fetch_sub(1, SeqCst);
3285 if breaks - 1 > 0 || cont == 0 {
3286 break;
3287 }
3288 crate::ported::builtin::CONTFLAG.store(0, SeqCst);
3289 continue;
3290 }
3291 if break_legacy {
3292 break;
3293 }
3294 }
3295
3296 Value::Status(last_status)
3297 });
3298
3299 // Magic special-parameter assoc lookup. Synthesizes values from
3300 // shell state for zsh's shell-introspection assocs:
3301 // commands, aliases, galiases, saliases, dis_aliases, dis_galiases,
3302 // dis_saliases, functions, dis_functions, builtins, dis_builtins,
3303 // reswords, options, parameters, jobtexts, jobdirs, jobstates,
3304 // nameddirs, userdirs, modules.
3305 // Returns None if `name` isn't a recognized magic name.
3306
3307 // `${arr[idx]}` — pop name, then idx_str. zsh is 1-based for positive
3308 // indices; we honor that. `@`/`*` return the whole array as Value::Array
3309 // so Op::Exec splice produces N argv slots. For `${foo[key]}` where foo
3310 // is an assoc, the idx is a string key — we check assoc_arrays first
3311 // when the idx isn't `@`/`*` and the name has an assoc binding.
3312 // BUILTIN_ARRAY_INDEX — `${name[idx]}` paramsubst dispatch.
3313 // PURE PASSTHRU: pops the idx + name, hands the canonical
3314 // `${name[idx]}` form to `subst::paramsubst` (C port of
3315 // `Src/subst.c::paramsubst`). All subscript-flag dispatch
3316 // ((I)pat / (R)pat / (i)/(r)/(K)/(k), range slices `[N,M]`,
3317 // negative indices, magic-assoc shape lookup, DQ-join collapse)
3318 // lives inside paramsubst → fetchvalue → getarg in params.rs.
3319 //
3320 // Outer-flag dispatch (`(@)` / `(@k)` / `(v)NAME[(I)pat]` / etc.)
3321 // routes through BUILTIN_BRIDGE_BRACE_ARRAY at the compile path
3322 // (canonical paramsubst flag parser owns dispatch at Src/subst.c:2147+),
3323 // so BUILTIN_ARRAY_INDEX receives clean name+key with no sentinel
3324 // prefixes.
3325 vm.register_builtin(BUILTIN_ARRAY_INDEX, |vm, _argc| {
3326 let idx = vm.pop().to_str();
3327 let name = vm.pop().to_str();
3328 array_index_lookup(&name, &idx)
3329 });
3330 // BUILTIN_ARRAY_INDEX_UNBRACED — bare `$name[idx]` (no braces).
3331 // Same subscript dispatch as BUILTIN_ARRAY_INDEX when KSHARRAYS
3332 // is unset, but under KSHARRAYS the UNBRACED form does NOT
3333 // subscript at all:
3334 // c:Src/subst.c:2800-2802 — fetchvalue's bracket-parse arg is
3335 // `(unset(KSHARRAYS) || inbrace) ? 1 : -1`; -1 inhibits
3336 // subscript parsing for the bare form under KSHARRAYS.
3337 // c:Src/subst.c:2867 — the bracket-consuming loop only runs
3338 // `while (v || ((inbrace || (unset(KSHARRAYS) && vunset)) &&
3339 // isbrack(*s)))` — bare + KSHARRAYS leaves `[...]` as literal
3340 // trailing text.
3341 // The bare `$name` expands (first element for identifier-named
3342 // arrays per c:Src/params.c:2293-2296 `v->end = 1, v->isarr = 0`),
3343 // the literal `[idx]` (+ any literal suffix) joins the last word,
3344 // and the word undergoes filename generation: `[...]` is a glob
3345 // char class, so unquoted it hits the c:Src/glob.c:1873-1886
3346 // nomatch/nullglob dispatch (reused via exec.expand_glob).
3347 // Operands: [name, idx, suffix, quoted] — `quoted` set when the
3348 // word carries DQ markers (no filename generation in DQ; zsh 5.9:
3349 // `setopt ksharrays; a=(x y z); print "$a[0]"` → `x[0]`).
3350 // Verbatim zsh 5.9 ground truth for the unquoted form:
3351 // `setopt ksharrays; a=(x y z); print -- $a[0]` →
3352 // stderr `zsh:1: no matches found: x[0]`, rc=1, empty stdout.
3353 vm.register_builtin(BUILTIN_ARRAY_INDEX_UNBRACED, |vm, _argc| {
3354 let quoted = vm.pop().to_str() == "1";
3355 let suffix = vm.pop().to_str();
3356 let idx = vm.pop().to_str();
3357 let name = vm.pop().to_str();
3358 if !crate::ported::zsh_h::isset(crate::ported::zsh_h::KSHARRAYS) {
3359 let v = array_index_lookup(&name, &idx);
3360 if suffix.is_empty() {
3361 return v;
3362 }
3363 // Mirrors the previous compile-shape `ARRAY_INDEX +
3364 // Op::Concat` exactly: Concat stringifies via as_str_cow
3365 // (fusevm value.rs:132-146, arrays join with " ").
3366 return Value::str(format!("{}{}", v.to_str(), suffix));
3367 }
3368 // KSHARRAYS bare form: no subscript. Bare-`$name` words +
3369 // literal `[idx]suffix` glued onto the last word.
3370 let mut words = ksharrays_bare_words(&name);
3371 let last = format!("{}[{}]{}", words.pop().unwrap_or_default(), idx, suffix);
3372 if quoted {
3373 // DQ context — no filename generation, bracket text stays
3374 // literal.
3375 words.push(last);
3376 return Value::str(words.join(" "));
3377 }
3378 // c:Src/glob.c:1873-1886 — expand_glob handles nullglob /
3379 // NOMATCH (zerr "no matches found" + errflag + the
3380 // current_command_glob_failed cell consumed at the command
3381 // dispatch boundary) / literal passthrough for glob-free text.
3382 let matches = with_executor(|exec| exec.expand_glob(&last));
3383 let mut out: Vec<Value> = words.into_iter().map(Value::str).collect();
3384 out.extend(matches.into_iter().map(Value::str));
3385 if out.len() == 1 {
3386 return out.pop().unwrap();
3387 }
3388 Value::Array(out)
3389 });
3390 // BUILTIN_ASSOC_HAS_KEY — `${(k)assoc[name]}` key-existence query.
3391 // Pops [assoc_name, key]; returns key (Str) if present in the
3392 // assoc, empty Str otherwise. Mirrors zsh's `${(k)h[name]}`
3393 // documented semantics in zshparam(1) "Parameter Expansion Flags".
3394 // Distinct from BUILTIN_ARRAY_INDEX (which returns the VALUE) and
3395 // from `${+h[name]}` (which returns "0"/"1"). Bug #145.
3396 vm.register_builtin(BUILTIN_ASSOC_HAS_KEY, |vm, _argc| {
3397 let key = vm.pop().to_str();
3398 let name = vm.pop().to_str();
3399 // c:Src/params.c getindex — the subscript text is substituted
3400 // (singsub) before the lookup; `${(k)H[$k]}` must resolve $k.
3401 // The compiler hands this opcode the RAW subscript text, so a
3402 // dynamic key arrived literally ("$k") and matched nothing.
3403 // singsub is identity for plain keys.
3404 let key = if key.contains('$') || key.contains('`') || key.contains('\u{8c}') {
3405 crate::ported::subst::singsub(&key)
3406 } else {
3407 key
3408 };
3409 // c:Src/params.c:3131 gethkparam covers ordinary PM_HASHED
3410 // paramtab entries only. Special/magic hashes (`parameters`,
3411 // `options`, … — the zsh/parameter module's partab-backed
3412 // params, Src/Modules/parameter.c) aren't in that storage, so
3413 // a None here doesn't mean "no such assoc". Route those
3414 // through paramsubst, whose assoc materialization handles the
3415 // magic hashes and whose getarg port (c:Src/params.c:1591 +
3416 // Src/subst.c:2922) returns the KEY for `(k)` on a plain
3417 // subscript. zsh 5.9: `${(k)parameters[PATH]}` → "PATH".
3418 match crate::ported::params::gethkparam(&name) {
3419 Some(keys) => {
3420 if keys.iter().any(|k| k == &key) {
3421 Value::str(key)
3422 } else {
3423 Value::str("")
3424 }
3425 }
3426 None => paramsubst_to_value(&format!("${{(k){}[{}]}}", name, key)),
3427 }
3428 });
3429 vm.register_builtin(BUILTIN_BRIDGE_BRACE_ARRAY, |vm, _argc| {
3430 // Inner body of `${(...)...}` (already stripped of `${`/`}` by
3431 // the caller). The compiler optionally prefixes Qstring
3432 // (\u{8c}) to signal "expanded in DQ context" — strip it
3433 // here and bump in_dq_context for the paramsubst call so the
3434 // SUB_ZIP and other qt-aware paths fire.
3435 let body = vm.pop().to_str();
3436 let (dq, inner) = if let Some(rest) = body.strip_prefix('\u{8c}') {
3437 (true, rest.to_string())
3438 } else {
3439 (false, body)
3440 };
3441 if dq {
3442 with_executor(|exec| exec.in_dq_context += 1);
3443 }
3444 let v = paramsubst_to_value(&format!("${{{}}}", inner));
3445 if dq {
3446 with_executor(|exec| exec.in_dq_context -= 1);
3447 }
3448 v
3449 });
3450
3451 // BUILTIN_PARAM_FLAG — `${(flags)name}` paramsubst dispatch.
3452 // PURE PASSTHRU: pops sentinel-tagged flags + name, hands the
3453 // canonical `${(flags)name}` form to `subst::paramsubst` (C port
3454 // of `Src/subst.c::paramsubst`). The bridge does no flag
3455 // walking, no DQ-context branching, no array/scalar shape
3456 // selection — all of that lives inside paramsubst. Compile-time
3457 // context (DQ / scalar-assign-RHS) flows through executor cells
3458 // (in_dq_context, in_scalar_assign) bumped by BUILTIN_EXPAND_TEXT.
3459 vm.register_builtin(BUILTIN_PARAM_FLAG, |vm, _argc| {
3460 let flags = vm.pop().to_str();
3461 let name = vm.pop().to_str();
3462 let body = format!("${{({}){}}}", flags, name);
3463 paramsubst_to_value(&body)
3464 });
3465
3466 // `foo[key]=val` — single-key set on an assoc array. Stack: [name, key, value].
3467 // PURE PASSTHRU: assignsparam with `name[key]` form (C port of
3468 // `Src/params.c::assignsparam` subscript path at c:3210-3231)
3469 // already does the indexed-array vs assoc decision, PM_HASHED
3470 // auto-vivification, numeric-subscript bounds handling, and
3471 // PM_READONLY rejection.
3472 vm.register_builtin(BUILTIN_SET_ASSOC, |vm, _argc| {
3473 // `${~spec}` carrier: an assignment statement is a word-
3474 // pipeline boundary too — restore the user's GLOB_SUBST
3475 // before the NEXT word expands (`Z[d]=${~Z[d]}; print
3476 // ${options[globsubst]}` must read the user value).
3477 consume_tilde_globsubst_carrier();
3478 // argc 4 = compile flagged the subscript as DYNAMIC (`H[$k]`):
3479 // an EXPANDED-empty key is then a legal assoc key (C's
3480 // assignsparam isident gate sees the raw `$k` text and the
3481 // empty key stores at getindex time — zinit's
3482 // ZINIT_SICE[$1…$2] relies on it). argc 3 = source-literal
3483 // key; `H[]` stays the "not an identifier" error.
3484 let key_is_dynamic = if _argc == 4 {
3485 vm.pop().to_int() != 0
3486 } else {
3487 false
3488 };
3489 let value = vm.pop().to_str();
3490 let key = vm.pop().to_str();
3491 let name = vm.pop().to_str();
3492 if key_is_dynamic && key.is_empty() {
3493 with_executor(|exec| {
3494 let _ = exec;
3495 });
3496 // Mirror assignsparam's PM_HASHED tail directly (the
3497 // textual `name[]` reconstruction can't pass isident).
3498 if let Ok(mut store) = crate::ported::params::paramtab_hashed_storage().lock() {
3499 let entry = store.entry(name.clone()).or_default();
3500 let newval = if let Some(old) = entry.get("") {
3501 // `+=` arrives pre-concatenated by the compile
3502 // read-modify-write; plain `=` overwrites.
3503 let _ = old;
3504 value.clone()
3505 } else {
3506 value.clone()
3507 };
3508 entry.insert(String::new(), newval);
3509 }
3510 return Value::Status(0);
3511 }
3512 with_executor(|exec| {
3513 #[cfg(feature = "recorder")]
3514 if crate::recorder::is_enabled() && exec.local_scope_depth == 0 {
3515 let ctx = exec.recorder_ctx();
3516 let attrs = exec.recorder_attrs_for(&name);
3517 crate::recorder::emit_assoc_assign(
3518 &name,
3519 vec![(key.clone(), value.clone())],
3520 attrs,
3521 true,
3522 ctx,
3523 );
3524 }
3525 let _ = exec;
3526 });
3527 // Build `name[key]=value` shape for assignsparam's subscript
3528 // dispatch. Arith-evaluate numeric subscripts on an existing
3529 // indexed array (`a[i+1]=v` form) before handing off — the
3530 // canonical port currently only handles literal int / string
3531 // keys, so pre-resolve here.
3532 let resolved_key = with_executor(|exec| {
3533 let is_indexed = exec.array(&name).is_some();
3534 let is_assoc = exec.assoc(&name).is_some();
3535 let is_scalar = !is_indexed && !is_assoc && exec.scalar(&name).is_some();
3536 // c:Src/params.c::getindex — `(i)pat` / `(I)pat` / `(R)pat`
3537 // / `(r)pat` subscript flags on an indexed array LHS resolve
3538 // to a numeric index (first / last match of pat). On a
3539 // SCALAR LHS the same flags resolve to a CHAR position
3540 // (1-based first/last match of pat in the scalar string)
3541 // for the c:2748+ char-splice assignment. zshrs's
3542 // read-form `${a[(i)pat]}` already implements both shapes;
3543 // the LHS assignment path silently stored the literal
3544 // "(i)pat" as an assoc key (for scalar: auto-vivified to
3545 // PM_HASHED via the assignsparam unknown-subscript
3546 // fallback). Bug #293 (array) / scalar sibling.
3547 //
3548 // Detect the `(flags)pat` shape and resolve to a numeric
3549 // index before assignsparam.
3550 if is_indexed || is_scalar {
3551 if let Some(rest) = key.strip_prefix('(') {
3552 if let Some(close) = rest.find(')') {
3553 let flags = &rest[..close];
3554 let pat = &rest[close + 1..];
3555 if !flags.is_empty()
3556 && flags
3557 .chars()
3558 .all(|c| matches!(c, 'I' | 'R' | 'i' | 'r' | 'n' | 'e'))
3559 {
3560 // Resolve via the array's contents.
3561 if let Some(arr) = exec.array(&name) {
3562 let return_index = true; // LHS write — index needed
3563 let down = flags.contains('I') || flags.contains('R');
3564 let exact = flags.contains('e');
3565 let iter: Box<dyn Iterator<Item = (usize, &String)>> = if down {
3566 Box::new(arr.iter().enumerate().rev())
3567 } else {
3568 Box::new(arr.iter().enumerate())
3569 };
3570 let mut found: Option<usize> = None;
3571 for (idx, elem) in iter {
3572 let matched = if exact {
3573 elem == pat
3574 } else {
3575 crate::ported::pattern::patcompile(
3576 &{
3577 let mut __pat_tok = (pat).to_string();
3578 crate::ported::glob::tokenize(&mut __pat_tok);
3579 __pat_tok
3580 },
3581 crate::ported::zsh_h::PAT_HEAPDUP as i32,
3582 None,
3583 )
3584 .map_or(false, |p| crate::ported::pattern::pattry(&p, elem))
3585 };
3586 if matched {
3587 found = Some(idx);
3588 break;
3589 }
3590 }
3591 let _ = return_index;
3592 // (i)/(r) return 1-based index of match,
3593 // arr.len()+1 (or 1 for I/R) on miss
3594 // per zsh docs. We mirror the read-form
3595 // semantics from subst.rs.
3596 let idx_1based = match found {
3597 Some(i) => (i + 1) as i64,
3598 None => (arr.len() + 1) as i64,
3599 };
3600 return idx_1based.to_string();
3601 }
3602 // Scalar LHS — resolve to a CHAR position
3603 // (1-based first/last match of pat in the
3604 // string). c:Src/params.c:1411-1418 — the
3605 // scalar path returns the char index from
3606 // sliding-window pattern match against
3607 // pm.u_str. Same algorithm as the read-form
3608 // at subst.rs:5283-5306. Bug (scalar
3609 // sibling of #293): `a=hello; a[(I)l]=X`
3610 // previously auto-vivified `a` into
3611 // PM_HASHED with key "(I)l" instead of
3612 // splicing X at the last 'l' position
3613 // (yielding "helXo").
3614 if is_scalar {
3615 let s = exec.scalar(&name).unwrap_or_default();
3616 let s_chars: Vec<char> = s.chars().collect();
3617 let n = s_chars.len();
3618 let want_last = flags.contains('I') || flags.contains('R');
3619 let exact = flags.contains('e');
3620 let mut found: Option<usize> = None;
3621 'outer: for start in 0..=n {
3622 let lengths: Box<dyn Iterator<Item = usize>> = if want_last {
3623 Box::new((1..=(n - start)).rev())
3624 } else {
3625 Box::new(1..=(n - start))
3626 };
3627 for len in lengths {
3628 let cand: String =
3629 s_chars[start..start + len].iter().collect();
3630 let matched = if exact {
3631 cand == pat
3632 } else {
3633 crate::ported::pattern::patcompile(
3634 &{
3635 let mut __pat_tok = (pat).to_string();
3636 crate::ported::glob::tokenize(&mut __pat_tok);
3637 __pat_tok
3638 },
3639 crate::ported::zsh_h::PAT_HEAPDUP as i32,
3640 None,
3641 )
3642 .map_or(false, |p| {
3643 crate::ported::pattern::pattry(&p, &cand)
3644 })
3645 };
3646 if matched {
3647 found = Some(start);
3648 if !want_last {
3649 break 'outer;
3650 }
3651 break;
3652 }
3653 }
3654 }
3655 // (I)/(R): scan again to find LAST.
3656 if want_last {
3657 let mut last_found: Option<usize> = found;
3658 for start in (0..=n).rev() {
3659 for len in 1..=(n - start) {
3660 let cand: String =
3661 s_chars[start..start + len].iter().collect();
3662 let matched = if exact {
3663 cand == pat
3664 } else {
3665 crate::ported::pattern::patcompile(
3666 &{
3667 let mut __pat_tok = (pat).to_string();
3668 crate::ported::glob::tokenize(
3669 &mut __pat_tok,
3670 );
3671 __pat_tok
3672 },
3673 crate::ported::zsh_h::PAT_HEAPDUP as i32,
3674 None,
3675 )
3676 .map_or(false, |p| {
3677 crate::ported::pattern::pattry(&p, &cand)
3678 })
3679 };
3680 if matched {
3681 last_found = Some(start);
3682 break;
3683 }
3684 }
3685 if last_found.is_some() && last_found.unwrap() >= start {
3686 break;
3687 }
3688 }
3689 found = last_found;
3690 }
3691 let idx_1based = match found {
3692 Some(i) => (i + 1) as i64,
3693 // (i) miss → len+1 (one past end).
3694 None => (n + 1) as i64,
3695 };
3696 return idx_1based.to_string();
3697 }
3698 }
3699 }
3700 }
3701 }
3702 if is_indexed && key.trim().parse::<i64>().is_err() {
3703 crate::ported::math::mathevali(&crate::ported::subst::singsub(&key))
3704 .map(|n| n.to_string())
3705 .unwrap_or(key.clone())
3706 } else {
3707 key.clone()
3708 }
3709 });
3710 let subscripted = format!("{}[{}]", name, resolved_key);
3711 crate::ported::params::assignsparam(&subscripted, &value, crate::ported::zsh_h::ASSPM_WARN);
3712 Value::Status(0)
3713 });
3714
3715 // Brace expansion. Routes through executor.xpandbraces (already
3716 // implemented for the pre-fusevm executor). Returns Value::Array.
3717 // BUILTIN_ARRAY_DROP_EMPTY — filter out empty Value::Str entries
3718 // from a Value::Array on the stack. Used by `for x in $@` /
3719 // `for x in $*` unquoted forms which drop empty positionals
3720 // (POSIX-like) but do NOT IFS-split each element internally
3721 // (zsh-specific — scalar word splitting is off by default).
3722 // Distinct from BUILTIN_WORD_SPLIT which routes through
3723 // multsub PREFORK_SPLIT (full IFS-split). Bug #166.
3724 vm.register_builtin(BUILTIN_ARRAY_DROP_EMPTY, |vm, _argc| {
3725 let v = vm.pop();
3726 match v {
3727 Value::Array(items) => {
3728 let filtered: Vec<Value> = items
3729 .into_iter()
3730 .filter(|x| !x.to_str().is_empty())
3731 .collect();
3732 Value::Array(filtered)
3733 }
3734 Value::Str(s) if s.is_empty() => Value::Array(Vec::new()),
3735 other => other,
3736 }
3737 });
3738
3739 // BUILTIN_QUOTEDZPUTS — re-wrap top-of-stack scalar via the
3740 // canonical quotedzputs (Src/utils.c:6464). Non-printable bytes
3741 // come back as `$'…'` C-string form so the cond xtrace prefix
3742 // line preserves the source-quoting form for `[[ -n $'\C-[OP' ]]`
3743 // instead of leaking raw ESC + "OP" bytes through the terminal.
3744 vm.register_builtin(BUILTIN_QUOTEDZPUTS, |vm, _argc| {
3745 let s = vm.pop().to_str();
3746 Value::str(crate::ported::utils::quotedzputs(&s))
3747 });
3748
3749 // BUILTIN_QUOTE_TOKENIZED_OUTPUT — char-aware mirror of
3750 // c:Src/exec.c:2114 `quote_tokenized_output`. The canonical
3751 // port at exec::quote_tokenized_output operates on bytes
3752 // (zsh's metafied encoding); zshrs strings are UTF-8 so
3753 // `\u{87}` Star is `[0xC2, 0x87]`, and a byte walk writes
3754 // 0xC2 raw (invalid UTF-8 lead → U+FFFD on lossy decode).
3755 // Walk by char and dispatch the same switch the byte port
3756 // uses, but with the token chars matching the UTF-8 form.
3757 vm.register_builtin(BUILTIN_QUOTE_TOKENIZED_OUTPUT, |vm, _argc| {
3758 let s = vm.pop().to_str();
3759 let mut out = String::with_capacity(s.len());
3760 let chars: Vec<char> = s.chars().collect();
3761 let mut i = 0;
3762 while i < chars.len() {
3763 let c = chars[i];
3764 // c:2120 — Meta-quoted byte: emit `*++s ^ 32`.
3765 // In UTF-8 strings Meta is `\u{83}`; the next char is
3766 // the metafied payload.
3767 if c == '\u{83}' {
3768 if let Some(&n) = chars.get(i + 1) {
3769 if (n as u32) < 0x80 {
3770 out.push(((n as u8) ^ 32) as char);
3771 } else {
3772 out.push(n);
3773 }
3774 i += 2;
3775 continue;
3776 }
3777 i += 1;
3778 continue;
3779 }
3780 // c:2124 — Nularg: skip.
3781 if c == '\u{a1}' {
3782 i += 1;
3783 continue;
3784 }
3785 // c:2128-2143 — ASCII specials get backslash-prefixed
3786 // then fall through to emit the literal char.
3787 match c {
3788 '\\' | '<' | '>' | '(' | '|' | ')' | '^' | '#' | '~' | '[' | ']' | '*' | '?'
3789 | '$' | ' ' => {
3790 out.push('\\');
3791 out.push(c);
3792 i += 1;
3793 continue;
3794 }
3795 '\t' => {
3796 out.push_str("$'\\t'");
3797 i += 1;
3798 continue;
3799 }
3800 '\n' => {
3801 out.push_str("$'\\n'");
3802 i += 1;
3803 continue;
3804 }
3805 '\r' => {
3806 out.push_str("$'\\r'");
3807 i += 1;
3808 continue;
3809 }
3810 '=' => {
3811 if i == 0 {
3812 out.push('\\');
3813 }
3814 out.push(c);
3815 i += 1;
3816 continue;
3817 }
3818 _ => {}
3819 }
3820 // c:2163 — `if (itok(*s)) putc(ztokens[*s - Pound]);`
3821 // Map zsh token chars (`\u{84}`..`\u{a1}` range, the
3822 // ones the lexer emits for `#$^*()…`) back to their
3823 // source ASCII via the `ztokens` table.
3824 let cp = c as u32;
3825 if (0x84..=0xa1).contains(&cp) {
3826 let idx = (cp - 0x84) as usize;
3827 let ztokens = crate::ported::lex::ztokens.as_bytes();
3828 if idx < ztokens.len() {
3829 out.push(ztokens[idx] as char);
3830 i += 1;
3831 continue;
3832 }
3833 }
3834 out.push(c);
3835 i += 1;
3836 }
3837 Value::str(out)
3838 });
3839
3840 // BUILTIN_WORD_SPLIT — `${=var}` IFS-split runtime.
3841 // PURE PASSTHRU: route through canonical `subst::multsub` with
3842 // PREFORK_SPLIT flag (C port of `Src/subst.c::multsub` at c:544
3843 // — the IFS-split walker with whitespace-vs-non-whitespace
3844 // gating, quote-aware parsing, and empty-field handling).
3845 vm.register_builtin(BUILTIN_WORD_SPLIT, |vm, _argc| {
3846 let s = vm.pop().to_str();
3847 let (_joined, parts, _isarr, _flags) =
3848 crate::ported::subst::multsub(&s, crate::ported::zsh_h::PREFORK_SPLIT);
3849 // Empty single-string special case → empty Array (drop empty arg).
3850 if parts.len() == 1 && parts[0].is_empty() {
3851 return Value::Array(Vec::new());
3852 }
3853 nodes_to_value(parts)
3854 });
3855
3856 vm.register_builtin(BUILTIN_BRACE_EXPAND, |vm, _argc| {
3857 // c:Src/glob.c::xpandbraces — brace expansion runs per word.
3858 // When the upstream produced an array (e.g. `${a:e}` splat),
3859 // expand braces on each element separately so the splat
3860 // survives. `pop().to_str()` would join with space and lose
3861 // the array shape. Parity bug #28 cousin: the BRACE_EXPAND
3862 // emit always fires for any word containing `{` (including
3863 // `${...}` param-expansion braces), so its collapse hit even
3864 // pure-paramsubst args.
3865 let raw = vm.pop();
3866 // c:Src/options.c — `no_brace_expand` (negated braceexpand)
3867 // disables brace expansion entirely. When set, `{a,b}` stays
3868 // literal. Mirror by short-circuiting xpandbraces; pass the
3869 // input through unchanged.
3870 let brace_expand = opt_state_get("braceexpand").unwrap_or(true);
3871 let brace_ccl = opt_state_get("braceccl").unwrap_or(false);
3872 let inputs: Vec<String> = match raw {
3873 Value::Array(items) => items.into_iter().map(|v| v.to_str()).collect(),
3874 other => vec![other.to_str()],
3875 };
3876 if !brace_expand {
3877 return nodes_to_value(inputs);
3878 }
3879 let mut out: Vec<String> = Vec::with_capacity(inputs.len());
3880 for s in inputs {
3881 for w in crate::ported::glob::xpandbraces(&s, brace_ccl) {
3882 out.push(w);
3883 }
3884 }
3885 nodes_to_value(out)
3886 });
3887
3888 // `*(qual)` glob qualifier filter. Stack: [pattern, qualifier].
3889 // Pattern is glob-expanded normally, then each result is filtered by the
3890 // qualifier predicate. Common qualifiers:
3891 // . — regular files only
3892 // / — directories only
3893 // @ — symlinks
3894 // x — executable
3895 // r/w/x — readable/writable/executable
3896 // N — nullglob (no error if no match)
3897 // L+N / L-N — size > N / size < N (in bytes)
3898 // mh-N / mh+N — modified within N hours / older than N hours
3899 // md-N / md+N — modified within N days / older than N days
3900 // on/On — sort by name asc/desc (default)
3901 // oL/OL — sort by length
3902 // om/Om — sort by mtime
3903 // Pop a scalar pattern, run expand_glob, push Value::Array. Used
3904 // by the segment-concat compile path for `$D/*`-style words.
3905 vm.register_builtin(BUILTIN_GLOB_EXPAND, |vm, _argc| {
3906 // c:Src/glob.c:1872 — honour `setopt noglob` / `noglob CMD`
3907 // precommand. When the option is on, the word stays literal
3908 // (zsh skips the glob expansion entirely). Without this, the
3909 // segment-fast-path BUILTIN_GLOB_EXPAND fired even after
3910 // `noglob` set the option, so `noglob echo *.xyz` saw the
3911 // NOMATCH error instead of the literal pass-through.
3912 let raw = vm.pop();
3913 let noglob =
3914 opt_state_get("noglob").unwrap_or(false) || !opt_state_get("glob").unwrap_or(true);
3915 glob_expand_word_value(raw, noglob)
3916 });
3917 // Redirect-target variant of BUILTIN_GLOB_EXPAND. c:Src/glob.c:
3918 // 2161-2167 xpandredir — `prefork(&fake, isset(MULTIOS) ? 0 :
3919 // PREFORK_SINGLE, NULL)` then "Globbing is only done for
3920 // multios.": a redirect target word is only globbed when the
3921 // MULTIOS option is set. With it unset, `echo hi > *.txt`
3922 // creates the literal file `*.txt`, and `wc -c < *.txt` errors
3923 // "no such file or directory: *.txt". Bug #36 follow-up in
3924 // docs/BUGS.md.
3925 vm.register_builtin(BUILTIN_REDIR_GLOB_EXPAND, |vm, _argc| {
3926 let raw = vm.pop();
3927 let noglob =
3928 opt_state_get("noglob").unwrap_or(false) || !opt_state_get("glob").unwrap_or(true);
3929 let multios = opt_state_get("multios").unwrap_or(true);
3930 glob_expand_word_value(raw, noglob || !multios)
3931 });
3932 // Clear the default-word glob-pending carrier before the word's
3933 // expansion runs, so a flag set by a prior word never leaks in.
3934 vm.register_builtin(BUILTIN_DEFAULT_WORD_GLOB_RESET, |_vm, _argc| {
3935 crate::ported::subst::DEFAULT_WORD_GLOB_PENDING.with(|c| c.set(false));
3936 Value::Status(0)
3937 });
3938 // After the word is assembled, run filename generation ONLY if the
3939 // default/alternate paramsubst arm flagged a source-glob default
3940 // (DEFAULT_WORD_GLOB_PENDING). Otherwise pass the word through
3941 // literally — a parameter VALUE must not glob. c:Src/subst.c globlist.
3942 vm.register_builtin(BUILTIN_DEFAULT_WORD_GLOB, |vm, _argc| {
3943 let raw = vm.pop();
3944 let pending = crate::ported::subst::DEFAULT_WORD_GLOB_PENDING.with(|c| {
3945 let v = c.get();
3946 c.set(false); // read + clear
3947 v
3948 });
3949 if !pending {
3950 return raw;
3951 }
3952 let noglob =
3953 opt_state_get("noglob").unwrap_or(false) || !opt_state_get("glob").unwrap_or(true);
3954 glob_expand_word_value(raw, noglob)
3955 });
3956
3957 // `break`/`continue` from a sub-VM body. The compile path emits
3958 // these when the keyword appears at chunk top-level (no enclosing
3959 // for/while in the current chunk's patch lists). Outer-loop
3960 // builtins (BUILTIN_RUN_SELECT and any future loop-via-builtin
3961 // construct) drain canonical BREAKS/CONTFLAG after each iteration.
3962 //
3963 // Writes match `bin_break`'s c:5836+ pattern:
3964 // continue: contflag = 1; breaks++ (Src/builtin.c::bin_break)
3965 // break: breaks++
3966 vm.register_builtin(BUILTIN_SET_BREAK, |_vm, _argc| {
3967 use std::sync::atomic::Ordering::SeqCst;
3968 crate::ported::builtin::BREAKS.fetch_add(1, SeqCst);
3969 Value::Status(0)
3970 });
3971 vm.register_builtin(BUILTIN_SET_CONTINUE, |_vm, _argc| {
3972 use std::sync::atomic::Ordering::SeqCst;
3973 crate::ported::builtin::CONTFLAG.store(1, SeqCst);
3974 crate::ported::builtin::BREAKS.fetch_add(1, SeqCst);
3975 Value::Status(0)
3976 });
3977
3978 // `${arr[*]}` — join array elements with the first IFS char into
3979 // a single string. Matches zsh: in DQ context this preserves the
3980 // join; in array context too the result is one Value::Str.
3981 // Set or clear a shell option directly. Used by `noglob CMD ...`
3982 // precommand wrapping — the compiler emits SET_RAW_OPT to flip the
3983 // option ON before compiling the inner words and OFF after, so glob
3984 // expansion of the inner args sees the temporary state.
3985 vm.register_builtin(BUILTIN_SET_RAW_OPT, |vm, _argc| {
3986 let on = vm.pop().to_int() != 0;
3987 let opt = vm.pop().to_str();
3988 // Pure passthru: canonical port lives in
3989 // src/ported/options.rs::opt_state_set_via_alias and
3990 // handles negation-alias resolution per c:Src/options.c.
3991 crate::ported::options::opt_state_set_via_alias(&opt, on);
3992 Value::Status(0)
3993 });
3994
3995 // c:Src/options.c GLOB_SUBST — runtime glob expansion of
3996 // substituted words. Pop a Value (Str or Array); when
3997 // GLOB_SUBST is ON, run expand_glob on each string element;
3998 // when OFF, pass through unchanged. Bug #119 in docs/BUGS.md.
3999 vm.register_builtin(BUILTIN_GLOB_SUBST_EXPAND, |vm, _argc| {
4000 let raw = vm.pop();
4001 let glob_subst = crate::ported::zsh_h::isset(crate::ported::zsh_h::GLOBSUBST);
4002 if !glob_subst {
4003 return raw;
4004 }
4005 // Collect input strings (Str → vec![s]; Array → multiple).
4006 let inputs: Vec<String> = match raw {
4007 Value::Array(items) => items.into_iter().map(|v| v.to_str()).collect(),
4008 other => vec![other.to_str()],
4009 };
4010 // Run expand_glob on each. Empty matches collapse to a
4011 // single literal pass-through to mirror nullglob-off default.
4012 let mut out: Vec<String> = Vec::with_capacity(inputs.len());
4013 for pattern in inputs {
4014 // c:Src/subst.c — GLOB_SUBST subjects the value to the FULL
4015 // filename-generation pipeline: `filesub` (tilde/`=` expansion)
4016 // BEFORE globbing (prefork runs filesub then globlist). zshrs
4017 // globbed but skipped filesub, so `${~x}` / `setopt globsubst`
4018 // left `~/foo` un-expanded. filesubstr matches the Tilde TOKEN,
4019 // so shtokenize the value first (`~`→Tilde, glob metas active),
4020 // run filesub, then untokenize the surviving glob metas back to
4021 // raw for expand_glob (which re-tokenizes internally).
4022 let pattern = if pattern.contains('~') || pattern.contains('=') {
4023 let mut tok = pattern.clone();
4024 crate::ported::glob::shtokenize(&mut tok);
4025 let fs = crate::ported::subst::filesub(&tok, 0);
4026 crate::ported::lex::untokenize(&fs)
4027 } else {
4028 pattern
4029 };
4030 let matches = with_executor(|exec| exec.expand_glob(&pattern));
4031 if matches.is_empty() {
4032 // No match: keep the literal (like nullglob off).
4033 out.push(pattern);
4034 } else {
4035 for m in matches {
4036 out.push(m);
4037 }
4038 }
4039 }
4040 if out.len() == 1 {
4041 Value::str(out.into_iter().next().unwrap())
4042 } else {
4043 Value::Array(out.into_iter().map(Value::str).collect())
4044 }
4045 });
4046
4047 // c:Src/math.c:337 — `getmathparam` for ArithCompiler pre-load.
4048 // Pop a variable name, return its math-coerced value. Mirrors
4049 // the routing in math::getmathparam: try i64, then f64, then
4050 // recursive arith-eval, else 0. Bug #118 in docs/BUGS.md.
4051 vm.register_builtin(BUILTIN_GET_MATH_VAR, |vm, _argc| {
4052 let name = vm.pop().to_str();
4053 let raw = crate::ported::params::getsparam(&name).unwrap_or_default();
4054 // Empty / unset → 0.
4055 if raw.is_empty() {
4056 return Value::Int(0);
4057 }
4058 // Direct int / float parse.
4059 if let Ok(n) = raw.parse::<i64>() {
4060 return Value::Int(n);
4061 }
4062 if let Ok(f) = raw.parse::<f64>() {
4063 return Value::Float(f);
4064 }
4065 // Recursive arith eval (matches getmathparam fallback at
4066 // Src/math.c:337). If that fails too, return 0 — C's
4067 // mathevall returns 0 with errflag set on parse failure.
4068 match crate::ported::math::mathevali(&raw) {
4069 Ok(n) => Value::Int(n),
4070 Err(_) => Value::Int(0),
4071 }
4072 });
4073
4074 // c:Src/options.c GLOB_SUBST + Src/cond.c:552 cond_match.
4075 // Pop pattern string; when GLOB_SUBST is OFF, escape every glob
4076 // metachar with `\` so the downstream StrMatch + patcompile
4077 // treat them as literals (matching C's tokenization-based
4078 // gate). When GLOB_SUBST is ON, pass through unchanged.
4079 // See BUILTIN_GLOB_SUBST_GUARD docs above for full rationale.
4080 vm.register_builtin(BUILTIN_GLOB_SUBST_GUARD, |vm, _argc| {
4081 let p = vm.pop().to_str();
4082 let glob_subst = crate::ported::zsh_h::isset(crate::ported::zsh_h::GLOBSUBST);
4083 if glob_subst {
4084 return Value::str(p);
4085 }
4086 let mut out = String::with_capacity(p.len() * 2);
4087 for c in p.chars() {
4088 match c {
4089 '*' | '?' | '[' | ']' | '(' | ')' | '|' | '<' | '>' | '#' | '^' | '~' | '\\' => {
4090 out.push('\\');
4091 out.push(c);
4092 }
4093 _ => out.push(c),
4094 }
4095 }
4096 Value::str(out)
4097 });
4098
4099 vm.register_builtin(BUILTIN_ARRAY_JOIN_STAR, |vm, _argc| {
4100 let name = vm.pop().to_str();
4101 let (joined, ifs_full, in_dq) = with_executor(|exec| {
4102 // c:Src/params.c — `"$*"` joins by IFS[0]. zsh
4103 // distinguishes IFS=unset (→ default `" "`) from
4104 // IFS="" (→ EMPTY separator → fields concatenate).
4105 // chars().next() collapsed both into the default, so
4106 // IFS="" was treated as IFS=" ".
4107 let ifs_full = exec.scalar("IFS").unwrap_or_else(|| " \t\n".to_string());
4108 let sep = ifs_full
4109 .chars()
4110 .next()
4111 .map(|c| c.to_string())
4112 .unwrap_or_default();
4113 let in_dq = exec.in_dq_context > 0;
4114 let joined = if name == "@" || name == "*" || name == "argv" {
4115 exec.pparams().join(&sep)
4116 } else if let Some(assoc_map) = exec.assoc(&name) {
4117 // c:Src/params.c — assoc-splat values for
4118 // `"${h[@]}"` / `"${h[*]}"`. Bug #109 in
4119 // docs/BUGS.md.
4120 assoc_map.values().cloned().collect::<Vec<_>>().join(&sep)
4121 } else if let Some(arr) = exec.array(&name) {
4122 arr.join(&sep)
4123 } else {
4124 exec.get_variable(&name)
4125 };
4126 (joined, ifs_full, in_dq)
4127 });
4128 // c:Src/subst.c — UNQUOTED `${name[*]}` (or `$*`) goes
4129 // through the canonical "join via IFS[0], then word-split
4130 // via IFS" pipeline. The fast-path bypassed paramsubst
4131 // entirely so it never word-split, producing one joined
4132 // string instead of N argv entries. Bug #428.
4133 //
4134 // In QUOTED (`"${name[*]}"`) context, the result IS a
4135 // single scalar — return it as Str without splitting.
4136 if in_dq {
4137 return Value::str(joined);
4138 }
4139 if joined.is_empty() {
4140 return Value::Array(Vec::new());
4141 }
4142 // IFS word-split — every IFS char is a separator. Empty
4143 // resulting fields are dropped (the canonical
4144 // "remove empty unquoted words" pass from
4145 // Src/subst.c::prefork c:184-187).
4146 let parts: Vec<String> = joined
4147 .split(|c: char| ifs_full.contains(c))
4148 .filter(|s| !s.is_empty())
4149 .map(String::from)
4150 .collect();
4151 if parts.is_empty() {
4152 Value::Array(Vec::new())
4153 } else if parts.len() == 1 {
4154 Value::str(parts.into_iter().next().unwrap())
4155 } else {
4156 Value::Array(parts.into_iter().map(Value::str).collect())
4157 }
4158 });
4159
4160 vm.register_builtin(BUILTIN_ARRAY_ALL, |vm, _argc| {
4161 let name = vm.pop().to_str();
4162 with_executor(|exec| {
4163 // Special positional names — splice the positional list.
4164 if name == "@" || name == "*" || name == "argv" {
4165 return Value::Array(exec.pparams().iter().map(Value::str).collect());
4166 }
4167 // c:Src/Modules/parameter.c — funcstack/funcfiletrace/
4168 // funcsourcetrace/functrace are PM_ARRAY|PM_READONLY
4169 // specials backed by the canonical FUNCSTACK Vec.
4170 // `${funcstack[@]}` inside a function call should splat
4171 // the innermost-first names; without this branch the
4172 // runtime fell to the scalar fallback (get_variable
4173 // returns empty for these specials) and `[@]` came out
4174 // empty. Bug #276 in docs/BUGS.md. Mirrors the parallel
4175 // arrays_get handler at src/ported/subst.rs ~10685.
4176 // c:Src/Modules/datetime.c:256 — `epochtime` PM_ARRAY|
4177 // PM_READONLY backed by getcurrenttime(). Same parallel
4178 // arrangement as the FUNCSTACK-backed specials below.
4179 if name == "epochtime" {
4180 let arr = crate::ported::modules::datetime::getcurrenttime();
4181 return Value::Array(arr.into_iter().map(Value::str).collect());
4182 }
4183 if matches!(
4184 name.as_str(),
4185 "funcstack" | "funcfiletrace" | "funcsourcetrace" | "functrace"
4186 ) {
4187 // Route the three trace arrays through the canonical
4188 // ported getfns (Src/Modules/parameter.c:648/:679/:711)
4189 // — the previous inline copy emitted wrong shapes
4190 // (bare filename for funcfiletrace, `name:lineno` for
4191 // functrace instead of `caller:lineno`); same dedup as
4192 // the parallel arrays_get handler in subst.rs.
4193 let vals: Vec<String> = match name.as_str() {
4194 "funcstack" => crate::ported::modules::parameter::FUNCSTACK
4195 .lock()
4196 .map(|f| f.iter().rev().map(|fs| fs.name.clone()).collect())
4197 .unwrap_or_default(),
4198 "funcfiletrace" => crate::ported::modules::parameter::funcfiletracegetfn(
4199 std::ptr::null_mut(),
4200 ),
4201 "funcsourcetrace" => crate::ported::modules::parameter::funcsourcetracegetfn(
4202 std::ptr::null_mut(),
4203 ),
4204 _ => crate::ported::modules::parameter::functracegetfn(std::ptr::null_mut()),
4205 };
4206 return Value::Array(vals.into_iter().map(Value::str).collect());
4207 }
4208 // c:Src/params.c — `${assoc[@]}` enumerates VALUES (per
4209 // params.c:1696-1750 hashparam splat). Check assoc
4210 // storage BEFORE the scalar fallback so an associative
4211 // array named X resolves `${X[@]}` to the values, not
4212 // empty. Bug #109 in docs/BUGS.md: `${h[@]}` on an
4213 // assoc routed through BUILTIN_ARRAY_ALL, which only
4214 // consulted `exec.array(name)` (the indexed-array map)
4215 // — that lookup missed for assocs, fell through to
4216 // `get_variable("h")` (also empty for an assoc-only
4217 // name), and returned `Array(vec![])`. zsh's expected
4218 // behavior is to enumerate values.
4219 if let Some(assoc_map) = exec.assoc(&name) {
4220 return Value::Array(
4221 assoc_map.values().cloned().map(Value::str).collect(),
4222 );
4223 }
4224 match exec.array(&name) {
4225 Some(v) => Value::Array(v.iter().map(Value::str).collect()),
4226 None => {
4227 // Fall back to scalar lookup. zsh (unlike bash)
4228 // does NOT IFS-split a scalar variable in a for
4229 // list — `for w in $scalar` iterates ONCE with the
4230 // scalar value. Word-splitting requires either
4231 // sh_word_split option or explicit `${(s.,.)scalar}`.
4232 let val = exec.get_variable(&name);
4233 if val.is_empty() && !exec.has_scalar(&name) && env::var(&name).is_err() {
4234 // c:Src/subst.c:3480-3485 — `${arr[@]}` on a genuinely
4235 // UNSET parameter under NO_UNSET is a "parameter not set"
4236 // error (vunset > 0 && unset(UNSET)), exactly like the
4237 // scalar `$arr`, the `${arr[*]}` splat, and `${arr[1]}`
4238 // — all of which already fire it via GET_VAR. The `[@]`
4239 // splat path returned an empty array silently, so
4240 // `setopt NO_UNSET; print "${arr[@]}"` exited 0 where zsh
4241 // exits 1. A DECLARED-but-empty array (`arr=()`) resolves
4242 // to `Some(vec![])` above and never reaches here, so it
4243 // still splats to nothing without erroring — matching zsh.
4244 if opt_state_get("nounset").unwrap_or(false) {
4245 crate::ported::utils::zerr(&format!("{}: parameter not set", name));
4246 crate::ported::utils::errflag.fetch_or(
4247 crate::ported::zsh_h::ERRFLAG_ERROR,
4248 std::sync::atomic::Ordering::Relaxed,
4249 );
4250 exec.set_last_status(1);
4251 }
4252 Value::Array(vec![])
4253 } else if opt_state_get("shwordsplit").unwrap_or(false) {
4254 // bash-compat: under setopt sh_word_split, do
4255 // split scalars on IFS chars.
4256 let ifs = exec.scalar("IFS").unwrap_or_else(|| " \t\n".to_string());
4257 let parts: Vec<Value> = val
4258 .split(|c: char| ifs.contains(c))
4259 .filter(|s| !s.is_empty())
4260 .map(Value::str)
4261 .collect();
4262 Value::Array(parts)
4263 } else {
4264 Value::Array(vec![Value::str(val)])
4265 }
4266 }
4267 }
4268 })
4269 });
4270
4271 // BUILTIN_ARRAY_FLATTEN(N): pops N values, flattens one level of Array
4272 // nesting, pushes the resulting Array AND its length as a separate Int.
4273 // The two-value return shape lets the caller (for-loop compile path)
4274 // SetSlot the length before SetSlot'ing the array, without re-deriving
4275 // the length from the array via a second builtin call.
4276 // `coproc [name] { body }` — bidirectional pipe to backgrounded body.
4277 // Stack discipline (top first): [name (str, "" for default), sub_idx (int)].
4278 // On success: parent's `executor.arrays[name]` becomes [write_fd, read_fd]
4279 // and Status(0) is returned. The caller writes to the child's stdin via
4280 // write_fd, reads its stdout via read_fd, and closes both when done.
4281 //
4282 // Bash's coproc convention is `${NAME[0]}` = read_fd, `${NAME[1]}` =
4283 // write_fd. We follow that: arrays[name] = [read_fd_str, write_fd_str].
4284 vm.register_builtin(BUILTIN_RUN_COPROC, |vm, _argc| {
4285 let sub_idx = vm.pop().to_int() as usize;
4286 let job_text = vm.pop().to_str();
4287 let raw_name = vm.pop().to_str();
4288 let name = if raw_name.is_empty() {
4289 "COPROC".to_string()
4290 } else {
4291 raw_name
4292 };
4293 let chunk = match vm.chunk.sub_chunks.get(sub_idx).cloned() {
4294 Some(c) => c,
4295 None => return Value::Status(1),
4296 };
4297
4298 // c:Src/exec.c:1710-1712 — starting a new coproc closes the
4299 // previous one's fds FIRST:
4300 // if (coprocin >= 0) { zclose(coprocin); zclose(coprocout); }
4301 // The old coproc child then sees EOF on its stdin and exits on
4302 // its own schedule (its job-table entry stays until it's
4303 // reaped) — zsh does NOT deletejob it here. This is also what
4304 // makes the `exec 4<&p; coproc exit; read -u4` EOF idiom work:
4305 // the replacement coproc closes the shell's write end to the
4306 // old one.
4307 {
4308 use std::sync::atomic::Ordering;
4309 let old_in = crate::ported::modules::clone::coprocin.load(Ordering::Relaxed);
4310 if old_in >= 0 {
4311 let old_out = crate::ported::modules::clone::coprocout.load(Ordering::Relaxed);
4312 unsafe {
4313 libc::close(old_in);
4314 if old_out >= 0 {
4315 libc::close(old_out);
4316 }
4317 }
4318 crate::ported::modules::clone::coprocin.store(-1, Ordering::Relaxed);
4319 crate::ported::modules::clone::coprocout.store(-1, Ordering::Relaxed);
4320 }
4321 }
4322
4323 // (parent_read ← child_stdout)
4324 let mut p2c = [0i32; 2]; // parent writes, child reads
4325 let mut c2p = [0i32; 2]; // child writes, parent reads
4326 if unsafe { libc::pipe(p2c.as_mut_ptr()) } < 0 {
4327 return Value::Status(1);
4328 }
4329 if unsafe { libc::pipe(c2p.as_mut_ptr()) } < 0 {
4330 unsafe {
4331 libc::close(p2c[0]);
4332 libc::close(p2c[1]);
4333 }
4334 return Value::Status(1);
4335 }
4336 // c:Src/exec.c:5160 mpipe — both pipes' fds are moved above
4337 // the user-visible range (movefd → F_DUPFD ≥ 10) so the
4338 // coproc fds never collide with explicit user fds like
4339 // `exec 3>&p`.
4340 for fd in p2c.iter_mut().chain(c2p.iter_mut()) {
4341 *fd = crate::ported::utils::movefd(*fd);
4342 }
4343
4344 match unsafe { libc::fork() } {
4345 -1 => {
4346 unsafe {
4347 libc::close(p2c[0]);
4348 libc::close(p2c[1]);
4349 libc::close(c2p[0]);
4350 libc::close(c2p[1]);
4351 }
4352 Value::Status(1)
4353 }
4354 0 => {
4355 // Child: stdin from p2c[0], stdout to c2p[1]. Close all
4356 // unused fds. setsid so SIGINT to fg doesn't hit us.
4357 unsafe {
4358 libc::dup2(p2c[0], libc::STDIN_FILENO);
4359 libc::dup2(c2p[1], libc::STDOUT_FILENO);
4360 libc::close(p2c[0]);
4361 libc::close(p2c[1]);
4362 libc::close(c2p[0]);
4363 libc::close(c2p[1]);
4364 libc::setsid();
4365 }
4366 crate::fusevm_disasm::maybe_print_stdout("coproc:child", &chunk);
4367 let mut co_vm = fusevm::VM::new(chunk);
4368 register_builtins(&mut co_vm);
4369 let _ = co_vm.run();
4370 let _ = std::io::stdout().flush();
4371 let _ = std::io::stderr().flush();
4372 std::process::exit(co_vm.last_status);
4373 }
4374 pid => {
4375 // Parent: close child ends, store [read_fd, write_fd] in NAME.
4376 unsafe {
4377 libc::close(p2c[0]);
4378 libc::close(c2p[1]);
4379 }
4380 let read_fd = c2p[0];
4381 let write_fd = p2c[1];
4382 with_executor(|exec| {
4383 exec.unset_scalar(&name);
4384 exec.set_array(name, vec![read_fd.to_string(), write_fd.to_string()]);
4385 });
4386 // c:Src/exec.c — `coprocin`/`coprocout` are the
4387 // canonical globals that bin_read's `-p` arm
4388 // (Src/builtin.c:6510) and bin_print's `-p` arm
4389 // (Src/builtin.c:4827) read to find the
4390 // coprocess fds. The Rust port has the atomic
4391 // declarations at src/ported/modules/clone.rs:262
4392 // but the coproc-launch path never updated them,
4393 // so `read -p` / `print -p` always errored with
4394 // "-p: no coprocess" even when a coproc was
4395 // running. Bug #388 in docs/BUGS.md. Update them
4396 // here so the canonical builtins find the live
4397 // pipe.
4398 crate::ported::modules::clone::coprocin
4399 .store(read_fd, std::sync::atomic::Ordering::Relaxed);
4400 crate::ported::modules::clone::coprocout
4401 .store(write_fd, std::sync::atomic::Ordering::Relaxed);
4402 // c:Src/exec.c:1725 — `fdtable[coprocin] =
4403 // fdtable[coprocout] = FDT_UNUSED;`: the two kept ends
4404 // are user-reachable (via `>&p` / `<&p`), so they drop
4405 // the FDT_INTERNAL mark movefd gave them.
4406 crate::ported::utils::fdtable_set(read_fd, crate::ported::zsh_h::FDT_UNUSED);
4407 crate::ported::utils::fdtable_set(write_fd, crate::ported::zsh_h::FDT_UNUSED);
4408 // c:Src/exec.c:2837 — `lastpid = (zlong) pid;`. zsh
4409 // sets the `$!` global to the coproc child's PID so
4410 // subsequent `$!` reads return it. The Rust port at
4411 // exec.rs:6773 mirrors this for regular background
4412 // jobs but the coproc launch path was missing the
4413 // assignment, leaving `$!` at 0 after `coproc cmd`.
4414 crate::ported::modules::clone::lastpid
4415 .store(pid, std::sync::atomic::Ordering::Relaxed);
4416 // c:Src/exec.c:1700-1758 — the coproc rides the SAME
4417 // Z_ASYNC job-table path as `cmd &`: `thisjob = newjob
4418 // = initjob()` (c:1700), addproc hangs the pid+text
4419 // proc entry off the job, `jobtab[thisjob].stat |=
4420 // STAT_NOSTTY` (c:1746), `clearoldjobtab()` (c:1744)
4421 // and `spawnjob()` (c:1758) promote it to curjob. This
4422 // is what makes `jobs` list the coproc as
4423 // `[1] + running cat` and `kill %1` resolve it.
4424 // Mirrors the BUILTIN_RUN_BG parent arm exactly.
4425 {
4426 use crate::ported::jobs;
4427 use std::sync::Mutex;
4428 let table = jobs::JOBTAB.get_or_init(|| Mutex::new(Vec::new()));
4429 let idx = {
4430 let mut tab = table.lock().unwrap_or_else(|e| e.into_inner());
4431 let idx = jobs::initjob(&mut tab); // c:exec.c:1700
4432 jobs::addproc(
4433 &mut tab[idx],
4434 pid,
4435 &job_text,
4436 false,
4437 Some(std::time::Instant::now()),
4438 -1,
4439 -1,
4440 );
4441 tab[idx].stat |= crate::ported::zsh_h::STAT_NOSTTY; // c:exec.c:1746
4442 idx
4443 };
4444 jobs::clearoldjobtab(); // c:exec.c:1744
4445 if let Ok(mut tj) = jobs::THISJOB.get_or_init(|| Mutex::new(-1)).lock() {
4446 *tj = idx as i32;
4447 }
4448 jobs::spawnjob(); // c:exec.c:1758
4449 }
4450 with_executor(|exec| {
4451 exec.jobs
4452 .add_pid_job(pid, job_text.clone(), JobState::Running);
4453 });
4454 Value::Status(0)
4455 }
4456 }
4457 });
4458
4459 vm.register_builtin(BUILTIN_ARRAY_FLATTEN, |vm, argc| {
4460 let n = argc as usize;
4461 let start = vm.stack.len().saturating_sub(n);
4462 let raw: Vec<Value> = vm.stack.drain(start..).collect();
4463 let mut flat: Vec<Value> = Vec::with_capacity(raw.len());
4464 for v in raw {
4465 match v {
4466 Value::Array(items) => flat.extend(items),
4467 other => flat.push(other),
4468 }
4469 }
4470 let len = flat.len() as i64;
4471 // Push the array first; the Int(len) becomes the builtin's return
4472 // value (which CallBuiltin already pushes). Caller consumes in
4473 // reverse: SetSlot(len_slot) pops Int, SetSlot(arr_slot) pops Array.
4474 vm.push(Value::Array(flat));
4475 Value::Int(len)
4476 });
4477
4478 // Shell variable get/set — routes through executor.variables so nested
4479 // VMs (function calls) and tree-walker callers see the same storage.
4480 // GET_VAR / GET_VAR_DQ share one body via `get_var_impl`; the only
4481 // difference is `force_dq`, which the compiler sets for QUOTED simple
4482 // reads (`"$name"`) so an array's empty elements are preserved (the
4483 // `in_dq_context` runtime flag is 0 for these compiler-direct reads).
4484 fn get_var_impl(vm: &mut fusevm::VM, argc: u8, force_dq: bool) -> Value {
4485 let args = pop_args(vm, argc);
4486 let name = args.into_iter().next().unwrap_or_default();
4487 let live_status = vm.last_status;
4488 // Suppress sync when a deferred subshell exit just landed:
4489 // LASTVAL holds the correct deferred status, vm.last_status
4490 // is stale (post-subshell vm doesn't propagate status). See
4491 // SUBSHELL_EXIT_STATUS_PENDING TLS declaration for rationale.
4492 let suppress_sync = SUBSHELL_EXIT_STATUS_PENDING.with(|c| {
4493 let prev = c.get();
4494 c.set(false);
4495 prev
4496 });
4497 // `$@` and `$*` need splice semantics — return Value::Array of
4498 // positional params so for-loop's BUILTIN_ARRAY_FLATTEN spreads them
4499 // and pop_args splits them into argv slots. zsh's `"$@"` bslashquote-each-
4500 // word semantics matches: each pos-param becomes its own arg.
4501 // Same for arrays accessed by name (e.g. `$arr` in some contexts).
4502 let sync_status = |exec: &mut ShellExecutor| {
4503 if !suppress_sync {
4504 exec.set_last_status(live_status);
4505 }
4506 };
4507 if name == "@" || name == "*" {
4508 // Quoting decides empty-word retention (c:Src/subst.c:
4509 // 184-187): the COMPILE site knows it and emits
4510 // BUILTIN_ARRAY_DROP_EMPTY after this read for the
4511 // unquoted form only — in_dq_context is NOT a valid
4512 // discriminator here (the quoted "$@" fast path emits
4513 // GET_VAR directly without an EXPAND_TEXT wrapper).
4514 return with_executor(|exec| {
4515 sync_status(exec);
4516 Value::Array(exec.pparams().iter().map(Value::str).collect())
4517 });
4518 }
4519 // RC_EXPAND_PARAM: when the option is set and `name` refers to
4520 // an array, return Value::Array so the enclosing word's
4521 // BUILTIN_CONCAT_DISTRIBUTE distributes element-wise. Without
4522 // the option, arrays still join to a space-separated scalar
4523 // (zsh's default unquoted-array-as-scalar semantics).
4524 let rc_expand = with_executor(|exec| opt_state_get("rcexpandparam").unwrap_or(false));
4525 // c:Src/subst.c — under KSHARRAYS a bare `$name` (no [@]/[*] subscript;
4526 // this GET_VAR path only handles the bare form) is element 1 ONLY — a
4527 // scalar. RC_EXPAND_PARAM then has a single value to distribute, so
4528 // `$acc` → "p1", NOT the whole array. Skip the whole-array rc_expand
4529 // shortcut when KSHARRAYS is set and fall through to the normal path,
4530 // which applies the element-1 collapse. Without this gate,
4531 // `setopt KSH_ARRAYS rc_expand_param; print -r -- $acc` splatted every
4532 // element while zsh prints just "p1".
4533 let ksh_arrays = crate::ported::zsh_h::isset(crate::ported::zsh_h::KSHARRAYS);
4534 if rc_expand && !ksh_arrays {
4535 let arr_val = with_executor(|exec| {
4536 sync_status(exec);
4537 exec.array(&name)
4538 });
4539 if let Some(arr) = arr_val {
4540 return Value::Array(arr.into_iter().map(Value::str).collect());
4541 }
4542 }
4543 // Magic-assoc fallback FIRST — `${aliases}` / `${functions}`
4544 // / `${commands}` / etc. should return the value list per
4545 // zsh's bare-assoc semantics. Without this, those names fell
4546 // through to `get_variable` which is empty (they live in
4547 // separate executor tables, not `assoc_arrays`). Return as
4548 // a Value::Array so `arr=(${aliases})` distributes into
4549 // multiple elements, matching zsh's array-context word
4550 // splitting for assoc-bare references.
4551 let magic_vals = with_executor(|exec| {
4552 sync_status(exec);
4553 // Canonical PARTAB dispatch (Src/Modules/parameter.c:2235-
4554 // 2298 + SPECIALPMDEFs in mapfile/terminfo/termcap/system/
4555 // zleparameter): PARTAB_ARRAY entries → whole-array getfn;
4556 // PARTAB entries → scan keys + per-key getpm/scanpm fn
4557 // pointers.
4558 let _ = exec;
4559 if let Some(values) = partab_array_get(&name) {
4560 Some(values)
4561 } else if let Some(keys) = partab_scan_keys(&name) {
4562 Some(
4563 keys.iter()
4564 .map(|k| partab_get(&name, k).unwrap_or_default())
4565 .collect::<Vec<_>>(),
4566 )
4567 } else {
4568 None
4569 }
4570 });
4571 if let Some(vals) = magic_vals {
4572 // Distinguish "name IS a magic-assoc with no entries"
4573 // (return Array(empty)) from "name is unknown — fall
4574 // through to get_variable".
4575 // c:Src/params.c:2293-2296 — KSHARRAYS bare reference
4576 // collapses to the FIRST element in scan order
4577 // (`v->end = 1, v->isarr = 0`). For `options` the scan
4578 // order is optiontab bucket order (OPTIONTAB), so zsh 5.9
4579 // prints `off` (posixargzero) for
4580 // `setopt ksharrays; print $options`.
4581 if opt_state_get("ksharrays").unwrap_or(false) {
4582 return Value::str(vals.into_iter().next().unwrap_or_default());
4583 }
4584 return Value::Array(vals.into_iter().map(Value::str).collect());
4585 }
4586 // Indexed-array path: return Value::Array so pop_args splats
4587 // each element into its own argv slot. Direct port of zsh's
4588 // unquoted `$arr` semantics — each element becomes a separate
4589 // word in command-arg position.
4590 //
4591 // DQ context exception: inside `"...$arr..."`, zsh joins with
4592 // the first char of $IFS (default space) so the DQ word stays
4593 // a single argv slot. Detect via in_dq_context (bumped by
4594 // BUILTIN_EXPAND_TEXT mode 1) and return the joined scalar.
4595 // Direct port of Src/subst.c:1759-1813 nojoin/sepjoin: in DQ
4596 // (qt=1) without explicit `(@)`, sepjoin runs and the result
4597 // is one word.
4598 let arr_assoc_data = with_executor(|exec| {
4599 sync_status(exec);
4600 let in_dq = force_dq || exec.in_dq_context > 0;
4601 // KSH_ARRAYS: bare `$arr` returns ONLY arr[0] (zero-
4602 // based first-element-only semantics). Direct port of
4603 // Src/params.c getstrvalue's KSH_ARRAYS gate which
4604 // returns aval[0] instead of the whole array.
4605 let ksh_arrays = opt_state_get("ksharrays").unwrap_or(false);
4606 if let Some(arr) = exec.array(&name) {
4607 if ksh_arrays {
4608 return Some((vec![arr.first().cloned().unwrap_or_default()], in_dq));
4609 }
4610 return Some((arr.clone(), in_dq));
4611 }
4612 if exec.assoc(&name).is_some() {
4613 // c:Src/hashtable.c scanhashtable — a bare `$assoc` joins its
4614 // VALUES in zsh hash-BUCKET order (the same order `(k)`/`(v)`
4615 // enumerate), NOT sorted or insertion order. `assoc_get`
4616 // rebuilds zsh's bucket layout; use it so `$as` matches
4617 // `${(v)as}` (`as=(zebra 9 apple 1)` → `9 1`, not the
4618 // alphabetical `1 9`). Under KSHARRAYS the bare form collapses
4619 // to the bucket-FIRST value (`9`), matching zsh.
4620 let values: Vec<String> = crate::ported::subst::assoc_get(&name)
4621 .map(|m| m.values().cloned().collect())
4622 .unwrap_or_default();
4623 if ksh_arrays {
4624 return Some((vec![values.into_iter().next().unwrap_or_default()], in_dq));
4625 }
4626 return Some((values, in_dq));
4627 }
4628 None
4629 });
4630 if let Some((items, in_dq)) = arr_assoc_data {
4631 // c:Src/subst.c:184-187 — prefork's `else if (!keep)
4632 // uremnode(list, node)`: UNQUOTED expansion drops empty
4633 // list nodes before they reach argv, so `a=(y '' x);
4634 // print -- $a` passes TWO args in zsh (`y x`), while the
4635 // quoted "${a[@]}" splat keeps the empty slot. The
4636 // paramsubst splat path already does this (Bug #578
4637 // retain); this GET_VAR fast path bypassed it and leaked
4638 // empty argv slots (visible double-space, wrong arg
4639 // counts in `for`/`print -l`).
4640 let items: Vec<String> = if in_dq {
4641 items
4642 } else {
4643 items.into_iter().filter(|s| !s.is_empty()).collect()
4644 };
4645 if in_dq {
4646 // c:Src/utils.c:3936-3945 sepjoin default-sep rule:
4647 // set-but-empty IFS joins with "" (`IFS=""; echo
4648 // "$arr"` concatenates); only unset / space-leading
4649 // IFS yields " ". The previous get_variable read
4650 // couldn't distinguish unset from set-empty.
4651 return Value::str(crate::ported::utils::sepjoin(&items, None));
4652 }
4653 return Value::Array(items.into_iter().map(Value::str).collect());
4654 }
4655 let (val, in_dq, is_known) = with_executor(|exec| {
4656 sync_status(exec);
4657 let v = exec.get_variable(&name);
4658 // For nounset detection: a name is "known" when it has a
4659 // paramtab/array/assoc/env entry. Special chars ($?, $#,
4660 // $@, $*, $-, $$, $!, $_, $0) always count as known
4661 // regardless of value. Pure-digit positional params
4662 // count as known iff index <= $# (set -- has populated
4663 // that slot). c:Src/subst.c:1689 — NOUNSET fires on
4664 // unset positional param too: `set --; echo "$1"` with
4665 // nounset must diagnose.
4666 let is_special_single = name.len() == 1
4667 && matches!(
4668 name.chars().next().unwrap(),
4669 '?' | '#' | '@' | '*' | '-' | '$' | '!' | '_' | '0'
4670 );
4671 let is_pure_digit = !name.is_empty() && name.chars().all(|c| c.is_ascii_digit());
4672 let positional_known = if is_pure_digit {
4673 let idx: usize = name.parse().unwrap_or(0);
4674 if idx == 0 {
4675 true // $0 always set
4676 } else {
4677 idx <= exec.pparams().len()
4678 }
4679 } else {
4680 false
4681 };
4682 let known = !v.is_empty()
4683 || name.is_empty()
4684 || is_special_single
4685 || positional_known
4686 || crate::ported::params::paramtab()
4687 .read()
4688 .ok()
4689 .map(|t| t.contains_key(&name))
4690 .unwrap_or(false)
4691 || env::var(&name).is_ok();
4692 (v, force_dq || exec.in_dq_context > 0, known)
4693 });
4694 // c:Src/subst.c:1689 — NO_UNSET / nounset: reading an unset
4695 // parameter fires "parameter not set" diagnostic and aborts
4696 // the substitution. Direct port of the noerrs gate at c:1689
4697 // (zerr + errflag). Matches `set -u` POSIX semantics.
4698 if !is_known && opt_state_get("nounset").unwrap_or(false) {
4699 crate::ported::utils::zerr(&format!("{}: parameter not set", name));
4700 crate::ported::utils::errflag.fetch_or(
4701 crate::ported::zsh_h::ERRFLAG_ERROR,
4702 std::sync::atomic::Ordering::Relaxed,
4703 );
4704 with_executor(|exec| exec.set_last_status(1));
4705 return Value::str("");
4706 }
4707 // Empty unquoted scalar → drop the arg (zsh "remove empty
4708 // unquoted words" rule). Returning empty Value::Array makes
4709 // pop_args contribute zero items. DQ context keeps the empty
4710 // string so "$a" stays a single empty arg. Direct port of
4711 // subst.c's elide-empty pass.
4712 if val.is_empty() && !in_dq {
4713 return Value::Array(Vec::new());
4714 }
4715 // c:Src/subst.c:1759 SH_WORD_SPLIT — when shwordsplit is set and
4716 // we're in unquoted command-arg position (not DQ), split scalar
4717 // value on IFS into multiple words. Matches BUILTIN_ARRAY_ALL's
4718 // shwordsplit arm (fusevm_bridge.rs:2200). Without this, bare
4719 // `$s` in `print $s` stayed a single arg even with the option
4720 // set, breaking POSIX-style scalar word-splitting.
4721 if !in_dq && opt_state_get("shwordsplit").unwrap_or(false) {
4722 let ifs =
4723 with_executor(|exec| exec.scalar("IFS").unwrap_or_else(|| " \t\n".to_string()));
4724 let parts: Vec<Value> = val
4725 .split(|c: char| ifs.contains(c))
4726 .filter(|s| !s.is_empty())
4727 .map(|s| Value::str(s.to_string()))
4728 .collect();
4729 if parts.is_empty() {
4730 return Value::Array(Vec::new());
4731 } else if parts.len() == 1 {
4732 return parts.into_iter().next().unwrap();
4733 } else {
4734 return Value::Array(parts);
4735 }
4736 }
4737 Value::str(val)
4738 }
4739 vm.register_builtin(BUILTIN_GET_VAR, |vm, argc| get_var_impl(vm, argc, false));
4740 vm.register_builtin(BUILTIN_GET_VAR_DQ, |vm, argc| get_var_impl(vm, argc, true));
4741
4742 // `name+=val` (no parens) — runtime dispatch:
4743 // - if `name` is in `arrays` → push `val` as new element
4744 // - if `name` is in `assoc_arrays` → refuse (zsh errors here)
4745 // - else → scalar concat (existing behavior)
4746 // Stack: [name, value].
4747 vm.register_builtin(BUILTIN_APPEND_SCALAR_OR_PUSH, |vm, argc| {
4748 let args = pop_args(vm, argc);
4749 let mut iter = args.into_iter();
4750 let name = iter.next().unwrap_or_default();
4751 let value = iter.next().unwrap_or_default();
4752 with_executor(|exec| {
4753 // Array form: `arr+=elem` pushes a single element.
4754 // Routes through canonical assignaparam(name, [value],
4755 // ASSPM_AUGMENT) — Src/params.c:3357 c:3402-3412 augment
4756 // path prepends prior scalar / appends to existing array.
4757 if exec.array(&name).is_some() {
4758 // c:Src/params.c — under KSHARRAYS a bare array name
4759 // addresses element 0 (ksh), so `a+=X` (scalar augment)
4760 // CONCATENATES onto the first element ("firstlast second"),
4761 // it does NOT push a new element. C routes scalar `+=`
4762 // through assignsparam (which targets the elem-0 value);
4763 // zshrs's APPEND_SCALAR_OR_PUSH would otherwise push.
4764 if crate::ported::zsh_h::isset(crate::ported::zsh_h::KSHARRAYS) {
4765 let mut arr = exec.array(&name).unwrap_or_default();
4766 if arr.is_empty() {
4767 arr.push(value.clone());
4768 } else {
4769 arr[0] = format!("{}{}", arr[0], value);
4770 }
4771 exec.set_array(name.clone(), arr);
4772 #[cfg(feature = "recorder")]
4773 if crate::recorder::is_enabled() && exec.local_scope_depth == 0 {
4774 let ctx = exec.recorder_ctx();
4775 let attrs = exec.recorder_attrs_for(&name);
4776 emit_path_or_assign(&name, std::slice::from_ref(&value), attrs, true, &ctx);
4777 }
4778 return;
4779 }
4780 let _ = crate::ported::params::assignaparam(
4781 &name,
4782 vec![value.clone()],
4783 crate::ported::zsh_h::ASSPM_AUGMENT,
4784 );
4785 #[cfg(feature = "recorder")]
4786 if crate::recorder::is_enabled() && exec.local_scope_depth == 0 {
4787 let ctx = exec.recorder_ctx();
4788 let attrs = exec.recorder_attrs_for(&name);
4789 emit_path_or_assign(&name, std::slice::from_ref(&value), attrs, true, &ctx);
4790 }
4791 return;
4792 }
4793 if exec.assoc(&name).is_some() {
4794 eprintln!("zshrs: {}: cannot use += on assoc without (key val)", name);
4795 return;
4796 }
4797 // Scalar / integer / float form: route through canonical
4798 // assignsparam(name, value, ASSPM_AUGMENT) which
4799 // dispatches PM_TYPE — PM_SCALAR concats, PM_INTEGER
4800 // arith-adds (c:2775-2778), PM_FLOAT float-adds.
4801 let _ = crate::ported::params::assignsparam(
4802 &name,
4803 &value,
4804 crate::ported::zsh_h::ASSPM_AUGMENT,
4805 );
4806 #[cfg(feature = "recorder")]
4807 if crate::recorder::is_enabled() && exec.local_scope_depth == 0 {
4808 let ctx = exec.recorder_ctx();
4809 let attrs = exec.recorder_attrs_for(&name);
4810 // Re-read the canonical value via get_variable for the
4811 // recorder bundle (assignsparam may have transformed it
4812 // through integer/float arithmetic).
4813 let final_val = exec.get_variable(&name);
4814 let lower = name.to_ascii_lowercase();
4815 if matches!(
4816 lower.as_str(),
4817 "path" | "fpath" | "manpath" | "module_path" | "cdpath"
4818 ) {
4819 emit_path_or_assign(&name, std::slice::from_ref(&final_val), attrs, true, &ctx);
4820 } else {
4821 crate::recorder::emit_assign_typed(&name, &final_val, attrs, ctx);
4822 }
4823 }
4824 });
4825 Value::Status(0)
4826 });
4827
4828 // BUILTIN_SET_VAR — `name=value` runtime scalar assignment.
4829 // PURE PASSTHRU: hand to canonical `setsparam` (C port of
4830 // `Src/params.c::setsparam`). That walks assignsparam →
4831 // assignstrvalue which already does:
4832 // - readonly rejection (zerr + errflag at c:2701)
4833 // - PM_INTEGER math evaluation (mathevali at c:3590)
4834 // - PM_EFLOAT / PM_FFLOAT float coercion (c:3608)
4835 // - PM_LOWER / PM_UPPER case fold (via setstrvalue)
4836 // - GSU special-param dispatch (homesetfn / ifssetfn / etc.)
4837 // - allexport env mirror via the PM_EXPORTED setfn
4838 //
4839 // Bridge-only concerns kept here:
4840 // - inline_env_stack (zsh `X=foo cmd` scoped env)
4841 // - recorder emission (PFA-SMR)
4842 // - vm.last_status propagation for `a=$(cmd)` exit-code chaining
4843 // Sets the GLOB_ASSIGN-eligibility flag consumed by the NEXT BUILTIN_SET_VAR.
4844 // Emitted only when a scalar-assign RHS had an unquoted glob token. Takes no
4845 // args; its pushed return is discarded by a following Op::Pop.
4846 vm.register_builtin(BUILTIN_MARK_GLOB_ELIGIBLE, |_vm, _argc| {
4847 SET_VAR_GLOB_ELIGIBLE.with(|c| c.set(true));
4848 fusevm::Value::Int(0)
4849 });
4850 vm.register_builtin(BUILTIN_SET_VAR, |vm, argc| {
4851 // `${~spec}` carrier: an assignment statement is a word-
4852 // pipeline boundary too — restore the user's GLOB_SUBST
4853 // before the NEXT word expands (`Z[d]=${~Z[d]}; print
4854 // ${options[globsubst]}` must read the user value).
4855 consume_tilde_globsubst_carrier();
4856 // Snapshot the raw Values BEFORE pop_args's to_str
4857 // flattening — needed to distinguish Int (arith assignment,
4858 // integer-typed param) from Str (scalar assignment).
4859 let mut raw_values: Vec<fusevm::Value> = Vec::with_capacity(argc as usize);
4860 for _ in 0..argc {
4861 raw_values.push(vm.pop());
4862 }
4863 raw_values.reverse();
4864 let name = raw_values.first().map(|v| v.to_str()).unwrap_or_default();
4865 let value_raw = raw_values.get(1).cloned();
4866 let value = value_raw.as_ref().map(|v| v.to_str()).unwrap_or_default();
4867 // c:Src/params.c — when the bytecode hands us an Int value
4868 // (only the arith assignment paths emit this — `(( X = N ))`
4869 // is the canonical site), route through setiparam so the
4870 // param ends up PM_INTEGER + inherits the math layer's
4871 // `lastbase` for display formatting (`(( X = 16#ff ));
4872 // echo \$X` → `16#FF`). Scalar `X=val` and `$((expr))`
4873 // assignments still take the setsparam path below.
4874 let int_assign = matches!(value_raw, Some(fusevm::Value::Int(_)));
4875 let float_assign = matches!(value_raw, Some(fusevm::Value::Float(_)));
4876 let mut assign_failed = false;
4877 with_executor(|exec| {
4878 // c:Src/params.c assignsparam — PM_READONLY rejection
4879 // BEFORE any env mutation. The inline-env-prefix path
4880 // (`X=2 env`) called env::set_var unconditionally before
4881 // the readonly check fired in setsparam, so the OS env
4882 // got X=2 even though the assignment errored. env then
4883 // inherited the polluted env from fork, leaking the
4884 // attempted override past the readonly guard. Mirror
4885 // C's order: readonly check → zerr → bail; only mutate
4886 // env when the assignment is admissible. Bug #551
4887 // (security-relevant).
4888 if exec.is_readonly_param(&name) {
4889 crate::ported::utils::zerr(&format!("read-only variable: {}", name));
4890 return;
4891 }
4892 // Inline-assignment frame tracking (`X=foo cmd` reverts on
4893 // command return).
4894 if !exec.inline_env_stack.is_empty() {
4895 let prev_var = crate::ported::params::getsparam(&name);
4896 let prev_env = env::var(&name).ok();
4897 exec.inline_env_stack
4898 .last_mut()
4899 .unwrap()
4900 .push((name.clone(), prev_var, prev_env));
4901 let _ = crate::ported::params::zputenv(&format!("{}={}", &name, &value));
4902 // c:Src/params.c:5354
4903 }
4904 // Canonical setsparam handles readonly, integer math, case
4905 // fold, GSU dispatch. For Int values (arith assigns) route
4906 // through setiparam so the param is PM_INTEGER + inherits
4907 // the math layer's lastbase for display formatting. For
4908 // Float (arith assigns producing MN_FLOAT) route through
4909 // setnparam so the param is PM_FFLOAT — `(( b = a * 2 ))`
4910 // with scalar `a="3.14"` should create b as typeset -F,
4911 // not a scalar holding "6.28".
4912 if int_assign {
4913 if let Some(fusevm::Value::Int(i)) = value_raw {
4914 crate::ported::params::setiparam(&name, i);
4915 } else {
4916 assign_failed = crate::ported::params::setsparam(&name, &value).is_none();
4917 }
4918 } else if float_assign {
4919 if let Some(fusevm::Value::Float(f)) = value_raw {
4920 // ArithCompiler returns Value::Float whenever any
4921 // operand came through Str (BUILTIN_GET_VAR yields
4922 // Value::Str even for integer-shaped scalars). To
4923 // avoid forcing every `(( b = a + 3 ))` to PM_FFLOAT
4924 // when `a="5"` (integer-shaped), detect integer-
4925 // valued floats and route through setiparam instead.
4926 // True floats (non-integral) reach setnparam →
4927 // PM_FFLOAT so `typeset -p b` shows `typeset -F …`.
4928 if f.fract() == 0.0 && f.is_finite() && f.abs() <= i64::MAX as f64 {
4929 crate::ported::params::setiparam(&name, f as i64);
4930 } else {
4931 let mnval = crate::ported::math::mnumber {
4932 l: 0,
4933 d: f,
4934 type_: crate::ported::math::MN_FLOAT,
4935 };
4936 crate::ported::params::setnparam(&name, mnval);
4937 }
4938 } else {
4939 assign_failed = crate::ported::params::setsparam(&name, &value).is_none();
4940 }
4941 } else {
4942 // c:Src/exec.c:2554-2567 — GLOB_ASSIGN. When the
4943 // `globassign` option is on and the scalar RHS is a glob
4944 // pattern, glob it and recreate the parameter as a scalar
4945 // (≤1 match) or array (>1) — csh-style assignment. The
4946 // bridge hands `value` UNTOKENIZED, so re-tokenize
4947 // (shtokenize) before haswilds/globlist; zsh's wordcode
4948 // value arrives pre-tokenized via `htok`. The
4949 // `isset(GLOBASSIGN)` gate is first and cheap (option off
4950 // by default), so the common path is unchanged.
4951 let mut globbed = false;
4952 // Only glob the RHS when the compiler flagged an UNQUOTED glob
4953 // token in the literal wordcode (SET_VAR_GLOB_ELIGIBLE). zsh's
4954 // GLOB_ASSIGN (Src/exec.c:2554) globs literal patterns only —
4955 // `x="/tmp/*"`, `x='/tmp/*'`, `x=$param`, `x=$(cmd)` all assign
4956 // verbatim. The value arrives here untokenized (DQ-wrapped by
4957 // the compiler), so this compile-time flag is the only surviving
4958 // signal of whether the pattern was quote-protected.
4959 let glob_eligible = SET_VAR_GLOB_ELIGIBLE.with(|c| c.replace(false));
4960 if glob_eligible && crate::ported::zsh_h::isset(crate::ported::zsh_h::GLOBASSIGN) {
4961 let mut tv = value.clone();
4962 crate::ported::glob::shtokenize(&mut tv);
4963 if crate::ported::pattern::haswilds(&tv) {
4964 // Committed to the glob path: never fall back to
4965 // assigning the literal pattern (zsh errors on
4966 // no-match instead).
4967 globbed = true;
4968 // globlist tokenizes its input internally (for
4969 // haswilds + glob_path) and prints the ORIGINAL
4970 // string verbatim in its "no matches found" error,
4971 // so feed it the UNtokenized value — passing the
4972 // tokenized form would leak the Star/Quest token
4973 // bytes into the error message.
4974 let mut ll: crate::ported::linklist::LinkList<String> = Default::default();
4975 ll.push_back(value.clone());
4976 crate::ported::subst::globlist(&mut ll, 0); // c:2556
4977 if crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed)
4978 == 0
4979 {
4980 let matches: Vec<String> = ll
4981 .nodes
4982 .iter()
4983 .map(|s| crate::ported::lex::untokenize(s).to_string())
4984 .collect();
4985 crate::ported::params::unsetparam(&name); // c:2562
4986 if matches.len() <= 1 {
4987 let v = matches.into_iter().next().unwrap_or_default();
4988 assign_failed =
4989 crate::ported::params::setsparam(&name, &v).is_none();
4990 } else {
4991 crate::ported::params::setaparam(&name, matches);
4992 }
4993 }
4994 // errflag set → globlist already reported
4995 // "no matches found"; leave the param unassigned
4996 // to match zsh's abort.
4997 }
4998 }
4999 // c:Src/exec.c addvars — a NULL return from
5000 // assignsparam (e.g. nameref resolving out of scope,
5001 // createparam refusal at c:1108-1118) fails the
5002 // assignment with status 1.
5003 if !globbed {
5004 assign_failed = crate::ported::params::setsparam(&name, &value).is_none();
5005 }
5006 }
5007 // PM_EXPORTED / allexport env mirror — read AFTER setsparam
5008 // so the flag bit reflects any GSU setfn side-effects.
5009 let allexport = opt_state_get("allexport").unwrap_or(false);
5010 let already_exported =
5011 (exec.param_flags(&name) as u32 & crate::ported::zsh_h::PM_EXPORTED) != 0;
5012 if allexport || already_exported {
5013 let _ = crate::ported::params::zputenv(&format!("{}={}", &name, &value));
5014 // c:Src/params.c:5354
5015 }
5016 #[cfg(feature = "recorder")]
5017 if crate::recorder::is_enabled()
5018 && exec.local_scope_depth == 0
5019 && !matches!(
5020 name.as_str(),
5021 "PPID" | "LINENO" | "ZSH_ARGZERO" | "argv0" | "ARGC" | "?" | "_" | "RANDOM"
5022 )
5023 {
5024 let ctx = exec.recorder_ctx();
5025 let attrs = exec.recorder_attrs_for(&name);
5026 crate::recorder::emit_assign_typed(&name, &value, attrs, ctx);
5027 }
5028 });
5029 Value::Status(vm.last_status)
5030 });
5031
5032 // c:Src/exec.c execfor → Src/params.c:6362 setloopvar — the
5033 // for-loop variable bind. Distinct from BUILTIN_SET_VAR because a
5034 // PM_NAMEREF loop variable REBINDS (new refname) instead of
5035 // assigning through the resolved chain.
5036 vm.register_builtin(BUILTIN_SET_LOOP_VAR, |vm, argc| {
5037 let args = pop_args(vm, argc);
5038 let name = args.first().cloned().unwrap_or_default();
5039 let value = args.get(1).cloned().unwrap_or_default();
5040 if crate::vm_helper::is_nameref(&name) {
5041 let ef_before =
5042 crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed);
5043 crate::ported::params::setloopvar(&name, &value); // c:6362
5044 let ef_after = crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed);
5045 if (ef_after & crate::ported::utils::ERRFLAG_ERROR) != 0 && ef_after != ef_before {
5046 // zerr fired (read-only reference / invalid self
5047 // reference) — abort the loop, status 1 (C errflag).
5048 vm.last_status = 1;
5049 return Value::Bool(false);
5050 }
5051 return Value::Bool(true);
5052 }
5053 // Plain loop var — canonical scalar path (same shape as
5054 // BUILTIN_SET_VAR's setsparam arm).
5055 with_executor(|exec| {
5056 if exec.is_readonly_param(&name) {
5057 crate::ported::utils::zerr(&format!("read-only variable: {}", name));
5058 return;
5059 }
5060 crate::ported::params::setsparam(&name, &value);
5061 let allexport = opt_state_get("allexport").unwrap_or(false);
5062 let already_exported =
5063 (exec.param_flags(&name) as u32 & crate::ported::zsh_h::PM_EXPORTED) != 0;
5064 if allexport || already_exported {
5065 let _ = crate::ported::params::zputenv(&format!("{}={}", &name, &value));
5066 // c:Src/params.c:5354
5067 }
5068 });
5069 Value::Bool(true)
5070 });
5071
5072 // Pre-compiled function registration — used by compile_zsh.rs's
5073 // FuncDef path. Stack: [name, base64-bincode-of-Chunk]. We decode
5074 // the base64, deserialize the Chunk, and store directly in
5075 // executor.functions_compiled. Bypasses the ShellCommand JSON layer.
5076 // BUILTIN_VAR_EXISTS — `[[ -v name ]]` set-test.
5077 // PURE PASSTHRU: build `${+name}` and route through canonical
5078 // `subst::paramsubst` which returns "1" for set / "0" for unset
5079 // (C port of `Src/subst.c::paramsubst` plus-prefix arm).
5080 // paramsubst handles all the shapes the 48-line hand-roll did:
5081 // - bare scalar / array / assoc
5082 // - subscripted `a[N]` / `h[key]`
5083 // - positional params (any digit-only name)
5084 // - env-var fallback (`HOME` set via getsparam → lookup_special_var)
5085 vm.register_builtin(BUILTIN_VAR_EXISTS, |vm, _argc| {
5086 let name = vm.pop().to_str();
5087 // c:Src/cond.c:361 `case 'v': return !issetvar(left)`. `-v` is
5088 // NOT `${+name}` — issetvar (params.c:751) additionally rejects
5089 // trailing chars after the parsed name/subscript (`arr[3]extra`,
5090 // nested `arr[2][1]`) and validates array-slice bounds (an
5091 // out-of-range `(i)`-not-found index is "unset"). `${+}` is
5092 // lenient and reported those as set.
5093 Value::Bool(crate::ported::params::issetvar(&name) != 0)
5094 });
5095
5096 // `time { compound; ... }` — runs the sub-chunk and prints elapsed
5097 // wall-clock time. zsh's full `time` also tracks user/system CPU via
5098 // getrusage on the *child*; we approximate via wall-time only since
5099 // the sub-chunk runs in-process (no fork). Output format matches
5100 // `time simple-cmd` (already implemented elsewhere via exectime).
5101 vm.register_builtin(BUILTIN_TIME_SUBLIST, |vm, argc| {
5102 let sub_idx = vm.pop().to_int() as usize;
5103 // c:Src/jobs.c:1028-1029 — `pn->text` arg to printtime. argc==2
5104 // means the compiler also pushed a desc string (bug #66 fix);
5105 // older callers with argc==1 push only sub_idx and we synthesize
5106 // an empty desc for backward compat with cached bytecode that
5107 // predates the desc-threading patch.
5108 let desc = if argc >= 2 {
5109 vm.pop().to_str().to_string()
5110 } else {
5111 String::new()
5112 };
5113 let chunk_opt = vm.chunk.sub_chunks.get(sub_idx).cloned();
5114 let Some(chunk) = chunk_opt else {
5115 return Value::Status(0);
5116 };
5117 // c:Src/jobs.c:1968 — `getrusage(RUSAGE_CHILDREN, &ti)` before
5118 // and after the timed sublist gives accurate per-stage user/sys
5119 // CPU. Wall-time-only approximation (0.7×/0.1× fudge factors)
5120 // produced bogus user/sys columns and ignored TIMEFMT. Bug #66
5121 // in docs/BUGS.md.
5122 let ru_before: libc::rusage = unsafe {
5123 let mut r: libc::rusage = std::mem::zeroed();
5124 libc::getrusage(libc::RUSAGE_CHILDREN, &mut r);
5125 r
5126 };
5127 // c:Src/jobs.c — zsh's `time` reports only for JOBS (forked
5128 // work). Builtins/brace-groups/functions run in the shell
5129 // process with no job, so `zsh -fc 'time true'` emits NOTHING.
5130 // Snapshot the fork-event counter; report only if the timed
5131 // body forked (external command or subshell).
5132 let forks_before = crate::vm_helper::FORK_EVENTS.load(std::sync::atomic::Ordering::Relaxed);
5133 let start = Instant::now();
5134 crate::fusevm_disasm::maybe_print_stdout("time_sublist", &chunk);
5135 let mut sub_vm = fusevm::VM::new(chunk);
5136 register_builtins(&mut sub_vm);
5137 let _ = sub_vm.run();
5138 let status = sub_vm.last_status;
5139 let elapsed = start.elapsed();
5140 let ru_after: libc::rusage = unsafe {
5141 let mut r: libc::rusage = std::mem::zeroed();
5142 libc::getrusage(libc::RUSAGE_CHILDREN, &mut r);
5143 r
5144 };
5145 // Delta children rusage = timed work's CPU.
5146 let mut delta = ru_after;
5147 let sub = |a: libc::timeval, b: libc::timeval| -> libc::timeval {
5148 let mut sec = a.tv_sec - b.tv_sec;
5149 let mut usec = a.tv_usec as i64 - b.tv_usec as i64;
5150 if usec < 0 {
5151 sec -= 1;
5152 usec += 1_000_000;
5153 }
5154 libc::timeval {
5155 tv_sec: sec,
5156 tv_usec: usec as libc::suseconds_t,
5157 }
5158 };
5159 delta.ru_utime = sub(ru_after.ru_utime, ru_before.ru_utime);
5160 delta.ru_stime = sub(ru_after.ru_stime, ru_before.ru_stime);
5161 let ti = crate::ported::zsh_h::timeinfo::from_rusage(&delta);
5162 // c:Src/jobs.c:808-809 — `s = getsparam("TIMEFMT"); s ||
5163 // DEFAULT_TIMEFMT`. Honor user-set TIMEFMT, fall back to the
5164 // canonical default.
5165 let fmt = crate::ported::params::getsparam("TIMEFMT")
5166 .unwrap_or_else(|| crate::ported::zsh_system_h::DEFAULT_TIMEFMT.to_string());
5167 // c:Src/jobs.c:768 `desc` arg — for the `time { sublist }` /
5168 // `time simple-cmd` keyword path, zsh passes the sublist's
5169 // source text (used by %J via printtime). The compiler now
5170 // threads the rendered source text through as the desc operand
5171 // (compile_zsh.rs Time arm, argc==2 form). Bug #66.
5172 // c:Src/jobs.c — no job, no report (see forks_before above).
5173 let forked = crate::vm_helper::FORK_EVENTS.load(std::sync::atomic::Ordering::Relaxed)
5174 != forks_before;
5175 if forked {
5176 let line = crate::ported::jobs::printtime(elapsed.as_secs_f64(), &ti, &fmt, &desc);
5177 eprintln!("{}", line);
5178 }
5179 Value::Status(status)
5180 });
5181
5182 // `{name}>file` / `{name}<file` / `{name}>>file` — named-fd allocator.
5183 // Stack: [path, varid, op_byte]. Opens path with the appropriate mode
5184 // and stores the resulting fd number in $varid as a string. We use
5185 // a high starting fd (10+) by allocating then dup'ing — matches zsh's
5186 // "fresh fd >= 10" promise so subsequent commands don't collide on
5187 // stdin/out/err.
5188 vm.register_builtin(BUILTIN_OPEN_NAMED_FD, |vm, _argc| {
5189 use std::sync::atomic::Ordering;
5190 let op_byte = vm.pop().to_int() as u8;
5191 let varid = vm.pop().to_str();
5192 let path = vm.pop().to_str();
5193 // Param introspection used by both the open and close forms.
5194 let param_flags = crate::ported::params::paramtab()
5195 .read()
5196 .ok()
5197 .and_then(|t| t.get(&varid).map(|p| p.node.flags));
5198 let param_readonly = param_flags
5199 .map(|f| (f & crate::ported::zsh_h::PM_READONLY as i32) != 0)
5200 .unwrap_or(false);
5201 // `{varid}>&-` / `{varid}<&-` — REDIR_CLOSE with varid.
5202 // Direct port of Src/exec.c:3805-3850.
5203 if matches!(
5204 op_byte,
5205 b if b == fusevm::op::redirect_op::DUP_WRITE
5206 || b == fusevm::op::redirect_op::DUP_READ
5207 ) {
5208 let n = path.trim_start_matches('&');
5209 if n == "-" {
5210 let val = with_executor(|exec| exec.scalar(&varid)).unwrap_or_default();
5211 let fd1 = val.parse::<i32>();
5212 // c:3811-3816 — bad=1: parameter doesn't contain an fd.
5213 let Ok(fd1) = fd1 else {
5214 crate::ported::utils::zwarn(&format!(
5215 "parameter {} does not contain a file descriptor",
5216 varid
5217 ));
5218 with_executor(|exec| exec.redirect_failed = true);
5219 return Value::Status(1);
5220 };
5221 // c:3813-3814 — bad=2: readonly parameter.
5222 if param_readonly {
5223 crate::ported::utils::zwarn(&format!(
5224 "can't close file descriptor from readonly parameter {}",
5225 varid
5226 ));
5227 with_executor(|exec| exec.redirect_failed = true);
5228 return Value::Status(1);
5229 }
5230 // c:3830-3835 — bad=3: fd >= 10 marked FDT_INTERNAL.
5231 if fd1 >= 10
5232 && fd1 <= crate::ported::utils::MAX_ZSH_FD.load(Ordering::Relaxed)
5233 && crate::ported::utils::fdtable_get(fd1) == crate::ported::zsh_h::FDT_INTERNAL
5234 {
5235 crate::ported::utils::zwarn(&format!(
5236 "file descriptor {} used by shell, not closed",
5237 fd1
5238 ));
5239 with_executor(|exec| exec.redirect_failed = true);
5240 return Value::Status(1);
5241 }
5242 // c:3870-3873 — close; report failure (varid form
5243 // always reports, unlike bare `N>&-`).
5244 if crate::ported::utils::zclose(fd1) < 0 {
5245 crate::ported::utils::zwarn(&format!(
5246 "failed to close file descriptor {}: {}",
5247 fd1,
5248 std::io::Error::last_os_error()
5249 ));
5250 return Value::Status(1);
5251 }
5252 return Value::Status(0);
5253 }
5254 // `{varid}>&N` — dup N to a fresh fd >= 10, store in varid.
5255 if let Ok(src) = n.parse::<i32>() {
5256 if param_readonly {
5257 crate::ported::utils::zwarn(&format!(
5258 "can't allocate file descriptor to readonly parameter {}",
5259 varid
5260 ));
5261 with_executor(|exec| exec.redirect_failed = true);
5262 return Value::Status(1);
5263 }
5264 let dup = unsafe { libc::dup(src) };
5265 if dup < 0 {
5266 crate::ported::utils::zwarn(&format!("{}: bad file descriptor", src));
5267 with_executor(|exec| exec.redirect_failed = true);
5268 return Value::Status(1);
5269 }
5270 // c:2404-2412 addfd varid arm — movefd + FDT_EXTERNAL.
5271 let final_fd = crate::ported::utils::movefd(dup);
5272 crate::ported::utils::fdtable_set(final_fd, crate::ported::zsh_h::FDT_EXTERNAL);
5273 with_executor(|exec| {
5274 exec.set_scalar(varid, final_fd.to_string());
5275 });
5276 return Value::Status(0);
5277 }
5278 return Value::Status(1);
5279 }
5280 // `{varid}<<HERE` / `{varid}<<<str` — op byte 255 (zshrs-side
5281 // contract with compile_redir; fusevm's redirect_op stops at
5282 // 8). C path: gethere/getherestr write the body to a temp
5283 // file (Src/exec.c:4660-4682), then addfd's varid arm moves
5284 // the read fd >= 10, marks FDT_EXTERNAL and sets the param
5285 // (c:2402-2412). `path` carries the BODY text here.
5286 if op_byte == 255 {
5287 if param_readonly {
5288 crate::ported::utils::zwarn(&format!(
5289 "can't allocate file descriptor to readonly parameter {}",
5290 varid
5291 ));
5292 with_executor(|exec| exec.redirect_failed = true);
5293 return Value::Status(1);
5294 }
5295 let body = format!("{}\n", path.trim_end_matches('\n'));
5296 let mut tmpl: Vec<u8> = b"/tmp/zshrs_hd_XXXXXX\0".to_vec();
5297 let write_fd = unsafe { libc::mkstemp(tmpl.as_mut_ptr() as *mut libc::c_char) };
5298 if write_fd < 0 {
5299 crate::ported::utils::zwarn(&format!(
5300 "can't create temp file for here document: {}",
5301 std::io::Error::last_os_error()
5302 ));
5303 return Value::Status(1);
5304 }
5305 let bytes = body.as_bytes();
5306 let mut off = 0;
5307 while off < bytes.len() {
5308 let n = unsafe {
5309 libc::write(
5310 write_fd,
5311 bytes[off..].as_ptr() as *const libc::c_void,
5312 bytes.len() - off,
5313 )
5314 };
5315 if n <= 0 {
5316 unsafe { libc::close(write_fd) };
5317 return Value::Status(1);
5318 }
5319 off += n as usize;
5320 }
5321 unsafe { libc::close(write_fd) };
5322 let read_fd =
5323 unsafe { libc::open(tmpl.as_ptr() as *const libc::c_char, libc::O_RDONLY) };
5324 unsafe { libc::unlink(tmpl.as_ptr() as *const libc::c_char) };
5325 if read_fd < 0 {
5326 return Value::Status(1);
5327 }
5328 let final_fd = crate::ported::utils::movefd(read_fd);
5329 if final_fd < 0 {
5330 return Value::Status(1);
5331 }
5332 crate::ported::utils::fdtable_set(final_fd, crate::ported::zsh_h::FDT_EXTERNAL);
5333 with_executor(|exec| {
5334 exec.set_scalar(varid, final_fd.to_string());
5335 });
5336 return Value::Status(0);
5337 }
5338 // Open form: `{varid}>file` etc.
5339 // c:Src/exec.c:2177-2215 checkclobberparam — gate BEFORE open.
5340 if param_readonly {
5341 // c:2191-2197
5342 crate::ported::utils::zwarn(&format!(
5343 "can't allocate file descriptor to readonly parameter {}",
5344 varid
5345 ));
5346 with_executor(|exec| exec.redirect_failed = true);
5347 return Value::Status(1);
5348 }
5349 // c:2199-2213 — NO_CLOBBER refuses to overwrite a parameter
5350 // already holding an OPEN fd (decimal value, fdtable says
5351 // FDT_EXTERNAL).
5352 if !isset(crate::ported::zsh_h::CLOBBER) && op_byte != fusevm::op::redirect_op::CLOBBER {
5353 if let Some(val) = with_executor(|exec| exec.scalar(&varid)) {
5354 if let Ok(fd) = val.parse::<i32>() {
5355 if fd >= 0
5356 && fd <= crate::ported::utils::MAX_ZSH_FD.load(Ordering::Relaxed)
5357 && crate::ported::utils::fdtable_get(fd)
5358 == crate::ported::zsh_h::FDT_EXTERNAL
5359 {
5360 crate::ported::utils::zwarn(&format!(
5361 "can't clobber parameter {} containing file descriptor {}",
5362 varid, fd
5363 ));
5364 with_executor(|exec| exec.redirect_failed = true);
5365 return Value::Status(1);
5366 }
5367 }
5368 }
5369 }
5370 let path_c = match CString::new(path.clone()) {
5371 Ok(c) => c,
5372 Err(_) => return Value::Status(1),
5373 };
5374 let flags = match op_byte {
5375 b if b == fusevm::op::redirect_op::READ => libc::O_RDONLY,
5376 b if b == fusevm::op::redirect_op::WRITE || b == fusevm::op::redirect_op::CLOBBER => {
5377 libc::O_WRONLY | libc::O_CREAT | libc::O_TRUNC
5378 }
5379 b if b == fusevm::op::redirect_op::APPEND => {
5380 libc::O_WRONLY | libc::O_CREAT | libc::O_APPEND
5381 }
5382 b if b == fusevm::op::redirect_op::READ_WRITE => libc::O_RDWR | libc::O_CREAT,
5383 _ => return Value::Status(1),
5384 };
5385 let fd = unsafe { libc::open(path_c.as_ptr(), flags, 0o666) };
5386 if fd < 0 {
5387 return Value::Status(1);
5388 }
5389 // c:2404-2412 addfd varid arm — `fd1 = movefd(fd2);
5390 // fdtable[fd1] = FDT_EXTERNAL; setiparam(varid, fd1);`.
5391 // FDT_EXTERNAL (not INTERNAL): the user owns this fd — the
5392 // NO_CLOBBER gate above and `{fd}>&-` close both key off it.
5393 let final_fd = crate::ported::utils::movefd(fd);
5394 if final_fd < 0 {
5395 crate::ported::utils::zerr(&format!(
5396 "cannot move fd {}: {}",
5397 fd,
5398 std::io::Error::last_os_error()
5399 ));
5400 return Value::Status(1);
5401 }
5402 crate::ported::utils::fdtable_set(final_fd, crate::ported::zsh_h::FDT_EXTERNAL);
5403 let _ = Ordering::Relaxed;
5404 with_executor(|exec| {
5405 exec.set_scalar(varid, final_fd.to_string());
5406 });
5407 Value::Status(0)
5408 });
5409
5410 // BUILTIN_SET_TRY_BLOCK_ERROR — capture the try-block's exit
5411 // status into `__zshrs_try_block_saved_status` (a scratch
5412 // scalar) so the always-arm can later restore it. Also set
5413 // `TRY_BLOCK_ERROR` per zsh semantics: it stays at -1 unless
5414 // the try-block fired an explicit error (errflag), per
5415 // c:Src/exec.c execlist's WC_TRYBLOCK arm.
5416 vm.register_builtin(BUILTIN_SET_TRY_BLOCK_ERROR, |vm, _argc| {
5417 use std::sync::atomic::Ordering;
5418 let vm_status = vm.last_status;
5419 let errored = (crate::ported::utils::errflag.load(Ordering::Relaxed)
5420 & crate::ported::zsh_h::ERRFLAG_ERROR)
5421 != 0;
5422 // c:Src/exec.c WC_TRYBLOCK — the always-arm runs with a
5423 // clean escape state. Snapshot RETFLAG / BREAKS / CONTFLAG /
5424 // EXIT_PENDING here and clear them; RESTORE_TRY_BLOCK_STATUS
5425 // re-applies them at always-arm exit so the propagation jump
5426 // emitted by compile_zsh fires correctly.
5427 let ret_save = crate::ported::builtin::RETFLAG.swap(0, Ordering::Relaxed);
5428 let brk_save = crate::ported::builtin::BREAKS.swap(0, Ordering::Relaxed);
5429 let cont_save = crate::ported::builtin::CONTFLAG.swap(0, Ordering::Relaxed);
5430 let exit_save = crate::ported::builtin::EXIT_PENDING.swap(0, Ordering::Relaxed);
5431 TRY_ESCAPE_SAVE.with(|s| {
5432 s.borrow_mut()
5433 .push((ret_save, brk_save, cont_save, exit_save));
5434 });
5435 with_executor(|exec| {
5436 exec.set_scalar(
5437 "__zshrs_try_block_saved_status".to_string(),
5438 vm_status.to_string(),
5439 );
5440 // c:Src/loop.c:765-766 — `try_errflag = errflag &
5441 // ERRFLAG_ERROR; try_interrupt = (errflag & ERRFLAG_INT) ?
5442 // 1 : 0`. Updates the canonical loop.c globals that
5443 // `$TRY_BLOCK_ERROR` / `$TRY_BLOCK_INTERRUPT` read from
5444 // (port at src/ported/loop.rs::try_errflag etc.).
5445 // Mirror into paramtab too via set_scalar so
5446 // `${parameters[TRY_BLOCK_ERROR]}` and direct assignment
5447 // shapes stay in sync.
5448 let try_err = if errored { vm_status as i64 } else { 0 };
5449 crate::ported::r#loop::try_errflag.store(try_err, Ordering::Relaxed);
5450 crate::ported::r#loop::try_interrupt.store(0, Ordering::Relaxed);
5451 if errored {
5452 exec.set_scalar("TRY_BLOCK_ERROR".to_string(), vm_status.to_string());
5453 // Clear errflag so always-arm runs cleanly. c:768.
5454 crate::ported::utils::errflag
5455 .fetch_and(!crate::ported::zsh_h::ERRFLAG_ERROR, Ordering::Relaxed);
5456 } else {
5457 exec.set_scalar("TRY_BLOCK_ERROR".to_string(), "0".to_string());
5458 }
5459 });
5460 Value::Status(0)
5461 });
5462
5463 // BUILTIN_BEGIN_INLINE_ENV / END_INLINE_ENV — wrap an
5464 // inline-assignment-prefixed command (`X=foo Y=bar cmd`):
5465 // BEGIN pushes a save frame; SET_VAR fires for each assign and
5466 // ALSO env::set_var's the value (visible to cmd's child); the
5467 // command runs; END pops the frame and restores both shell-var
5468 // and process-env state. Direct port of zsh's addvars() →
5469 // execute_simple → restore-after-exec contract.
5470 vm.register_builtin(BUILTIN_BEGIN_INLINE_ENV, |_vm, _argc| {
5471 with_executor(|exec| {
5472 exec.inline_env_stack.push(Vec::new());
5473 });
5474 Value::Status(0)
5475 });
5476 vm.register_builtin(BUILTIN_END_INLINE_ENV, |_vm, _argc| {
5477 with_executor(|exec| {
5478 if let Some(frame) = exec.inline_env_stack.pop() {
5479 for (name, prev_var, prev_env) in frame.into_iter().rev() {
5480 match prev_var {
5481 Some(v) => {
5482 exec.set_scalar(name.clone(), v);
5483 }
5484 None => {
5485 exec.unset_scalar(&name);
5486 }
5487 }
5488 match prev_env {
5489 Some(v) => env::set_var(&name, &v),
5490 None => env::remove_var(&name),
5491 }
5492 }
5493 }
5494 });
5495 Value::Status(0)
5496 });
5497 // c:Src/exec.c:3969-3976 — bare-exec assignment epilogue: see the
5498 // const's doc block. POSIX_BUILTINS → assignments persist (pop the
5499 // frame, discard the saved state); otherwise → restore_params
5500 // (same walk as END_INLINE_ENV).
5501 vm.register_builtin(BUILTIN_EXEC_INLINE_ENV_DONE, |_vm, _argc| {
5502 let persist = isset(crate::ported::zsh_h::POSIXBUILTINS);
5503 with_executor(|exec| {
5504 if let Some(frame) = exec.inline_env_stack.pop() {
5505 if persist {
5506 return; // c:3971 — no save/restore under POSIX_BUILTINS
5507 }
5508 for (name, prev_var, prev_env) in frame.into_iter().rev() {
5509 match prev_var {
5510 Some(v) => {
5511 exec.set_scalar(name.clone(), v);
5512 }
5513 None => {
5514 exec.unset_scalar(&name);
5515 }
5516 }
5517 match prev_env {
5518 Some(v) => env::set_var(&name, &v),
5519 None => env::remove_var(&name),
5520 }
5521 }
5522 }
5523 });
5524 Value::Status(0)
5525 });
5526
5527 // BUILTIN_RESTORE_TRY_BLOCK_STATUS — emitted at the end of an
5528 // `always` arm. Per zshmisc, the exit status of the entire
5529 // `{ try } always { finally }` construct is the try-list's
5530 // status, regardless of what happens in the always-list (the
5531 // exception is `return`/`exit` inside always, which short-
5532 // circuits and the cleanup is the only thing that runs). So
5533 // restore TRY_BLOCK_ERROR unconditionally — the always-list's
5534 // exit status is discarded for the construct.
5535 vm.register_builtin(BUILTIN_RESTORE_TRY_BLOCK_STATUS, |_vm, _argc| {
5536 use std::sync::atomic::Ordering;
5537 // c:Src/exec.c — the entire `{try} always {…}` construct's
5538 // exit status is the try-block's last status. Per zsh
5539 // semantics this carries through regardless of what the
5540 // always-arm did (including reads/writes of TRY_BLOCK_ERROR
5541 // — those affect later commands' visible value but don't
5542 // override the construct's exit). The "swallow" idiom in
5543 // C is gated on errflag state at always-arm exit, not on
5544 // TBE's literal value; full fidelity needs more state and
5545 // is deferred.
5546 let saved = with_executor(|exec| {
5547 exec.scalar("__zshrs_try_block_saved_status")
5548 .and_then(|s| s.parse::<i32>().ok())
5549 .unwrap_or(0)
5550 });
5551 // Re-apply the escape flags captured by SET_TRY_BLOCK_ERROR.
5552 // If the always-arm itself fired return/break/continue/exit,
5553 // its handler already overwrote the canonical atomics; let
5554 // those win — the always-arm's own escape always takes
5555 // priority over the try-block's deferred one.
5556 if let Some((ret, brk, cont, exit_p)) = TRY_ESCAPE_SAVE.with(|s| s.borrow_mut().pop()) {
5557 if crate::ported::builtin::RETFLAG.load(Ordering::Relaxed) == 0 {
5558 crate::ported::builtin::RETFLAG.store(ret, Ordering::Relaxed);
5559 }
5560 if crate::ported::builtin::BREAKS.load(Ordering::Relaxed) == 0 {
5561 crate::ported::builtin::BREAKS.store(brk, Ordering::Relaxed);
5562 }
5563 if crate::ported::builtin::CONTFLAG.load(Ordering::Relaxed) == 0 {
5564 crate::ported::builtin::CONTFLAG.store(cont, Ordering::Relaxed);
5565 }
5566 if crate::ported::builtin::EXIT_PENDING.load(Ordering::Relaxed) == 0 {
5567 crate::ported::builtin::EXIT_PENDING.store(exit_p, Ordering::Relaxed);
5568 }
5569 }
5570 Value::Status(saved)
5571 });
5572
5573 // `[[ -r/-w/-x file ]]` — the cond path must use access(2) (the
5574 // C-faithful doaccess), NOT fusevm's generic Op::TestFile which only
5575 // checks existence for -r/-w (so a `chmod 000` file read as readable;
5576 // C02cond.ztst:13). Stack: [path, mode]; mode is the access(2) bit
5577 // (R_OK=4, W_OK=2, X_OK=1). Mirrors cond.rs:232/238/267 `doaccess`.
5578 vm.register_builtin(BUILTIN_COND_ACCESS, |vm, _argc| {
5579 let mode = vm.pop().to_int() as i32;
5580 let path = vm.pop().to_str();
5581 Value::Bool(crate::ported::cond::doaccess(&path, mode) != 0)
5582 });
5583
5584 vm.register_builtin(BUILTIN_IS_TTY, |vm, _argc| {
5585 let fd_str = vm.pop().to_str();
5586 let fd: i32 = fd_str.trim().parse().unwrap_or(-1);
5587 let is_tty = if fd < 0 {
5588 false
5589 } else {
5590 unsafe { libc::isatty(fd) != 0 }
5591 };
5592 Value::Bool(is_tty)
5593 });
5594
5595 // Set $LINENO before executing the next statement. Direct
5596 // port of zsh's `lineno` global tracking from Src/input.c
5597 // (`if ((inbufflags & INP_LINENO) || !strin) && c == '\n')
5598 // lineno++;`). The compiler emits one of these before each
5599 // top-level pipe in `compile_sublist`, carrying the line
5600 // number captured by the parser at `ZshPipe.lineno`. Pops
5601 // [n], updates `$LINENO` in the variable table.
5602 vm.register_builtin(BUILTIN_SET_LINENO, |vm, _argc| {
5603 let n = vm.pop().to_int();
5604 // c:Src/exec.c:lineno = N — direct write to the param's
5605 // u_val. Cannot go through setsparam because LINENO carries
5606 // PM_READONLY (so `(t)LINENO` reads `integer-readonly-special`
5607 // per zsh); setsparam → assignstrvalue's PM_READONLY guard
5608 // would reject the internal write. C zsh handles this via the
5609 // PM_SPECIAL GSU vtable's setfn callback which bypasses the
5610 // generic readonly check; the Rust port writes the canonical
5611 // field directly instead.
5612 if let Ok(mut tab) = crate::ported::params::paramtab().write() {
5613 if let Some(pm) = tab.get_mut("LINENO") {
5614 pm.u_val = n;
5615 pm.node.flags &= !(crate::ported::zsh_h::PM_UNSET as i32);
5616 }
5617 }
5618 // Mirror to the file-static `lineno` (utils.c:121) that
5619 // zerrmsg reads at utils.c:301 for the `:N: msg` prefix.
5620 crate::ported::utils::set_lineno(n as i32);
5621 // Also drive lex::LEX_LINENO — zerrmsg (utils.rs:376) reads
5622 // THAT counter for the `name:N:` prefix. C zsh interleaves
5623 // parse and execute per top-level list, so its single
5624 // `lineno` global serves both; zshrs compiles the whole
5625 // script before running, leaving LEX_LINENO parked at EOF.
5626 // Without this write, every runtime zwarn/zerr reported the
5627 // script's LAST line instead of the failing statement's.
5628 crate::ported::lex::set_lineno(n as u64);
5629 // DAP hook — checks breakpoints / step mode / pause-request
5630 // for the line we just landed on. O(1) no-op when DAP is off
5631 // (single atomic load on a OnceLock). Inside `--dap` mode
5632 // this is the call that blocks the executor on a Condvar
5633 // until the IDE sends `continue`. Mirrors strykelang's
5634 // `debugger.should_stop(line) → debugger.prompt(...)` flow.
5635 crate::extensions::dap::check_line(n as u32);
5636 Value::Status(0)
5637 });
5638
5639 // Direct port of Src/prompt.c:1623 cmdpush. Token is a `CS_*`
5640 // value (zsh.h:2775-2806) emitted by compile_zsh around each
5641 // compound command (if/while/[[…]]/((…))/$(…)) and consumed by
5642 // `%_` in PS4 / prompt expansion.
5643 vm.register_builtin(BUILTIN_CMD_PUSH, |vm, _argc| {
5644 let token = vm.pop().to_int() as u8;
5645 // Route through canonical cmdpush (Src/prompt.c:1623). The
5646 // prompt expander reads from the file-static `CMDSTACK` at
5647 // `prompt.rs:2006`, not `exec.cmd_stack` — without this,
5648 // `%_` in PS4 saw an empty stack during xtrace.
5649 if (token as i32) < crate::ported::zsh_h::CS_COUNT {
5650 crate::ported::prompt::cmdpush(token);
5651 }
5652 Value::Status(0)
5653 });
5654
5655 // Direct port of Src/prompt.c:1631 cmdpop.
5656 vm.register_builtin(BUILTIN_CMD_POP, |_vm, _argc| {
5657 crate::ported::prompt::cmdpop();
5658 Value::Status(0)
5659 });
5660
5661 vm.register_builtin(BUILTIN_OPTION_SET, |vm, _argc| {
5662 let name = vm.pop().to_str();
5663 // Direct port of `optison(char *name, char *s)` at Src/cond.c:502 — `[[ -o NAME ]]`
5664 // reads through the same `opts[]` array that `setopt NAME`
5665 // writes via `dosetopt`. Earlier code read a duplicate Executor
5666 // HashMap which never saw `bin_setopt`'s writes (those land in
5667 // `OPTS_LIVE` via `opt_state_set`). Routing through the canonical
5668 // C port restores the single-store invariant: one `opts[]`,
5669 // shared between setopt/unsetopt and `[[ -o ]]`.
5670 let r = crate::ported::cond::optison(None, &name); // c:cond.c:502 (fromtest=NULL for [[ -o ]])
5671 match r {
5672 0 => Value::Bool(true), // c:cond.c:520 set
5673 1 => Value::Bool(false), // c:cond.c:518/520 unset
5674 _ => {
5675 // c:cond.c:514 — unknown option. optison already emitted the
5676 // diagnostic (via zwarn, now that fromtest=NULL for
5677 // `[[ -o ]]`); re-printing here double-emitted for
5678 // `[[ ! -o bad ]]` / `[[ -o a || -o b ]]`.
5679 Value::Bool(false)
5680 }
5681 }
5682 });
5683 // Tri-state `-o` for compile_cond's direct status path. Returns
5684 // 0 / 1 / 3 as a Value::Int that compile_cond consumes via
5685 // Op::SetStatus. Mirrors zsh's `[[ -o invalid ]]` returning $?=3.
5686 vm.register_builtin(BUILTIN_OPTION_CHECK_TRISTATE, |vm, _argc| {
5687 let name = vm.pop().to_str();
5688 let r = crate::ported::cond::optison(None, &name); // c:cond.c:502 (fromtest=NULL for [[ -o ]])
5689 // optison itself prints the diagnostic via zwarnnam when r=3
5690 // and POSIXBUILTINS is unset (the canonical path). Don't
5691 // double-emit here. r is already 0/1/3.
5692 Value::Int(r as i64)
5693 });
5694
5695 // BUILTIN_PARAM_FILTER — `${var:#pat}` / `${var:|name}` etc.
5696 // PURE PASSTHRU: rebuild `${name:#pat}` and route to paramsubst.
5697 vm.register_builtin(BUILTIN_PARAM_FILTER, |vm, _argc| {
5698 let pattern = vm.pop().to_str();
5699 let name = vm.pop().to_str();
5700 let body = format!("${{{}:#{}}}", name, pattern);
5701 paramsubst_to_value(&body)
5702 });
5703
5704 // `a[i]=(elements)` / `a[i,j]=(elements)` / `a[i]=()`
5705 // — subscripted-array assign with array RHS. Stack pushed by
5706 // compile_assign as: [elem0, elem1, …, elemN-1, name, key].
5707 vm.register_builtin(BUILTIN_SET_SUBSCRIPT_RANGE, |vm, argc| {
5708 let n = argc as usize;
5709 let mut popped: Vec<Value> = Vec::with_capacity(n);
5710 for _ in 0..n {
5711 popped.push(vm.pop());
5712 }
5713 popped.reverse();
5714 if popped.len() < 3 {
5715 return Value::Status(1);
5716 }
5717 // c:Src/params.c:3511-3526 — trailing append flag. The ARRAY
5718 // subscript path (`a[N]+=(v)` / `a[lo,hi]+=(v)`) sets it so the
5719 // AUGMENT transform below collapses the range to an empty range
5720 // after the slice end and inserts ONLY the new value. The scalar
5721 // path pre-concats the old slice (ARRAY_INDEX+Concat) and passes
5722 // 0, so it keeps plain-replace semantics.
5723 let append = popped.pop().map_or(false, |v| v.to_str() == "1");
5724 let key = popped.pop().unwrap().to_str();
5725 let name = popped.pop().unwrap().to_str();
5726 let mut values: Vec<String> = Vec::new();
5727 for v in popped {
5728 match v {
5729 Value::Array(items) => {
5730 for it in items {
5731 values.push(it.to_str());
5732 }
5733 }
5734 other => values.push(other.to_str()),
5735 }
5736 }
5737 with_executor(|exec| {
5738 // Parse subscript: slice `lo,hi` or single index `i`.
5739 // setarrvalue (Src/params.c:2895) expects 1-based start/
5740 // end inclusive where start==end means replace one
5741 // element. Negative bounds translate to len+n+1 (1-based).
5742 //
5743 // c:Src/params.c — the END side accepts 0 as a valid value
5744 // that signals "insert BEFORE start position" (the canonical
5745 // `a[N,N-1]=val` prepend / mid-insert idiom). Bug #275 in
5746 // docs/BUGS.md: the previous Rust port clamped end up to 1,
5747 // collapsing `a[1,0]=(X Y)` into `a[1,1]=(X Y)` which
5748 // OVERWRITES position 1 instead of prepending. Provide two
5749 // translators — start_translate clamps to 1 (1-based);
5750 // end_translate keeps 0 intact so the splice in
5751 // setarrvalue (start_idx=0..end_idx=0) inserts at the front.
5752 // Bug #589: for scalars (no array), use the scalar's char
5753 // count as `len` so negative-index translation (`a[2,-1]`)
5754 // computes against the actual string length, not 0.
5755 let len = exec
5756 .array(&name)
5757 .map(|a| a.len() as i64)
5758 .or_else(|| {
5759 crate::ported::params::paramtab().read().ok().and_then(|t| {
5760 t.get(&name).and_then(|pm| {
5761 if crate::ported::zsh_h::PM_TYPE(pm.node.flags as u32)
5762 == crate::ported::zsh_h::PM_SCALAR
5763 {
5764 pm.u_str.as_ref().map(|s| s.chars().count() as i64)
5765 } else {
5766 None
5767 }
5768 })
5769 })
5770 })
5771 .unwrap_or(0);
5772 let start_translate = |raw: i64| -> i32 {
5773 if raw < 0 {
5774 (len + raw + 1).max(1) as i32
5775 } else {
5776 raw.max(1) as i32
5777 }
5778 };
5779 let end_translate = |raw: i64| -> i32 {
5780 if raw < 0 {
5781 (len + raw + 1).max(0) as i32
5782 } else {
5783 raw.max(0) as i32
5784 }
5785 };
5786 // c:Src/params.c — KSH_ARRAYS option flips array subscripts
5787 // from 1-based to 0-based. setarrvalue expects 1-based
5788 // inclusive bounds, so under KSH_ARRAYS we shift positive
5789 // inputs by +1 before translation. Negative bounds left
5790 // alone (count from end). Sibling of #610/#611/#612.
5791 // Bug #613.
5792 let ksh_arrays = crate::ported::zsh_h::isset(crate::ported::zsh_h::KSHARRAYS);
5793 let ksh_shift = |raw: i64| -> i64 {
5794 if ksh_arrays && raw >= 0 {
5795 raw + 1
5796 } else {
5797 raw
5798 }
5799 };
5800 // c:Src/params.c getindex — subscript bounds are MATH
5801 // expressions, not bare integers: `a[(( ${#a}+1 ))]=(x)`,
5802 // `a[n+1]=(x)`. Plain `parse::<i64>()` returned 0 on any
5803 // arithmetic subscript, which the `i == 0` guard turned into
5804 // a silent no-op (computed-index append never landed). Parse
5805 // the literal fast-path first, then fall back to mathevali
5806 // (which handles `(( ))` grouping, var refs, and operators).
5807 let eval_bound = |s: &str| -> i64 {
5808 let t = s.trim();
5809 t.parse::<i64>()
5810 .ok()
5811 .or_else(|| crate::ported::math::mathevali(t).ok())
5812 .unwrap_or(0)
5813 };
5814 let (start, end) = if let Some((s_str, e_str)) = key.split_once(',') {
5815 let s = ksh_shift(eval_bound(s_str));
5816 let e = ksh_shift(eval_bound(e_str));
5817 (start_translate(s), end_translate(e))
5818 } else {
5819 let i = ksh_shift(eval_bound(&key));
5820 if i == 0 {
5821 return;
5822 }
5823 let n = start_translate(i);
5824 (n, n)
5825 };
5826 // c:Src/params.c:3518-3520 (assignaparam ASSPM_AUGMENT) — a
5827 // subscripted `+=` to an array does NOT prepend the old slice;
5828 // it collapses the range to an EMPTY range positioned right
5829 // AFTER the slice end (`v->start = v->end--`) and splices in
5830 // ONLY the new value: `a[2]+=(d)` on (a b c) → (a b d c);
5831 // `a[2,3]+=(x)` on (1 2 3 4) → (1 2 3 x 4). In setarrvalue's
5832 // 1-based convention here that means start = end+1 (so
5833 // start_idx == end_idx == end → splice arr[end..end]).
5834 let (start, end) = if append && end > 0 {
5835 (end + 1, end)
5836 } else {
5837 (start, end)
5838 };
5839 // c:Src/params.c:392-430 IPDEF9("argv"/"@"/"*", &pparams) —
5840 // the positional parameters live in the `pparams` vector, NOT
5841 // paramtab, so a subscript splice (`argv[2]=(X Y Z)`,
5842 // `2=(X Y Z)`) must read/write pparams. Splice a synthetic
5843 // array param holding the current positionals via the
5844 // canonical setarrvalue, then store the result back to
5845 // pparams — mirroring assignaparam's argv/@/* special-case
5846 // (params.rs:6937) for the whole-array form.
5847 if name == "argv" || name == "@" || name == "*" {
5848 let mut pm = {
5849 crate::ported::params::createparam(
5850 &name,
5851 crate::ported::zsh_h::PM_ARRAY as i32,
5852 );
5853 crate::ported::params::paramtab()
5854 .write()
5855 .ok()
5856 .and_then(|mut t| t.remove(&name))
5857 };
5858 if let Some(ref mut p) = pm {
5859 p.u_arr = Some(exec.pparams());
5860 }
5861 let mut v = crate::ported::zsh_h::value {
5862 pm,
5863 arr: Vec::new(),
5864 scanflags: 0,
5865 valflags: 0,
5866 start,
5867 end,
5868 };
5869 crate::ported::params::setarrvalue(&mut v, values);
5870 let result = v.pm.and_then(|p| p.u_arr).unwrap_or_default();
5871 exec.set_pparams(result);
5872 return;
5873 }
5874 // Route through canonical setarrvalue (Src/params.c:2895).
5875 // It handles PM_READONLY rejection, PM_HASHED slice-error,
5876 // PM_ARRAY splice + bounds clamp + padding (c:2980+).
5877 let taken = match crate::ported::params::paramtab().write() {
5878 Ok(mut tab) => tab.remove(&name),
5879 Err(_) => None,
5880 };
5881 // c:Src/exec.c:2640 / getvalue(…, 1) — a subscript assignment to
5882 // a NONEXISTENT parameter auto-creates it. getindex/fetchvalue
5883 // with the create flag calls createparam(name, PM_ARRAY) so the
5884 // splice has an array to write into; `unset u; u[1,2]=(a z)`
5885 // then yields the array (a z). Without this, setarrvalue saw
5886 // v.pm == None and silently stored nothing. The single-index
5887 // scalar-value path (SET_ASSOC/SET_ARRAY_AT) already vivifies;
5888 // this brings the range/array-value path to parity.
5889 let taken = taken.or_else(|| {
5890 crate::ported::params::createparam(&name, crate::ported::zsh_h::PM_ARRAY as i32);
5891 crate::ported::params::paramtab()
5892 .write()
5893 .ok()
5894 .and_then(|mut t| t.remove(&name))
5895 });
5896 // c:Src/params.c:2748+ — PM_SCALAR with subscript range
5897 // SPLICES the value into the scalar's char string. Bug
5898 // #589: zshrs's slice handler always called setarrvalue,
5899 // erroring "attempt to assign array value to non-array"
5900 // for `a=hello; a[2,3]=XYZ`. Detect PM_SCALAR and route
5901 // through assignstrvalue (which does scalar splice via
5902 // the PM_SCALAR arm at params.rs:3709-3789).
5903 let is_scalar = taken.as_ref().map_or(false, |pm| {
5904 crate::ported::zsh_h::PM_TYPE(pm.node.flags as u32)
5905 == crate::ported::zsh_h::PM_SCALAR
5906 });
5907 let mut v = crate::ported::zsh_h::value {
5908 pm: taken,
5909 arr: Vec::new(),
5910 scanflags: 0,
5911 valflags: 0,
5912 start,
5913 end,
5914 };
5915 if is_scalar {
5916 // Scalar splice — concat values, route through
5917 // assignstrvalue which dispatches by PM_TYPE.
5918 // start_translate returns 1-based positions; assignstrvalue's
5919 // PM_SCALAR arm at params.rs:3735+ expects 0-based start
5920 // (chars before start are kept) and 0-based end-exclusive
5921 // (chars from end are kept). Convert: start-=1.
5922 if v.start > 0 {
5923 v.start -= 1;
5924 }
5925 let val: String = values.join("");
5926 crate::ported::params::assignstrvalue(Some(&mut v), Some(val), 0);
5927 } else {
5928 crate::ported::params::setarrvalue(&mut v, values);
5929 }
5930 // Write the mutated Param back to paramtab — setarrvalue
5931 // mutated v.pm in-place; the prior `tab.remove(&name)` at
5932 // the top of this handler took ownership, so we re-insert
5933 // here. setarrvalue + this re-insert IS the canonical
5934 // store (Src/params.c:2895). No further mirror needed.
5935 if let Some(pm) = v.pm {
5936 if let Ok(mut tab) = crate::ported::params::paramtab().write() {
5937 tab.insert(name, pm);
5938 }
5939 }
5940 });
5941 Value::Status(0)
5942 });
5943
5944 // BUILTIN_CONCAT_SPLICE — word-segment concat with first/last
5945 // sticking (default zsh splice semantics for `${arr[@]}`, `$@`).
5946 vm.register_builtin(BUILTIN_CONCAT_SPLICE, |vm, _argc| {
5947 let rhs = vm.pop();
5948 let lhs = vm.pop();
5949 match (lhs, rhs) {
5950 (Value::Array(mut la), Value::Array(ra)) => {
5951 if la.is_empty() {
5952 return Value::Array(ra);
5953 }
5954 if ra.is_empty() {
5955 return Value::Array(la);
5956 }
5957 // Last of la merges with first of ra; rest unchanged.
5958 let last_l = la.pop().unwrap();
5959 let mut ra_iter = ra.into_iter();
5960 let first_r = ra_iter.next().unwrap();
5961 let l_s = last_l.as_str_cow();
5962 let r_s = first_r.as_str_cow();
5963 let mut merged = String::with_capacity(l_s.len() + r_s.len());
5964 merged.push_str(&l_s);
5965 merged.push_str(&r_s);
5966 la.push(Value::str(merged));
5967 la.extend(ra_iter);
5968 Value::Array(la)
5969 }
5970 (Value::Array(mut la), rhs_scalar) => {
5971 // c:Src/subst.c paramsubst splice — empty array on
5972 // either side preserves the empty (zero words),
5973 // doesn't collapse into a single-empty-string scalar.
5974 // Bug #120 in docs/BUGS.md: empty array slice
5975 // concatenated with empty literal returned
5976 // Value::str("") which surfaced as one empty arg
5977 // instead of zero args.
5978 let rhs_s = rhs_scalar.as_str_cow();
5979 if la.is_empty() {
5980 if rhs_s.is_empty() {
5981 return Value::Array(Vec::new());
5982 }
5983 return Value::str(rhs_s.to_string());
5984 }
5985 let last = la.pop().unwrap();
5986 let l_s = last.as_str_cow();
5987 let mut s = String::with_capacity(l_s.len() + rhs_s.len());
5988 s.push_str(&l_s);
5989 s.push_str(&rhs_s);
5990 la.push(Value::str(s));
5991 Value::Array(la)
5992 }
5993 (lhs_scalar, Value::Array(mut ra)) => {
5994 let lhs_s = lhs_scalar.as_str_cow();
5995 if ra.is_empty() {
5996 // Empty-array RHS — preserve emptiness when the
5997 // LHS is also empty (no prefix to attach). Bug
5998 // #120 in docs/BUGS.md.
5999 if lhs_s.is_empty() {
6000 return Value::Array(Vec::new());
6001 }
6002 return Value::str(lhs_s.to_string());
6003 }
6004 let first = ra.remove(0);
6005 let r_s = first.as_str_cow();
6006 let mut s = String::with_capacity(lhs_s.len() + r_s.len());
6007 s.push_str(&lhs_s);
6008 s.push_str(&r_s);
6009 let mut out = Vec::with_capacity(ra.len() + 1);
6010 out.push(Value::str(s));
6011 out.extend(ra);
6012 Value::Array(out)
6013 }
6014 (lhs_s, rhs_s) => {
6015 let l = lhs_s.as_str_cow();
6016 let r = rhs_s.as_str_cow();
6017 let mut s = String::with_capacity(l.len() + r.len());
6018 s.push_str(&l);
6019 s.push_str(&r);
6020 Value::str(s)
6021 }
6022 }
6023 });
6024
6025 // BUILTIN_CONCAT_DISTRIBUTE — word-segment concat. With
6026 // rcexpandparam (zsh option), distributes element-wise (cartesian
6027 // product). Default mode: joins arrays with IFS first char to a
6028 // single scalar before concat, matching zsh's default unquoted
6029 // and DQ semantics. Direct port of Src/subst.c sepjoin path
6030 // (line ~1813) which gates element-vs-join on the rc_expand_param
6031 // option, defaulting to join.
6032 // BUILTIN_CONCAT_DISTRIBUTE_FORCED — same shape as
6033 // CONCAT_DISTRIBUTE, but always cartesian-distributes when one
6034 // side is Array. Used for compile-time-detected explicit
6035 // distribution forms (`${^arr}` etc.) where the source flag
6036 // overrides the rcexpandparam option default.
6037 vm.register_builtin(BUILTIN_CONCAT_DISTRIBUTE_FORCED, |vm, _argc| {
6038 let rhs = vm.pop();
6039 let lhs = vm.pop();
6040 match (lhs, rhs) {
6041 (Value::Array(la), Value::Array(ra)) => {
6042 if ra.is_empty() {
6043 return Value::Array(la);
6044 }
6045 if la.is_empty() {
6046 return Value::Array(ra);
6047 }
6048 let mut out = Vec::with_capacity(la.len() * ra.len());
6049 for a in &la {
6050 let a_s = a.as_str_cow();
6051 for b in &ra {
6052 let b_s = b.as_str_cow();
6053 let mut s = String::with_capacity(a_s.len() + b_s.len());
6054 s.push_str(&a_s);
6055 s.push_str(&b_s);
6056 out.push(Value::str(s));
6057 }
6058 }
6059 Value::Array(out)
6060 }
6061 (Value::Array(la), rhs_scalar) => {
6062 // An EMPTY array contributes nothing to a concatenated
6063 // word — the surrounding scalar text survives. zsh:
6064 // `x${^a}y` (a=()) / `x${(P)scalar-empty}y` → "xy", NOT
6065 // a dropped word. Without this, a `(P)` indirect to an
6066 // unset/empty scalar (which nodes_to_value collapses to
6067 // Value::Array([]) for standalone-removal semantics)
6068 // cartesian-dropped the whole word — p10k's
6069 // `typeset -g _$2=${(P)2}` then arrived as a bare
6070 // `typeset` and dumped every parameter (~217× → 19 MB
6071 // terminal flood → startup hang).
6072 if la.is_empty() {
6073 return rhs_scalar;
6074 }
6075 let r = rhs_scalar.as_str_cow();
6076 let out: Vec<Value> = la
6077 .into_iter()
6078 .map(|a| {
6079 let a_s = a.as_str_cow();
6080 let mut s = String::with_capacity(a_s.len() + r.len());
6081 s.push_str(&a_s);
6082 s.push_str(&r);
6083 Value::str(s)
6084 })
6085 .collect();
6086 Value::Array(out)
6087 }
6088 (lhs_scalar, Value::Array(ra)) => {
6089 // Symmetric empty-array-contributes-nothing rule; see
6090 // the (Array, scalar) arm above.
6091 if ra.is_empty() {
6092 return lhs_scalar;
6093 }
6094 let l = lhs_scalar.as_str_cow();
6095 let out: Vec<Value> = ra
6096 .into_iter()
6097 .map(|b| {
6098 let b_s = b.as_str_cow();
6099 let mut s = String::with_capacity(l.len() + b_s.len());
6100 s.push_str(&l);
6101 s.push_str(&b_s);
6102 Value::str(s)
6103 })
6104 .collect();
6105 Value::Array(out)
6106 }
6107 (lhs_s, rhs_s) => {
6108 let l = lhs_s.as_str_cow();
6109 let r = rhs_s.as_str_cow();
6110 let mut s = String::with_capacity(l.len() + r.len());
6111 s.push_str(&l);
6112 s.push_str(&r);
6113 Value::str(s)
6114 }
6115 }
6116 });
6117
6118 vm.register_builtin(BUILTIN_CONCAT_DISTRIBUTE, |vm, argc| {
6119 let rhs = vm.pop();
6120 let lhs = vm.pop();
6121 // c:Src/options.c — RC_EXPAND_PARAM applies to UNQUOTED
6122 // expansions only; inside DQ `"$foo${arr}bar"` joins via
6123 // $IFS[0] regardless of the option. The compiler emits
6124 // CallBuiltin(BUILTIN_CONCAT_DISTRIBUTE, 1) when the parent
6125 // word is DQ-wrapped (compile_zsh.rs parent_is_dq); the
6126 // default UNQUOTED path emits argc=2 (lhs + rhs). Treat
6127 // argc==1 as "force rc_expand off." Bug #246 in docs/BUGS.md.
6128 let dq_suppress = argc == 1;
6129 let rc_expand =
6130 !dq_suppress && with_executor(|exec| opt_state_get("rcexpandparam").unwrap_or(false));
6131 // Helper: join an Array to scalar via sepjoin's IFS default.
6132 // c:Src/utils.c:3936-3945 — set-but-empty IFS joins with ""
6133 // (`IFS=""; echo "x$*y"` → `xabcy`); only unset /
6134 // space-leading IFS yields " ".
6135 let join_arr = |arr: Vec<Value>| -> String {
6136 let strs: Vec<String> = arr.iter().map(|v| v.as_str_cow().into_owned()).collect();
6137 crate::ported::utils::sepjoin(&strs, None)
6138 };
6139 if !rc_expand {
6140 // Default: join any Array side to scalar, then concat.
6141 let l = match lhs {
6142 Value::Array(a) => join_arr(a),
6143 other => other.as_str_cow().into_owned(),
6144 };
6145 let r = match rhs {
6146 Value::Array(a) => join_arr(a),
6147 other => other.as_str_cow().into_owned(),
6148 };
6149 let mut s = String::with_capacity(l.len() + r.len());
6150 s.push_str(&l);
6151 s.push_str(&r);
6152 return Value::str(s);
6153 }
6154 match (lhs, rhs) {
6155 (Value::Array(la), Value::Array(ra)) => {
6156 // Cartesian product: [a + b for a in la for b in ra].
6157 let mut out = Vec::with_capacity(la.len() * ra.len().max(1));
6158 if ra.is_empty() {
6159 return Value::Array(la);
6160 }
6161 if la.is_empty() {
6162 return Value::Array(ra);
6163 }
6164 for a in &la {
6165 let a_s = a.as_str_cow();
6166 for b in &ra {
6167 let b_s = b.as_str_cow();
6168 let mut s = String::with_capacity(a_s.len() + b_s.len());
6169 s.push_str(&a_s);
6170 s.push_str(&b_s);
6171 out.push(Value::str(s));
6172 }
6173 }
6174 Value::Array(out)
6175 }
6176 (Value::Array(la), rhs_scalar) => {
6177 // An EMPTY array contributes nothing to a concatenated
6178 // word — the surrounding scalar text survives. zsh:
6179 // `x${^a}y` (a=()) / `x${(P)scalar-empty}y` → "xy", NOT
6180 // a dropped word. Without this, a `(P)` indirect to an
6181 // unset/empty scalar (which nodes_to_value collapses to
6182 // Value::Array([]) for standalone-removal semantics)
6183 // cartesian-dropped the whole word — p10k's
6184 // `typeset -g _$2=${(P)2}` then arrived as a bare
6185 // `typeset` and dumped every parameter (~217× → 19 MB
6186 // terminal flood → startup hang).
6187 if la.is_empty() {
6188 return rhs_scalar;
6189 }
6190 let r = rhs_scalar.as_str_cow();
6191 let out: Vec<Value> = la
6192 .into_iter()
6193 .map(|a| {
6194 let a_s = a.as_str_cow();
6195 let mut s = String::with_capacity(a_s.len() + r.len());
6196 s.push_str(&a_s);
6197 s.push_str(&r);
6198 Value::str(s)
6199 })
6200 .collect();
6201 Value::Array(out)
6202 }
6203 (lhs_scalar, Value::Array(ra)) => {
6204 // Symmetric empty-array-contributes-nothing rule; see
6205 // the (Array, scalar) arm above.
6206 if ra.is_empty() {
6207 return lhs_scalar;
6208 }
6209 let l = lhs_scalar.as_str_cow();
6210 let out: Vec<Value> = ra
6211 .into_iter()
6212 .map(|b| {
6213 let b_s = b.as_str_cow();
6214 let mut s = String::with_capacity(l.len() + b_s.len());
6215 s.push_str(&l);
6216 s.push_str(&b_s);
6217 Value::str(s)
6218 })
6219 .collect();
6220 Value::Array(out)
6221 }
6222 (lhs_s, rhs_s) => {
6223 // Fast path: both scalar → identical to Op::Concat.
6224 let l = lhs_s.as_str_cow();
6225 let r = rhs_s.as_str_cow();
6226 let mut s = String::with_capacity(l.len() + r.len());
6227 s.push_str(&l);
6228 s.push_str(&r);
6229 Value::str(s)
6230 }
6231 }
6232 });
6233
6234 // `[[ a -ef b ]]` — same-inode test. Resolves both paths via fs::metadata
6235 // (follows symlinks the way zsh's -ef does) and compares (dev, inode).
6236 // Returns false on any I/O error (path missing, permission denied, etc.).
6237 vm.register_builtin(BUILTIN_SAME_FILE, |vm, _argc| {
6238 let b = vm.pop().to_str();
6239 let a = vm.pop().to_str();
6240 let same = match (fs::metadata(&a), fs::metadata(&b)) {
6241 (Ok(ma), Ok(mb)) => ma.dev() == mb.dev() && ma.ino() == mb.ino(),
6242 _ => false,
6243 };
6244 Value::Bool(same)
6245 });
6246
6247 // `[[ -c path ]]` — character device.
6248 vm.register_builtin(BUILTIN_IS_CHARDEV, |vm, _argc| {
6249 let path = vm.pop().to_str();
6250 let result = fs::metadata(&path)
6251 .map(|m| m.file_type().is_char_device())
6252 .unwrap_or(false);
6253 Value::Bool(result)
6254 });
6255 // `[[ -b path ]]` — block device.
6256 vm.register_builtin(BUILTIN_IS_BLOCKDEV, |vm, _argc| {
6257 let path = vm.pop().to_str();
6258 let result = fs::metadata(&path)
6259 .map(|m| m.file_type().is_block_device())
6260 .unwrap_or(false);
6261 Value::Bool(result)
6262 });
6263 // `[[ -p path ]]` — FIFO (named pipe).
6264 vm.register_builtin(BUILTIN_IS_FIFO, |vm, _argc| {
6265 let path = vm.pop().to_str();
6266 let result = fs::metadata(&path)
6267 .map(|m| m.file_type().is_fifo())
6268 .unwrap_or(false);
6269 Value::Bool(result)
6270 });
6271 // `[[ -S path ]]` — socket.
6272 vm.register_builtin(BUILTIN_IS_SOCKET, |vm, _argc| {
6273 let path = vm.pop().to_str();
6274 let result = fs::symlink_metadata(&path)
6275 .map(|m| m.file_type().is_socket())
6276 .unwrap_or(false);
6277 Value::Bool(result)
6278 });
6279
6280 // `[[ -k path ]]` / `-u` / `-g` — sticky / setuid / setgid bit.
6281 vm.register_builtin(BUILTIN_HAS_STICKY, |vm, _argc| {
6282 let path = vm.pop().to_str();
6283 let result = fs::metadata(&path)
6284 .map(|m| m.permissions().mode() & libc::S_ISVTX as u32 != 0)
6285 .unwrap_or(false);
6286 Value::Bool(result)
6287 });
6288 vm.register_builtin(BUILTIN_HAS_SETUID, |vm, _argc| {
6289 let path = vm.pop().to_str();
6290 let result = fs::metadata(&path)
6291 .map(|m| m.permissions().mode() & libc::S_ISUID as u32 != 0)
6292 .unwrap_or(false);
6293 Value::Bool(result)
6294 });
6295 vm.register_builtin(BUILTIN_HAS_SETGID, |vm, _argc| {
6296 let path = vm.pop().to_str();
6297 let result = fs::metadata(&path)
6298 .map(|m| m.permissions().mode() & libc::S_ISGID as u32 != 0)
6299 .unwrap_or(false);
6300 Value::Bool(result)
6301 });
6302 vm.register_builtin(BUILTIN_OWNED_BY_USER, |vm, _argc| {
6303 let path = vm.pop().to_str();
6304 let euid = unsafe { libc::geteuid() };
6305 let result = fs::metadata(&path)
6306 .map(|m| m.uid() == euid)
6307 .unwrap_or(false);
6308 Value::Bool(result)
6309 });
6310 vm.register_builtin(BUILTIN_OWNED_BY_GROUP, |vm, _argc| {
6311 let path = vm.pop().to_str();
6312 let egid = unsafe { libc::getegid() };
6313 let result = fs::metadata(&path)
6314 .map(|m| m.gid() == egid)
6315 .unwrap_or(false);
6316 Value::Bool(result)
6317 });
6318
6319 // `[[ -N path ]]` — file's access time is NOT newer than its
6320 // modification time (zsh man: "true if file exists and its
6321 // access time is not newer than its modification time"). Used
6322 // by zsh's mailbox-watching code. The semantic is `atime <=
6323 // mtime` (equivalent to `mtime >= atime`) — equal counts as
6324 // true, which a strict `mtime > atime` check missed for newly
6325 // created files where both stamps are identical.
6326 vm.register_builtin(BUILTIN_FILE_MODIFIED_SINCE_ACCESS, |vm, _argc| {
6327 let path = vm.pop().to_str();
6328 let result = fs::metadata(&path)
6329 .map(|m| m.atime() <= m.mtime())
6330 .unwrap_or(false);
6331 Value::Bool(result)
6332 });
6333
6334 // `[[ a -nt b ]]` — true if `a`'s mtime is strictly later than `b`'s.
6335 // BOTH files must exist; if either is missing the result is false.
6336 // (Earlier behavior was bash's "missing == infinitely-old"; zsh
6337 // strictly requires both files to exist.)
6338 vm.register_builtin(BUILTIN_FILE_NEWER, |vm, _argc| {
6339 let b = vm.pop().to_str();
6340 let a = vm.pop().to_str();
6341 // Use SystemTime modified() for nanosecond precision —
6342 // MetadataExt::mtime() returns seconds only, so two files
6343 // touched within the same second compared equal even when
6344 // 500ms apart. zsh tracks ns and uses `>=` for ties (touching
6345 // a then b in quick succession should still report b newer).
6346 let ta = fs::metadata(&a).and_then(|m| m.modified()).ok();
6347 let tb = fs::metadata(&b).and_then(|m| m.modified()).ok();
6348 let result = match (ta, tb) {
6349 (Some(ta), Some(tb)) => ta > tb,
6350 _ => false,
6351 };
6352 Value::Bool(result)
6353 });
6354
6355 // `[[ a -ot b ]]` — mirror of -nt. Same both-must-exist contract.
6356 vm.register_builtin(BUILTIN_FILE_OLDER, |vm, _argc| {
6357 let b = vm.pop().to_str();
6358 let a = vm.pop().to_str();
6359 let ta = fs::metadata(&a).and_then(|m| m.modified()).ok();
6360 let tb = fs::metadata(&b).and_then(|m| m.modified()).ok();
6361 let result = match (ta, tb) {
6362 (Some(ta), Some(tb)) => ta < tb,
6363 _ => false,
6364 };
6365 Value::Bool(result)
6366 });
6367
6368 // `set -e` / `setopt errexit` post-command check. Compiler emits
6369 // this after each top-level command's SetStatus (skipped inside
6370 // conditionals/pipelines/&&||/`!`). If errexit is on AND the last
6371 // command exited non-zero AND it's not a `return` from a function,
6372 // exit the shell with that status.
6373 // `set -x` / `setopt xtrace` — print each command before it runs.
6374 // The compiler emits this BEFORE the actual builtin/external call
6375 // with the command's literal text as a single string arg. We
6376 // print to stderr if xtrace is on. Honors `$PS4` (default `+ `).
6377 //
6378 // ── XTRACE flow control ────────────────────────────────────────
6379 // Mirror of C zsh's `doneps4` flag in execcmd_exec (Src/exec.c).
6380 // When an assignment trace fires (XTRACE_ASSIGN), it emits PS4
6381 // and sets this flag so the subsequent XTRACE_ARGS skips its own
6382 // PS4 emission — the assignment + command end up on the SAME
6383 // line: `<PS4>a=1 echo hello\n`. XTRACE_ARGS / XTRACE_NEWLINE
6384 // reset the flag after emitting the trailing `\n`.
6385 vm.register_builtin(BUILTIN_XTRACE_IS_ON, |_vm, _argc| {
6386 // Push live xtrace state. Caller pairs this with JumpIfFalse
6387 // to skip the trace-string-building block when xtrace is off,
6388 // avoiding side-effectful operand re-evaluation. Bug #159 in
6389 // docs/BUGS.md.
6390 let on = with_executor(|_| opt_state_get("xtrace").unwrap_or(false));
6391 Value::Int(if on { 1 } else { 0 })
6392 });
6393
6394 vm.register_builtin(BUILTIN_XTRACE_LINE, |vm, _argc| {
6395 let cmd_text = vm.pop().to_str();
6396 // Sync exec.last_status with the live vm.last_status BEFORE
6397 // the next command runs. Direct port of the zsh exec.c
6398 // contract — `$?` reads the exit status of the *most recent*
6399 // command. XTRACE_LINE is emitted by the compiler BEFORE
6400 // every simple command, so it's the natural sync point.
6401 let live = vm.last_status;
6402 with_executor(|exec| {
6403 exec.set_last_status(live);
6404 });
6405 // C zsh emits xtrace for `(( … ))` / `[[ … ]]` / `case` /
6406 // `if/while/until/for/repeat` head expressions via
6407 // `printprompt4(); fprintf(xtrerr, "%s\n", expr)` at
6408 // Src/exec.c:5240 (math), c:5286 (cond), c:4117 (for), etc.
6409 // The compiler emits BUILTIN_XTRACE_LINE only at those
6410 // construct boundaries (compile_arith / compile_cond /
6411 // compile_if / compile_while / compile_for / compile_case);
6412 // simple commands route to BUILTIN_XTRACE_ARGS instead. So
6413 // this handler always emits when xtrace is on — no prefix-
6414 // string heuristic.
6415 let on = with_executor(|exec| opt_state_get("xtrace").unwrap_or(false));
6416 if on {
6417 let already = XTRACE_DONE_PS4.with(|f| f.get());
6418 if !already {
6419 printprompt4();
6420 }
6421 // c:exec.c:5240/5286 — `fprintf(xtrerr, "%s\n", expr)`. Buffer
6422 // the line + newline, flush once (single write).
6423 xtrerr_fputs(&cmd_text);
6424 xtrerr_fputs("\n");
6425 xtrerr_flush();
6426 XTRACE_DONE_PS4.with(|f| f.set(false));
6427 }
6428 Value::Status(0)
6429 });
6430
6431 // Like XTRACE_LINE but reads the top `argc - 1` values from the
6432 // VM stack WITHOUT consuming them (peek), then pops a prefix
6433 // string at the top. Joins prefix + peeked args with spaces using
6434 // zsh's quotedzputs-equivalent quoting. Direct port of
6435 // Src/exec.c:2055-2066 — emit AFTER expansion, with each arg
6436 // shell-quoted, so `for i in a b; echo for $i` traces as
6437 // `echo for a` / `echo for b`, not `echo for $i`.
6438 //
6439 // Stack contract on entry: [arg1, arg2, ..., argN, prefix].
6440 // Pops prefix; peeks argN..arg1 below. argc = N + 1.
6441 vm.register_builtin(BUILTIN_XTRACE_ARGS, |vm, argc| {
6442 let prefix = vm.pop().to_str();
6443 let live = vm.last_status;
6444 with_executor(|exec| {
6445 exec.set_last_status(live);
6446 });
6447 let on = with_executor(|exec| opt_state_get("xtrace").unwrap_or(false));
6448 if on {
6449 let n_args = argc.saturating_sub(1) as usize;
6450 let len = vm.stack.len();
6451 // c:Src/exec.c:2055 — argv is the POST-expansion word
6452 // list, so an arg that expanded to multiple words splats
6453 // into multiple trace tokens AND an arg that expanded to
6454 // zero words (empty unquoted `${UNSET}`) emits nothing.
6455 // pop_args (line 6243) already does this splat for the
6456 // real handler; mirror the same Array → splat / empty →
6457 // drop logic here so xtrace renders `echo ${UNSET}` as
6458 // `echo` (zsh) instead of `echo ''` (the previous
6459 // single-arg stringify path returned "" and then
6460 // quotedzputs wrapped it in `''`).
6461 let arg_strs: Vec<String> = if n_args > 0 && len >= n_args {
6462 let mut out = Vec::new();
6463 for v in &vm.stack[len - n_args..] {
6464 match v {
6465 Value::Array(items) => {
6466 for item in items {
6467 out.push(quotedzputs(&item.to_str()));
6468 }
6469 }
6470 other => out.push(quotedzputs(&other.to_str())),
6471 }
6472 }
6473 out
6474 } else {
6475 Vec::new()
6476 };
6477 // Builtins dispatch through `execbuiltin` (Src/builtin.c:442)
6478 // which emits its own PS4 + name + args xtrace. To avoid
6479 // double-emission, skip our emission here when the first
6480 // arg is a known builtin with a registered HandlerFunc —
6481 // those go through execbuiltin and will trace themselves.
6482 // Externals + builtins-not-yet-routed-through-execbuiltin
6483 // keep our emission as a stand-in.
6484 let goes_through_execbuiltin = crate::ported::builtin::BUILTINS
6485 .iter()
6486 .any(|b| b.node.nam == prefix && b.handlerfunc.is_some());
6487 if !goes_through_execbuiltin {
6488 let line = if arg_strs.is_empty() {
6489 prefix
6490 } else {
6491 format!("{} {}", prefix, arg_strs.join(" "))
6492 };
6493 // Mirrors Src/exec.c:2055 xtrace emission. C does:
6494 // if (!doneps4) printprompt4();
6495 // ... emit args + spaces ...
6496 // fputc('\n', xtrerr); fflush(xtrerr);
6497 // printprompt4 + the args + `\n` all land in the xtrerr
6498 // buffer; the single fflush below writes the whole line in
6499 // one syscall so concurrent pipeline stages never
6500 // interleave (c:makecline:2122-2123).
6501 let already_ps4 = XTRACE_DONE_PS4.with(|f| f.get());
6502 if !already_ps4 {
6503 printprompt4();
6504 }
6505 xtrerr_fputs(&line);
6506 xtrerr_fputs("\n"); // c:2122 fputc('\n', xtrerr)
6507 xtrerr_flush(); // c:2123 fflush(xtrerr)
6508 }
6509 XTRACE_DONE_PS4.with(|f| f.set(false));
6510 }
6511 Value::Status(0)
6512 });
6513
6514 // BUILTIN_XTRACE_ASSIGN — direct port of the per-assignment
6515 // trace block at Src/exec.c:2517-2582. C body excerpt:
6516 // xtr = isset(XTRACE);
6517 // if (xtr) { printprompt4(); doneps4 = 1; }
6518 // while (assign) {
6519 // if (xtr) fprintf(xtrerr, "%s+=" or "%s=", name);
6520 // ... eval value into `val` ...
6521 // if (xtr) { quotedzputs(val, xtrerr); fputc(' ', xtrerr); }
6522 // ...
6523 // }
6524 //
6525 // Stack on entry: [..., name, value]. PEEKS both (they're left
6526 // on stack for SET_VAR to pop). Emits `name=<quoted-val> ` with
6527 // no newline; trailing `\n` comes from XTRACE_ARGS (cmd path)
6528 // or XTRACE_NEWLINE (assignment-only path).
6529 vm.register_builtin(BUILTIN_XTRACE_ASSIGN, |vm, _argc| {
6530 let on = with_executor(|exec| opt_state_get("xtrace").unwrap_or(false));
6531 if on {
6532 // PEEK [..., name, value] — argc==2 by contract.
6533 let len = vm.stack.len();
6534 if len >= 2 {
6535 let name = vm.stack[len - 2].to_str();
6536 let value = vm.stack[len - 1].to_str();
6537 let already_ps4 = XTRACE_DONE_PS4.with(|f| f.get());
6538 if !already_ps4 {
6539 printprompt4();
6540 XTRACE_DONE_PS4.with(|f| f.set(true));
6541 }
6542 // C: `fprintf(xtrerr, "%s=", name)` then `quotedzputs
6543 // (val); fputc(' ', xtrerr);`. Append to the xtrerr buffer
6544 // (no newline / no flush — the line continues with the
6545 // command via XTRACE_ARGS, or ends at XTRACE_NEWLINE).
6546 xtrerr_fputs(&format!("{}={} ", name, quotedzputs(&value)));
6547 }
6548 }
6549 Value::Status(0)
6550 });
6551
6552 // BUILTIN_XTRACE_NEWLINE — emit trailing `\n` + flush iff a
6553 // prior XTRACE_ASSIGN this line already emitted PS4. Mirrors
6554 // C's `fputc('\n', xtrerr); fflush(xtrerr);` at exec.c:3398
6555 // (the assignment-only path through execcmd_exec).
6556 vm.register_builtin(BUILTIN_XTRACE_NEWLINE, |_vm, _argc| {
6557 let on = with_executor(|exec| opt_state_get("xtrace").unwrap_or(false));
6558 if on {
6559 let already_ps4 = XTRACE_DONE_PS4.with(|f| f.get());
6560 if already_ps4 {
6561 xtrerr_fputs("\n"); // c:3398 fputc('\n', xtrerr)
6562 xtrerr_flush(); // c:3398 fflush(xtrerr)
6563 XTRACE_DONE_PS4.with(|f| f.set(false));
6564 }
6565 }
6566 Value::Status(0)
6567 });
6568
6569 // c:Src/exec.c WC_TRYBLOCK — post-always re-jump probes. Each
6570 // returns 1 + consumes the atomic when the corresponding
6571 // escape flag is set; the try-block compile pairs each with
6572 // a JumpIfFalse + Jump → outer scope's return / break /
6573 // continue patches.
6574 vm.register_builtin(BUILTIN_RETFLAG_CHECK, |_vm, _argc| {
6575 use std::sync::atomic::Ordering;
6576 let r = crate::ported::builtin::RETFLAG.load(Ordering::Relaxed);
6577 if r != 0 {
6578 // Don't clear here — doshfunc owns the clear at c:6047
6579 // when the function unwinds. Leaving it set propagates
6580 // through nested `eval`/`source` callers correctly.
6581 Value::Int(1)
6582 } else {
6583 Value::Int(0)
6584 }
6585 });
6586 vm.register_builtin(BUILTIN_BREAKS_CHECK, |_vm, _argc| {
6587 use std::sync::atomic::Ordering;
6588 let b = crate::ported::builtin::BREAKS.load(Ordering::Relaxed);
6589 let c = crate::ported::builtin::CONTFLAG.load(Ordering::Relaxed);
6590 // `break` sets BREAKS but NOT CONTFLAG; `continue` sets both.
6591 // Filter out the continue path here so the two checks are
6592 // mutually exclusive.
6593 if b != 0 && c == 0 {
6594 // Consume BREAKS so the outer loop's break_patches
6595 // landing doesn't double-decrement.
6596 crate::ported::builtin::BREAKS.store(0, Ordering::Relaxed);
6597 Value::Int(1)
6598 } else {
6599 Value::Int(0)
6600 }
6601 });
6602 vm.register_builtin(BUILTIN_CONTFLAG_CHECK, |_vm, _argc| {
6603 use std::sync::atomic::Ordering;
6604 let c = crate::ported::builtin::CONTFLAG.load(Ordering::Relaxed);
6605 if c != 0 {
6606 crate::ported::builtin::CONTFLAG.store(0, Ordering::Relaxed);
6607 crate::ported::builtin::BREAKS.store(0, Ordering::Relaxed);
6608 Value::Int(1)
6609 } else {
6610 Value::Int(0)
6611 }
6612 });
6613 vm.register_builtin(BUILTIN_NOEXEC_CHECK, |_vm, _argc| {
6614 // c:Src/exec.c:1390 — `set -n` / `noexec` option: parse but
6615 // don't execute. Returns Int(1) when noexec is set so the
6616 // emit-side JumpIfTrue skips the statement body.
6617 if opt_state_get("noexec").unwrap_or(false) {
6618 return Value::Int(1);
6619 }
6620 // c:Src/exec.c:1390 — execlist's list-loop gate:
6621 // `while (wc_code(code) == WC_LIST && !breaks && !retflag
6622 // && !errflag)`
6623 // — once errflag is set, the NEXT sublist never starts, so
6624 // lastval survives untouched to the shell exit. Without this
6625 // prologue gate the follow-up statement RAN, its dispatch
6626 // saw errflag, returned 1, and SetStatus clobbered lastval —
6627 // `[[ x == [a- ]]; print rc=$?` exited 1 instead of zsh's 2
6628 // (the cond syntax error set lastval=2 per exec.c:5216-5221).
6629 if (crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed)
6630 & crate::ported::zsh_h::ERRFLAG_ERROR)
6631 != 0
6632 {
6633 return Value::Int(1);
6634 }
6635 Value::Int(0)
6636 });
6637 vm.register_builtin(BUILTIN_DONETRAP_RESET, |_vm, _argc| {
6638 // c:Src/exec.c:1455 — `donetrap = 0;` at sublist start.
6639 // Reset before each top-level statement so the next
6640 // sublist's ERREXIT_CHECK fires the ZERR trap on its FIRST
6641 // non-zero command. Carries the "already fired" state
6642 // across function-call returns within the SAME outer
6643 // sublist (per C semantics — donetrap is process-global).
6644 // Bug #303 in docs/BUGS.md.
6645 crate::ported::exec::DONETRAP.store(0, std::sync::atomic::Ordering::Relaxed);
6646 Value::Status(0)
6647 });
6648
6649 // `[[ -z X ]]` / `[[ -n X ]]` — pop one Value, route through
6650 // canonical `src/ported/cond.rs::evalcond` so the actual
6651 // empty/non-empty test reuses the C-port at `cond.rs:270-271`
6652 // (`'n' => !arg.is_empty()`, `'z' => arg.is_empty()`).
6653 //
6654 // The Array→args conversion lives at the bridge because cond.rs
6655 // expects `&[&str]` (C `cond_str` signature equivalent). For
6656 // `"${arr[@]}"` in DQ context the splice yields `Value::Array`
6657 // — an empty array still expands to one implicit empty word
6658 // (per zsh's "${arr[@]}" splat preserving at least one slot
6659 // in cond context), so:
6660 // - Array(0) → ["-z", ""] → evalcond → 0 (true)
6661 // - Array(1) → ["-z", word] → evalcond → 0/1
6662 // - Array(2+) → ["-z", w1, w2, ...] → evalcond → 2 (parse
6663 // error: too many ops)
6664 // → coerced to false
6665 // - Str(s) → ["-z", s] → evalcond → 0/1
6666 //
6667 // Bug #185 in docs/BUGS.md.
6668 fn run_cond_str_empty(v: Value, op: &str) -> Value {
6669 let words: Vec<String> = match v {
6670 Value::Array(arr) => arr.into_iter().map(|x| x.to_str()).collect(),
6671 Value::Str(s) => vec![s.to_string()],
6672 other => vec![other.to_str()],
6673 };
6674 let mut args: Vec<&str> = vec![op];
6675 if words.is_empty() {
6676 args.push("");
6677 } else {
6678 args.extend(words.iter().map(|s| s.as_str()));
6679 }
6680 let opts: std::collections::HashMap<String, bool> = std::collections::HashMap::new();
6681 let vars: std::collections::HashMap<String, String> = std::collections::HashMap::new();
6682 // c:Src/cond.c:62-66 — `evalcond` returns 0=true, 1=false,
6683 // 2=syntax-error. Coerce error to false (observable behavior
6684 // in zsh: `[[ -z a b ]]` errors and the test as a whole
6685 // returns non-zero).
6686 // `[[ ]]` dispatch — C's `evalcond(state, NULL)` calling convention.
6687 // `None` for from_test → mathevali integer-compare coercion path.
6688 let ret = crate::ported::cond::evalcond(&args, &opts, &vars, false, None);
6689 Value::Int(if ret == 0 { 1 } else { 0 })
6690 }
6691 vm.register_builtin(BUILTIN_COND_STR_EMPTY, |vm, _argc| {
6692 let v = vm.pop();
6693 run_cond_str_empty(v, "-z")
6694 });
6695 vm.register_builtin(BUILTIN_COND_STR_NONEMPTY, |vm, _argc| {
6696 let v = vm.pop();
6697 run_cond_str_empty(v, "-n")
6698 });
6699
6700 // `exec N<<<"str"` — herestring redirect to explicit fd, applied
6701 // permanently. Direct port of `Src/exec.c:4655 getherestr` +
6702 // `addfd(forked, save, mfds, fn->fd1, fil, 0, ...)` at c:3766-
6703 // 3780 for the nullexec=1 bare-exec-redir path. Bug #205 in
6704 // docs/BUGS.md.
6705 vm.register_builtin(BUILTIN_EXEC_HERESTR_FD, |vm, _argc| {
6706 let fd = vm.pop().to_int() as i32;
6707 let content = vm.pop().to_str();
6708 // c:4671-4672 — append `\n` for "real" herestrings (not
6709 // heredoc-derived). zshrs's bare-exec path only fires for
6710 // the `<<<` syntax (REDIR_HERESTR), so always append.
6711 let body = format!("{}\n", content);
6712 // c:4673-4679 — gettempfile → write_loop → close → reopen
6713 // read-only → unlink. Rust equivalent via tempfile crate or
6714 // explicit O_TMPFILE; use mkstemp + unlink-immediately to
6715 // mirror C exactly.
6716 use std::ffi::CString;
6717 let mut tmpl: Vec<u8> = b"/tmp/zshrs_hs_XXXXXX\0".to_vec();
6718 let write_fd = unsafe { libc::mkstemp(tmpl.as_mut_ptr() as *mut libc::c_char) };
6719 if write_fd < 0 {
6720 crate::ported::utils::zwarn(&format!(
6721 "can't create temp file for here document: {}",
6722 std::io::Error::last_os_error()
6723 ));
6724 return Value::Status(1);
6725 }
6726 // c:4675 — write_loop(fd, t, len)
6727 let bytes = body.as_bytes();
6728 let mut off = 0;
6729 while off < bytes.len() {
6730 let n = unsafe {
6731 libc::write(
6732 write_fd,
6733 bytes[off..].as_ptr() as *const libc::c_void,
6734 bytes.len() - off,
6735 )
6736 };
6737 if n <= 0 {
6738 unsafe { libc::close(write_fd) };
6739 return Value::Status(1);
6740 }
6741 off += n as usize;
6742 }
6743 unsafe { libc::close(write_fd) }; // c:4676
6744 // Path null-terminated by mkstemp; reopen for reading.
6745 let read_fd = unsafe { libc::open(tmpl.as_ptr() as *const libc::c_char, libc::O_RDONLY) };
6746 // c:4678 — unlink immediately so the file disappears on
6747 // close, leaving only the fd reference.
6748 unsafe { libc::unlink(tmpl.as_ptr() as *const libc::c_char) };
6749 if read_fd < 0 {
6750 return Value::Status(1);
6751 }
6752 // c:3779 addfd → dup2 to target fd, close intermediate.
6753 let r = unsafe { libc::dup2(read_fd, fd) };
6754 unsafe { libc::close(read_fd) };
6755 if r < 0 {
6756 return Value::Status(1);
6757 }
6758 Value::Status(0)
6759 });
6760 // c:Src/exec.c:2418 + addfd splice — MULTIOS fan-out. Stack
6761 // layout pushed by compile_zsh's coalescing pass:
6762 // [target_1, op_byte_1, target_2, op_byte_2, …, target_N,
6763 // op_byte_N, fd]
6764 // argc = 2N + 1. Pops, opens every target, sets up a pipe +
6765 // splitter thread that reads pipe → writes every chunk to
6766 // every opened target, dup2's pipe-write-end onto fd. The
6767 // splitter is closed + joined by host_redirect_scope_end.
6768 // Bug #36 in docs/BUGS.md.
6769 vm.register_builtin(BUILTIN_MULTIOS_REDIRECT, |vm, argc| {
6770 if argc < 3 || argc % 2 == 0 {
6771 // Bad shape — bail.
6772 return Value::Status(1);
6773 }
6774 // Pop fd first (top of stack).
6775 let fd = vm.pop().to_int() as i32;
6776 // Then pop (op, target) pairs in reverse compile order. Keep
6777 // targets as Values — a glob-bearing target arrives as a
6778 // Value::Array of matches.
6779 let n_targets = ((argc - 1) / 2) as usize;
6780 let mut pairs: Vec<(u8, Value)> = Vec::with_capacity(n_targets);
6781 for _ in 0..n_targets {
6782 let op_byte = vm.pop().to_int() as u8;
6783 let target = vm.pop();
6784 pairs.push((op_byte, target));
6785 }
6786 // Restore compile order (target_1 first).
6787 pairs.reverse();
6788
6789 // c:Src/glob.c:2195-2203 xpandredir — "Loop over matches,
6790 // duplicating the redirection for each file found": a glob
6791 // target with N matches becomes N members of the same multio
6792 // (`echo hi > *.txt` with two matches writes both files).
6793 let mut entries: Vec<(u8, String)> = Vec::with_capacity(pairs.len());
6794 for (op_byte, target) in pairs {
6795 match target {
6796 Value::Array(items) => {
6797 for item in items {
6798 entries.push((op_byte, item.to_str()));
6799 }
6800 }
6801 other => entries.push((op_byte, other.to_str())),
6802 }
6803 }
6804 if entries.is_empty() {
6805 return Value::Status(1);
6806 }
6807
6808 // c:Src/exec.c:2418 — `else if (!mfds[fd1] || unset(MULTIOS))`:
6809 // with MULTIOS unset every redirect takes the REPLACE path in
6810 // script order — each target is still opened (created /
6811 // truncated) and dup2'd over the fd, so the LAST one wins and
6812 // earlier files end up empty (`unsetopt multios; print x > a
6813 // > b` leaves `a` empty, `x` in `b`). host_apply_redirect is
6814 // exactly one replace step, noclobber gate included.
6815 let multios_on = opt_state_get("multios").unwrap_or(true);
6816 if !multios_on {
6817 with_executor(|exec| {
6818 for (op_byte, target) in &entries {
6819 exec.host_apply_redirect(fd as u8, *op_byte, target);
6820 if exec.redirect_failed {
6821 // c:Src/exec.c execerr — abort the remaining
6822 // redirect list on failure.
6823 break;
6824 }
6825 }
6826 });
6827 return Value::Status(0);
6828 }
6829
6830 if entries.len() == 1 {
6831 // Single member after splicing — a plain replace
6832 // (c:2418 new-multio arm). Route through
6833 // host_apply_redirect so the noclobber gate, the
6834 // pipeline-output split partial, and error handling all
6835 // apply exactly as for an un-bagged redirect.
6836 let (op_byte, target) = &entries[0];
6837 with_executor(|exec| {
6838 exec.host_apply_redirect(fd as u8, *op_byte, target);
6839 });
6840 return Value::Status(0);
6841 }
6842
6843 // c:Src/exec.c:3722-3724 — when this command's stdout IS the
6844 // pipeline output, C seeds mfds[1] with the pipe BEFORE
6845 // walking the redirect list, so the pipe is the multio's
6846 // first member (`print x >&1 > f | cat` sends `x` down the
6847 // pipe TWICE: once for the seed, once for the `>&1` dup).
6848 let pipe_seed = fd == 1
6849 && with_executor(|exec| {
6850 exec.pipe_output_scope
6851 .is_some_and(|d| d + 1 == exec.redirect_scope_stack.len())
6852 });
6853
6854 // Save current fd state for scope-end restoration — BEFORE
6855 // the first member's replace dup2 below.
6856 let saved = unsafe { libc::dup(fd) };
6857 if saved >= 0 {
6858 with_executor(|exec| {
6859 if let Some(top) = exec.redirect_scope_stack.last_mut() {
6860 top.push((fd, saved));
6861 } else {
6862 unsafe { libc::close(saved) };
6863 }
6864 });
6865 }
6866
6867 // Accumulate member fds in redirect order. c:Src/exec.c:
6868 // 2447-2480 addfd — the FIRST member REPLACES the fd
6869 // (c:2448-2450 `mfds[fd1]->ct=1; mfds[fd1]->fds[0]=fd1;`), so
6870 // a later numeric `>&N` self-dup resolves against the fd's
6871 // value at that point in the sequence: `print x > f >&1`
6872 // writes f TWICE; `print x >&1 > f` writes the ORIGINAL
6873 // stdout + f.
6874 let mut target_fds: Vec<i32> = Vec::with_capacity(entries.len() + 1);
6875 if pipe_seed {
6876 let p = unsafe { libc::dup(fd) };
6877 if p >= 0 {
6878 target_fds.push(p);
6879 }
6880 }
6881 let noclobber = opt_state_get("noclobber").unwrap_or(false)
6882 || !opt_state_get("clobber").unwrap_or(true);
6883 for (i, (op_byte, target)) in entries.iter().enumerate() {
6884 let open_result: std::io::Result<i32> = match *op_byte {
6885 r::DUP_WRITE | r::DUP_READ => {
6886 // Numeric `>&N` — dup the LIVE fd N (after any
6887 // earlier member's replace).
6888 match target.trim_start_matches('&').parse::<i32>() {
6889 Ok(src) => {
6890 let d = unsafe { libc::dup(src) };
6891 if d >= 0 {
6892 Ok(d)
6893 } else {
6894 Err(std::io::Error::last_os_error())
6895 }
6896 }
6897 Err(_) => Err(std::io::Error::from_raw_os_error(libc::EBADF)),
6898 }
6899 }
6900 r::WRITE => {
6901 // c:Src/exec.c clobber_open — noclobber applies
6902 // to multio file targets too; failure aborts the
6903 // remaining redirect list (execerr), so `setopt
6904 // noclobber; touch a; print x > a > b` errors on
6905 // `a` and never creates `b`.
6906 let target_meta = std::fs::metadata(target).ok();
6907 let target_is_regular_file = target_meta
6908 .as_ref()
6909 .map(|m| m.file_type().is_file())
6910 .unwrap_or(false);
6911 // c:Src/exec.c:2313 clobber_open — CLOBBER_EMPTY re-uses
6912 // an empty regular file under noclobber (same allowance
6913 // as the single-redirect path).
6914 let clobber_empty_ok = opt_state_get("clobberempty").unwrap_or(false)
6915 && target_meta.as_ref().map(|m| m.len() == 0).unwrap_or(false);
6916 if noclobber && target_is_regular_file && !clobber_empty_ok {
6917 eprintln!(
6918 "{}:{}: file exists: {}",
6919 shname(),
6920 crate::ported::lex::lineno(),
6921 target
6922 );
6923 for prev in &target_fds {
6924 unsafe {
6925 libc::close(*prev);
6926 }
6927 }
6928 with_executor(|exec| {
6929 exec.redirect_failed = true;
6930 });
6931 // Sink the upcoming command's output (mirrors
6932 // the single-redirect noclobber arm in
6933 // host_apply_redirect).
6934 if let Ok(file) = fs::OpenOptions::new().write(true).open("/dev/null") {
6935 let new_fd = file.into_raw_fd();
6936 unsafe {
6937 libc::dup2(new_fd, fd);
6938 libc::close(new_fd);
6939 }
6940 }
6941 return Value::Status(1);
6942 }
6943 fs::OpenOptions::new()
6944 .write(true)
6945 .create(true)
6946 .truncate(true)
6947 .open(target)
6948 .map(|f| f.into_raw_fd())
6949 }
6950 r::APPEND => fs::OpenOptions::new()
6951 .write(true)
6952 .create(true)
6953 .append(true)
6954 .open(target)
6955 .map(|f| f.into_raw_fd()),
6956 _ => fs::OpenOptions::new()
6957 .write(true)
6958 .create(true)
6959 .truncate(true)
6960 .open(target)
6961 .map(|f| f.into_raw_fd()),
6962 };
6963 match open_result {
6964 Ok(tfd) => {
6965 if i == 0 && !pipe_seed {
6966 // c:2448-2450 — first member replaces the fd.
6967 unsafe {
6968 libc::dup2(tfd, fd);
6969 }
6970 }
6971 target_fds.push(tfd);
6972 }
6973 Err(e) => {
6974 // c:Src/exec.c:3741 — `zwarn("%e: %s", errno, fname)`:
6975 // zwarning supplies the `name:LINE:` prefix with the
6976 // REAL current lineno; redir_errno_msg builds the `%e`
6977 // errno message (was a hardcoded ErrorKind match that
6978 // showed generic "redirect failed" for EROFS/etc.).
6979 let msg = redir_errno_msg(&e);
6980 crate::ported::utils::zwarn(&format!("{}: {}", msg, target));
6981 // Close already-opened fds to avoid leaks.
6982 for prev in &target_fds {
6983 unsafe {
6984 libc::close(*prev);
6985 }
6986 }
6987 with_executor(|exec| {
6988 exec.redirect_failed = true;
6989 });
6990 return Value::Status(1);
6991 }
6992 }
6993 }
6994
6995 // Create the splitter pipe.
6996 let (read_end, write_end) = match os_pipe::pipe() {
6997 Ok(p) => p,
6998 Err(_) => {
6999 for f in &target_fds {
7000 unsafe {
7001 libc::close(*f);
7002 }
7003 }
7004 return Value::Status(1);
7005 }
7006 };
7007 let pipe_write_raw = AsRawFd::as_raw_fd(&write_end);
7008 // Spawn the splitter thread: read pipe → write every chunk
7009 // to every target fd. Each write inside the thread uses
7010 // libc::write directly on the raw fd (no Rust File ownership
7011 // so the splitter can close after EOF without racing main).
7012 let target_fds_for_thread = target_fds.clone();
7013 let handle = std::thread::spawn(move || {
7014 let mut r = read_end;
7015 let mut buf = [0u8; 8192];
7016 loop {
7017 match std::io::Read::read(&mut r, &mut buf) {
7018 Ok(0) => break,
7019 Ok(n) => {
7020 for &tfd in &target_fds_for_thread {
7021 let mut off = 0;
7022 while off < n {
7023 let w = unsafe {
7024 libc::write(
7025 tfd,
7026 buf[off..n].as_ptr() as *const libc::c_void,
7027 n - off,
7028 )
7029 };
7030 if w <= 0 {
7031 break;
7032 }
7033 off += w as usize;
7034 }
7035 }
7036 }
7037 Err(_) => break,
7038 }
7039 }
7040 // Close every target so file contents flush.
7041 for tfd in target_fds_for_thread {
7042 unsafe {
7043 libc::close(tfd);
7044 }
7045 }
7046 });
7047
7048 // Dup the pipe write-end onto the target fd; close the
7049 // original write_end so EOF arrives when host_redirect_scope_end
7050 // closes our tracked pipe_write_fd.
7051 let write_dup = unsafe { libc::dup(pipe_write_raw) };
7052 drop(write_end);
7053 if write_dup < 0 {
7054 return Value::Status(1);
7055 }
7056 unsafe {
7057 libc::dup2(write_dup, fd);
7058 libc::close(write_dup);
7059 }
7060 // Track the running splitter so scope-end can drain + join.
7061 // The "write_fd" we store is the user-visible fd (e.g. 1).
7062 // Closing that fd at scope-end isn't quite right; we need a
7063 // way to send EOF. Solution: track the write_dup we just
7064 // closed; instead keep a second dup for the close-on-end.
7065 let close_on_end = unsafe { libc::dup(fd) };
7066 with_executor(|exec| {
7067 if let Some(top) = exec.multios_scope_stack.last_mut() {
7068 top.push((close_on_end, handle));
7069 } else {
7070 // No scope — leak the dup; thread will keep running
7071 // until process exit. Should not happen because
7072 // host_redirect_scope_begin pushed a frame.
7073 unsafe { libc::close(close_on_end) };
7074 }
7075 });
7076 Value::Status(0)
7077 });
7078 // c:Src/exec.c:2418 input-arm — MULTIOS read fan-in. Stack
7079 // layout pushed by compile_zsh (mirrors the write side):
7080 // [source_1, op_1, source_2, op_2, …, source_N, op_N, fd]
7081 // argc = 2N + 1; op distinguishes file opens (READ) from numeric
7082 // `<&N` dups (DUP_READ); a glob source arrives as Value::Array
7083 // and splices into one member per match (c:Src/glob.c:2195-2203).
7084 // Opens every source, sets up a pipe + producer thread that
7085 // reads each source in order and writes to the pipe write-end,
7086 // then closes its write-end so the consumer gets EOF. dup2 the
7087 // pipe read-end onto fd. Bug #36 input side in docs/BUGS.md.
7088 vm.register_builtin(BUILTIN_MULTIOS_READ, |vm, argc| {
7089 if argc < 3 || argc % 2 == 0 {
7090 return Value::Status(1);
7091 }
7092 let fd = vm.pop().to_int() as i32;
7093 let n_sources = ((argc - 1) / 2) as usize;
7094 let mut pairs: Vec<(u8, Value)> = Vec::with_capacity(n_sources);
7095 for _ in 0..n_sources {
7096 let op_byte = vm.pop().to_int() as u8;
7097 let source = vm.pop();
7098 pairs.push((op_byte, source));
7099 }
7100 pairs.reverse();
7101
7102 // Splice glob match arrays (c:Src/glob.c:2195-2203).
7103 let mut entries: Vec<(u8, String)> = Vec::with_capacity(pairs.len());
7104 for (op_byte, source) in pairs {
7105 match source {
7106 Value::Array(items) => {
7107 for item in items {
7108 entries.push((op_byte, item.to_str()));
7109 }
7110 }
7111 other => entries.push((op_byte, other.to_str())),
7112 }
7113 }
7114 if entries.is_empty() {
7115 return Value::Status(1);
7116 }
7117
7118 // c:Src/exec.c:2418 — `unset(MULTIOS)`: sequential replace,
7119 // last source wins (`unsetopt multios; cat < a < b` reads
7120 // only b; a is still opened — and errors still surface).
7121 let multios_on = opt_state_get("multios").unwrap_or(true);
7122 if !multios_on {
7123 with_executor(|exec| {
7124 for (op_byte, source) in &entries {
7125 exec.host_apply_redirect(fd as u8, *op_byte, source);
7126 if exec.redirect_failed {
7127 break;
7128 }
7129 }
7130 });
7131 return Value::Status(0);
7132 }
7133
7134 if entries.len() == 1 {
7135 // Single member after splicing — plain replace.
7136 let (op_byte, source) = &entries[0];
7137 with_executor(|exec| {
7138 exec.host_apply_redirect(fd as u8, *op_byte, source);
7139 });
7140 return Value::Status(0);
7141 }
7142
7143 // Save current fd state for scope-end restoration — BEFORE
7144 // the first member's replace dup2 below.
7145 let saved = unsafe { libc::dup(fd) };
7146 if saved >= 0 {
7147 with_executor(|exec| {
7148 if let Some(top) = exec.redirect_scope_stack.last_mut() {
7149 top.push((fd, saved));
7150 } else {
7151 unsafe { libc::close(saved) };
7152 }
7153 });
7154 }
7155
7156 // Open every source in redirect order; numeric `<&N` dups
7157 // resolve against the LIVE fd table. First member replaces
7158 // the fd (c:2448-2450) so later self-dups see it.
7159 let mut source_fds: Vec<i32> = Vec::with_capacity(entries.len());
7160 for (i, (op_byte, source)) in entries.iter().enumerate() {
7161 let open_result: std::io::Result<i32> = match *op_byte {
7162 r::DUP_READ | r::DUP_WRITE => match source.trim_start_matches('&').parse::<i32>() {
7163 Ok(src) => {
7164 let d = unsafe { libc::dup(src) };
7165 if d >= 0 {
7166 Ok(d)
7167 } else {
7168 Err(std::io::Error::last_os_error())
7169 }
7170 }
7171 Err(_) => Err(std::io::Error::from_raw_os_error(libc::EBADF)),
7172 },
7173 _ => fs::File::open(source).map(|f| f.into_raw_fd()),
7174 };
7175 match open_result {
7176 Ok(tfd) => {
7177 if i == 0 {
7178 unsafe {
7179 libc::dup2(tfd, fd);
7180 }
7181 }
7182 source_fds.push(tfd);
7183 }
7184 Err(e) => {
7185 let msg = match e.kind() {
7186 std::io::ErrorKind::PermissionDenied => "permission denied",
7187 std::io::ErrorKind::NotFound => "no such file or directory",
7188 _ => "open failed",
7189 };
7190 // c:Src/exec.c:3741 — zwarn with real lineno prefix.
7191 crate::ported::utils::zwarn(&format!("{}: {}", msg, source));
7192 for prev in &source_fds {
7193 unsafe {
7194 libc::close(*prev);
7195 }
7196 }
7197 with_executor(|exec| {
7198 exec.redirect_failed = true;
7199 });
7200 return Value::Status(1);
7201 }
7202 }
7203 }
7204
7205 // Create the concatenator pipe.
7206 let (read_end, write_end) = match os_pipe::pipe() {
7207 Ok(p) => p,
7208 Err(_) => {
7209 for f in &source_fds {
7210 unsafe {
7211 libc::close(*f);
7212 }
7213 }
7214 return Value::Status(1);
7215 }
7216 };
7217 // dup the pipe read-end onto fd before spawning the
7218 // producer; close the original read_end so the consumer
7219 // (reading via fd) is the sole reference until scope-end.
7220 let read_dup = unsafe { libc::dup(AsRawFd::as_raw_fd(&read_end)) };
7221 drop(read_end);
7222 if read_dup < 0 {
7223 for f in &source_fds {
7224 unsafe {
7225 libc::close(*f);
7226 }
7227 }
7228 return Value::Status(1);
7229 }
7230 unsafe {
7231 libc::dup2(read_dup, fd);
7232 libc::close(read_dup);
7233 }
7234 // Spawn the producer.
7235 let source_fds_for_thread = source_fds.clone();
7236 let handle = std::thread::spawn(move || {
7237 let mut w = write_end;
7238 let mut buf = [0u8; 8192];
7239 for sfd in source_fds_for_thread {
7240 loop {
7241 let n = unsafe {
7242 libc::read(sfd, buf.as_mut_ptr() as *mut libc::c_void, buf.len())
7243 };
7244 if n <= 0 {
7245 break;
7246 }
7247 let n = n as usize;
7248 if std::io::Write::write_all(&mut w, &buf[..n]).is_err() {
7249 break;
7250 }
7251 }
7252 unsafe {
7253 libc::close(sfd);
7254 }
7255 }
7256 // Closing w (the write_end) at scope drop signals EOF
7257 // to the consumer.
7258 });
7259 with_executor(|exec| {
7260 // Track using a closed-write sentinel — the producer
7261 // owns write_end so we just need to join. Use -1 fd
7262 // marker meaning "no fd to close".
7263 if let Some(top) = exec.multios_scope_stack.last_mut() {
7264 top.push((-1, handle));
7265 } else {
7266 let _ = handle.join();
7267 }
7268 });
7269 Value::Status(0)
7270 });
7271 // c:Src/exec.c:3978-3986 — nullexec==1 marker. See the const's
7272 // doc block. Arg: 1 = entering a bare-exec redirect, 0 = leaving.
7273 vm.register_builtin(BUILTIN_EXEC_PERM_REDIRS, |vm, _argc| {
7274 let on = vm.pop().to_int() != 0;
7275 with_executor(|exec| exec.exec_redirs_permanent = on);
7276 Value::Status(0)
7277 });
7278 // Bare-exec redirect epilogue — see the const's doc block.
7279 // c:Src/exec.c:252-259 (execerr) + c:4367-4386 (done: POSIX gate).
7280 vm.register_builtin(BUILTIN_EXEC_REDIR_DONE, |vm, _argc| {
7281 use std::sync::atomic::Ordering;
7282 let failed = with_executor(|exec| {
7283 let f = exec.redirect_failed;
7284 exec.redirect_failed = false;
7285 f
7286 });
7287 if !failed {
7288 return Value::Status(0);
7289 }
7290 // c:255 — `redir_err = lastval = 1`.
7291 vm.last_status = 1;
7292 if isset(crate::ported::zsh_h::POSIXBUILTINS) && !isset(crate::ported::zsh_h::INTERACTIVE) {
7293 // c:4379-4383 — non-interactive POSIX fatal: exit(1).
7294 // In-process equivalent: arm EXIT_PENDING/EXIT_VAL so the
7295 // next BUILTIN_ERREXIT_CHECK (trigger 2) unwinds the
7296 // script with status 1 — same deferred-exit shape the
7297 // `exit` builtin uses inside subshell contexts.
7298 crate::ported::builtin::EXIT_VAL.store(1, Ordering::Relaxed);
7299 crate::ported::builtin::EXIT_PENDING.store(1, Ordering::Relaxed);
7300 }
7301 Value::Status(1)
7302 });
7303 // c:Src/exec.c:3722-3724 — see the const's doc block. No args.
7304 vm.register_builtin(BUILTIN_PIPE_OUTPUT_MARK, |_vm, _argc| {
7305 with_executor(|exec| exec.pipe_output_pending = true);
7306 Value::Status(0)
7307 });
7308 // c:Src/exec.c — block-level redirect-failure gate. When a
7309 // compound command (`{ … } < file`, `( … ) > file`, etc.) has a
7310 // failing redirect (e.g. `< /nonexistent`), zsh skips the entire
7311 // body AND sets lastval to 1. The simple-command path's
7312 // redirect_failed check (line 215-221 above) only catches the
7313 // failure when a builtin dispatches and is consumed by that
7314 // single builtin call — so a multi-statement block kept running
7315 // its remaining statements after the redir error. Emit-side at
7316 // compile_zsh.rs::compile_command's Redirected arm pairs this
7317 // with a JumpIfTrue → WithRedirectsEnd to abandon the body.
7318 vm.register_builtin(BUILTIN_REDIRECT_FAILED_CHECK, |vm, _argc| {
7319 let failed = with_executor(|exec| {
7320 let f = exec.redirect_failed;
7321 exec.redirect_failed = false;
7322 f
7323 });
7324 if failed {
7325 vm.last_status = 1;
7326 Value::Int(1)
7327 } else {
7328 Value::Int(0)
7329 }
7330 });
7331 // c:Src/exec.c — drop-in replacement for fusevm's Op::Exec used by
7332 // the dynamic-first-word path (`$cmd`, `$(cmd)`, glob-named cmds).
7333 // fusevm's Op::Exec returns Value::Status(0) when post-expansion
7334 // argv is empty (vm.rs:1722) — that clobbers \$? for the
7335 // `\$(exit 1); echo \$?` case where the cmd-subst left
7336 // last_status = 1 but the empty expansion gets exec'd to 0.
7337 // Mirror C zsh: when the word list is empty after expansion,
7338 // \$? becomes whatever the inner cmd-subst's last_status is
7339 // (preserved here by returning Value::Status(last_status)).
7340 // c:Src/cond.c:308-316 — `if (!(pprog = patcompile(right, ...)))
7341 // { zwarnnam(fromtest, "bad pattern: %s", right); return 2; }`.
7342 // The cond path must NOT use str_match/glob_match_static: the
7343 // case-statement consumer of those follows Src/loop.c:667 zerr
7344 // semantics (errflag abort), while cond is a zwarn + status-2
7345 // soft failure. COND_BAD_PATTERN carries the 2 across the
7346 // Bool-shaped stack contract (so `!=`'s LogNot can't lose it).
7347 thread_local! {
7348 static COND_BAD_PATTERN: std::cell::Cell<bool> =
7349 const { std::cell::Cell::new(false) };
7350 }
7351 vm.register_builtin(BUILTIN_COND_STRMATCH, |vm, _argc| {
7352 let pat = vm.pop().to_str();
7353 let s = vm.pop().to_str();
7354 let mut pat_tok = pat.clone();
7355 crate::ported::glob::tokenize(&mut pat_tok);
7356 if crate::ported::pattern::patcompile(
7357 &pat_tok,
7358 crate::ported::zsh_h::PAT_STATIC as i32,
7359 None,
7360 )
7361 .is_none()
7362 {
7363 // c:314 — zwarnnam(fromtest, "bad pattern: %s", right).
7364 crate::ported::utils::zwarn(&format!("bad pattern: {}", pat));
7365 COND_BAD_PATTERN.with(|c| c.set(true));
7366 return Value::Bool(false);
7367 }
7368 // Match via the shared engine so `(#b)`/`(#m)` backref and
7369 // MATCH-variable population stays in one place.
7370 Value::Bool(crate::vm_helper::glob_match_static(&s, &pat))
7371 });
7372 vm.register_builtin(BUILTIN_COND_UNKNOWN, |vm, _argc| {
7373 // c:Src/cond.c:150-188 — `zwarnnam(fromtest, "unknown condition: %s",
7374 // name)` for a `-X` op with no matching cond module. Like a cond
7375 // syntax error it yields status 2 and aborts: arm COND_BAD_PATTERN so
7376 // the downstream BUILTIN_COND_STATUS_FROM_BOOL carries the 2 across the
7377 // Bool-shaped stack and runs the shared errflag+set_last_status(2)+abort
7378 // path (c:Src/exec.c:5216-5221). Returns Bool(false) as the operand.
7379 let op = vm.pop().to_str();
7380 crate::ported::utils::zerr(&format!("unknown condition: {}", op));
7381 COND_BAD_PATTERN.with(|c| c.set(true));
7382 Value::Bool(false)
7383 });
7384 vm.register_builtin(BUILTIN_COND_STATUS_FROM_BOOL, |vm, _argc| {
7385 // `${~pat}` / `${(P)~pat}` inside a `[[ … ]]` operand flips
7386 // GLOB_SUBST on via the tilde carrier so the pattern match sees
7387 // active metacharacters. In C that flag is prefork-scoped and
7388 // gone once the operand is consumed; zshrs restores it at the
7389 // next command-dispatch boundary, but a bare `[[ … ]]` has no
7390 // trailing assignment to trigger that — so globsubst leaked ON
7391 // into the NEXT command's word expansion, filename-generating a
7392 // scalar value it should not (p10k `_p9k_set_prompt`: line 45
7393 // `[[ … != ${(P)~disabled} ]]` leaked into line 46's
7394 // `local val=$arr[idx]`, whose glob-char-laden value then hit
7395 // "no matches found" and aborted the whole prompt build →
7396 // garbled 25-line prompt / interactive hang). Consume the
7397 // carrier here: this builtin ends EVERY `[[ … ]]`, and runs
7398 // after the operands (and their pattern match) are done.
7399 consume_tilde_globsubst_carrier();
7400 let ok = vm.pop().to_int() != 0;
7401 let bad = COND_BAD_PATTERN.with(|c| {
7402 let b = c.get();
7403 c.set(false);
7404 b
7405 });
7406 if bad {
7407 // c:Src/exec.c:5216-5221 — `stat = evalcond(...);
7408 // /* 2 indicates a syntax error. For compatibility,
7409 // turn this into a shell error. */
7410 // if (stat == 2) errflag |= ERRFLAG_ERROR;`
7411 // The errflag abort exits the script with lastval (2),
7412 // matching `zsh -fc '[[ x == [a- ]]; print rc=$?'`
7413 // printing nothing after the diagnostic and exiting 2.
7414 crate::ported::utils::errflag.fetch_or(
7415 crate::ported::zsh_h::ERRFLAG_ERROR,
7416 std::sync::atomic::Ordering::Relaxed,
7417 );
7418 with_executor(|exec| exec.set_last_status(2));
7419 return Value::Int(2); // c:Src/cond.c:316 `return 2;`
7420 }
7421 let status: i32 = if ok { 0 } else { 1 };
7422 // c:Src/exec.c:5216 — `lastval = evalcond(...)`: the conditional's
7423 // result IS the command's lastval, and c:Src/cond.c's evalcond
7424 // never inspects errflag while evaluating. So when a `[[ … ]]`
7425 // operand raised errflag (e.g. a nounset "parameter not set" zerr
7426 // on `${arr[99]}` under NO_UNSET), zsh STILL completes the test and
7427 // exits with the cond result; the errflag only aborts the FOLLOWING
7428 // commands. Sync the result to the executor's live lastval HERE —
7429 // the nounset site left it at a transient 1, and the next
7430 // BUILTIN_ERREXIT_CHECK reads the executor (not vm.last_status), so
7431 // without this sync `setopt NO_UNSET; [[ -z ${arr[99]} ]]` exited 1
7432 // instead of 0. The Op::SetStatus that follows sets vm.last_status;
7433 // this keeps the executor coherent with it before the abort check.
7434 with_executor(|exec| exec.set_last_status(status));
7435 Value::Int(status as i64)
7436 });
7437 vm.register_builtin(BUILTIN_USE_CMDOUTVAL_RESET, |_vm, _argc| {
7438 crate::ported::exec::use_cmdoutval.store(0, std::sync::atomic::Ordering::Relaxed);
7439 Value::Status(0)
7440 });
7441
7442 vm.register_builtin(BUILTIN_EXEC_DYNAMIC, |vm, argc| {
7443 let raw = pop_args(vm, argc);
7444 // Flatten Array entries into argv slots (matches fusevm
7445 // Op::Exec's flatten at vm.rs:1660-1665) so `${arr[@]}` /
7446 // splice expansions produce one argv slot per element.
7447 let args: Vec<String> = raw.into_iter().collect();
7448 // c:Src/subst.c paramsubst — when `${var:?msg}` or
7449 // `${var?msg}` set errflag, the expansion may produce empty
7450 // argv[0] which would fall into the EACCES/permission-denied
7451 // path below, masking the real paramsubst diagnostic with a
7452 // spurious "permission denied:" line and rc=126. Honour
7453 // errflag so the simple command ends with the paramsubst
7454 // error as the sole diagnostic, rc=1. Bug #86.
7455 if (crate::ported::utils::errflag.load(std::sync::atomic::Ordering::SeqCst)
7456 & crate::ported::zsh_h::ERRFLAG_ERROR)
7457 != 0
7458 {
7459 return Value::Status(1);
7460 }
7461 if args.is_empty() {
7462 // c:Src/exec.c:3442 — a command whose words expand to ZERO
7463 // words is a NULL command: `cmdoutval = use_cmdoutval ?
7464 // lastval : 0`. `use_cmdoutval` is set (below, in
7465 // BUILTIN_CMD_SUBST_TEXT) only when a command substitution
7466 // ran during this command's word expansion, so:
7467 // `false; $(exit 5)` → keep the subst status (5)
7468 // `false; $nonexistent` → reset to 0 (null command).
7469 // The previous port unconditionally kept `$?`, so
7470 // `false; $unset` wrongly stayed 1 (A01grammar.ztst:5).
7471 let keep =
7472 crate::ported::exec::use_cmdoutval.load(std::sync::atomic::Ordering::Relaxed) != 0;
7473 let status = if keep { vm.last_status } else { 0 };
7474 crate::ported::exec::use_cmdoutval.store(0, std::sync::atomic::Ordering::Relaxed);
7475 return Value::Status(status);
7476 }
7477 if args[0].is_empty() {
7478 // Explicit empty command word — exec returns EACCES.
7479 let script_name =
7480 crate::ported::utils::scriptname_get().unwrap_or_else(|| "zshrs".to_string());
7481 let lineno: u64 = with_executor(|exec| {
7482 exec.scalar("LINENO")
7483 .and_then(|s| s.parse::<u64>().ok())
7484 .unwrap_or(1)
7485 });
7486 eprintln!("{}:{}: permission denied: ", script_name, lineno);
7487 return Value::Status(126);
7488 }
7489 // AOP intercepts (zshrs extension, no C counterpart) — same
7490 // gate as host_exec_external (the static-head path): dynamic
7491 // command names (`cmd=/bin/echo; $cmd payload`) must consult
7492 // registered intercepts before dispatch, else `intercept
7493 // before /bin/echo ...` fires for the literal spelling but
7494 // not the variable one. run_intercepts runs before-advice
7495 // in-place and returns None to continue; Some(status) means
7496 // an around/after advice fully handled the command.
7497 let intercepted = with_executor(|exec| {
7498 if exec.intercepts.is_empty() {
7499 return None;
7500 }
7501 let full_cmd = if args.len() == 1 {
7502 args[0].clone()
7503 } else {
7504 args.join(" ")
7505 };
7506 let rest: Vec<String> = args[1..].to_vec();
7507 exec.run_intercepts(&args[0], &full_cmd, &rest)
7508 });
7509 if let Some(result) = intercepted {
7510 return Value::Status(result.unwrap_or(127));
7511 }
7512 // c:Src/exec.c:2900 execcmd_exec — canonical simple-command
7513 // dispatcher. Runs precmd-modifier walk (c:3013-3091), then
7514 // dispatches to execbuiltin (c:4233) / runshfunc (c:3431+) /
7515 // execute (c:4314) per the resolved head. zshrs's bytecode VM
7516 // expanded the args before reaching here; we feed them in via
7517 // eparams.args and let execcmd_exec do the rest exactly as C
7518 // does for static heads. Without this, `c=builtin; $c source X`
7519 // skipped the precmd walk and emitted "command not found:
7520 // builtin".
7521 let mut state = crate::ported::zsh_h::estate {
7522 prog: Box::<crate::ported::zsh_h::eprog>::default(),
7523 pc: 0,
7524 strs: None,
7525 strs_offset: 0,
7526 };
7527 let mut eparams = crate::ported::zsh_h::execcmd_params {
7528 args: Some(args),
7529 redir: None,
7530 beg: 0,
7531 varspc: None,
7532 assignspc: None,
7533 typ: crate::ported::zsh_h::WC_SIMPLE as i32,
7534 postassigns: 0,
7535 htok: 0,
7536 };
7537 // input/output=0 → no pipe redirection (use shell stdio
7538 // directly); `output != 0` at c:2988 forks immediately. last1=2
7539 // (c:Src/exec.c:2014 `last1 ? 1 : 2`): terminal pipe stage but
7540 // the shell IS needed afterward — the VM keeps executing
7541 // bytecode after this op. last1=1 would arm the fake-exec
7542 // optimization (c:3646-3651, gate at c:3662 `last1 != 1`),
7543 // making `execute()` execve THIS process for external heads:
7544 // `p=/bin/echo; $p hi; echo after` replaced the shell and
7545 // `after` never ran (D04parameter chunk 11 shell-killer).
7546 // c:Src/exec.c:1690-1700 — execpline's job frame: save thisjob
7547 // (`pj = thisjob`) and allocate the jobtab slot that
7548 // execcmd_fork's addproc (c:2853) hangs the child pid off.
7549 // Without a live thisjob, the fork at c:3662 (last1 != 1 →
7550 // external must fork) registers no proc, nothing waits, and
7551 // the child races the rest of the script.
7552 let pj = {
7553 use crate::ported::jobs;
7554 *jobs::THISJOB
7555 .get_or_init(|| std::sync::Mutex::new(-1))
7556 .lock()
7557 .unwrap_or_else(|e| e.into_inner())
7558 };
7559 let newjob = {
7560 use crate::ported::jobs;
7561 let table = jobs::JOBTAB.get_or_init(|| std::sync::Mutex::new(Vec::new()));
7562 let mut tab = table.lock().unwrap_or_else(|e| e.into_inner());
7563 jobs::initjob(&mut tab) // c:1700 `thisjob = newjob = initjob()`
7564 };
7565 {
7566 use crate::ported::jobs;
7567 *jobs::THISJOB
7568 .get_or_init(|| std::sync::Mutex::new(-1))
7569 .lock()
7570 .unwrap_or_else(|e| e.into_inner()) = newjob as i32;
7571 }
7572 crate::ported::exec::execcmd_exec(
7573 &mut state,
7574 &mut eparams,
7575 0, // input (c:2989)
7576 0, // output (c:2988)
7577 crate::ported::zsh_h::Z_SYNC as i32, // how
7578 2, // last1=2 — shell continues (c:2014)
7579 -1, // close_if_forked
7580 );
7581 // c:Src/exec.c:1828-1835 — execpline's Z_SYNC tail: waitjobs()
7582 // reaps the forked external. c:Src/jobs.c:487-495 + 551-552 —
7583 // the job's LAST proc sets lastval (0200|sig when signalled,
7584 // else WEXITSTATUS). Builtin/shfunc heads never forked (job
7585 // has no procs) — LASTVAL was already set by execbuiltin /
7586 // doshfunc inside execcmd_exec; skip the wait.
7587 {
7588 use crate::ported::jobs;
7589 let table = jobs::JOBTAB.get_or_init(|| std::sync::Mutex::new(Vec::new()));
7590 let mut tab = table.lock().unwrap_or_else(|e| e.into_inner());
7591 if jobs::hasprocs(&tab, newjob) {
7592 jobs::waitjobs(&mut tab, newjob); // c:1835
7593 if let Some(p) = tab[newjob].procs.last() {
7594 let val = if p.is_signaled() {
7595 0o200 | p.term_sig() // c:Src/jobs.c:489-490
7596 } else {
7597 p.exit_status() // c:Src/jobs.c:494
7598 };
7599 crate::ported::builtin::LASTVAL
7600 .store(val, std::sync::atomic::Ordering::Relaxed);
7601 }
7602 }
7603 // c:1977-1979 — `deletejob(jn, 0)` once done; c:1981
7604 // `thisjob = pj` restores the caller's job.
7605 if newjob < tab.len() {
7606 jobs::deletejob(&mut tab[newjob], false);
7607 }
7608 *jobs::THISJOB
7609 .get_or_init(|| std::sync::Mutex::new(-1))
7610 .lock()
7611 .unwrap_or_else(|e| e.into_inner()) = pj;
7612 }
7613 let status = crate::ported::builtin::LASTVAL.load(std::sync::atomic::Ordering::Relaxed);
7614 let mut synth = crate::ported::zsh_h::job::default();
7615 crate::ported::jobs::waitonejob(&mut synth);
7616 Value::Status(status)
7617 });
7618 // c:Src/exec.c:3340-3364 — `< file` / `> file` with no command
7619 // word. Resolves NULLCMD/READNULLCMD at runtime then routes
7620 // through host_exec_external. Redirects are already applied by
7621 // the surrounding WithRedirectsBegin scope.
7622 vm.register_builtin(BUILTIN_NULLCMD_EXEC, |vm, argc| {
7623 let args = pop_args(vm, argc);
7624 let is_single_read = args
7625 .first()
7626 .map(|s| s != "0" && !s.is_empty())
7627 .unwrap_or(false);
7628 // c:Src/exec.c — when the surrounding redir-open failed
7629 // (e.g. `< /nonexistent`), zerr already printed the diag
7630 // and set redirect_failed. Don't invoke NULLCMD — return
7631 // status 1 like the wordcode path does.
7632 let redir_failed = with_executor(|exec| {
7633 let f = exec.redirect_failed;
7634 exec.redirect_failed = false;
7635 f
7636 });
7637 if redir_failed {
7638 crate::ported::builtin::LASTVAL.store(1, std::sync::atomic::Ordering::Relaxed);
7639 return Value::Status(1);
7640 }
7641 let nullcmd = crate::ported::params::getsparam("NULLCMD");
7642 let nc_str = nullcmd.as_deref().unwrap_or("");
7643 let nc_empty = nc_str.is_empty();
7644 // c:3340-3344 — CSHNULLCMD or no NULLCMD set → diagnostic.
7645 if nc_empty || crate::ported::zsh_h::isset(crate::ported::zsh_h::CSHNULLCMD) {
7646 let script_name =
7647 crate::ported::utils::scriptname_get().unwrap_or_else(|| "zshrs".to_string());
7648 let lineno: u64 = with_executor(|exec| {
7649 exec.scalar("LINENO")
7650 .and_then(|s| s.parse::<u64>().ok())
7651 .unwrap_or(1)
7652 });
7653 eprintln!("{}:{}: redirection with no command", script_name, lineno);
7654 return Value::Status(1);
7655 }
7656 // c:3350 — SHNULLCMD → run `:`.
7657 let cmd: String = if crate::ported::zsh_h::isset(crate::ported::zsh_h::SHNULLCMD) {
7658 ":".to_string()
7659 } else if is_single_read {
7660 // c:3354-3359 — single REDIR_READ + READNULLCMD set → readnullcmd.
7661 let rnc = crate::ported::params::getsparam("READNULLCMD");
7662 let rnc_str = rnc.as_deref().unwrap_or("");
7663 if !rnc_str.is_empty() {
7664 rnc_str.to_string()
7665 } else {
7666 nc_str.to_string() // c:3360-3363 fallback
7667 }
7668 } else {
7669 nc_str.to_string() // c:3360-3363
7670 };
7671 let status = with_executor(|exec| exec.host_exec_external(&[cmd]));
7672 crate::ported::builtin::LASTVAL.store(status, std::sync::atomic::Ordering::Relaxed);
7673 Value::Status(status)
7674 });
7675 // c:Src/exec.c:3342 — `zerr("redirection with no command")`.
7676 // Bare prefix-keyword (`builtin`, `command`, `exec`, `noglob`,
7677 // `nocorrect`) with a redirect but no command word. Emits the
7678 // canonical diagnostic via zerr (which sets errflag) and
7679 // returns Status(1). Bug #534.
7680 vm.register_builtin(BUILTIN_REDIR_NO_CMD, |_vm, _argc| {
7681 crate::ported::utils::zerr("redirection with no command");
7682 Value::Status(1)
7683 });
7684 vm.register_builtin(BUILTIN_DEBUG_TRAP, |vm, _argc| {
7685 // c:Src/signals.c:1245 dotrap(SIGDEBUG) — fires the DEBUG
7686 // trap body once per statement. The body sees the parent
7687 // shell's $? (LASTVAL). Guard against re-entry: commands
7688 // inside the DEBUG trap body would otherwise trigger
7689 // DEBUG_TRAP recursively → stack overflow. zsh guards via
7690 // its in_trap counter; we mirror with a thread-local Cell.
7691 //
7692 // c:Src/exec.c::trapcmd — before dotrap, the C source sets
7693 // `ZSH_DEBUG_CMD` to the about-to-run command text via
7694 // `dupstring(text)`. The trap body reads the parameter;
7695 // C unsets it after the trap returns. compile_list emits
7696 // the rendered statement text as the single arg here so the
7697 // shell-visible parameter reflects the command. Bug #263 in
7698 // docs/BUGS.md.
7699 let cmd_text = vm.pop().to_str();
7700 DEBUG_TRAP_REENTRY.with(|c| {
7701 if c.get() {
7702 return Value::Status(0);
7703 }
7704 // c:Src/exec.c:1423 — `if (sigtrapped[SIGDEBUG] &&
7705 // isset(DEBUGBEFORECMD) && !intrap)`. Bug #573: without
7706 // this gate, every sublist boundary called
7707 // setsparam("ZSH_DEBUG_CMD", ...) even when no DEBUG trap
7708 // was set, polluting the param table and (under
7709 // WARN_CREATE_GLOBAL) emitting a spurious
7710 // `scalar parameter ZSH_DEBUG_CMD created globally`
7711 // warning at every function call.
7712 //
7713 // Two trap registries exist (per signals.rs:1481-1511 dotrap):
7714 // - settrap path → sigtrapped[SIGDEBUG] bits set
7715 // - bin_trap path → traps_table["DEBUG"] populated, sigtrapped untouched
7716 // Mirror the dotrap dispatch decision: skip only when BOTH
7717 // are absent.
7718 let sig_debug = crate::ported::signals_h::SIGDEBUG as usize;
7719 let debug_trapped = crate::ported::signals::sigtrapped
7720 .lock()
7721 .map(|v| v.get(sig_debug).copied().unwrap_or(0))
7722 .unwrap_or(0);
7723 let debug_in_table = crate::ported::builtin::traps_table()
7724 .lock()
7725 .map(|t| t.contains_key("DEBUG"))
7726 .unwrap_or(false);
7727 if debug_trapped == 0 && !debug_in_table {
7728 return Value::Status(0);
7729 }
7730 c.set(true);
7731 // c:Src/exec.c — set ZSH_DEBUG_CMD scalar (PM_READONLY
7732 // is NOT set on ZSH_DEBUG_CMD, so the canonical
7733 // setsparam path is fine here — no direct paramtab
7734 // mutation needed).
7735 crate::ported::params::setsparam("ZSH_DEBUG_CMD", &cmd_text);
7736 let _ = crate::ported::signals::dotrap(crate::ported::signals_h::SIGDEBUG);
7737 // c:Src/exec.c::trapcmd — `unsetparam("ZSH_DEBUG_CMD")`
7738 // after the trap returns. Mirror that.
7739 crate::ported::params::unsetparam("ZSH_DEBUG_CMD");
7740 c.set(false);
7741 Value::Status(0)
7742 })
7743 });
7744
7745 // Fatal-only abort check emitted between the pipes of an `&&` / `||`
7746 // chain, where the full errexit check is suppressed. Mirrors ONLY the
7747 // errflag arm of BUILTIN_ERREXIT_CHECK below: an errflag abandons the
7748 // list in zsh, and no connector can consume it.
7749 vm.register_builtin(BUILTIN_FATAL_ABORT_CHECK, |vm, _argc| {
7750 use std::sync::atomic::Ordering;
7751 let errflag_set = (crate::ported::utils::errflag.load(Ordering::Relaxed)
7752 & crate::ported::zsh_h::ERRFLAG_ERROR)
7753 != 0;
7754 if !errflag_set || isset(crate::ported::zsh_h::INTERACTIVE) {
7755 return Value::Int(0);
7756 }
7757 // CONTINUE_ON_ERROR: clear and keep going, as the full check does.
7758 if isset(crate::ported::zsh_h::CONTINUEONERROR) {
7759 crate::ported::utils::errflag
7760 .fetch_and(!crate::ported::zsh_h::ERRFLAG_ERROR, Ordering::Relaxed);
7761 return Value::Int(0);
7762 }
7763 // Abort the chain with the failing command's own status intact —
7764 // a cond syntax error left lastval=2 (c:Src/exec.c:5216-5221), and
7765 // that 2 is what zsh exits with. Reading the executor's live
7766 // lastval (not forcing 1) is the same rule the full check uses.
7767 vm.last_status = with_executor(|exec| exec.last_status());
7768 Value::Int(1)
7769 });
7770 vm.register_builtin(BUILTIN_ERREXIT_CHECK, |vm, _argc| {
7771 // Returns Value::Int(1) when the caller should jump to the
7772 // current scope's return-patch landing (subshell-end / func-
7773 // end / chunk-end). Returns Value::Int(0) otherwise. Emit
7774 // side at `emit_errexit_check` pairs this with a JumpIfTrue
7775 // → return_patches pattern so the caller can short-circuit.
7776 //
7777 // Four triggers:
7778 // 1. RETFLAG set by a nested `return` / `exit` (eval,
7779 // sourced file, called function). Unwind THIS scope so
7780 // the flag propagates outward until something clears it.
7781 // 2. EXIT_PENDING set (mostly subshell-context exits). Same
7782 // propagation logic.
7783 // 3. `set -e` + nonzero status — the classic errexit path.
7784 // 4. errflag set in non-interactive mode — readonly
7785 // reassign, bad redirect, parse error mid-expansion etc.
7786 // Aborts the script (c:Src/init.c loop()).
7787 use std::sync::atomic::Ordering;
7788 let retflag = crate::ported::builtin::RETFLAG.load(Ordering::Relaxed);
7789 let exit_pending = crate::ported::builtin::EXIT_PENDING.load(Ordering::Relaxed);
7790 if retflag != 0 || exit_pending != 0 {
7791 if exit_pending != 0 {
7792 // c:Src/builtin.c zexit — the deferred exit carries its
7793 // status in EXIT_VAL; sync it into the VM counter so
7794 // the top-level unwind reports it as the script's exit
7795 // (run_chunk returns vm.last_status). Without this, a
7796 // POSIX-fatal `.` failure exited 127 (bin_dot's return)
7797 // instead of C's exit(1) at Src/exec.c:4383.
7798 vm.last_status = crate::ported::builtin::EXIT_VAL.load(Ordering::Relaxed) & 0xFF;
7799 }
7800 return Value::Int(1);
7801 }
7802 let errflag_set = (crate::ported::utils::errflag.load(Ordering::Relaxed)
7803 & crate::ported::zsh_h::ERRFLAG_ERROR)
7804 != 0;
7805 // c:Src/init.c:1931 — `if (errflag && !interact &&
7806 // !isset(CONTINUEONERROR)) { errexit = 1; break; }` — with
7807 // CONTINUE_ON_ERROR set, the top-level do-while re-enters
7808 // loop() and the NEXT list runs instead of the shell exiting.
7809 // Clear the flag so the next statement starts clean (the
7810 // failed statement's lastval is already in place).
7811 if errflag_set
7812 && !crate::ported::zsh_h::isset(crate::ported::zsh_h::INTERACTIVE)
7813 && crate::ported::zsh_h::isset(crate::ported::zsh_h::CONTINUEONERROR)
7814 {
7815 crate::ported::utils::errflag
7816 .fetch_and(!crate::ported::zsh_h::ERRFLAG_ERROR, Ordering::Relaxed);
7817 return Value::Int(0);
7818 }
7819 if errflag_set && !crate::ported::zsh_h::isset(crate::ported::zsh_h::INTERACTIVE) {
7820 // c:Src/exec.c execlist — every enclosing list loop runs
7821 // `while (... && !errflag)`, so a set errflag breaks the
7822 // CURRENT scope and the check in the enclosing scope
7823 // breaks THAT one, all the way out. Leave errflag SET —
7824 // do NOT convert it to EXIT_PENDING: a process-exit
7825 // signal tunnels through the containment boundaries C
7826 // has, namely eval (Src/builtin.c:6221 `errflag &=
7827 // ~ERRFLAG_ERROR`), source (Src/init.c:1663 same), fork
7828 // boundaries (subshell/cmdsubst — child's errflag dies
7829 // with the child), and the interactive toplevel
7830 // (Src/init.c:139). Those boundaries clear errflag
7831 // themselves and execution continues past them; with
7832 // EXIT_PENDING armed here, `eval 'assoc=(odd)'; echo
7833 // after` aborted the whole script where zsh 5.9 prints
7834 // `after` (eval status 1). Bug #74's function case
7835 // (`f() { local -r x=5; x=10; }; f; echo after`) still
7836 // aborts: the function scope unwinds on THIS check, and
7837 // the caller's next ERREXIT_CHECK sees the still-set
7838 // errflag and unwinds too — exactly C's propagation.
7839 //
7840 // c:Src/init.c:234 — loop() BREAKS on errflag and
7841 // zsh_main exits with the UNTOUCHED lastval, NOT a
7842 // forced 1: `typeset -i x=3#8` (math error during the
7843 // assignment, before typeset sets a status) exits 0 in
7844 // zsh; a cond syntax error set lastval=2 (exec.c:5216-
7845 // 5221) and zsh exits 2; the readonly-reassign case
7846 // exits 1 because ITS lastval is 1. Sync the VM counter
7847 // from the executor's live lastval instead of
7848 // overwriting.
7849 vm.last_status = with_executor(|exec| exec.last_status());
7850 return Value::Int(1);
7851 }
7852 let last = vm.last_status;
7853 if last == 0 {
7854 return Value::Int(0);
7855 }
7856 // c:Src/exec.c:1598 `if (!this_noerrexit && !donetrap &&
7857 // !this_donetrap)` — gate the ZERR trap fire on DONETRAP so
7858 // an inner sublist (e.g. `false` inside a function) that
7859 // already fired ZERR doesn't fire it AGAIN at the outer
7860 // sublist's post-command check (after the function
7861 // returned non-zero). Bug #303 in docs/BUGS.md. DONETRAP
7862 // is reset at top-level statement boundaries via
7863 // BUILTIN_DONETRAP_RESET (compile_list emit at
7864 // compile_zsh.rs).
7865 let already_done = crate::ported::exec::DONETRAP.load(Ordering::Relaxed) != 0;
7866 if !already_done {
7867 // c:Src/signals.c:1245 dotrap(SIGZERR) — canonical ZERR
7868 // trap dispatch. Fires whenever a command exits
7869 // non-zero.
7870 let _ = crate::ported::signals::dotrap(crate::ported::signals_h::SIGZERR);
7871 // c:1602 — `donetrap = 1;` after firing.
7872 crate::ported::exec::DONETRAP.store(1, Ordering::Relaxed);
7873 }
7874 // c:Src/exec.c:1605-1610 — compute errreturn / errexit.
7875 // errreturn = ERRRETURN && (INTERACTIVE || locallevel || sourcelevel)
7876 // && !(noerrexit & NOERREXIT_RETURN)
7877 // errexit = (ERREXIT || (ERRRETURN && !errreturn))
7878 // && !(noerrexit & NOERREXIT_EXIT)
7879 let no_err = crate::ported::exec::noerrexit.load(Ordering::Relaxed);
7880 let locallvl = crate::ported::params::locallevel.load(Ordering::Relaxed);
7881 let sourcelvl = crate::ported::init::sourcelevel.load(Ordering::Relaxed);
7882 let errreturn_opt = isset(crate::ported::zsh_h::ERRRETURN);
7883 let in_unwindable_scope =
7884 isset(crate::ported::zsh_h::INTERACTIVE) || locallvl != 0 || sourcelvl != 0;
7885 let errreturn = errreturn_opt
7886 && in_unwindable_scope
7887 && (no_err & crate::ported::zsh_h::NOERREXIT_RETURN) == 0;
7888 if errreturn {
7889 // c:1620-1623 — `retflag = 1; breaks = loops;` — unwind to
7890 // function boundary without exiting the shell.
7891 crate::ported::builtin::RETFLAG.store(1, Ordering::Relaxed);
7892 let loops = crate::ported::builtin::LOOPS.load(Ordering::Relaxed);
7893 crate::ported::builtin::BREAKS.store(loops, Ordering::Relaxed);
7894 return Value::Int(1);
7895 }
7896 let (errexit_on, in_subshell) = with_executor(|exec| {
7897 let on_canonical = isset(ERREXIT) || (errreturn_opt && !errreturn); // c:1608-1609
7898 let on_legacy = opt_state_get("errexit").unwrap_or(false);
7899 (
7900 (on_canonical || on_legacy) && (no_err & crate::ported::zsh_h::NOERREXIT_EXIT) == 0,
7901 !exec.subshell_snapshots.is_empty(),
7902 )
7903 });
7904 if !errexit_on {
7905 return Value::Int(0);
7906 }
7907 // c:Src/exec.c — `set -e` fires shell exit, NOT a function-only
7908 // unwind. When LOCAL_OPTIONS restores the option mid-fn, the
7909 // restoration would otherwise mask the trigger and let the
7910 // outer scope continue. Setting EXIT_PENDING + EXIT_VAL here
7911 // (for ALL scope kinds, not just subshells) makes the fn-exit
7912 // path propagate to the shell-exit boundary at c:6135-6155.
7913 crate::ported::builtin::EXIT_VAL.store(last, Ordering::Relaxed);
7914 crate::ported::builtin::EXIT_PENDING.store(1, Ordering::Relaxed);
7915 let _ = in_subshell;
7916 // Function scope and top-level scope both branch to their
7917 // respective return_patches; top-level lands at chunk-end,
7918 // so execute_script returns `last` as the script's exit
7919 // status (same observable behavior as a process::exit).
7920 Value::Int(1)
7921 });
7922
7923 // BUILTIN_ASSIGN_ONLY_STATUS — status of an assignment-only
7924 // simple command. c:Src/exec.c:3393-3396 (execcmd_exec, no
7925 // command word + varspc): `if (errflag) lastval = 1; else
7926 // lastval = cmdoutval;`; same shape at c:1322 (execsimple
7927 // WC_ASSIGN: `lv = (errflag ? errflag : cmdoutval)`) and
7928 // c:3977 (nullexec=2 redir variant). cmdoutval is the exit of
7929 // a `$()` that ran in an RHS (already in vm.last_status via
7930 // compile_assign's per-assign SetStatus), 0 otherwise. The
7931 // store goes to the canonical LASTVAL too — that IS C's single
7932 // `lastval` global; without it the errflag-abort path
7933 // (BUILTIN_ERREXIT_CHECK trigger 4) syncs vm.last_status from
7934 // a stale LASTVAL and `readonly r=1; r=2` exited 0, not 1.
7935 vm.register_builtin(BUILTIN_ASSIGN_ONLY_STATUS, |vm, _argc| {
7936 use std::sync::atomic::Ordering;
7937 let had_cmd_subst = vm.pop().to_int() != 0;
7938 let errflag_set = (crate::ported::utils::errflag.load(Ordering::Relaxed)
7939 & crate::ported::zsh_h::ERRFLAG_ERROR)
7940 != 0;
7941 // c:Src/exec.c addvars — `if (!pm) { lastval = 1; if
7942 // (!cmdoutval) cmdoutval = 1; }` (assignment-failed cheat).
7943 let assign_failed = ASSIGN_FAILED_FLAG.swap(false, std::sync::atomic::Ordering::Relaxed);
7944 let status = if errflag_set || assign_failed {
7945 1 // c:Src/exec.c:3394 `lastval = 1` / addvars cmdoutval=1
7946 } else if had_cmd_subst {
7947 vm.last_status // c:3396 `lastval = cmdoutval` (subst exit)
7948 } else {
7949 0 // c:3396 `lastval = cmdoutval` (cmdoutval = 0)
7950 };
7951 with_executor(|exec| exec.set_last_status(status));
7952 Value::Status(status)
7953 });
7954
7955 // `${var:-default}` / `${var:=default}` / `${var:?error}` / `${var:+alt}`
7956 // Pops [name, op_byte, rhs] (rhs popped first). Returns the modified
7957 // value as Value::Str. Handles unset/empty distinction (`:-` etc.
7958 // treat empty same as unset, matching POSIX).
7959 // BUILTIN_PARAM_DEFAULT_FAMILY — `${var-x}` / `${var:-x}` / `${var=x}` /
7960 // `${var:=x}` / `${var?x}` / `${var:?x}` / `${var+x}` / `${var:+x}`.
7961 // PURE PASSTHRU: pop name + op + rhs, reconstruct the canonical
7962 // brace expression, hand to `subst::paramsubst` (C port of
7963 // `Src/subst.c::paramsubst`). All "missing vs empty" gating,
7964 // nounset suppression, default-evaluation, and elide-empty-words
7965 // semantics live inside paramsubst.
7966 vm.register_builtin(BUILTIN_PARAM_DEFAULT_FAMILY, |vm, _argc| {
7967 let rhs = vm.pop().to_str();
7968 let op = vm.pop().to_int() as u8;
7969 let name = vm.pop().to_str();
7970 // op=8 is the `${+name}` set-test prefix form (distinct from the
7971 // `${name+rhs}` substitute-if-set suffix form which is op=7).
7972 // Per compile_zsh.rs::parse_param_modifier: the `+` is emitted as
7973 // a leading sigil and `rhs` is empty.
7974 let body = if op == 8 {
7975 format!("${{+{}}}", name)
7976 } else {
7977 let op_str = match op {
7978 0 => ":-",
7979 1 => ":=",
7980 2 => ":?",
7981 3 => ":+",
7982 4 => "-",
7983 5 => "=",
7984 6 => "?",
7985 7 => "+",
7986 _ => "-",
7987 };
7988 format!("${{{}{}{}}}", name, op_str, rhs)
7989 };
7990 paramsubst_to_value(&body)
7991 });
7992
7993 // `${var:offset[:length]}` — substring. Pops [name, offset, length].
7994 // length == -1 means "rest of string". Negative offset counts from end.
7995 // BUILTIN_PARAM_SUBSTRING — `${var:offset:length}` literal-int form.
7996 // PURE PASSTHRU: reconstruct `${name:offset:length}` and route
7997 // through `subst::paramsubst`. Length sentinel `i64::MIN` =
7998 // "no length given" (omit the `:length` portion).
7999 //
8000 // c:Src/subst.c:1571,3781 — `${name:-N}` is the colon-default
8001 // operator, NOT a substring with negative offset. zsh's lexical
8002 // rule disambiguates via a literal space: `${name: -N}` (space
8003 // before `-`) is the substring form. The reconstructed body MUST
8004 // preserve that space when offset < 0; otherwise paramsubst's
8005 // `:-` dispatch fires on the synthesized `${name:-N}` body and
8006 // returns N as the unset-default instead of slicing the last N
8007 // chars. Length-form `${name:-N:M}` has the same trap.
8008 vm.register_builtin(BUILTIN_PARAM_SUBSTRING, |vm, _argc| {
8009 let length = vm.pop().to_int();
8010 let offset = vm.pop().to_int();
8011 let name = vm.pop().to_str();
8012 let off_sep = if offset < 0 { " " } else { "" };
8013 let body = if length == i64::MIN {
8014 format!("${{{}:{}{}}}", name, off_sep, offset)
8015 } else {
8016 format!("${{{}:{}{}:{}}}", name, off_sep, offset, length)
8017 };
8018 paramsubst_to_value(&body)
8019 });
8020
8021 // BUILTIN_PARAM_SUBSTRING_EXPR — `${var:offset_expr[:length_expr]}` form.
8022 // PURE PASSTHRU: rebuild `${name:offset:length}` using the
8023 // expression text verbatim (paramsubst's offset/length
8024 // parser evaluates arith / param refs itself).
8025 //
8026 // c:Src/subst.c:1571,3781 — same `:-` disambiguation trap as
8027 // BUILTIN_PARAM_SUBSTRING. The expression text may itself start
8028 // with `-` (e.g. `${VAR:$((-1))}` arith resolves at the body-
8029 // assembly layer in some upstream paths, leaving `-1` in
8030 // off_expr). Insert a leading space when off_expr starts with
8031 // `-` so paramsubst's check_colon_subscript (subst.c:1571)
8032 // accepts the operand as a math expression instead of the
8033 // `:-` operator catching it.
8034 vm.register_builtin(BUILTIN_PARAM_SUBSTRING_EXPR, |vm, _argc| {
8035 let has_len = vm.pop().to_int() != 0;
8036 let len_expr = vm.pop().to_str();
8037 let off_expr = vm.pop().to_str();
8038 let name = vm.pop().to_str();
8039 let off_sep = if off_expr.starts_with('-') { " " } else { "" };
8040 let body = if has_len {
8041 format!("${{{}:{}{}:{}}}", name, off_sep, off_expr, len_expr)
8042 } else {
8043 format!("${{{}:{}{}}}", name, off_sep, off_expr)
8044 };
8045 paramsubst_to_value(&body)
8046 });
8047
8048 // `${var#pat}` / `${var##pat}` / `${var%pat}` / `${var%%pat}`
8049 // Pops [name, pattern, op_byte]. op: 0=`#` short-prefix, 1=`##` long,
8050 // 2=`%` short-suffix, 3=`%%` long. Glob-pattern matching via the
8051 // existing glob_match_static helper.
8052 // BUILTIN_PARAM_STRIP — `${var#pat}` / `${var##pat}` / `${var%pat}` /
8053 // `${var%%pat}`. PURE PASSTHRU: reconstruct the brace expression
8054 // and route through `subst::paramsubst`. (M)/(S) flags arrive
8055 // through SUB_FLAGS (already inside paramsubst's scope), so we
8056 // just clear the bridge-side cached read.
8057 vm.register_builtin(BUILTIN_PARAM_STRIP, |vm, _argc| {
8058 let _dq_flag = vm.pop().to_int() != 0;
8059 let op = vm.pop().to_int() as u8;
8060 let pattern = vm.pop().to_str();
8061 let name = vm.pop().to_str();
8062 let op_str = match op {
8063 0 => "#",
8064 1 => "##",
8065 2 => "%",
8066 3 => "%%",
8067 _ => "#",
8068 };
8069 let body = format!("${{{}{}{}}}", name, op_str, pattern);
8070 paramsubst_to_value(&body)
8071 });
8072
8073 // `$((expr))` — pops [expr_string], evaluates via MathEval which
8074 // honors integer-vs-float distinction (zsh-compatible). Returns
8075 // the result as Value::Str so it can be Concat'd into surrounding
8076 // word context.
8077 vm.register_builtin(BUILTIN_ARITH_EVAL, |vm, _argc| {
8078 // Pure path: evaluate expr, return string. errflag may be
8079 // set by arithsubst on math error; the caller decides
8080 // whether to clear it. For `(( ... ))` (math command) the
8081 // compile_arith path clears via BUILTIN_ARITH_CMD_FINISH;
8082 // for `$((... ))` (substitution inside another command)
8083 // errflag stays set so the surrounding command aborts —
8084 // matches c:Src/math.c "math errors propagate as errflag
8085 // through the containing word expansion".
8086 let expr = vm.pop().to_str();
8087 let result = crate::ported::subst::arithsubst(&expr, "", "");
8088 let _ = vm; // silence unused warning when no math error path mutates
8089 Value::str(result)
8090 });
8091
8092 // After-call hook used by compile_arith's `(( ... ))` path: when
8093 // arithsubst set errflag (math error), clear it and signal
8094 // status=2 in vm.last_status — matches zsh's c:exec.c arith-
8095 // failure: the math command exits 2 and the script continues.
8096 vm.register_builtin(BUILTIN_ARITH_CMD_FINISH, |vm, _argc| {
8097 use std::sync::atomic::Ordering;
8098 let live = crate::ported::utils::errflag.load(Ordering::Relaxed);
8099 let err = live & crate::ported::zsh_h::ERRFLAG_ERROR;
8100 let hard = live & crate::ported::zsh_h::ERRFLAG_HARD;
8101 if err != 0 {
8102 // c:Src/subst.c:3344 — when `${var:?msg}` fires, errflag
8103 // is OR'd with ERRFLAG_HARD to signal a script-abort
8104 // error (vs a recoverable math error like `$((1/0))`).
8105 // Clear only the ERRFLAG_ERROR bit; preserve
8106 // ERRFLAG_HARD so the next ERREXIT_CHECK aborts the
8107 // script. Bug #193 in docs/BUGS.md.
8108 if hard != 0 {
8109 // Keep ERRFLAG_HARD AND ERRFLAG_ERROR set so the
8110 // script-abort gate downstream still fires.
8111 vm.last_status = 2;
8112 Value::Status(2)
8113 } else {
8114 crate::ported::utils::errflag
8115 .fetch_and(!crate::ported::zsh_h::ERRFLAG_ERROR, Ordering::Relaxed);
8116 vm.last_status = 2;
8117 Value::Status(2)
8118 }
8119 } else {
8120 Value::Status(vm.last_status)
8121 }
8122 });
8123
8124 // `$(cmd)` — pops [cmd_string], routes through
8125 // run_command_substitution which performs an in-process pipe-capture.
8126 // Avoids the Op::CmdSubst sub-chunk word-emit bug
8127 // (`printf "a\nb"` produced "anb" via that path). Returns trimmed
8128 // output (trailing newlines stripped per POSIX cmd-sub semantics).
8129 vm.register_builtin(BUILTIN_CMD_SUBST_TEXT, |vm, _argc| {
8130 let cmd = vm.pop().to_str();
8131 // Inherit live $? into the inner shell so cmd-subst sees the
8132 // parent's most recent exit. Same rationale as the mode-3
8133 // backtick path above.
8134 let live_status = vm.last_status;
8135 let result = with_executor(|exec| {
8136 exec.set_last_status(live_status);
8137 exec.run_command_substitution(&cmd)
8138 });
8139 // Mirror run_command_substitution's exec.last_status side
8140 // effect into the VM's live counter so a containing
8141 // assignment's BUILTIN_SET_VAR — which reads vm.last_status
8142 // — sees the cmd-subst's exit. Without this, `a=$(false);
8143 // echo $?` reads stale 0 (vm.last_status was zeroed by
8144 // compile_assign's prelude SetStatus, and run_cmd_subst only
8145 // updated exec.last_status). Pull the value back through
8146 // exec since it owns the canonical post-subst record.
8147 let cs_status = with_executor(|exec| exec.last_status());
8148 vm.last_status = cs_status;
8149 // c:Src/exec.c — a command substitution running during a
8150 // command's word expansion makes its exit the status of an
8151 // otherwise-empty command (`$(exit 5)` → 5). Flag it so
8152 // BUILTIN_EXEC_DYNAMIC's null-command branch keeps `$?` instead
8153 // of resetting to 0.
8154 crate::ported::exec::use_cmdoutval.store(1, std::sync::atomic::Ordering::Relaxed);
8155 Value::str(result)
8156 });
8157
8158 // Text-based word expansion. Pops [preserved_text, mode_byte].
8159 // mode_byte:
8160 // 0 = Default — expand_string + xpandbraces + expand_glob
8161 // 1 = DoubleQuoted — strip outer `"…"`, expand_string only
8162 // (no brace, no glob — DQ semantics)
8163 // 2 = SingleQuoted — strip outer `'…'`, no expansion
8164 // (kept for symmetry; Snull early-return covers most SQ)
8165 // 3 = AltBackquote — strip backticks, run as cmd-sub
8166 // 7 = RedirTarget — same as Default but glob gated on MULTIOS
8167 // (c:Src/glob.c:2161-2167 xpandredir)
8168 // Single result → Value::str; multi → Value::Array.
8169 vm.register_builtin(BUILTIN_EXPAND_TEXT, |vm, _argc| {
8170 let mode = vm.pop().to_int() as u8;
8171 let text = vm.pop().to_str();
8172 // Sync vm.last_status → exec.last_status so cmd-subst (mode 3)
8173 // and any nested $? reads inside singsub see the live `$?`
8174 // from the most recent VM op. Without this, cmd-subst inside
8175 // arg-eval saw a stale exec.last_status that was zeroed at
8176 // the start of the current statement. Direct port of zsh's
8177 // pre-cmdsubst lastval propagation per Src/exec.c:4770.
8178 let live_status = vm.last_status;
8179 with_executor(|exec| exec.set_last_status(live_status));
8180 let result_value = with_executor(|exec| match mode {
8181 // Mode 1 = DoubleQuoted (argument context).
8182 // Mode 5 = DoubleQuoted in scalar-assignment context.
8183 // Both share the same DQ unescape pre-processing; mode 5
8184 // additionally bumps `in_scalar_assign` so subst_port's
8185 // paramsubst sees ssub=true and suppresses split flags
8186 // `(f)` / `(s:STR:)` / `(0)` per Src/subst.c:1759 +
8187 // Src/exec.c::addvars line 2546 (the PREFORK_SINGLE bit
8188 // C zsh sets when prefork-ing the assignment RHS).
8189 1 | 5 => {
8190 // DoubleQuoted: strip outer `"…"` if present. In DQ
8191 // context, `\` escapes the DQ-special chars `$`, `` ` ``,
8192 // `"`, `\`. zsh's expand_string expects the lexer's
8193 // `\0X` literal-marker for an already-escaped char, so
8194 // we pre-process: `\$` → `\0$`, `\\` → `\0\`, etc. Then
8195 // expand_string handles the rest.
8196 let inner = if text.len() >= 2 && text.starts_with('"') && text.ends_with('"') {
8197 &text[1..text.len() - 1]
8198 } else {
8199 text.as_str()
8200 };
8201 // The lexer's dquote_parse (Src/lex.c) already tokenized
8202 // DQ contents: `$` → Qstring (\u{8c}), `\$`/`\\`/`\"`/
8203 // `` \` `` → Bnull (\u{9f}) + literal. Stringsubst /
8204 // multsub recognize these markers natively. We pass
8205 // `inner` through verbatim — no re-tokenization needed.
8206 let prepped: String = inner.to_string();
8207 // Tell parameter-flag application that we're inside
8208 // double quotes — array-only flags ((o), (O), (n),
8209 // (i), (M), (u)) must be no-ops here per zsh.
8210 exec.in_dq_context += 1;
8211 if mode == 5 {
8212 exec.in_scalar_assign += 1;
8213 }
8214 // Mode 1 = argv DQ word; mode 5 = scalar-assign RHS.
8215 // In C zsh, the corresponding prefork-on-list paths
8216 // are: argv → `prefork(argv_list, 0)` returns multi-
8217 // word LinkList (Src/exec.c::execcmd), assignment →
8218 // `prefork(rhs_list, PREFORK_SINGLE|PREFORK_ASSIGN)`
8219 // returns single-word (Src/exec.c::addvars line
8220 // 2546). zshrs's `multsub` (Src/subst.c:544) is the
8221 // multi-result variant; `singsub` (Src/subst.c:514)
8222 // asserts ≤1 node. Mode 5 keeps singsub; mode 1
8223 // switches to multsub so `"${(@)arr}"`/`"$@"`/
8224 // `"${arr[@]}"` in argv context emit multiple words
8225 // as the C path would.
8226 // c:Src/lex.c untokenize — the final argv pass C runs
8227 // on every expanded word (glob.c:1862 / exec.c) drops
8228 // the Nularg empty-word sentinel remnulargs left in
8229 // place and folds any remaining token chars. Without
8230 // it, quoted splits with empty pieces
8231 // ("${(s:|:)x}" on "|a|b|") leak U+00A1 into argv.
8232 let result_value = if mode == 5 {
8233 let out = crate::ported::subst::singsub(&prepped);
8234 Value::str(crate::ported::lex::untokenize(&out))
8235 } else {
8236 let (_first, nodes, _ms_ws, _ret) = crate::ported::subst::multsub(&prepped, 0);
8237 // c:Src/subst.c:655 — multsub returns Vec::new()
8238 // for zero-word results (quoted array splat that
8239 // resolved to empty array). Surface as
8240 // Value::Array(vec![]) so the downstream array
8241 // assignment / argv flattening sees ZERO args.
8242 // Previous Rust port returned Value::str("") which
8243 // surfaced as ONE empty arg. Bug #120 in
8244 // docs/BUGS.md.
8245 if nodes.is_empty() {
8246 Value::Array(Vec::new())
8247 } else if nodes.len() == 1 {
8248 Value::str(crate::ported::lex::untokenize(
8249 &nodes.into_iter().next().unwrap(),
8250 ))
8251 } else {
8252 Value::Array(
8253 nodes
8254 .into_iter()
8255 .map(|n| Value::str(crate::ported::lex::untokenize(&n)))
8256 .collect(),
8257 )
8258 }
8259 };
8260 if mode == 5 {
8261 exec.in_scalar_assign -= 1;
8262 }
8263 exec.in_dq_context -= 1;
8264 result_value
8265 }
8266 2 => {
8267 // SingleQuoted: pure literal, strip outer `'…'`.
8268 let inner = if text.len() >= 2 && text.starts_with('\'') && text.ends_with('\'') {
8269 &text[1..text.len() - 1]
8270 } else {
8271 text.as_str()
8272 };
8273 Value::str(inner.to_string())
8274 }
8275 3 => {
8276 // Backquote command sub: strip outer backticks.
8277 // Word-split the result on IFS when the surrounding
8278 // word is unquoted — zsh: `print -l \`echo a b c\``
8279 // emits one arg per word. The $(…) path applies the
8280 // same split via BUILTIN_WORD_SPLIT after capture; do
8281 // the equivalent here for the `…` form.
8282 let inner = if text.len() >= 2 && text.starts_with('`') && text.ends_with('`') {
8283 &text[1..text.len() - 1]
8284 } else {
8285 text.as_str()
8286 };
8287 // Apply the live VM status before running the inner
8288 // shell so the inherited $? matches zsh's lastval
8289 // propagation.
8290 exec.set_last_status(live_status);
8291 let captured = exec.run_command_substitution(inner);
8292 let trimmed = captured.trim_end_matches('\n');
8293 if exec.in_dq_context > 0 {
8294 Value::str(trimmed.to_string())
8295 } else {
8296 let ifs = exec.scalar("IFS").unwrap_or_else(|| " \t\n".to_string());
8297 let parts: Vec<Value> = trimmed
8298 .split(|c: char| ifs.contains(c))
8299 .filter(|s| !s.is_empty())
8300 .map(|s| Value::str(s.to_string()))
8301 .collect();
8302 if parts.is_empty() {
8303 Value::str(String::new())
8304 } else if parts.len() == 1 {
8305 parts.into_iter().next().unwrap()
8306 } else {
8307 Value::Array(parts)
8308 }
8309 }
8310 }
8311 4 => {
8312 // HeredocBody: expand variables / command-subst / arith
8313 // but NOT glob or brace. Heredoc lines like `[42]` must
8314 // pass through verbatim — running them through the
8315 // default pipeline triggers NOMATCH on the literal.
8316 Value::str(crate::ported::subst::singsub(&text))
8317 }
8318 _ => {
8319 // Default (unquoted): the lexer's gettokstr already
8320 // tokenized backslash-escapes (`\$` → Bnull+$, etc).
8321 // Pass `text` through verbatim — multsub/stringsubst
8322 // recognize the markers natively. No bridge-side
8323 // re-tokenization needed.
8324 //
8325 // Mode 6 = unquoted RHS in scalar-assign context.
8326 // Pass PREFORK_ASSIGN so prefork's filesub colon-walk
8327 // fires per c:Src/exec.c:2546.
8328 let prepped: String = text.clone();
8329 if std::env::var("ZSHRS_TRACE_DEFP").is_ok() {
8330 eprintln!(
8331 "[TRACE_DEFP] text={:?} prepped={:?} mode={}",
8332 text, prepped, mode
8333 );
8334 }
8335 let pf_flags = if mode == 6 {
8336 crate::ported::zsh_h::PREFORK_ASSIGN
8337 } else {
8338 0
8339 };
8340 // c:Src/subst.c:544+ — `multsub(&prepped, 0)` is the
8341 // unquoted-argv equivalent of zsh's `prefork(list,
8342 // 0, NULL)` for a single-element list. Returns the
8343 // post-expansion node list (Vec<String>) so array-
8344 // shape results (e.g. `${a:e}`, `${a[@]}`,
8345 // `${(s::)str}`) splat into multiple argv words.
8346 // singsub() collapses to one string and discards the
8347 // splat — parity bug #28 (whole-array modifier).
8348 let (_first, nodes, _ms_ws, _ret) =
8349 crate::ported::subst::multsub(&prepped, pf_flags);
8350 if std::env::var("ZSHRS_TRACE_MULTSUB").is_ok() {
8351 eprintln!("[TRACE_MULTSUB] prepped={:?} nodes={:?}", prepped, nodes);
8352 }
8353 // c:Src/subst.c:166 — xpandbraces runs AFTER prefork's
8354 // substitution pass and BEFORE untokenize/glob. Per
8355 // word, scan for Inbrace TOKEN and expand. Words that
8356 // don't contain Inbrace TOKEN pass through unchanged.
8357 // Brace expansion is done here (inside the bridge
8358 // default arm) instead of via a post-EXPAND_TEXT
8359 // BRACE_EXPAND emit because untokenize (line below)
8360 // strips TOKEN bytes, after which the strict-TOKEN
8361 // xpandbraces gate would no longer match.
8362 let brace_ccl = opt_state_get("braceccl").unwrap_or(false);
8363 // c:Src/options.c — `no_brace_expand` (negated
8364 // `braceexpand`) gates brace expansion entirely.
8365 // When off, `{a,b}` stays literal.
8366 let brace_expand = opt_state_get("braceexpand").unwrap_or(true);
8367 let pre_brace: Vec<String> = if nodes.is_empty() {
8368 vec![String::new()]
8369 } else {
8370 nodes
8371 };
8372 let brace_expanded: Vec<String> = pre_brace
8373 .into_iter()
8374 .flat_map(|w| {
8375 if brace_expand && w.contains('\u{8f}') {
8376 crate::ported::glob::xpandbraces(&w, brace_ccl)
8377 } else {
8378 vec![w]
8379 }
8380 })
8381 .collect();
8382 // zsh stores the option as `glob` (default ON);
8383 // `setopt noglob` writes `glob=false`. Honor either
8384 // form so the dispatcher behaves the same as zsh.
8385 // Mode 7 = redirect-target word: glob only under
8386 // MULTIOS (c:Src/glob.c:2161-2167 xpandredir,
8387 // "Globbing is only done for multios.").
8388 let noglob = opt_state_get("noglob").unwrap_or(false)
8389 || opt_state_get("GLOB").map(|v| !v).unwrap_or(false)
8390 || !opt_state_get("glob").unwrap_or(true)
8391 || (mode == 7 && !opt_state_get("multios").unwrap_or(true));
8392 let parts: Vec<String> = brace_expanded
8393 .into_iter()
8394 .flat_map(|s| {
8395 // The lexer leaves glob metacharacters in their
8396 // META-encoded form: `*` → `\u{87}`, `?` →
8397 // `\u{86}`, `[` → `\u{91}`, etc. expand_string
8398 // doesn't untokenize them, so the literal-char
8399 // checks below (`s.contains('*')`) would miss
8400 // every real glob and skip expand_glob — that
8401 // bug let `echo *.toml` print the literal
8402 // `*.toml` because the META `\u{87}` never
8403 // matched the literal `*`. Untokenize once so
8404 // the metacharacter checks see the canonical
8405 // form. zsh's pattern.c expects `*` etc. as
8406 // bare chars at the glob layer.
8407 // c:Src/pattern.c:4306 haswilds on the still-
8408 // TOKENIZED word (pre-untokenize), matching C's
8409 // zglob entry gate (Src/glob.c:1230) which runs
8410 // on the lexer-tokenized string. haswilds
8411 // matches ONLY token codes: source-level
8412 // `*.toml` carries Star and fires; bare literal
8413 // `[`/`*`/`?` from `$'...'` decode, `:-`
8414 // default values, or nested-substitution
8415 // results were never shtokenize'd (C
8416 // subst.c:3231 sets globsubst=0 in the `:-`
8417 // arm) and stay literal — bug #625. Plain
8418 // multibyte text (`↔`) never matches a token
8419 // codepoint — bug #627.
8420 let is_glob_pre = !noglob && crate::ported::pattern::haswilds(&s);
8421 let s = crate::lex::untokenize(&s);
8422 // Skip glob expansion for assignment-shaped
8423 // words (`NAME=value`). zsh doesn't expand the
8424 // RHS of an assignment as a path glob unless
8425 // `setopt globassign` is set, and feeding such
8426 // words through expand_glob makes NOMATCH
8427 // (default ON) fire spuriously on
8428 // `integer i=2*3+1`, `path=*.rs`, etc.
8429 let is_assignment_shape = {
8430 let bytes = s.as_bytes();
8431 let mut i = 0;
8432 if !bytes.is_empty()
8433 && (bytes[0] == b'_' || bytes[0].is_ascii_alphabetic())
8434 {
8435 i += 1;
8436 while i < bytes.len()
8437 && (bytes[i] == b'_' || bytes[i].is_ascii_alphanumeric())
8438 {
8439 i += 1;
8440 }
8441 i < bytes.len() && bytes[i] == b'='
8442 } else {
8443 false
8444 }
8445 };
8446 // Glob-trigger decision: pre-untokenize
8447 // haswilds_tokens_only result (computed above
8448 // before the untokenize that collapses META
8449 // tokens to their ASCII forms). The TOKEN-only
8450 // gate matches C `Src/pattern.c:4306-4376`
8451 // exactly — only Inbrack/Star/Quest/Inpar/Bar/
8452 // Inang/Pound/Hat token codes count as wild,
8453 // not their literal ASCII counterparts. Source-
8454 // level `*.toml` carries Star token so globs;
8455 // `$'…'`-decoded `[abc]` carries bare `[` so
8456 // stays literal. Bug #625.
8457 if is_glob_pre && !is_assignment_shape {
8458 exec.expand_glob(&s)
8459 } else if is_assignment_shape
8460 && crate::ported::zsh_h::isset(crate::ported::zsh_h::MAGICEQUALSUBST)
8461 {
8462 // c:Src/exec.c:3353 — when MAGIC_EQUAL_SUBST is set
8463 // on a non-typeset command, esprefork = PREFORK_TYPESET,
8464 // so every NAME=value arg runs through
8465 // filesub(PREFORK_TYPESET): the `~`/`=` after the
8466 // first `=` (and after each `:`) undergo filename
8467 // expansion. `print foo=~/bar` → `foo=$HOME/bar`.
8468 // filesubstr (subst.c:741) keys on the Tilde TOKEN,
8469 // not literal `~`; this `s` was already untokenized
8470 // above, so re-tokenize (as BUILTIN_MAGIC_EQUALS_PREFORK
8471 // does) before filesub, then untokenize the result.
8472 let mut tokd = s.clone();
8473 crate::ported::glob::shtokenize(&mut tokd);
8474 let exp = crate::ported::subst::filesub(
8475 &tokd,
8476 crate::ported::zsh_h::PREFORK_TYPESET,
8477 );
8478 vec![crate::lex::untokenize(&exp).to_string()]
8479 } else {
8480 vec![s]
8481 }
8482 })
8483 .collect();
8484 if parts.len() == 1 {
8485 let only = parts.into_iter().next().unwrap_or_default();
8486 // Empty unquoted expansion → drop the arg entirely
8487 // (zsh "remove empty unquoted words" rule). Returning
8488 // an empty Value::Array makes pop_args contribute zero
8489 // items. Direct port of subst.c's empty-elide pass at
8490 // the end of multsub which removes empty linknodes
8491 // from unquoted contexts. Quoted DQ/SQ paths (modes
8492 // 1/2/5) take separate arms above and always emit
8493 // Value::Str so the empty arg survives.
8494 if only.is_empty() {
8495 Value::Array(Vec::new())
8496 } else {
8497 Value::str(only)
8498 }
8499 } else {
8500 Value::Array(parts.into_iter().map(Value::str).collect())
8501 }
8502 }
8503 });
8504 // Pull any inner cmd-subst (`` `cmd` `` via mode 3 or via
8505 // mode 0/6 multsub → getoutput, `$(cmd)` via the default
8506 // arm's multsub path, nested `$()`s reached through
8507 // stringsubst) back into vm.last_status so a containing
8508 // assignment's BUILTIN_SET_VAR — which reads vm.last_status —
8509 // sees the cmd-subst's exit. Without this, backtick
8510 // assignments (`a=\`false\`; echo $?`) reported 0 because the
8511 // ported LASTVAL update never reached the VM-side counter.
8512 let cs_status = with_executor(|exec| exec.last_status());
8513 vm.last_status = cs_status;
8514 result_value
8515 });
8516
8517 // `${#name}` — pops [name]. Returns the value's element count for
8518 // arrays (indexed and assoc) or character length for scalars.
8519 // BUILTIN_PARAM_LENGTH — `${#name}`. PURE PASSTHRU.
8520 vm.register_builtin(BUILTIN_PARAM_LENGTH, |vm, _argc| {
8521 let name = vm.pop().to_str();
8522 // PARAM_LENGTH's empty-result semantics differ from
8523 // paramsubst_to_value: 0 nodes → "0" (numeric length), not
8524 // empty array. paramsubst on `${#X}` always returns at least
8525 // one node in practice (the length string); the empty case
8526 // is defensive.
8527 let mut ret_flags: i32 = 0;
8528 let (_full, _pos, nodes) = crate::ported::subst::paramsubst(
8529 &format!("${{#{}}}", name),
8530 0,
8531 false,
8532 0i32,
8533 &mut ret_flags,
8534 );
8535 if crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed) != 0 {
8536 with_executor(|exec| exec.set_last_status(1));
8537 }
8538 if nodes.is_empty() {
8539 Value::str("0")
8540 } else {
8541 nodes_to_value(nodes)
8542 }
8543 });
8544
8545 // `${var/pat/repl}` / `${var//pat/repl}` / `${var/#pat/repl}` /
8546 // `${var/%pat/repl}` — Pops [name, pattern, replacement, op_byte].
8547 // op: 0=first, 1=all, 2=anchor-prefix (`/#`), 3=anchor-suffix (`/%`).
8548 // BUILTIN_PARAM_REPLACE — `${var/pat/repl}` / `${var//pat/repl}` /
8549 // `${var/#pat/repl}` / `${var/%pat/repl}`. PURE PASSTHRU.
8550 vm.register_builtin(BUILTIN_PARAM_REPLACE, |vm, _argc| {
8551 let dq_flag = vm.pop().to_int() != 0;
8552 let op = vm.pop().to_int() as u8;
8553 let repl = vm.pop().to_str();
8554 let pattern = vm.pop().to_str();
8555 let name = vm.pop().to_str();
8556 // DQ context: C's lexer marks every `$` inside double quotes
8557 // as the Qstring token (Src/lex.c dquote_parse) and keeps `'`
8558 // a plain char — so a DQ replacement's `$'…'` is LITERAL in
8559 // C (Src/subst.c:301 decodes only the tokenized Snull form;
8560 // `"${a/X/$'\0'}"` keeps the five chars `$'\0'`). The body
8561 // rebuilt below re-enters stringsubst as raw text, which
8562 // would mis-decode `$'…'` as ANSI-C; stamp the Qstring
8563 // marker on the repl's `$` so stringsubst sees the same DQ
8564 // signal C's tokens carry. The PATTERN side keeps decoding
8565 // (matches observed zsh: the pattern's `$'\0'` matches a
8566 // real NUL while the repl's stays literal).
8567 let repl = if dq_flag {
8568 repl.replace('$', "\u{8c}")
8569 } else {
8570 repl
8571 };
8572 // op encoding: 0 = first `/`, 1 = all `//`, 2 = anchor-prefix
8573 // `/#`, 3 = anchor-suffix `/%`. The brace form distinguishes
8574 // first-vs-all by single vs doubled slash, and anchored by
8575 // a `#` or `%` immediately after the slash(es).
8576 let body = match op {
8577 0 => format!("${{{}/{}/{}}}", name, pattern, repl),
8578 1 => format!("${{{}//{}/{}}}", name, pattern, repl),
8579 2 => format!("${{{}/#{}/{}}}", name, pattern, repl),
8580 3 => format!("${{{}/%{}/{}}}", name, pattern, repl),
8581 _ => format!("${{{}/{}/{}}}", name, pattern, repl),
8582 };
8583 // c:Src/subst.c:1625 — paramsubst's qt flag. The compiler
8584 // threads the word's DQ context onto the stack; dropping it
8585 // (the old `let _dq_flag`) ran the rebuilt body with qt=false
8586 // whenever the opcode fired outside an EXPAND_TEXT scope, so
8587 // DQ-only semantics inside the replacement (e.g. `$'` staying
8588 // literal per Src/subst.c:301 — `"${a/x/$'\t'q}"`) were lost.
8589 // Bump in_dq_context exactly like EXPAND_TEXT mode 1 so
8590 // paramsubst_to_value's qt probe sees the right context.
8591 if dq_flag {
8592 with_executor(|exec| exec.in_dq_context += 1);
8593 }
8594 let ret = paramsubst_to_value(&body);
8595 if dq_flag {
8596 with_executor(|exec| exec.in_dq_context -= 1);
8597 }
8598 ret
8599 });
8600
8601 vm.register_builtin(BUILTIN_REGISTER_COMPILED_FN, |vm, argc| {
8602 let args = pop_args(vm, argc);
8603 let mut iter = args.into_iter();
8604 let name = iter.next().unwrap_or_default();
8605 let body_b64 = iter.next().unwrap_or_default();
8606 let body_source = iter.next().unwrap_or_default();
8607 let line_base_str = iter.next().unwrap_or_default();
8608 let line_base: i64 = line_base_str.parse().unwrap_or(0);
8609 let bytes = base64_decode(&body_b64);
8610 let status = match bincode::deserialize::<fusevm::Chunk>(&bytes) {
8611 Ok(chunk) => with_executor(|exec| {
8612 // c:Src/exec.c:5383 — `shf->filename =
8613 // ztrdup(scriptfilename);` — the function's
8614 // definition-file is read from the canonical
8615 // file-scope `scriptfilename` global at compile
8616 // time, NOT from a per-executor struct field.
8617 // exec.scriptfilename is seeded once at
8618 // bins/zshrs.rs:1717 to the bin basename ("zsh")
8619 // and never updates on source/dot, so reading from
8620 // it left every user function's def_file as "zsh".
8621 // Route through scriptfilename_get() so source /
8622 // dot's set_scriptfilename calls propagate.
8623 let def_file = crate::ported::utils::scriptfilename_get()
8624 .or_else(|| exec.scriptfilename.clone());
8625 if !body_source.is_empty() {
8626 exec.function_source
8627 .insert(name.clone(), body_source.clone());
8628 }
8629 exec.function_line_base.insert(name.clone(), line_base);
8630 exec.function_def_file.insert(name.clone(), def_file);
8631 // PFA-SMR aspect: every `name() {}` / `function name { }`
8632 // funnels through here at compile time. Emit one record
8633 // with the function name + raw body source.
8634 #[cfg(feature = "recorder")]
8635 if crate::recorder::is_enabled() {
8636 let ctx = exec.recorder_ctx();
8637 let body = if body_source.is_empty() {
8638 None
8639 } else {
8640 Some(body_source.as_str())
8641 };
8642 crate::recorder::emit_function(&name, body, ctx);
8643 }
8644 // Mirror into canonical shfunctab so scanfunctions /
8645 // ${(k)functions} / functions builtin see user defs.
8646 // C: exec.c:funcdef → shfunctab->addnode(ztrdup(name),shf).
8647 if let Ok(mut tab) = crate::ported::hashtable::shfunctab_lock().write() {
8648 let mut shf = crate::ported::hashtable::shfunc_with_body(&name, &body_source);
8649 // c:Src/exec.c:5409 — `shf->lineno = lineno;`. Use
8650 // the same max(1, line_base) clamp as the synth_shf
8651 // in vm_helper::dispatch_function_call. Bug #396.
8652 shf.lineno = std::cmp::max(1, line_base);
8653 tab.add(shf);
8654 }
8655 // c:Src/exec.c:5460-5475 — `TRAP<SIG>() { ... }` is the
8656 // function-named trap install. zsh detects the `TRAP`
8657 // prefix at func-def time and calls
8658 // `settrap(signum, NULL, ZSIG_FUNC)` so the next
8659 // dispatch of that signal routes to the named shfunc.
8660 // Bug #157 in docs/BUGS.md — fusevm_bridge's funcdef
8661 // opcode skipped this dispatch entirely, so TRAPEXIT /
8662 // TRAPUSR1 / TRAPZERR / TRAPDEBUG never fired.
8663 if name.len() > 4 && name.starts_with("TRAP") {
8664 if let Some(sn) = crate::ported::jobs::getsigidx(&name[4..]) {
8665 let _ = crate::ported::signals::settrap(
8666 sn,
8667 None,
8668 crate::ported::zsh_h::ZSIG_FUNC as i32,
8669 );
8670 }
8671 }
8672 exec.functions_compiled.insert(name, chunk);
8673 0
8674 }),
8675 Err(_) => 1,
8676 };
8677 Value::Status(status)
8678 });
8679
8680 // Wire the ShellHost so direct shell ops (Op::Glob, Op::TildeExpand,
8681 // Op::ExpandParam, Op::CmdSubst, Op::CallFunction, etc.) route through
8682 // ZshrsHost back into the executor.
8683 vm.set_shell_host(Box::new(ZshrsHost));
8684}
8685
8686impl ZshrsHost {
8687 /// True iff `c` can be a `(j:…:)` / `(s:…:)` delimiter — non-alphanumeric,
8688 /// non-underscore. Restricting to punctuation avoids `(jL)` consuming `L`
8689 /// as a delim instead of as the next flag.
8690 fn is_zsh_flag_delim(c: char) -> bool {
8691 !c.is_ascii_alphanumeric() && c != '_'
8692 }
8693}
8694
8695/// Shared `${name[idx]}` subscript dispatch for BUILTIN_ARRAY_INDEX
8696/// and the KSHARRAYS-unset arm of BUILTIN_ARRAY_INDEX_UNBRACED.
8697///
8698/// c:Src/subst.c subscript parsing — when paramsubst re-parses the
8699/// synthesized `${name[idx]}` body, characters like `'` `"` `\` `$`
8700/// etc. are LEXER-active inside the `[…]` and get reinterpreted
8701/// (quote-strip, paramsubst recursion, …). For PRE-EVALUATED key
8702/// strings (the dynamic-key fast path at compile_zsh.rs:3234 already
8703/// expanded `$k` via EXPAND_TEXT), the idx is a literal string that
8704/// must match the stored key byte-for-byte — no further
8705/// reinterpretation. Direct assoc lookup bypasses the lexer for this
8706/// case, avoiding the quote-strip bug where `h[a'b]` failed to
8707/// resolve because paramsubst's subscript lexer treated the `'` as a
8708/// quote. Bug #338. Only fires for simple assoc-name + non-flag idx
8709/// (no outer-flag sentinels, no `(…)` flag prefix on idx, no splat
8710/// operator). Other paths (slice, splat, flag-based search,
8711/// magic-assoc) still flow through paramsubst.
8712fn array_index_lookup(name: &str, idx: &str) -> Value {
8713 let idx_is_simple = !idx.starts_with('(') && idx != "@" && idx != "*" && !idx.contains(',');
8714 if idx_is_simple {
8715 if let Some(v) = with_executor(|exec| exec.assoc(name).and_then(|m| m.get(idx).cloned())) {
8716 return Value::str(v);
8717 }
8718 }
8719 let body = format!("${{{}[{}]}}", name, idx);
8720 paramsubst_to_value(&body)
8721}
8722
8723/// KSHARRAYS bare-`$name` expansion words for the unbraced
8724/// no-subscript form (BUILTIN_ARRAY_INDEX_UNBRACED's KSHARRAYS arm).
8725///
8726/// - `@` / `*` stay the full positional list (the c:Src/params.c:
8727/// 2293-2296 first-element collapse is gated on
8728/// `itype_end(t, IIDENT, 1) != t` — an identifier-shaped name —
8729/// which `@`/`*` are not). The literal `[idx]` then joins the LAST
8730/// word, matching zsh 5.9: `setopt ksharrays; set -- p q;
8731/// print -- $@[0]` → `zsh:1: no matches found: q[0]`.
8732/// - Identifier-named arrays collapse to the FIRST element
8733/// (c:Src/params.c:2293-2296 `v->end = 1, v->isarr = 0`).
8734/// - Assocs collapse to the first value in scan order; the `options`
8735/// magic assoc's scan order is `OPTIONTAB` bucket order (first key
8736/// `posixargzero`), matching zsh 5.9: `emulate sh -L;
8737/// print $options[posixargzero]` → `off[posixargzero]`.
8738/// - Scalars / unset names expand to their value / empty (zsh 5.9:
8739/// `setopt ksharrays; print -- $unsetvar[0]` →
8740/// `zsh:1: no matches found: [0]`).
8741fn ksharrays_bare_words(name: &str) -> Vec<String> {
8742 if name == "@" || name == "*" {
8743 return with_executor(|exec| exec.pparams());
8744 }
8745 // Magic special-parameter lookups first — mirrors the
8746 // BUILTIN_GET_VAR precedence (partab before executor tables).
8747 if let Some(vals) = crate::vm_helper::partab_array_get(name) {
8748 return vec![vals.into_iter().next().unwrap_or_default()];
8749 }
8750 if let Some(keys) = crate::vm_helper::partab_scan_keys(name) {
8751 let v = keys
8752 .first()
8753 .and_then(|k| crate::vm_helper::partab_get(name, k))
8754 .unwrap_or_default();
8755 return vec![v];
8756 }
8757 let arr_or_assoc = with_executor(|exec| {
8758 if let Some(arr) = exec.array(name) {
8759 // c:Src/params.c:2293-2296 — first element only.
8760 return Some(arr.first().cloned().unwrap_or_default());
8761 }
8762 if let Some(map) = exec.assoc(name) {
8763 // Mirrors BUILTIN_GET_VAR's bare-assoc ordering
8764 // (sorted keys, first value).
8765 let mut keys: Vec<&String> = map.keys().collect();
8766 keys.sort();
8767 return Some(
8768 keys.first()
8769 .and_then(|k| map.get(*k).cloned())
8770 .unwrap_or_default(),
8771 );
8772 }
8773 None
8774 });
8775 if let Some(v) = arr_or_assoc {
8776 return vec![v];
8777 }
8778 vec![with_executor(|exec| exec.get_variable(name))]
8779}
8780
8781/// Run `body` through `crate::ported::subst::paramsubst` and convert
8782/// the resulting node list into a fusevm `Value`. Centralises the
8783/// pattern duplicated across ~10 BUILTIN_* handlers:
8784/// - build a `${...}` body string from opcode operands
8785/// - paramsubst the body
8786/// - propagate errflag to `exec.last_status`
8787/// - delegate the LinkList → Value conversion to `nodes_to_value`
8788///
8789/// **Extension** — Rust-only helper. No direct C analog because C
8790/// zsh uses LinkList everywhere; the conversion happens at the
8791/// boundary back into the VM's stack.
8792fn paramsubst_to_value(body: &str) -> Value {
8793 // c:Src/subst.c:1625 paramsubst's `qt` flag is the C signal that
8794 // the current expansion is inside `"…"`. The fast-path bridges
8795 // (BUILTIN_PARAM_*, BUILTIN_BRIDGE_BRACE_ARRAY) used to hardcode
8796 // qt=false, which silently broke DQ-only semantics inside
8797 // `${arr:^other}` / `${arr:^^other}` (Src/subst.c:3456-3520).
8798 // The executor's `in_dq_context` counter is bumped by EXPAND_TEXT
8799 // mode 1 / mode 5 before the bridge fires, so reading it here
8800 // propagates the DQ flag without changing every bridge call site.
8801 let qt = with_executor(|exec| exec.in_dq_context > 0);
8802 let mut ret_flags: i32 = 0;
8803 let (_full, _pos, nodes) = crate::ported::subst::paramsubst(body, 0, qt, 0i32, &mut ret_flags);
8804 if crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed) != 0 {
8805 with_executor(|exec| exec.set_last_status(1));
8806 }
8807 // c:Src/lex.c untokenize — the final argv pass C runs on every
8808 // expanded word (glob.c:1862 / exec.c) DROPS the Nularg
8809 // empty-word sentinel (c:2089 `if (c != Nularg)`) that
8810 // remnulargs faithfully leaves in place (glob.c:3673 re-adds it
8811 // for all-empty results). These fast-path bridges are terminal —
8812 // their output lands directly in argv slots — so apply it here.
8813 // Without it, quoted splits with empty pieces ("${(s:|:)x}" on
8814 // "|a|b|") leak U+00A1 into argv.
8815 let nodes: Vec<String> = nodes
8816 .into_iter()
8817 .map(|n| crate::ported::lex::untokenize(&n))
8818 .collect();
8819 nodes_to_value(nodes)
8820}
8821
8822/// Wrap a `Vec<String>` (e.g. paramsubst nodes, multsub parts,
8823/// xpandbraces output) into a fusevm `Value`: 0 → empty Array, 1 →
8824/// Str, >1 → Array. Same unwrap idiom every handler that calls a
8825/// canonical Vec-returning fn does.
8826fn nodes_to_value(nodes: Vec<String>) -> Value {
8827 // c:Src/glob.c:3649 remnulargs — strip the Nularg (`\u{a1}`)
8828 // sentinel and other INULL bytes that paramsubst's splat block
8829 // emits for empty array elements (so prefork's empty-node-delete
8830 // pass doesn't drop them). Downstream consumers (cond `-z`/`-n`,
8831 // command args, etc.) must see the post-remnulargs strings. Bug
8832 // #185 in docs/BUGS.md: `[[ -z "${b[@]}" ]]` for b=("") returned
8833 // false because the leftover `\u{a1}` had StringLen=1.
8834 let stripped: Vec<String> = nodes
8835 .into_iter()
8836 .map(|mut s| {
8837 crate::ported::glob::remnulargs(&mut s);
8838 s
8839 })
8840 .collect();
8841 if stripped.is_empty() {
8842 Value::Array(Vec::new())
8843 } else if stripped.len() == 1 {
8844 let only = stripped.into_iter().next().unwrap();
8845 // c:Src/subst.c:183-186 — `else if (!(flags & PREFORK_SINGLE)
8846 // && !(*ret_flags & PREFORK_KEY_VALUE) && !keep)
8847 // uremnode(list, node);`
8848 // C zsh's prefork removes empty linknodes from the result
8849 // list when in non-SINGLE (argv-context) mode. The ported
8850 // prefork at subst.rs:388-396 honors the same delete-empty
8851 // pass, but some paramsubst paths land here with a single-
8852 // empty-string Vec instead of an empty Vec (paramsubst's
8853 // slice / substring / parameter-flag branches allocate a
8854 // result before checking emptiness). Mirror the prefork
8855 // drop at this layer: single-empty under !in_dq_context
8856 // collapses to Value::Array(empty), and pop_args (line 6243)
8857 // splats the empty Array → zero argv words. DQ context
8858 // (in_dq_context > 0) keeps the empty string so
8859 // `echo "${UNSET}"` still produces an empty arg per zsh's
8860 // quoting rules (c:Src/subst.c:1650-1656 isarr comment).
8861 if only.is_empty() {
8862 let in_dq = with_executor(|exec| exec.in_dq_context > 0);
8863 if !in_dq {
8864 return Value::Array(Vec::new());
8865 }
8866 }
8867 Value::str(only)
8868 } else {
8869 Value::Array(stripped.into_iter().map(Value::str).collect())
8870 }
8871}
8872
8873fn pop_args(vm: &mut fusevm::VM, argc: u8) -> Vec<String> {
8874 let mut popped: Vec<Value> = Vec::with_capacity(argc as usize);
8875 for _ in 0..argc {
8876 popped.push(vm.pop());
8877 }
8878 popped.reverse();
8879 let mut args: Vec<String> = Vec::with_capacity(popped.len());
8880 for v in popped {
8881 match v {
8882 Value::Array(items) => {
8883 for item in items {
8884 args.push(item.to_str());
8885 }
8886 }
8887 other => args.push(other.to_str()),
8888 }
8889 }
8890 // `expand_glob` set the glob-failed cell when a no-match glob
8891 // triggered nomatch (c:Src/glob.c:1877). Signal the failure via
8892 // last_status + the per-command glob_failed cell; the dispatcher
8893 // (`host_exec_external`) consumes + clears it and returns status 1
8894 // without running the command body.
8895 if with_executor(|exec| exec.current_command_glob_failed.get()) {
8896 with_executor(|exec| exec.set_last_status(1));
8897 }
8898 // `$_` tracks the last argument of the PREVIOUSLY executed
8899 // command (zsh / bash convention). Promote the deferred value
8900 // into `$_` BEFORE this command runs (so `echo $_` reads the
8901 // prior command's last arg) then stash THIS command's last arg
8902 // for the next dispatch.
8903 let new_last = args.last().cloned();
8904 with_executor(|exec| {
8905 if let Some(prev) = exec.pending_underscore.take() {
8906 exec.set_scalar("_".to_string(), prev);
8907 }
8908 if let Some(last) = new_last {
8909 exec.pending_underscore = Some(last);
8910 }
8911 });
8912 args
8913}
8914
8915/// zsh dispatch order is alias → function → builtin → external. The
8916/// compiler emits direct CallBuiltin ops for known builtin names for
8917/// perf, which silently skips a user function that shadows the same
8918/// name (e.g. `echo() { ... }; echo hi` would run the C builtin
8919/// without this check). Returns Some(status) when the call is routed
8920/// to the user function; the builtin handler should fall through to
8921/// its native impl when None.
8922/// Fork+exec a system binary by name. Used by `reg_overridable!` as
8923/// the fall-through path when `[builtins].coreutils_shadows = off`
8924/// (the default) — runs the canonical `/bin/X` instead of zshrs's
8925/// in-process shadow so old scripts hit zero behavioral divergence.
8926///
8927/// Inherits stdin/stdout/stderr from the parent so pipelines work
8928/// transparently. Resolves the binary via PATH; mirrors what zsh's
8929/// own external-command dispatch would do. Returns the child's exit
8930/// status (or 127 if PATH lookup fails — the standard "command not
8931/// found" code).
8932fn exec_system_command(name: &str, args: &[String]) -> i32 {
8933 // c:Src/jobs.c — count the fork so `time` reports for an
8934 // overridable coreutils shadow run as an external (`time sleep 0`,
8935 // `time cat …`). This is a distinct spawn path from
8936 // execute_external_bg; without the bump BUILTIN_TIME_SUBLIST saw no
8937 // job and stayed silent. (Builtins that don't reach a spawn never
8938 // hit this fn.)
8939 crate::vm_helper::FORK_EVENTS.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
8940 let status = std::process::Command::new(name)
8941 .args(args)
8942 .stdin(std::process::Stdio::inherit())
8943 .stdout(std::process::Stdio::inherit())
8944 .stderr(std::process::Stdio::inherit())
8945 .status();
8946 match status {
8947 Ok(s) => s.code().unwrap_or(if s.success() { 0 } else { 1 }),
8948 Err(e) => {
8949 eprintln!("zshrs: {}: {}", name, e);
8950 127
8951 }
8952 }
8953}
8954
8955fn try_user_fn_override(name: &str, args: &[String]) -> Option<i32> {
8956 let has_fn = with_executor(|exec| {
8957 exec.functions_compiled.contains_key(name) || exec.function_exists(name)
8958 });
8959 if !has_fn {
8960 return None;
8961 }
8962 Some(with_executor(|exec| {
8963 exec.dispatch_function_call(name, args).unwrap_or(127)
8964 }))
8965}
8966
8967/// Builtin ID for `${name}` reads — routes through canonical
8968/// `getsparam` (Src/params.c:3076) via paramtab + env walk so nested
8969/// VMs (function calls) see the same storage.
8970pub const BUILTIN_GET_VAR: u16 = 283;
8971
8972/// Like `BUILTIN_GET_VAR` but forces double-quoted (DQ) semantics on
8973/// the read regardless of the runtime `in_dq_context`. The compiler
8974/// emits this for a QUOTED simple-var read (`"$name"`) — those compile
8975/// to a direct GET_VAR with no EXPAND_TEXT wrapper, so `in_dq_context`
8976/// is 0 and the plain GET_VAR would wrongly word-elide an array's empty
8977/// elements (`a=(1 "" 3); "$a"` must keep the empty → `1 3`, not
8978/// `1 3`). With force_dq the array joins via sepjoin keeping empties and
8979/// a scalar is returned verbatim (no empty-drop, no SH_WORD_SPLIT).
8980pub const BUILTIN_GET_VAR_DQ: u16 = 639;
8981
8982/// Builtin ID for `name=value` assignments — pops [name, value] and
8983/// routes through canonical `setsparam` (Src/params.c:3350).
8984pub const BUILTIN_SET_VAR: u16 = 284;
8985
8986/// Builtin ID that sets the thread-local [`SET_VAR_GLOB_ELIGIBLE`] flag true.
8987/// Emitted by the compiler immediately before a `BUILTIN_SET_VAR` whose scalar
8988/// RHS carried an UNQUOTED glob token, so the runtime knows the RHS is a literal
8989/// glob pattern eligible for GLOB_ASSIGN. Takes no stack args, pushes nothing.
8990pub const BUILTIN_MARK_GLOB_ELIGIBLE: u16 = 640;
8991
8992/// Builtin ID for pipeline execution. Pops N sub-chunk indices from the stack;
8993/// each index points into `vm.chunk.sub_chunks` (compiled stage bodies). Forks
8994/// N children, wires stdin/stdout between them via pipes, runs each stage's
8995/// bytecode on a fresh VM in its child, parent waits for all and pushes the
8996/// last stage's exit status. This is bytecode-native pipeline execution —
8997/// no tree-walker delegation.
8998pub const BUILTIN_RUN_PIPELINE: u16 = 285;
8999
9000/// Builtin ID for `Array → String` joining. Pops one value: if it's an Array,
9001/// joins its string-coerced elements with a single space; otherwise passes
9002/// through. Used after `Op::Glob` to convert the pattern's matched paths into
9003/// the single argv-token form the bytecode word model expects (no per-word
9004/// splitting yet — that's a future phase).
9005pub const BUILTIN_ARRAY_JOIN: u16 = 286;
9006
9007/// Builtin ID for `cmd &` background execution. IDs 287/288/289 are reserved
9008/// for the planned array work in Phase G1 (SET_ARRAY/SET_ASSOC/ARRAY_INDEX),
9009/// so this lands at 290. Pops the sub-chunk index then the job text; forks;
9010/// child detaches (`setsid`), runs the sub-chunk on a fresh VM, exits with
9011/// last_status; parent registers the job in the canonical JOBTAB
9012/// (initjob/addproc/spawnjob per c:Src/exec.c:1700-1758) so `jobs` / `wait
9013/// %N` / `kill %N` / `disown` and the zsh/parameter assocs all see it, then
9014/// returns Status(0) immediately.
9015pub const BUILTIN_RUN_BG: u16 = 290;
9016
9017/// Indexed-array assignment: `arr=(a b c)`. Compile_simple emits N element
9018/// pushes followed by name push, then `CallBuiltin(BUILTIN_SET_ARRAY, N+1)`.
9019/// The handler pops args (last popped = name in our pushing order) and stores
9020/// `Vec<String>` into `executor.arrays`. Tree-walker callers see the same
9021/// storage. Any prior scalar binding in `executor.variables` for `name` is
9022/// removed so `${name}` (scalar context) consistently reflects the array's
9023/// first element via `get_variable`.
9024pub const BUILTIN_SET_ARRAY: u16 = 287;
9025
9026/// Single-key set on an associative array: `foo[key]=val`. Stack (top-down):
9027/// [name, key, value]. Stores `value` into `executor.assoc_arrays[name][key]`,
9028/// creating the outer entry if missing. compile_simple detects `var[...]=...`
9029/// in assignments and emits this builtin.
9030pub const BUILTIN_SET_ASSOC: u16 = 288;
9031
9032/// `${arr[idx]}` — single-element array index. Pops two args:
9033/// stack: [name, idx_str]
9034/// Returns the indexed element as Value::str. Indexing semantics: zsh is
9035/// 1-based by default; bash is 0-based. We follow zsh.
9036/// Special idx values: `@` and `*` return the whole array as Value::Array
9037/// (which fuses correctly via the Op::Exec splice for argv splice).
9038pub const BUILTIN_ARRAY_INDEX: u16 = 289;
9039
9040/// `${#arr[@]}` and `${#arr}` (when arr is an array name) — array length.
9041/// Pops one arg: name. Returns Value::str of len.
9042
9043/// `${arr[@]}` — splice all elements as a Value::Array. Pops one arg: name.
9044/// The Array gets flattened by Op::Exec/ExecBg/CallFunction into argv.
9045pub const BUILTIN_ARRAY_ALL: u16 = 292;
9046
9047/// Flatten one level of Value::Array nesting. Pops N values; for each, if it's
9048/// a Value::Array, its elements are appended directly; otherwise the value is
9049/// appended as-is. Pushes a single Value::Array of the flattened result. Used
9050/// by the for-loop word-list compile path: when a word like `${arr[@]}`
9051/// produces a nested Array, this lets `for i in ${arr[@]}` iterate over the
9052/// inner elements rather than the outer single-element array.
9053pub const BUILTIN_ARRAY_FLATTEN: u16 = 293;
9054
9055/// `coproc [name] { body }` — bidirectional pipe to async child. Pops a name
9056/// (optional, "" for default) and a sub-chunk index. Creates two pipes, forks,
9057/// child redirects its fd 0/1 to the inner ends and runs the body, parent
9058/// stores [write_fd, read_fd] into the named array (default `COPROC`). Caller
9059/// closes the fds and `wait`s when done. Job-table integration deferred to
9060/// Phase G6 alongside the bg `&` work.
9061pub const BUILTIN_RUN_COPROC: u16 = 294;
9062
9063/// `arr+=(d e f)` — append N elements to an existing indexed array. Compile
9064/// emits N element pushes + name push, then `CallBuiltin(295, N+1)`. Handler
9065/// drains args (last popped = name), extends `executor.arrays[name]` (creates
9066/// the entry if missing). Mirrors zsh's `+=` semantics for indexed arrays.
9067pub const BUILTIN_APPEND_ARRAY: u16 = 295;
9068
9069/// `name[@]=(...)` / `name[*]=(...)` whole-array SET. Identical to
9070/// BUILTIN_SET_ARRAY for an indexed array / scalar (whole replace), but
9071/// rejects an associative target with "attempt to set slice of
9072/// associative array" (c:Src/params.c:3324-3327).
9073pub const BUILTIN_SET_ARRAY_AT: u16 = 633;
9074
9075/// `name[@]+=(...)` / `name[*]+=(...)` whole-array APPEND. Indexed
9076/// append (push), assoc target → same slice-of-assoc error as 633.
9077pub const BUILTIN_APPEND_ARRAY_AT: u16 = 634;
9078
9079/// `select var in words; do body; done` — interactive numbered-menu loop.
9080/// Compile emits N word pushes + var-name push + sub-chunk index push, then
9081/// `CallBuiltin(296, N+2)`. Handler prints `1) word1\n2) word2\n...` to
9082/// stderr, prints `$PROMPT3` (default `?# `) to stderr, reads a line from
9083/// stdin. On EOF returns 0. On a valid 1-based number, sets `var` to the
9084/// chosen word, runs the sub-chunk, then redisplays the menu and loops. On
9085/// invalid input redraws the menu without running the body. `break` from
9086/// inside the body exits the loop (handled by the body's own bytecode).
9087pub const BUILTIN_RUN_SELECT: u16 = 296;
9088
9089/// `m[k]+=value` — append onto an existing assoc-array value (string concat).
9090/// If the key doesn't exist, behaves like SET_ASSOC. Stack: [name, key, value].
9091
9092/// `break` from inside a body that runs on a sub-VM (select, future
9093/// loop-via-builtin constructs). Writes the canonical
9094/// `crate::ported::builtin::BREAKS` atomic (port of `Src/loop.c:46
9095/// breaks`). Outer-loop builtins drain BREAKS/CONTFLAG after each
9096/// body run, matching the loop.c:529-534 drain pattern.
9097pub const BUILTIN_SET_BREAK: u16 = 299;
9098
9099/// `continue` from inside a sub-VM body. Sets CONTFLAG=1 + bumps
9100/// BREAKS, matching `bin_break`'s WC_CONTINUE arm at Src/builtin.c
9101/// c:5836 `contflag = 1; FALLTHROUGH; breaks++;`.
9102pub const BUILTIN_SET_CONTINUE: u16 = 300;
9103
9104/// Brace expansion: `{a,b,c}` → 3 values, `{1..5}` → 5 values, `{01..05}` →
9105/// zero-padded numerics, `{a..e}` → letter range. Pops one string, returns
9106/// Value::Array of expansions (empty array → original string preserved).
9107pub const BUILTIN_BRACE_EXPAND: u16 = 301;
9108
9109/// Glob qualifier filter: `*(qualifier)` filters glob results by predicate.
9110/// Pops [pattern, qualifier_string]. Returns Value::Array of matching paths.
9111
9112/// Re-export the regex_match host method as a builtin so `[[ s =~ pat ]]`
9113/// works even when fusevm's Op::RegexMatch isn't routed (compat fallback).
9114
9115/// Word-split a string on IFS (default: whitespace). Pops one string,
9116/// returns Value::Array of fields. Used in array-literal context where
9117/// `arr=($(cmd))` should expand cmd's stdout into multiple elements.
9118pub const BUILTIN_WORD_SPLIT: u16 = 304;
9119
9120/// Register a pre-compiled fusevm chunk as a function. Stack: [name,
9121/// base64-bincode-of-Chunk]. Used by compile_zsh's compile_funcdef to
9122/// register functions parsed via parse_init+parse without going through the
9123/// ShellCommand JSON serialization path.
9124pub const BUILTIN_REGISTER_COMPILED_FN: u16 = 305;
9125/// `BUILTIN_VAR_EXISTS` constant.
9126pub const BUILTIN_VAR_EXISTS: u16 = 306;
9127/// Native param-modifier builtins. Each takes a fixed argv shape and
9128/// returns the modified value as Value::Str.
9129///
9130/// `${var:-default}` / `${var:=default}` / `${var:?error}` / `${var:+alt}`
9131/// — pop [name, op_byte, rhs]. op_byte: 0=`:-`, 1=`:=`, 2=`:?`, 3=`:+`.
9132pub const BUILTIN_PARAM_DEFAULT_FAMILY: u16 = 307;
9133/// `${var:offset[:length]}` — pop [name, offset, length] (length=-1 means
9134/// "rest of value"; negative offset counts from end).
9135pub const BUILTIN_PARAM_SUBSTRING: u16 = 308;
9136/// `${var#pat}` / `${var##pat}` / `${var%pat}` / `${var%%pat}` — pop
9137/// [name, pattern, op_byte]. op_byte: 0=`#`, 1=`##`, 2=`%`, 3=`%%`.
9138pub const BUILTIN_PARAM_STRIP: u16 = 309;
9139/// `${var/pat/repl}` / `${var//pat/repl}` / `${var/#pat/repl}` /
9140/// `${var/%pat/repl}` — pop [name, pattern, replacement, op_byte].
9141/// op_byte: 0=first, 1=all, 2=anchor-prefix, 3=anchor-suffix.
9142pub const BUILTIN_PARAM_REPLACE: u16 = 310;
9143/// `${#name}` — character length of a scalar value, or element count
9144/// of an indexed/assoc array. Pops \[name\], returns count as Value::Str.
9145pub const BUILTIN_PARAM_LENGTH: u16 = 311;
9146/// `$((expr))` arithmetic substitution. Pops \[expr_string\], evaluates
9147/// via the executor's MathEval (integer-aware), returns result as
9148/// Value::Str. Bypasses ArithCompiler's float-only Op::Div path so
9149/// `$((10/3))` returns "3" not "3.333...".
9150pub const BUILTIN_ARITH_EVAL: u16 = 312;
9151/// `(( ... ))` math command post-eval status hook. Pops nothing,
9152/// pushes Value::Status. If errflag is set (math error in the
9153/// preceding BUILTIN_ARITH_EVAL call), clears it and emits status=2
9154/// matching c:Src/math.c arith-failure semantics. Otherwise emits
9155/// the current vm.last_status. Used by compile_arith's `(( ... ))`
9156/// path so the math command swallows errors without halting the
9157/// script — `$((... ))` substitutions skip this hook so their
9158/// errflag propagates up to the containing command.
9159pub const BUILTIN_ARITH_CMD_FINISH: u16 = 527;
9160/// `$(cmd)` command substitution. Pops \[cmd_string\], runs through
9161/// `run_command_substitution` which compiles via parse_init+parse + ZshCompiler
9162/// and captures stdout via an in-process pipe. Returns trimmed output
9163/// as Value::Str. Avoids the sub-chunk word-emit quoting bug in the
9164/// raw Op::CmdSubst path.
9165pub const BUILTIN_CMD_SUBST_TEXT: u16 = 313;
9166/// Text-based word expansion. Pops \[preserved_text\]: the word with
9167/// quotes preserved (Dnull→`"`, Snull→`'`, Bnull→`\`), runs
9168/// `expand_string` (variable + cmd-sub + arith) then `xpandbraces`
9169/// then `expand_glob`. Returns Value::str (single match) or
9170/// Value::Array (multi-match brace/glob).
9171pub const BUILTIN_EXPAND_TEXT: u16 = 314;
9172
9173/// `[[ a -ef b ]]` — same-inode test. Stack: [a, b]. Pushes Bool true iff
9174/// both paths resolve to the same `(dev, inode)` pair (zsh + bash semantics).
9175pub const BUILTIN_SAME_FILE: u16 = 315;
9176
9177/// `[[ a -nt b ]]` — file `a` newer than file `b` (mtime strict).
9178/// Stack: [path_a, path_b]. Pushes Bool. zsh-compatible "missing"
9179/// rules: if both exist, compare mtime; if only `a` exists → true;
9180/// otherwise false.
9181pub const BUILTIN_FILE_NEWER: u16 = 324;
9182
9183/// `[[ a -ot b ]]` — mirror of `-nt`. If both exist, compare mtime;
9184/// if only `b` exists → true; otherwise false.
9185pub const BUILTIN_FILE_OLDER: u16 = 325;
9186
9187/// `[[ -k path ]]` — sticky bit (S_ISVTX) set on path.
9188pub const BUILTIN_HAS_STICKY: u16 = 326;
9189/// `[[ -u path ]]` — setuid bit (S_ISUID).
9190pub const BUILTIN_HAS_SETUID: u16 = 327;
9191/// `[[ -g path ]]` — setgid bit (S_ISGID).
9192pub const BUILTIN_HAS_SETGID: u16 = 328;
9193/// `[[ -O path ]]` — owned by effective UID.
9194pub const BUILTIN_OWNED_BY_USER: u16 = 329;
9195/// `[[ -G path ]]` — owned by effective GID.
9196pub const BUILTIN_OWNED_BY_GROUP: u16 = 330;
9197/// `[[ -N path ]]` — file modified since last accessed (atime <= mtime).
9198pub const BUILTIN_FILE_MODIFIED_SINCE_ACCESS: u16 = 341;
9199
9200/// `name+=val` (no parens) — runtime-dispatched append.
9201/// If name is an indexed array → push val as element.
9202/// If name is an assoc array → error (zsh requires `(k v)` form).
9203/// Else → scalar concat (existing SET_VAR behavior).
9204pub const BUILTIN_APPEND_SCALAR_OR_PUSH: u16 = 331;
9205
9206/// `[[ -c path ]]` — character device.
9207pub const BUILTIN_IS_CHARDEV: u16 = 332;
9208/// `[[ -b path ]]` — block device.
9209pub const BUILTIN_IS_BLOCKDEV: u16 = 333;
9210/// `[[ -p path ]]` — FIFO / named pipe.
9211pub const BUILTIN_IS_FIFO: u16 = 334;
9212/// `[[ -S path ]]` — socket.
9213pub const BUILTIN_IS_SOCKET: u16 = 335;
9214/// `BUILTIN_ERREXIT_CHECK` constant.
9215pub const BUILTIN_ERREXIT_CHECK: u16 = 336;
9216/// Fatal-error-only abort check, for use INSIDE an `&&` / `||` chain.
9217///
9218/// A chain suppresses the errexit check (a non-zero status is "consumed"
9219/// by the connector — `false && x` must not fire ERREXIT or the ZERR
9220/// trap). But an errflag — a *fatal* error such as a `[[ ]]` bad pattern
9221/// — is not a status the connector can consume: zsh abandons the whole
9222/// list. Without this, zshrs ran the `||` right-hand side after the
9223/// error (`[[ x = [a- ]] || touch f` created `f`; zsh does not) and the
9224/// aborted builtin then overwrote the cond's status 2 with 1.
9225///
9226/// Same errflag arm as `BUILTIN_ERREXIT_CHECK`, with the errexit/ZERR
9227/// half omitted.
9228pub const BUILTIN_FATAL_ABORT_CHECK: u16 = 641;
9229/// Post-`always`-arm checks for the canonical RETFLAG / BREAKS /
9230/// CONTFLAG atomics that mark try-block escapes. Each returns
9231/// Value::Int(1) when the corresponding atomic is set (and consumes
9232/// it so the next escape doesn't re-fire) and Value::Int(0) otherwise.
9233/// Paired with JumpIfFalse + Jump to outer return_patches /
9234/// break_patches / continue_patches by compile_zsh's `Try` arm.
9235pub const BUILTIN_RETFLAG_CHECK: u16 = 600;
9236/// `BUILTIN_BREAKS_CHECK` constant.
9237pub const BUILTIN_BREAKS_CHECK: u16 = 601;
9238/// `BUILTIN_CONTFLAG_CHECK` constant.
9239pub const BUILTIN_CONTFLAG_CHECK: u16 = 602;
9240/// Fire the DEBUG trap (SIGDEBUG) before each statement.
9241/// c:Src/exec.c:1357-1500 DEBUGBEFORECMD — when a "DEBUG" entry is
9242/// installed via `trap '...' DEBUG`, run the body just before the
9243/// next command. Cheap when no DEBUG trap is set (one hashmap lookup
9244/// returns None and we early-out).
9245pub const BUILTIN_DEBUG_TRAP: u16 = 603;
9246/// `set -n` / `set -o noexec` — parse but don't execute. Returns
9247/// Value::Int(1) when the noexec option is set so the caller's
9248/// JumpIfTrue skips the statement body. c:Src/exec.c:1390 main loop
9249/// check.
9250pub const BUILTIN_NOEXEC_CHECK: u16 = 604;
9251/// Block-level redirect-failure gate. Reads exec.redirect_failed
9252/// (set by host.redirect when a redirect open fails); returns
9253/// Value::Int(1) AND clears the flag if set, else 0. Emit-side at
9254/// compile_zsh.rs::compile_command's Redirected arm pairs with a
9255/// JumpIfTrue → WithRedirectsEnd to abandon the body. Without this,
9256/// a multi-statement block after a failed redir kept running every
9257/// statement after the first (the first builtin consumed the flag,
9258/// subsequent statements ran unimpeded).
9259pub const BUILTIN_REDIRECT_FAILED_CHECK: u16 = 605;
9260/// Drop-in replacement for fusevm's Op::Exec for the dynamic-first-
9261/// word path (`$cmd`, `$(cmd)`, `~/bin/foo`). Returns
9262/// Value::Status(vm.last_status) when post-expansion argv is empty
9263/// (preserves the inner cmd-subst's exit), Value::Status(126) with
9264/// "permission denied" when `argv[0]` is empty, otherwise routes
9265/// through executor.host_exec_external like Op::Exec did.
9266pub const BUILTIN_EXEC_DYNAMIC: u16 = 606;
9267/// Reset `use_cmdoutval` to 0 at the START of a dynamic command (before
9268/// its words expand), so a command substitution from a PREVIOUS command
9269/// can't leak into this command's null-command status decision
9270/// (c:Src/exec.c:3009 `use_cmdoutval = !args`). See BUILTIN_EXEC_DYNAMIC.
9271pub const BUILTIN_USE_CMDOUTVAL_RESET: u16 = 637;
9272/// `[[ lhs == pat ]]` / `!=` glob compare — cond-specific so the
9273/// bad-pattern diagnostic follows Src/cond.c:308-316: zwarnnam
9274/// "bad pattern: %s" WITHOUT errflag (the script continues) and the
9275/// cond statement exits 2. Stack: [lhs, pat] → Bool. On compile
9276/// failure pushes Bool(false) and arms COND_BAD_PATTERN so
9277/// BUILTIN_COND_STATUS_FROM_BOOL reports 2.
9278pub const BUILTIN_COND_STRMATCH: u16 = 624;
9279/// Pops the cond result Bool → Int status per Src/cond.c: true→0,
9280/// false→1, but 2 when COND_BAD_PATTERN was armed during this cond
9281/// (covers `!=` where LogNot flips the Bool before status time).
9282pub const BUILTIN_COND_STATUS_FROM_BOOL: u16 = 625;
9283/// `[[ ]]` unknown condition. Pops \[op_name\], emits `zerr("unknown
9284/// condition: %s")` and sets ERRFLAG_ERROR so the next BUILTIN_ERREXIT_CHECK
9285/// (trigger 4) aborts the input — matching zsh's COND_MODI "unknown condition"
9286/// path (Src/cond.c:150-188) for a `-X` op with no matching loadable module.
9287/// Returns Bool(false). Replaces a compile-time `eprintln!` hack that printed
9288/// the message but never set errflag (so the line didn't abort).
9289pub const BUILTIN_COND_UNKNOWN: u16 = 632;
9290/// Bare-`exec` redirect epilogue. Consumes `exec.redirect_failed` and
9291/// applies the C `done:` tail of execcmd_exec:
9292/// - c:Src/exec.c:252-259 execerr — `redir_err = lastval = 1` (the
9293/// failed redirect makes the exec statement exit 1, NOT fatal by
9294/// itself);
9295/// - c:Src/exec.c:4367-4386 — `if (isset(POSIXBUILTINS) && (cflags
9296/// & (BINF_PSPECIAL|BINF_EXEC)) ...) { if (redir_err || errflag)
9297/// { if (!isset(INTERACTIVE)) exit(1); } }` — POSIX_BUILTINS makes
9298/// a failed exec redirect fatal in a non-interactive shell.
9299/// Returns Value::Status(0|1) for the trailing SetStatus.
9300pub const BUILTIN_EXEC_REDIR_DONE: u16 = 626;
9301/// Assignment-prefix epilogue for bare `exec` redirects
9302/// (`x=$(cmd) exec >file`). c:Src/exec.c:3969-3976 — nullexec==1
9303/// runs addvars THEN, without POSIX_BUILTINS, restores the params
9304/// (`save_params` / `restore_params`): the RHS side effects fire but
9305/// the values don't persist. With POSIX_BUILTINS the assignments
9306/// stick. Pops the BEGIN_INLINE_ENV frame either way.
9307pub const BUILTIN_EXEC_INLINE_ENV_DONE: u16 = 627;
9308
9309/// `< file` / `> file` with no command word (NULLCMD path).
9310/// Resolves NULLCMD (default "cat") / READNULLCMD (default "more")
9311/// at runtime per Src/exec.c:3340-3364 and exec's it through
9312/// host_exec_external. Argc is 1: the int (0 or 1) on the stack
9313/// indicates whether this is a single REDIR_READ redirect
9314/// (selects READNULLCMD when set + non-empty).
9315pub const BUILTIN_NULLCMD_EXEC: u16 = 607;
9316/// `.` (dot) — alias of source/bin_dot but dispatches with the
9317/// literal name "." so the diagnostic prefix matches zsh's
9318/// (`zsh:.:1: …` vs source's `zsh:source:1: …`).
9319/// c:Src/builtin.c:9308 — `BUILTIN(".", BINF_PSPECIAL, bin_dot, …)`.
9320pub const BUILTIN_DOT: u16 = 608;
9321/// `logout` — fusevm maps this to BUILTIN_EXIT alongside `exit`/`bye`,
9322/// which drops the name and dispatches with BIN_EXIT funcid. zsh's
9323/// `logout` outside a login shell must emit "not login shell" + exit 1,
9324/// which only fires when bin_break sees BIN_LOGOUT funcid. Dedicated
9325/// opcode dispatches via BUILTINS table by literal name "logout".
9326pub const BUILTIN_LOGOUT: u16 = 610;
9327/// `BUILTIN_PARAM_SUBSTRING_EXPR` constant.
9328pub const BUILTIN_PARAM_SUBSTRING_EXPR: u16 = 337;
9329/// `BUILTIN_XTRACE_LINE` constant.
9330pub const BUILTIN_XTRACE_LINE: u16 = 338;
9331/// `BUILTIN_ARRAY_JOIN_STAR` constant.
9332pub const BUILTIN_ARRAY_JOIN_STAR: u16 = 339;
9333/// `BUILTIN_SET_RAW_OPT` constant.
9334pub const BUILTIN_SET_RAW_OPT: u16 = 340;
9335
9336/// `time { compound; ... }` — wall-clock-time the sub-chunk and print
9337/// elapsed seconds. Stack: [sub_chunk_idx as Int]. Runs the sub-chunk
9338/// on the current VM (so positional/local state is shared) and prints
9339/// the timing summary to stderr in zsh's format. Pushes Status.
9340pub const BUILTIN_TIME_SUBLIST: u16 = 316;
9341
9342/// `{name}>file` / `{name}<file` / `{name}>>file` — named-fd allocation.
9343/// Stack: [path, varid, op_byte]. Opens `path` per `op_byte`, gets the
9344/// new fd (≥10 in zsh; we use libc::open with O_CLOEXEC bit cleared so
9345/// the inherited fd survives Command::new spawns), stores the fd number
9346/// as a string in `$varid`. Pushes Status (0 success, 1 error).
9347pub const BUILTIN_OPEN_NAMED_FD: u16 = 317;
9348
9349/// Word-segment concat that does cartesian-product distribution over
9350/// arrays. Stack: [lhs, rhs]. Used for RC_EXPAND_PARAM `${arr}` and
9351/// explicit-distribute forms (`${^arr}`, `${(@)…}`).
9352///
9353/// - both scalar: `Value::str(a + b)` (fast path, identical to Op::Concat)
9354/// - lhs Array, rhs scalar: `Value::Array([a + rhs for a in lhs])`
9355/// - lhs scalar, rhs Array: `Value::Array([lhs + b for b in rhs])`
9356/// - both Array: cartesian product `[a + b for a in lhs for b in rhs]`
9357pub const BUILTIN_CONCAT_DISTRIBUTE: u16 = 318;
9358
9359/// Forced-distribute concat — like `BUILTIN_CONCAT_DISTRIBUTE` but
9360/// always distributes cartesian regardless of the `rcexpandparam`
9361/// option. Emitted by the segments fast-path when an
9362/// `is_distribute_expansion` segment is present (`${^arr}`,
9363/// `${(@)arr}`, `${(s.…)arr}` etc.) per zsh: the source-level
9364/// distribution flag overrides the option default.
9365/// Direct port of Src/subst.c:1875 `case Hat: nojoin = 1` and the
9366/// `rcexpandparam` test bypass for the explicit-distribute flags.
9367pub const BUILTIN_CONCAT_DISTRIBUTE_FORCED: u16 = 522;
9368
9369/// Capture current `last_status` into the `TRY_BLOCK_ERROR` variable.
9370/// Emitted between the try block and the always block of `{ … } always
9371/// { … }` so the finally arm can read $TRY_BLOCK_ERROR.
9372pub const BUILTIN_SET_TRY_BLOCK_ERROR: u16 = 320;
9373/// `BUILTIN_RESTORE_TRY_BLOCK_STATUS` constant.
9374pub const BUILTIN_RESTORE_TRY_BLOCK_STATUS: u16 = 432;
9375/// `BUILTIN_BEGIN_INLINE_ENV` constant.
9376pub const BUILTIN_BEGIN_INLINE_ENV: u16 = 433;
9377/// `BUILTIN_END_INLINE_ENV` constant.
9378pub const BUILTIN_END_INLINE_ENV: u16 = 434;
9379
9380/// `[[ -o option ]]` — shell-option-set test. Stack: \[option_name\].
9381/// Normalizes the name (strip underscores, lowercase) and reads
9382/// `exec.options`. Pushes Bool.
9383pub const BUILTIN_OPTION_SET: u16 = 321;
9384/// Tri-state `[[ -o NAME ]]` — same lookup as BUILTIN_OPTION_SET
9385/// but returns a Value::Int (0=set, 1=unset, 3=invalid-name). The
9386/// 3-state code matches zsh's `[[ -o invalid ]]` exit (Src/cond.c
9387/// :502 `optison()`). Used by compile_cond's `-o` arm to skip the
9388/// generic bool→status conversion and preserve the invalid-name
9389/// signal in `$?`.
9390pub const BUILTIN_OPTION_CHECK_TRISTATE: u16 = 609;
9391
9392/// `${var:#pattern}` — array filter: remove elements matching `pattern`.
9393/// Stack: [name, pattern]. For scalar `var`, returns empty if it matches
9394/// the pattern, else the value. For array `var`, returns Array of
9395/// non-matching elements.
9396pub const BUILTIN_PARAM_FILTER: u16 = 322;
9397
9398/// `a[i]=(elements)` / `a[i,j]=(elements)` / `a[i]=()` —
9399/// subscripted-array assign with array-literal RHS. Stack:
9400/// [...elements, name, key]. Empty elements + single-int key `a[i]=()`
9401/// removes that element. Comma-key `a[i,j]=(...)` splices.
9402pub const BUILTIN_SET_SUBSCRIPT_RANGE: u16 = 323;
9403
9404/// `[[ -X file ]]` for unknown unary test op `-X`. Stack: \[op_name\].
9405/// Emits zsh's `unknown condition: -X` diagnostic to stderr and
9406/// pushes Bool(false). Without this, unknown conditions silently
9407/// returned false matching neither zsh's error format nor the
9408/// expected status code (zsh returns 2 for parse error).
9409
9410/// `[[ -t fd ]]` — fd-is-a-tty check. Stack: \[fd_string\].
9411/// Routes through libc::isatty. Pushes Bool.
9412pub const BUILTIN_IS_TTY: u16 = 325;
9413/// `[[ -r/-w/-x file ]]` via access(2) (doaccess) — see handler.
9414pub const BUILTIN_COND_ACCESS: u16 = 638;
9415
9416/// Update `$LINENO` to track the source line of the next statement.
9417/// Stack: \[n\] (the line number from `ZshPipe.lineno`). Direct port
9418/// of zsh's `lineno` global tracking (Src/input.c:330) — the
9419/// compiler emits one of these per top-level pipe so `$LINENO`
9420/// reflects the source position at runtime. ID 342 picked because
9421/// the previous `326` collided with `BUILTIN_HAS_STICKY` (the file
9422/// has several other duplicate IDs — 325 has two as well — but
9423/// fixing those is out of scope for this port).
9424pub const BUILTIN_SET_LINENO: u16 = 342;
9425
9426/// Pop a scalar from the VM stack, run expand_glob on it, push the
9427/// result as Value::Array. Used by the segment-concat compile path
9428/// when var refs concatenate with glob meta literals (`$D/*`,
9429/// `${prefix}*`, etc.) — those skip the bridge's pathname-expansion
9430/// pass and would otherwise leak the glob meta to argv as a literal.
9431pub const BUILTIN_GLOB_EXPAND: u16 = 343;
9432
9433/// MULTIOS-gated glob expansion for redirect-target words
9434/// (c:Src/glob.c:2161-2167 xpandredir: "Globbing is only done for
9435/// multios."). Same stack shape as BUILTIN_GLOB_EXPAND; additionally
9436/// passes the word through literally when `unsetopt multios`.
9437// 624 is BUILTIN_COND_STRMATCH — the VM's builtin table is
9438// last-registration-wins, so a duplicate id silently shadows the
9439// earlier handler.
9440pub const BUILTIN_REDIR_GLOB_EXPAND: u16 = 628;
9441
9442/// Reset the default-word glob-pending carrier at the START of a word
9443/// whose source contains a glob metachar (so the flag never leaks from a
9444/// prior word/statement). Paired with BUILTIN_DEFAULT_WORD_GLOB.
9445pub const BUILTIN_DEFAULT_WORD_GLOB_RESET: u16 = 635;
9446
9447/// Filename-generate the ASSEMBLED word ONLY when the default/alternate
9448/// paramsubst arm took a SOURCE word carrying glob metachars
9449/// (subst::DEFAULT_WORD_GLOB_PENDING). A parameter VALUE never sets the
9450/// flag, so `x='*file'; ${x:-d}` stays literal while `${x:-*file}` /
9451/// `${x:-a*}bar` glob. Reads+clears the flag. c:Src/subst.c → globlist.
9452pub const BUILTIN_DEFAULT_WORD_GLOB: u16 = 636;
9453/// `BUILTIN_SET_LOOP_VAR` constant — for-loop variable binding via
9454/// `setloopvar` (Src/params.c:6362): a PM_NAMEREF loop var REBINDS
9455/// to each word (SETREFNAME + setscope) instead of assigning
9456/// through the chain. Returns Bool(false) when zerr fired
9457/// (read-only reference / invalid self reference) so the loop
9458/// driver aborts, mirroring C execfor's errflag check.
9459pub const BUILTIN_SET_LOOP_VAR: u16 = 629;
9460
9461/// EXTEND step of typeset paren-init packing. Pops `argc` values:
9462/// [base, e1, …, eN] — base is either the opener (`name=(` /
9463/// `name+=(`) or a previous EXTEND result. Pushes base with
9464/// `\u{1f}` + element appended per element. Array values SPLICE
9465/// their items as separate elements (`typeset b=( x $arr )` splat);
9466/// an empty Array contributes nothing (unquoted-empty elision).
9467/// CallBuiltin's argc is u8, so the compiler emits one EXTEND per
9468/// ≤200-element chunk — p10k's 408-element `__p9k_colors=( … )`
9469/// overflowed a single-shot pack (argc wrapped mod 256 and the
9470/// stack spilled into the arg list: "not an identifier: 173…").
9471/// BUILTIN_TYPESET_PAREN_CLOSE appends the final `\u{1f})`,
9472/// yielding the exact REJOIN_SEP-delimited one-arg form
9473/// bin_typeset's single-arg splitter consumes (builtin.rs ~4891,
9474/// empties preserved, leading/trailing sentinel-empties trimmed
9475/// once). One arg in → one arg out: bin_typeset's multi-arg rejoin
9476/// (paren-depth scan, unsafe on EXPANDED paren-literal elements
9477/// like p10k's `')' ''`) never runs.
9478pub const BUILTIN_TYPESET_PAREN_PACK: u16 = 630;
9479
9480/// CLOSE step — pops the EXTEND chain's result, pushes it with
9481/// `\u{1f})` appended. See BUILTIN_TYPESET_PAREN_PACK.
9482pub const BUILTIN_TYPESET_PAREN_CLOSE: u16 = 631;
9483
9484/// Shared body of BUILTIN_GLOB_EXPAND / BUILTIN_REDIR_GLOB_EXPAND.
9485/// c:Src/glob.c:1872 — `zglob` runs per-word in the argv pipeline.
9486/// When the upstream EXPAND_TEXT returned an array (e.g. `${a:e}`
9487/// splat → ["txt","md"]), glob each element separately, not a
9488/// sepjoin'd scalar. `skip_glob` short-circuits to a literal
9489/// pass-through (noglob, or a redirect target under
9490/// `unsetopt multios`).
9491fn glob_expand_word_value(raw: Value, skip_glob: bool) -> Value {
9492 let patterns: Vec<String> = match raw {
9493 Value::Array(items) => items.into_iter().map(|v| v.to_str()).collect(),
9494 other => vec![other.to_str()],
9495 };
9496 if skip_glob {
9497 return if patterns.is_empty() {
9498 Value::Array(Vec::new())
9499 } else if patterns.len() == 1 {
9500 Value::str(patterns.into_iter().next().unwrap())
9501 } else {
9502 Value::Array(patterns.into_iter().map(Value::str).collect())
9503 };
9504 }
9505 let mut out: Vec<String> = Vec::with_capacity(patterns.len());
9506 for pattern in &patterns {
9507 // c:Src/subst.c — filename generation runs `filesub` (tilde/`=`
9508 // expansion) BEFORE globbing. A `~`/`=` reaching this word-glob op
9509 // comes from `${~spec}` / GLOB_SUBST marking a substituted VALUE:
9510 // literal and quoted `~` words are filesub'd (or skip glob) upstream
9511 // and never arrive here. filesubstr matches the Tilde TOKEN, so
9512 // shtokenize first (raw `~`->Tilde; already-Tilde `${~a[@]}` results
9513 // pass through), run filesub, then untokenize surviving glob metas
9514 // for expand_glob. Gated on `~`/`=` (raw or token) so ordinary
9515 // substituted words skip the roundtrip. Fixes `${~x}` x="~/foo".
9516 let filesubbed = if pattern.contains('~')
9517 || pattern.contains('=')
9518 || pattern.contains(crate::ported::zsh_h::Tilde)
9519 || pattern.contains(crate::ported::zsh_h::Equals)
9520 {
9521 let mut tok = pattern.clone();
9522 crate::ported::glob::shtokenize(&mut tok);
9523 crate::ported::lex::untokenize(&crate::ported::subst::filesub(&tok, 0))
9524 } else {
9525 pattern.clone()
9526 };
9527 let matches = with_executor(|exec| exec.expand_glob(&filesubbed));
9528 if matches.is_empty() {
9529 // c:1872 nullglob — drop this word, don't emit a hole
9530 continue;
9531 }
9532 for m in matches {
9533 out.push(m);
9534 }
9535 }
9536 if out.is_empty() {
9537 return Value::Array(Vec::new());
9538 }
9539 if patterns.len() == 1 && out.len() == 1 && out[0] == patterns[0] {
9540 // No real matches; expand_glob returned the literal. Pass
9541 // back as scalar so downstream ops don't re-flatten.
9542 return Value::str(out.into_iter().next().unwrap());
9543 }
9544 Value::Array(out.into_iter().map(Value::str).collect())
9545}
9546
9547/// Push a `CmdState` token onto the command-context stack. Direct
9548/// port of zsh's `cmdpush(int cmdtok)` (Src/prompt.c:1623). The
9549/// stack is consulted by `%_` in PS4/prompt expansion to produce
9550/// the cumulative control-flow-context labels (`if`, `then`,
9551/// `cmdand`, `cmdor`, `cmdsubst`, …) that `zsh -x` xtrace shows
9552/// in the trace prefix. Compile_zsh emits push/pop pairs around
9553/// each compound command (if/while/[[…]]/((…))/$(…) etc.).
9554/// Token is a `CmdState as u8`.
9555pub const BUILTIN_CMD_PUSH: u16 = 344;
9556
9557/// Pop the top of the command-context stack. Direct port of zsh's
9558/// `cmdpop(void)` (Src/prompt.c:1631).
9559pub const BUILTIN_CMD_POP: u16 = 345;
9560
9561/// Emit an xtrace line built from the top `argc` values on the VM
9562/// stack, peeked WITHOUT consuming. Used to trace simple commands
9563/// AFTER expansion, so `echo for $i` shows as `echo for a` / `echo
9564/// for b`. Direct port of Src/exec.c:2055-2066.
9565pub const BUILTIN_XTRACE_ARGS: u16 = 346;
9566
9567/// Trace one assignment: emits `name=<quoted-value> ` (no newline)
9568/// to xtrerr if XTRACE is on. Coalesces with subsequent
9569/// XTRACE_ASSIGN / XTRACE_ARGS calls onto the SAME line via the
9570/// `XTRACE_DONE_PS4` flag so `a=1 b=2 echo $a $b` produces:
9571/// `<PS4>a=1 b=2 echo 1 2\n`
9572/// matching C zsh's `execcmd_exec` body (Src/exec.c:2517-2582):
9573/// xtr = isset(XTRACE);
9574/// if (xtr) { printprompt4(); doneps4 = 1; }
9575/// while (assign) {
9576/// if (xtr) fprintf(xtrerr, "%s=", name);
9577/// ... eval value ...
9578/// if (xtr) { quotedzputs(val, xtrerr); fputc(' ', xtrerr); }
9579/// }
9580///
9581/// Stack contract on entry: [..., name, value]. Both peeked, NOT
9582/// consumed (the matching SET_VAR call pops them after). argc = 2.
9583pub const BUILTIN_XTRACE_ASSIGN: u16 = 525;
9584
9585/// Emit a trailing `\n` + flush iff XTRACE is on AND PS4 was
9586/// emitted by an earlier XTRACE_ASSIGN this line. Used at the end
9587/// of compile_simple's assignment-only path so the trace line gets
9588/// terminated. Mirrors C's exec.c:3397-3399 (the assign-only return
9589/// path through execcmd_exec which does `fputc('\n', xtrerr);
9590/// fflush(xtrerr)`).
9591///
9592/// Stack: untouched. argc = 0.
9593pub const BUILTIN_XTRACE_NEWLINE: u16 = 526;
9594
9595/// Push the live `xtrace` opt-state as `Value::Int(1)` (on) or
9596/// `Value::Int(0)` (off). Used by `compile_cond` to gate the
9597/// trace-string-building block on xtrace state at runtime — without
9598/// this the trace path's `compile_word_str` on each operand re-
9599/// evaluates side-effectful expressions (`$((i++))`) once for the
9600/// trace string and once for the actual condition, doubling the
9601/// effective increment. Bug #159 in docs/BUGS.md.
9602///
9603/// Stack: pushes Int(0|1). argc = 0.
9604pub const BUILTIN_XTRACE_IS_ON: u16 = 611;
9605
9606/// Reset the `DONETRAP` flag at the start of each top-level statement
9607/// (sublist boundary). Mirrors C `Src/exec.c:1455` — `donetrap = 0`.
9608/// Stack: untouched. argc = 0. Bug #303 in docs/BUGS.md.
9609pub const BUILTIN_DONETRAP_RESET: u16 = 612;
9610
9611/// `[[ -z X ]]` / `[[ -n X ]]` operand-empty test that honours zsh's
9612/// array-splice semantics. C zsh evaluates `[[ -z X ]]` per
9613/// `Src/cond.c:347` (case 'z'): `s` is the SCALAR operand passed
9614/// through `cond_str`'s singsub. For `"${arr[@]}"` zsh expands per
9615/// `Src/subst.c:multsub` which yields each element as its own word
9616/// list node; cond.c then sees the joined-or-single-element form.
9617///
9618/// The compile-side `-z` shortcut at `compile_zsh.rs:5371` used
9619/// `Op::StringLen` which calls `Value::len` — for `Value::Array`
9620/// that returns ARRAY LENGTH, not string length. `b=("")` produced
9621/// `Value::Array([""])` → `len = 1` → `-z` returned false.
9622///
9623/// This builtin pops one `Value` and pushes `1` (empty) or `0`
9624/// (non-empty) per the cond context:
9625/// - `Value::Str(s)` → s.is_empty()
9626/// - `Value::Array([])` → true (zero words → vacuous-empty)
9627/// - `Value::Array([s])` → s.is_empty() (single-word case)
9628/// - `Value::Array([_; n>=2])` → false (multiple non-empty
9629/// words; zsh would raise "unknown condition" but the
9630/// observable test result is non-empty/false)
9631///
9632/// Companion to BUILTIN_COND_STR_NONEMPTY (#185 in docs/BUGS.md).
9633pub const BUILTIN_COND_STR_EMPTY: u16 = 613;
9634
9635/// `[[ -n X ]]` operand-non-empty test (logical complement of
9636/// BUILTIN_COND_STR_EMPTY).
9637pub const BUILTIN_COND_STR_NONEMPTY: u16 = 614;
9638
9639/// `exec N<<<"str"` — herestring redirect to explicit fd, applied
9640/// permanently to the shell (no scope restoration). Pops `[content,
9641/// fd]` from the stack; creates a temp file, writes
9642/// `content + "\n"`, reopens read-only, dup2's to `fd`, unlinks the
9643/// temp path so it disappears on close. Mirrors C `Src/exec.c:4655
9644/// getherestr` + `addfd(forked, save, mfds, fn->fd1, fil, 0, ...)`
9645/// at c:3766-3780 for the bare-exec-redir code path (nullexec=1).
9646/// Bug #205 in docs/BUGS.md.
9647///
9648/// Stack: pushes `Value::Status(0)` on success, `Status(1)` on
9649/// failure. argc = 2.
9650pub const BUILTIN_EXEC_HERESTR_FD: u16 = 615;
9651
9652/// MULTIOS write/append fan-out for `cmd > a > b` / `cmd > a >> b`
9653/// style redirects (Bug #36 in docs/BUGS.md). zsh's MULTIOS option
9654/// (Src/exec.c:2418 `mfds[fd1]` check + addfd splice) creates a
9655/// pipe at fd1, spawns an internal "tee" process that copies
9656/// stdin → every collected target file. Without this, only the
9657/// LAST redirect target survives because each dup2 overwrites the
9658/// previous binding.
9659///
9660/// Stack layout (pushed by compile_zsh's compile_redirs coalescing
9661/// pass): `[target_1, op_byte_1, target_2, op_byte_2, …, target_N,
9662/// op_byte_N, fd]`. Pops 2N+1 elements; `argc = 2*N + 1`. A target
9663/// may be a Value::Array of glob matches (spliced into one member
9664/// per match, c:Src/glob.c:2195-2203); an op may be DUP_WRITE for a
9665/// numeric `>&N` member (c:Src/exec.c:3895-3917).
9666///
9667/// Runtime (MULTIOS set):
9668/// 1. Seed the member list with `dup(1)` when this command's
9669/// stdout is the pipeline output (c:Src/exec.c:3722-3724).
9670/// 2. Open/dup all targets per their op_byte in redirect order
9671/// (WRITE truncate + noclobber gate / APPEND / DUP_WRITE live
9672/// dup); the first member replaces the fd (c:2448-2450).
9673/// 3. Save `dup(fd)` onto the active redirect_scope_stack so
9674/// `host_redirect_scope_end` restores the original fd.
9675/// 4. Create a pipe; spawn a thread that reads from the pipe
9676/// read-end and writes every chunk to every opened target.
9677/// 5. dup2 the pipe write-end onto `fd` so the command's writes
9678/// go through the splitter.
9679/// 6. Track `(pipe_write_fd, JoinHandle)` so scope-end can close
9680/// the pipe (draining the thread) and join before restoring.
9681///
9682/// MULTIOS unset (c:2418 `unset(MULTIOS)` replace arm): each entry
9683/// is applied as a plain sequential replace via host_apply_redirect
9684/// — every file still opened/truncated, last one wins.
9685pub const BUILTIN_MULTIOS_REDIRECT: u16 = 617;
9686
9687/// MULTIOS input-side concatenation for `cmd < a < b` shapes
9688/// (Bug #36 input arm). C zsh's `Src/exec.c:2418` mfds dispatch
9689/// also covers the read direction — when multiple `<` redirects
9690/// target the same fd, mfds[fd] grows and addfd splices a
9691/// concatenating cat into the pipe.
9692///
9693/// Stack layout (mirrors the write side): `[source_1, op_1,
9694/// source_2, op_2, …, source_N, op_N, fd]`. Pops 2N + 1 elements
9695/// (argc = 2N + 1). op is READ for file sources, DUP_READ for
9696/// numeric `<&N` members; a source may be a Value::Array of glob
9697/// matches (spliced, c:Src/glob.c:2195-2203).
9698///
9699/// Runtime (MULTIOS set):
9700/// 1. Open/dup every source in redirect order; first member
9701/// replaces the fd (c:Src/exec.c:2448-2450).
9702/// 2. Save `dup(fd)` onto the redirect_scope_stack.
9703/// 3. Create a pipe; spawn a thread that reads each source in
9704/// order and writes every chunk to the pipe write-end. Close
9705/// write-end when done so the consumer sees EOF.
9706/// 4. dup2 the pipe read-end onto `fd`.
9707/// 5. Track the JoinHandle so scope-end joins (no fd-close needed
9708/// here — the producer thread closes its own pipe write-end
9709/// on exit).
9710///
9711/// MULTIOS unset: sequential replace via host_apply_redirect — last
9712/// source wins (c:2418).
9713pub const BUILTIN_MULTIOS_READ: u16 = 618;
9714
9715/// Toggle `ShellExecutor::exec_redirs_permanent`. Emitted by
9716/// compile_zsh's bare-`exec`-with-redirects arm tightly around each
9717/// `Op::Redirect`: `LoadInt(1); CallBuiltin; …Redirect…; LoadInt(0);
9718/// CallBuiltin`. While set, `host_apply_redirect` skips pushing the
9719/// saved fd into the enclosing redirect scope, making the fd change
9720/// permanent.
9721///
9722/// c:Src/exec.c:3978-3986 — nullexec==1 (`exec` carrying only
9723/// redirections): "If nullexec is 1 we specifically *don't* restore
9724/// the original fd's before returning" — the per-execcmd `save[]`
9725/// dups are closed unrestored. An ENCLOSING group's own saves are a
9726/// different execcmd's `save[]` and still restore (verified:
9727/// `{ exec 2>/dev/null; } 2>&1; ls /nope` prints the ls error in zsh).
9728pub const BUILTIN_EXEC_PERM_REDIRS: u16 = 619;
9729
9730/// Set `ShellExecutor::pipe_output_pending`. Emitted by compile_pipe
9731/// at the head of a NON-LAST pipeline-stage sub-chunk when that
9732/// stage's top-level command carries redirects (`Simple` with redirs
9733/// or `Redirected` compound). The forked stage child runs the chunk
9734/// with stdout already dup2'd onto the pipe; the first
9735/// `host_redirect_scope_begin` (the stage command's own redirect
9736/// list) consumes the flag into `pipe_output_scope`, enabling the
9737/// MULTIOS stream-split for fd-1 write redirects in that list.
9738///
9739/// c:Src/exec.c:3722-3724 — `addfd(forked, save, mfds, 1, output, 1,
9740/// NULL)` registers the pipe in mfds[1] in the SAME execcmd that
9741/// walks the stage command's redirect list; mfds is per-execcmd, so
9742/// nested body commands (`{ echo a > f; } | cat`) never see it.
9743pub const BUILTIN_PIPE_OUTPUT_MARK: u16 = 620;
9744
9745/// Magic-equals prefork for a single arg word of a
9746/// `BINF_MAGICEQUALS` builtin head (`alias`). Direct port of
9747/// c:Src/exec.c:3298-3304 — `esprefork = PREFORK_TYPESET;
9748/// prefork(args, esprefork, NULL)` runs on the argv BEFORE the addfd
9749/// redirect loop at c:3720, so an expansion zerr (`alias bad===` →
9750/// equalsubstr "= not found" at Src/subst.c:726) prints to the
9751/// command's UN-redirected stderr. argc=1: pops the just-pushed
9752/// word value, runs shtokenize → prefork(PREFORK_TYPESET) →
9753/// untokenize on it (each element for Array splices), pushes the
9754/// result back. Emitted by compile_simple per arg word when the
9755/// dispatch head is `alias`; BUILTIN_ALIAS itself no longer
9756/// preforks (it would double-fire the diagnostic).
9757pub const BUILTIN_MAGIC_EQUALS_PREFORK: u16 = 621;
9758
9759/// Bare (unbraced) `$name[idx]` subscript — same dispatch as
9760/// `BUILTIN_ARRAY_INDEX` while KSHARRAYS is unset, but under
9761/// KSHARRAYS the unbraced form does NOT subscript (c:Src/subst.c:
9762/// 2800-2802 + 2867): `$name` expands bare and `[idx]` stays literal
9763/// trailing text that undergoes filename generation. Operands:
9764/// [name, idx, suffix, quoted].
9765pub const BUILTIN_ARRAY_INDEX_UNBRACED: u16 = 622;
9766
9767/// Assignment-only simple-command exit status. Direct port of
9768/// `lv = (errflag ? errflag : cmdoutval)` (c:Src/exec.c:1322,
9769/// execsimple's WC_ASSIGN arm) / `if (errflag) lastval = 1; else
9770/// lastval = cmdoutval;` (c:Src/exec.c:3393-3396, execcmd_exec's
9771/// no-command-word varspc path; redir variant at c:3977). Pops
9772/// [had_cmd_subst]; cmdoutval is the live vm.last_status when a
9773/// `$()` ran in any RHS of the chain, 0 otherwise. Writes the
9774/// canonical LASTVAL (C's single `lastval` global) so the
9775/// non-interactive errflag abort exits with this value per
9776/// Src/init.c:234. Caller pairs with SetStatus.
9777pub const BUILTIN_ASSIGN_ONLY_STATUS: u16 = 623;
9778
9779/// c:Src/exec.c addvars — `if (!pm) { lastval = 1; if (!cmdoutval)
9780/// cmdoutval = 1; }`. Set by BUILTIN_SET_VAR on assignsparam
9781/// failure, consumed by BUILTIN_ASSIGN_ONLY_STATUS so the
9782/// assignment-only command reports status 1. Process-global like
9783/// C's `cmdoutval` (function bodies may run on a different thread
9784/// than the opcode that reads the status back).
9785pub static ASSIGN_FAILED_FLAG: std::sync::atomic::AtomicBool =
9786 std::sync::atomic::AtomicBool::new(false);
9787
9788/// `redirection with no command` parse-time error for bare
9789/// `builtin 2>&1` / `command < file` / `exec >&-` precmd-keyword
9790/// shapes with a redirect but no following command. Direct port
9791/// of `Src/exec.c:3342 zerr("redirection with no command")`.
9792/// argc=0; pushes Value::Status(1).
9793pub const BUILTIN_REDIR_NO_CMD: u16 = 616;
9794
9795/// GLOB_SUBST guard for `[[ x == $pat ]]` pattern RHS coming from
9796/// parameter / command substitution. C-zsh's `[[ == ]]` semantics
9797/// (Src/options.c GLOB_SUBST default OFF + Src/cond.c:552
9798/// `cond_match` + Src/pattern.c patcompile tokenization-based
9799/// meta detection) treat chars from substitution as LITERAL
9800/// unless GLOB_SUBST is on. The Rust patcompile accepts both
9801/// tokenized and raw-ASCII meta chars, losing the distinction,
9802/// so `pat="h*"; [[ hello == $pat ]]` matched in zshrs but not
9803/// in zsh. Bug #116 in docs/BUGS.md.
9804///
9805/// Compile-time signal: emitted by `compile_cond_expr` ONLY when
9806/// the RHS contains `$` or backtick. Runtime checks the live
9807/// option state. If GLOB_SUBST is OFF, the popped string has
9808/// its glob metachars escaped with `\` so the downstream StrMatch
9809/// → patcompile treats them as literals. If GLOB_SUBST is ON,
9810/// the value passes through unchanged so `setopt glob_subst`
9811/// restores zsh's pattern-on-expansion behavior.
9812///
9813/// Stack: pops one string, pushes the (possibly escaped) result.
9814/// argc = 1.
9815pub const BUILTIN_GLOB_SUBST_GUARD: u16 = 528;
9816
9817/// Coerce a string parameter value to a math number (Int or Float)
9818/// for arithmetic-context reads, mirroring C-zsh's `getmathparam`
9819/// (Src/math.c:337). When the variable holds a string like "hello"
9820/// that isn't numeric, C falls back to recursively evaluating the
9821/// raw string as an arith expression; if that fails too, returns 0.
9822///
9823/// Used by the ArithCompiler pre-load path so `(( y = x ))` with
9824/// `x="hello"` reads `x` as integer 0, then assigns y as integer 0
9825/// — matching zsh's behaviour. The previous Rust port used
9826/// BUILTIN_GET_VAR which returned the raw string "hello"; the
9827/// ArithCompiler stored it verbatim in y's slot, and the post-sync
9828/// BUILTIN_SET_VAR wrote y="hello" as scalar instead of y=0 as
9829/// integer. Bug #118 in docs/BUGS.md.
9830///
9831/// Stack: pops `name` (string), pushes coerced numeric Value.
9832/// argc = 1.
9833pub const BUILTIN_GET_MATH_VAR: u16 = 529;
9834
9835/// GLOB_SUBST runtime gate for words containing parameter / command
9836/// substitution. C-zsh's `prefork` (Src/subst.c) runs `shtokenize`
9837/// on the substituted value when `GLOB_SUBST` is set, making the
9838/// substituted chars eligible for filename generation. With the
9839/// option off, substituted chars stay literal.
9840///
9841/// The Rust port's compile_zsh emits `compile_word_str` for words
9842/// like `/tmp/X/$pat`, which returns the post-expansion string but
9843/// never runs glob expansion (no path here triggers
9844/// BUILTIN_GLOB_EXPAND). Bug #119 in docs/BUGS.md: with `setopt
9845/// glob_subst`, `for f in /tmp/X/$pat` (pat="*.txt") never matched
9846/// `*.txt` files.
9847///
9848/// This opcode wraps the substitution result and dispatches at
9849/// runtime: when GLOB_SUBST is OFF, return unchanged; when ON,
9850/// pass the value through `expand_glob` so glob metas become
9851/// active. Emitted by `compile_for_words` (and similar sites)
9852/// after WORD_SPLIT for words with unquoted expansion.
9853///
9854/// Stack: pops a Value (Str or Array of Str), pushes the glob-
9855/// expanded result (still Str or Array depending on input shape).
9856/// argc = 1.
9857pub const BUILTIN_GLOB_SUBST_EXPAND: u16 = 530;
9858/// `BUILTIN_ASSOC_HAS_KEY` constant — `${(k)assoc[name]}` key-existence
9859/// query. Returns the key text on hit, empty string on miss. Bug #145.
9860pub const BUILTIN_ASSOC_HAS_KEY: u16 = 531;
9861/// `BUILTIN_ARRAY_DROP_EMPTY` constant — filter empty elements from
9862/// an Array on the stack. Used by `for x in $@` / `for x in $*`
9863/// unquoted forms. Bug #166.
9864pub const BUILTIN_ARRAY_DROP_EMPTY: u16 = 532;
9865/// `BUILTIN_QUOTEDZPUTS` constant — run top-of-stack value through
9866/// `crate::ported::utils::quotedzputs` and push the quoted result.
9867/// Used by the cond xtrace path so non-printable bytes (e.g.
9868/// `$'\C-[OP'` expanded ESC+OP) get re-wrapped in `$'…'` form for
9869/// the trace prefix line, matching zsh's `Src/exec.c` cond trace
9870/// which calls `quotedzputs(operand, xtrerr)` on each side. Bug
9871/// surfaced when `[[ -n $'\C-[OP' ]]` traced as `[[ -n OP ]]`
9872/// (raw bytes leaked through the terminal) vs zsh's
9873/// `[[ -n $'\C-[OP' ]]` source-form preservation.
9874pub const BUILTIN_QUOTEDZPUTS: u16 = 533;
9875/// `BUILTIN_QUOTE_TOKENIZED_OUTPUT` — port of
9876/// `crate::ported::exec::quote_tokenized_output` (Src/exec.c:2114)
9877/// applied to top-of-stack scalar. Used by cond xtrace for the RHS
9878/// of pattern-context comparisons (`=` / `==` / `!=`) where C zsh
9879/// emits the SOURCE form: untokenize lexer tokens (Star → `*`,
9880/// Inpar → `(`, …) and backslash-escape special chars, but
9881/// preserve literal ASCII unchanged. Distinct from quotedzputs
9882/// which wraps the whole string in `'…'` / `$'…'` based on
9883/// non-printability — that's wrong for `[[ x = a* ]]` which must
9884/// render as `[[ x = a* ]]`, not `'a*'`.
9885pub const BUILTIN_QUOTE_TOKENIZED_OUTPUT: u16 = 534;
9886
9887/// Bridge into subst_port::substitute_brace_array for nested forms
9888/// that need to PRESERVE array shape across the expand_string
9889/// boundary. Stack: `[content_string]`. Returns Value::Array of the
9890/// per-element words. Used by the compile path for
9891/// `${(@)<nested>...##pat}` shapes — the standard substitute_brace
9892/// returns String which collapses array→scalar; this builtin
9893/// preserves the multi-word output via paramsubst's third return
9894/// (`nodes` vec, the C source's `aval` thread).
9895pub const BUILTIN_BRIDGE_BRACE_ARRAY: u16 = 347;
9896
9897/// Word-segment concat with FIRST/LAST sticking. Stack: [lhs, rhs].
9898/// Used for default unquoted splice forms (`${arr[@]}`, `$@`, `$*`)
9899/// where prefix sticks to first element only and suffix to last only.
9900///
9901/// Distribution table:
9902/// - both scalar: `Value::str(a + b)` (fast path)
9903/// - lhs scalar, rhs Array(b₀..bₙ): `Value::Array([lhs+b₀, b₁, …, bₙ])`
9904/// - lhs Array(a₀..aₙ), rhs scalar: `Value::Array([a₀, …, aₙ₋₁, aₙ+rhs])`
9905/// - both Array: `Value::Array([a₀, …, aₙ₋₁, aₙ+b₀, b₁, …, bₙ])`
9906/// (last of lhs merges with first of rhs; the rest stay separate)
9907///
9908/// This is the default zsh semantics for `print -l X${arr[@]}Y` →
9909/// "Xa", "b", "cY" — three distinct args, surrounding text only on ends.
9910pub const BUILTIN_CONCAT_SPLICE: u16 = 319;
9911
9912/// `${(flags)name}` — zsh parameter expansion flags. Stack: [name, flags].
9913/// Flags applied left-to-right. Supported subset (high-value, used by zpwr):
9914///
9915/// `L` — lowercase the value (scalar; or each element if array)
9916/// `U` — uppercase
9917/// `j:sep:` — join array with `sep` (delim is the char after `j`)
9918/// `s:sep:` — split scalar on `sep` (returns Value::Array)
9919/// `f` — split on newlines (shorthand for `s.\n.`)
9920/// `o` — sort array ascending
9921/// `O` — sort array descending
9922/// `P` — indirect: read name's value as another var name, return that's value
9923/// `@` — keep as array (returns Value::Array — useful before `j` etc.)
9924/// `k` — keys of assoc array
9925/// `v` — values of assoc array
9926/// `#` — word count (array length as scalar)
9927///
9928/// Flags can stack: `(jL)` joins then lowercases; `(s.,.U)` splits on `,`
9929/// then uppercases each element. The long-tail flags (`q`, `qq`, `qqq` for
9930/// quoting, `A` for assoc, `%` for prompt expansion, `e`/`g` for re-eval,
9931/// `n`/`p` for numeric, `t` for type, etc.) are deferred — they hit the
9932/// runtime fallback via the catch-all expansion path.
9933pub const BUILTIN_PARAM_FLAG: u16 = 297;
9934
9935/// `ShellHost` implementation that delegates to the current `ShellExecutor`
9936/// via the `with_executor` thread-local.
9937///
9938/// Construct fresh on each VM run (it carries no state itself). The VM
9939/// dispatches host method calls during `vm.run()`, and `with_executor`
9940/// resolves to the executor pointer set by `ExecutorContext::enter`.
9941/// fusevm-host implementation tying bytecode ops to the
9942/// shell executor.
9943/// zshrs-original — no C counterpart. C zsh has no bytecode VM
9944/// to host; everything runs through `execlist()`/`execpline()`
9945/// directly (Src/exec.c lines 1349/1668).
9946pub struct ZshrsHost;
9947
9948impl fusevm::ShellHost for ZshrsHost {
9949 fn glob(&mut self, pattern: &str, _recursive: bool) -> Vec<String> {
9950 with_executor(|exec| exec.expand_glob(pattern))
9951 }
9952
9953 fn tilde_expand(&mut self, s: &str) -> String {
9954 with_executor(|exec| s.to_string())
9955 }
9956
9957 fn brace_expand(&mut self, s: &str) -> Vec<String> {
9958 // Direct call to the canonical brace expander
9959 // (Src/glob.c::xpandbraces port at glob.rs:1678). Was
9960 // routing through singsub which uses PREFORK_SINGLE — that
9961 // flag explicitly suppresses brace expansion in subst.c:166,
9962 // so `print X{1,2,3}Y` returned the literal string.
9963 //
9964 // brace_ccl: respect the BRACE_CCL option which the bracket-
9965 // class form `{a-z}` requires. Pull from executor options.
9966 let brace_ccl = with_executor(|exec| opt_state_get("braceccl").unwrap_or(false));
9967 crate::ported::glob::xpandbraces(s, brace_ccl)
9968 }
9969
9970 fn str_match(&mut self, s: &str, pattern: &str) -> bool {
9971 // Shell glob match — `*`, `?`, `[...]`, alternation. After the
9972 // cond path moved to BUILTIN_COND_STRMATCH, the consumer here
9973 // is the `case` arm dispatch, whose bad-pattern semantics are
9974 // Src/loop.c:663-667: `if (!(pprog = patcompile(pat, ...)))
9975 // zerr("bad pattern: %s", pat);` — errflag set, the arm
9976 // doesn't match, and the script aborts at the next command
9977 // boundary (matching `zsh -fc 'case x in [a-) ...'` printing
9978 // the diagnostic with exit 0 = untouched lastval).
9979 let mut pat_tok = pattern.to_string();
9980 crate::ported::glob::tokenize(&mut pat_tok);
9981 if crate::ported::pattern::patcompile(
9982 &pat_tok,
9983 crate::ported::zsh_h::PAT_STATIC as i32,
9984 None,
9985 )
9986 .is_none()
9987 {
9988 crate::ported::utils::zerr(&format!("bad pattern: {}", pattern)); // c:667
9989 return false;
9990 }
9991 glob_match_static(s, pattern)
9992 }
9993
9994 fn expand_param(&mut self, name: &str, _modifier: u8, _args: &[Value]) -> Value {
9995 // Sole funnel: route through `getsparam` matching C zsh's
9996 // `getsparam(name)` → `getvalue` → `getstrvalue` →
9997 // `Param.gsu->getfn` dispatch (Src/params.c:3076 / 2335).
9998 //
9999 // The lookup chain (GSU dispatch + variables + env + array-
10000 // join) lives in `params::getsparam`; subst.rs and this
10001 // bridge both call into it so the logic is in exactly one
10002 // place — mirroring C's "every read goes through getsparam"
10003 // architecture. fuseVM bytecode triggers this bridge when
10004 // the VM hits a PARAM opcode, equivalent to C's wordcode VM
10005 // resolving a parameter read during `exec.c` execution.
10006 //
10007 // Modifier handling: the `_modifier` / `_args` parameters
10008 // are populated by the bytecode compiler but applied by
10009 // separate VM opcodes (LENGTH/STRIP/SUBST/etc.) downstream
10010 // of this fetch — matching C's split between getsparam
10011 // (value fetch) and paramsubst's modifier-walk loop. This
10012 // bridge is the value-fetch step only.
10013 let val_str = crate::ported::params::getsparam(name).unwrap_or_default();
10014 Value::str(val_str)
10015 }
10016
10017 fn process_sub_in(&mut self, sub: &fusevm::Chunk) -> String {
10018 // c:Src/exec.c::getproc — `<(cmd)` uses pipe + fork + the
10019 // `/dev/fd/N` filesystem entry (where N is the read end of
10020 // the pipe held open in the parent). Consumer opens
10021 // `/dev/fd/N`, reads the cmd's stdout through the pipe.
10022 // Both macOS and Linux expose `/dev/fd` for held-open file
10023 // descriptors. Previous Rust port captured stdout into
10024 // `/tmp/zshrs_psub_*` tempfiles synchronously — works for
10025 // `diff <(a) <(b)` style readers that scan once but diverges
10026 // from zsh's observable path string and breaks any consumer
10027 // that introspects the path or expects a non-seekable pipe.
10028 let mut fds: [libc::c_int; 2] = [-1, -1];
10029 if unsafe { libc::pipe(fds.as_mut_ptr()) } != 0 {
10030 // Pipe creation failed — fall back to tempfile so we at
10031 // least return SOMETHING.
10032 let fifo_path = format!(
10033 "/tmp/zshrs_psub_fallback_{}_{}",
10034 std::process::id(),
10035 with_executor(|e| {
10036 let n = e.process_sub_counter;
10037 e.process_sub_counter += 1;
10038 n
10039 })
10040 );
10041 let _ = fs::remove_file(&fifo_path);
10042 return fifo_path;
10043 }
10044 let (read_end, write_end) = (fds[0], fds[1]);
10045 let sub_for_child = sub.clone();
10046 match unsafe { libc::fork() } {
10047 -1 => {
10048 unsafe {
10049 libc::close(read_end);
10050 libc::close(write_end);
10051 }
10052 return String::from("/dev/null");
10053 }
10054 0 => {
10055 // Child: close read end, dup write end to stdout,
10056 // run the sub-chunk, exit. The exit closes the
10057 // write end automatically, so the parent's reader
10058 // gets EOF when the cmd finishes.
10059 unsafe {
10060 libc::close(read_end);
10061 libc::dup2(write_end, libc::STDOUT_FILENO);
10062 libc::close(write_end);
10063 }
10064 crate::fusevm_disasm::maybe_print_stdout("process_subst_in", &sub_for_child);
10065 let mut vm = fusevm::VM::new(sub_for_child);
10066 register_builtins(&mut vm);
10067 vm.set_shell_host(Box::new(ZshrsHost));
10068 let _ = vm.run();
10069 let _ = std::io::stdout().flush();
10070 unsafe { libc::_exit(0) };
10071 }
10072 child_pid => {
10073 // c:Src/exec.c:5092 `procsubstpid = pid;` — record the
10074 // forked child's PID so `${sysparams[procsubstpid]}`
10075 // returns it (was reading the never-updated atomic, so it
10076 // always came back 0). p10k's gitstatus daemon reads
10077 // `sysparams[procsubstpid]` right after `sysopen <(cmd)`
10078 // to track its worker PID; with 0 the daemon's self-check
10079 // failed and gitstatus fell back to re-downloading
10080 // gitstatusd — surfacing as "no prebuilt gitstatusd".
10081 crate::ported::exec::procsubstpid
10082 .store(child_pid, std::sync::atomic::Ordering::Relaxed);
10083 // Parent: close write end, keep read end open under
10084 // the same fd value so `/dev/fd/N` resolves to the
10085 // pipe's read side. NOTE: FD_CLOEXEC must STAY clear
10086 // — consumers like `cat <(cmd)` and `diff <(a) <(b)`
10087 // discover the fd via exec inheritance, so closing
10088 // on exec defeats the whole point. C zsh's getproc
10089 // (Src/exec.c:5045+) leaves the fd open across exec.
10090 unsafe {
10091 libc::close(write_end);
10092 }
10093 // Park read_end for close-after-consuming-command,
10094 // exactly like process_sub_out does for its write_end
10095 // (c:Src/exec.c addfilelist(NULL, fd) → deletefilelist).
10096 // WITHOUT this the parent's read_end stayed open for
10097 // the whole shell lifetime: p10k's async worker /
10098 // realtime clock do `exec {fd}< <(cmd)` on every prompt,
10099 // so each keystroke/redraw leaked a pipe fd until the
10100 // ~256-fd limit was hit and the shell locked up
10101 // (107 leaked pipes + 107 unreaped children observed).
10102 let depth = PSUB_SCOPE_DEPTH.with(|d| d.get());
10103 PSUB_PENDING_FDS.with(|v| v.borrow_mut().push((depth, read_end)));
10104 // Reap the forked child so it doesn't linger as a
10105 // zombie. `<(cmd)` children are fire-and-forget (their
10106 // output flows through the pipe); C reaps them via the
10107 // job machinery. A non-blocking reap here is scheduled;
10108 // do a best-effort WNOHANG now and the rest drain on
10109 // subsequent proc-subs / prompt cycles.
10110 crate::fusevm_bridge::note_psub_child(child_pid);
10111 }
10112 }
10113 format!("/dev/fd/{}", read_end)
10114 }
10115
10116 fn process_sub_out(&mut self, sub: &fusevm::Chunk) -> String {
10117 // c:Src/exec.c:5025 getproc, PATH_DEV_FD branch — `>(cmd)`
10118 // (out == 0): `mpipe(pipes)`, fork; the CHILD `redup(pipes[0],
10119 // 0)` (pipe read end onto stdin) and `closem` drops the write
10120 // end; the PARENT closes pipes[0] and hands the consumer
10121 // `/dev/fd/<pipes[1]>` (the write end). The previous Rust port
10122 // used mkfifo + a child that BLOCKED in open(FIFO, O_RDONLY)
10123 // before running cmd — with no writer the child never started,
10124 // never exited, and kept its inherited stdout (e.g. a `$()`
10125 // capture pipe) open forever: `a=$(print -r -- >(true))` hung.
10126 // With the pipe shape the child runs immediately and exits,
10127 // releasing inherited fds exactly like zsh (verified: zsh
10128 // blocks ~2s on `a=$(print -r -- >(sleep 2))`, then EOFs).
10129 let mut fds: [libc::c_int; 2] = [-1, -1];
10130 if unsafe { libc::pipe(fds.as_mut_ptr()) } != 0 {
10131 // Pipe creation failed — fall back to a plain temp file so
10132 // the consumer at least has a writable path.
10133 let fallback = format!(
10134 "/tmp/zshrs_psub_out_{}_{}",
10135 std::process::id(),
10136 with_executor(|e| {
10137 let n = e.process_sub_counter;
10138 e.process_sub_counter += 1;
10139 n
10140 })
10141 );
10142 let _ = fs::write(&fallback, "");
10143 return fallback;
10144 }
10145 let (read_end, write_end) = (fds[0], fds[1]);
10146 let sub_for_child = sub.clone();
10147 match unsafe { libc::fork() } {
10148 -1 => {
10149 unsafe {
10150 libc::close(read_end);
10151 libc::close(write_end);
10152 }
10153 String::from("/dev/null")
10154 }
10155 0 => {
10156 // Child: close the write end (c: closem after redup),
10157 // dup the read end onto stdin (c: redup(pipes[0], 0)),
10158 // run the sub-chunk, exit. Other std fds stay
10159 // inherited — zsh's child keeps the surrounding
10160 // command's stdout/stderr.
10161 unsafe {
10162 libc::close(write_end);
10163 libc::dup2(read_end, libc::STDIN_FILENO);
10164 libc::close(read_end);
10165 }
10166 crate::fusevm_disasm::maybe_print_stdout("process_subst_out:child", &sub_for_child);
10167 let mut vm = fusevm::VM::new(sub_for_child);
10168 register_builtins(&mut vm);
10169 vm.set_shell_host(Box::new(ZshrsHost));
10170 let _ = vm.run();
10171 unsafe { libc::_exit(0) };
10172 }
10173 child_pid => {
10174 // c:Src/exec.c:5143 `procsubstpid = pid;` — same fix as
10175 // the `<(cmd)` in-path above: record the forked child's
10176 // PID for `${sysparams[procsubstpid]}` (was always 0).
10177 crate::ported::exec::procsubstpid
10178 .store(child_pid, std::sync::atomic::Ordering::Relaxed);
10179 // Parent: close the read end, keep the write end open
10180 // under its fd value so `/dev/fd/N` resolves to the
10181 // pipe's write side. FD_CLOEXEC must STAY clear —
10182 // consumers (`tee >(cmd)`) discover the fd via exec
10183 // inheritance, matching process_sub_in above and C's
10184 // fdtable[fd] = FDT_PROC_SUBST bookkeeping. Park the
10185 // fd for close-after-consuming-command (c: addfilelist
10186 // (NULL, fd) → deletefilelist) so the child's reader
10187 // EOFs — without this `tee >(wc -c) </dev/null` left
10188 // wc blocked until shell exit.
10189 unsafe {
10190 libc::close(read_end);
10191 }
10192 let depth = PSUB_SCOPE_DEPTH.with(|d| d.get());
10193 PSUB_PENDING_FDS.with(|v| v.borrow_mut().push((depth, write_end)));
10194 format!("/dev/fd/{}", write_end)
10195 }
10196 }
10197 }
10198
10199 fn subshell_begin(&mut self) {
10200 with_executor(|exec| {
10201 // libc::umask returns the previous mask AND sets the new
10202 // one; call with current value to read without changing.
10203 let cur_umask = unsafe {
10204 let m = libc::umask(0o022);
10205 libc::umask(m);
10206 m as u32
10207 };
10208 // Snapshot paramtab + hashed-storage too (step 1 of the
10209 // store unification mirrors writes there; restoring only
10210 // the HashMaps leaks subshell-scoped writes to the parent
10211 // via paramtab readers like `paramsubst → vars_get`).
10212 let paramtab_snap = crate::ported::params::paramtab()
10213 .read()
10214 .ok()
10215 .map(|t| t.clone())
10216 .unwrap_or_default();
10217 let paramtab_hashed_snap = crate::ported::params::paramtab_hashed_storage()
10218 .lock()
10219 .ok()
10220 .map(|m| m.clone())
10221 .unwrap_or_default();
10222 exec.subshell_snapshots.push(SubshellSnapshot {
10223 paramtab: paramtab_snap,
10224 paramtab_hashed_storage: paramtab_hashed_snap,
10225 positional_params: exec.pparams(),
10226 env_vars: env::vars().collect(),
10227 // Save the LOGICAL pwd ($PWD env), not `current_dir()`'s
10228 // symlink-resolved path. zsh's subshell isolation per
10229 // Src/exec.c at the `entersubsh` path treats `pwd` (the
10230 // shell-tracked logical PWD) as the carrier — see
10231 // `Src/builtin.c:1239-1242` where cd writes the logical
10232 // dest into `pwd`. Falling back to current_dir() only
10233 // when PWD is unset matches `setupvals` at
10234 // `Src/init.c:1100+`.
10235 cwd: env::var("PWD")
10236 .ok()
10237 .map(PathBuf::from)
10238 .or_else(|| env::current_dir().ok()),
10239 umask: cur_umask,
10240 // Snapshot canonical `traps_table` — bin_trap writes
10241 // there (`Src/builtin.c`).
10242 traps: crate::ported::builtin::traps_table()
10243 .lock()
10244 .map(|t| t.clone())
10245 .unwrap_or_default(),
10246 // Snapshot option store so `(set -e)` /
10247 // `(setopt extendedglob)` don't leak to parent.
10248 opts: crate::ported::options::opt_state_snapshot(),
10249 // c:Src/exec.c — fork() copies the alias table to
10250 // the subshell. `(alias x=y)` inside the subshell
10251 // dies with the child; the parent doesn't see x.
10252 // Snapshot here so subshell_end can restore.
10253 // Bug #209 in docs/BUGS.md.
10254 aliases: crate::ported::hashtable::aliastab_lock()
10255 .read()
10256 .ok()
10257 .map(|t| t.iter().map(|(k, v)| (k.clone(), v.text.clone())).collect())
10258 .unwrap_or_default(),
10259 // c:Src/exec.c::entersubsh — same fork-copy
10260 // semantics for shfunctab. `(f() { ... })` defined
10261 // inside the subshell dies with the child; parent's
10262 // `type f` reports "not found". Bug #208 in
10263 // docs/BUGS.md.
10264 shfuncs: crate::ported::hashtable::shfunctab_lock()
10265 .read()
10266 .ok()
10267 .map(|t| t.snapshot())
10268 .unwrap_or_default(),
10269 functions_compiled: exec.functions_compiled.clone(),
10270 function_source: exec.function_source.clone(),
10271 // c:Src/exec.c::entersubsh — subshell forks its own
10272 // modulestab. A `(zmodload zsh/X)` inside the
10273 // subshell flips MOD_INIT_B on the CHILD's
10274 // modulestab; when the child exits the change
10275 // dies with it. zshrs's in-process subshell would
10276 // otherwise leak the load to the parent.
10277 // Bug #210 in docs/BUGS.md. Snapshot just the
10278 // (name → flags) pairs since the only mutating
10279 // field is the flags bitmask (MOD_INIT_B for
10280 // loaded, MOD_UNLOAD for unloaded).
10281 modules: crate::ported::module::MODULESTAB
10282 .lock()
10283 .ok()
10284 .map(|t| {
10285 t.modules
10286 .iter()
10287 .map(|(k, v)| (k.clone(), v.node.flags))
10288 .collect()
10289 })
10290 .unwrap_or_default(),
10291 // c:Src/exec.c::entersubsh — fork-copy semantics for
10292 // THINGYTAB (ZLE widget registry). A subshell `zle -N`
10293 // / `zle -D` mutation dies with the child in C zsh;
10294 // mirror via in-process snapshot. Bug #453.
10295 thingytab: crate::ported::zle::zle_thingy::thingytab()
10296 .lock()
10297 .ok()
10298 .map(|t| t.clone())
10299 .unwrap_or_default(),
10300 // c:Src/exec.c::entersubsh — same fork-copy for the
10301 // KEYMAPNAMTAB (named keymap registry). `bindkey -N km`
10302 // / `bindkey -D km` inside a subshell dies with the
10303 // child. Bug #454.
10304 keymapnamtab: crate::ported::zle::zle_keymap::keymapnamtab()
10305 .lock()
10306 .ok()
10307 .map(|t| t.clone())
10308 .unwrap_or_default(),
10309 // c:Src/exec.c::entersubsh fork semantics — `$!`
10310 // (clone::lastpid) set by a `&` INSIDE the subshell
10311 // dies with the child: `( : & ); echo $!` -> 0.
10312 lastpid: crate::ported::modules::clone::lastpid
10313 .load(std::sync::atomic::Ordering::Relaxed),
10314 // c:Src/exec.c::entersubsh fork semantics — the
10315 // subshell gets a COPY of the job table; its disown/
10316 // wait/`&` mutations die with it. Bug #462.
10317 jobtab: crate::ported::jobs::JOBTAB
10318 .get_or_init(|| std::sync::Mutex::new(Vec::new()))
10319 .lock()
10320 .map(|t| t.clone())
10321 .unwrap_or_default(),
10322 curjob: *crate::ported::jobs::CURJOB
10323 .get_or_init(|| std::sync::Mutex::new(-1))
10324 .lock()
10325 .unwrap(),
10326 prevjob: *crate::ported::jobs::PREVJOB
10327 .get_or_init(|| std::sync::Mutex::new(-1))
10328 .lock()
10329 .unwrap(),
10330 maxjob: *crate::ported::jobs::MAXJOB
10331 .get_or_init(|| std::sync::Mutex::new(0))
10332 .lock()
10333 .unwrap(),
10334 thisjob: *crate::ported::jobs::THISJOB
10335 .get_or_init(|| std::sync::Mutex::new(-1))
10336 .lock()
10337 .unwrap(),
10338 // c:Src/exec.c entersubsh — fork copies the fd table;
10339 // the child's `exec >file` / `exec N<&-` mutations die
10340 // with it. Dup each user-range fd to >= 10 so
10341 // subshell_end can restore the parent's exact table.
10342 saved_fds: (0..10)
10343 .map(|fd| {
10344 let dup = unsafe { libc::fcntl(fd, libc::F_DUPFD, 10) };
10345 (fd, dup)
10346 })
10347 .collect(),
10348 });
10349 // C forks for `(...)` — count the fork-equivalent so
10350 // `time (builtin)` reports like zsh (see FORK_EVENTS).
10351 crate::vm_helper::FORK_EVENTS.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
10352 // c:Src/exec.c:2862 — subshell fork flags carry ESUB_PGRP,
10353 // so entersubsh runs `clearjobtab(monitor)` (c:1219): the
10354 // child gets an EMPTY job table plus the procless control
10355 // job grabbed at Src/jobs.c:1828 (`thisjob = initjob()`).
10356 // That's why zsh's `(jobs)` prints nothing and `(kill %1)`
10357 // hits the empty control job instead of the parent's job 1.
10358 // The snapshot pushed above restores the parent's table at
10359 // subshell_end. Bug #462.
10360 let monitor = crate::ported::zsh_h::isset(crate::ported::zsh_h::MONITOR) as i32;
10361 crate::ported::jobs::clearjobtab(&mut exec.jobs, monitor);
10362 // clearjobtab left THISJOB on the control job (Src/jobs.c:
10363 // 1828). In C the very next pipeline's execpline reassigns
10364 // thisjob (Src/exec.c:1700 `thisjob = newjob = initjob()`),
10365 // so by the time any builtin runs, thisjob never aliases
10366 // the control job. zshrs has no per-pipeline job slot —
10367 // model the between-pipelines state (-1) so getjob's
10368 // `jobnum != thisjob` (c:jobs.c:2107) doesn't reject %1 and
10369 // setcurjob doesn't demote an inherited curjob that
10370 // collides with the control slot.
10371 *crate::ported::jobs::THISJOB
10372 .get_or_init(|| std::sync::Mutex::new(-1))
10373 .lock()
10374 .unwrap() = -1;
10375 // Subshell starts with EXIT trap cleared so the parent's
10376 // EXIT handler doesn't fire when the subshell ends. zsh:
10377 // each subshell has its own trap context. Other signals
10378 // are inherited (well, parent's are still in place — but
10379 // a trap set INSIDE the subshell shouldn't leak out).
10380 if let Ok(mut t) = crate::ported::builtin::traps_table().lock() {
10381 t.remove("EXIT");
10382 }
10383 let level = exec
10384 .scalar("ZSH_SUBSHELL")
10385 .and_then(|s| s.parse::<i32>().ok())
10386 .unwrap_or(0);
10387 // c:Src/exec.c — ZSH_SUBSHELL carries PM_READONLY (declared
10388 // in params.rs special_params); setsparam would be rejected
10389 // by assignstrvalue's PM_READONLY guard. Write u_val
10390 // directly — same bypass pattern as BUILTIN_SET_LINENO at
10391 // line 2784. C zsh's PM_SPECIAL GSU vtable handles this
10392 // implicitly via the setfn callback.
10393 let new_level = (level + 1) as i64;
10394 if let Ok(mut tab) = crate::ported::params::paramtab().write() {
10395 if let Some(pm) = tab.get_mut("ZSH_SUBSHELL") {
10396 pm.u_val = new_level;
10397 pm.u_str = Some(new_level.to_string());
10398 pm.node.flags &= !(crate::ported::zsh_h::PM_UNSET as i32);
10399 }
10400 }
10401 });
10402 // Bump SUBSHELL_DEPTH so zexit defers process::exit (see
10403 // SUBSHELL_DEPTH declaration in src/ported/builtin.rs for
10404 // rationale).
10405 crate::ported::builtin::SUBSHELL_DEPTH.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
10406 // c:Src/exec.c::entersubsh — C zsh's subshell is a forked
10407 // child process: signals sent to the parent (via `kill $$`
10408 // inside the subshell, where `$$` is the parent's pid)
10409 // never reach the child's signal handlers. zshrs's
10410 // in-process subshell shares the process pid with the
10411 // parent, so without queueing the subshell's trap handler
10412 // fires for signals that zsh would deliver only to the
10413 // parent. Queue signals across the subshell body so the
10414 // parent's restored trap table sees them after
10415 // subshell_end's unqueue drain. Bug #450.
10416 crate::ported::signals_h::queue_signals();
10417 }
10418
10419 fn subshell_end(&mut self) -> Option<i32> {
10420 // Fire subshell's EXIT trap BEFORE restoring parent state so
10421 // the trap body sees the subshell's vars and exit status. zsh
10422 // forks for `(...)` so the trap runs in the child process,
10423 // before exit. We mirror by running it here, just before the
10424 // pop+restore. REMOVE the trap before firing so the inner
10425 // execute_script doesn't fire it again at its own end.
10426 let exit_trap_body = crate::ported::builtin::traps_table()
10427 .lock()
10428 .ok()
10429 .and_then(|mut t| t.remove("EXIT"));
10430 if let Some(body) = exit_trap_body {
10431 // Execute the trap body. Errors during trap execution
10432 // don't bubble — zsh ignores trap-body errors.
10433 with_executor(|exec| {
10434 let _ = exec.execute_script(&body);
10435 });
10436 }
10437 with_executor(|exec| {
10438 if let Some(snap) = exec.subshell_snapshots.pop() {
10439 // Restore paramtab + hashed storage so subshell-scoped
10440 // writes via setsparam/setaparam/sethparam don't leak
10441 // to the parent via paramtab readers.
10442 if let Some(tab) = crate::ported::params::paramtab()
10443 .write()
10444 .ok()
10445 .as_deref_mut()
10446 {
10447 *tab = snap.paramtab;
10448 }
10449 // c:Src/exec.c::entersubsh fork semantics — restore
10450 // the parent's `$!`; a background job inside `(...)`
10451 // dies with the child in C zsh.
10452 crate::ported::modules::clone::lastpid
10453 .store(snap.lastpid, std::sync::atomic::Ordering::Relaxed);
10454 // c:Src/exec.c::entersubsh fork semantics — restore the
10455 // parent's job table + curjob/prevjob/maxjob/thisjob.
10456 // The subshell mutated only its own copy. Bug #462.
10457 if let Ok(mut t) = crate::ported::jobs::JOBTAB
10458 .get_or_init(|| std::sync::Mutex::new(Vec::new()))
10459 .lock()
10460 {
10461 *t = snap.jobtab;
10462 }
10463 *crate::ported::jobs::CURJOB
10464 .get_or_init(|| std::sync::Mutex::new(-1))
10465 .lock()
10466 .unwrap() = snap.curjob;
10467 *crate::ported::jobs::PREVJOB
10468 .get_or_init(|| std::sync::Mutex::new(-1))
10469 .lock()
10470 .unwrap() = snap.prevjob;
10471 *crate::ported::jobs::MAXJOB
10472 .get_or_init(|| std::sync::Mutex::new(0))
10473 .lock()
10474 .unwrap() = snap.maxjob;
10475 *crate::ported::jobs::THISJOB
10476 .get_or_init(|| std::sync::Mutex::new(-1))
10477 .lock()
10478 .unwrap() = snap.thisjob;
10479 if let Some(m) = crate::ported::params::paramtab_hashed_storage()
10480 .lock()
10481 .ok()
10482 .as_deref_mut()
10483 {
10484 *m = snap.paramtab_hashed_storage;
10485 }
10486 exec.set_pparams(snap.positional_params);
10487 // Restore the OS env to its pre-subshell state.
10488 // Removes any `export` writes the subshell made, and
10489 // restores any vars the subshell unset. Without this
10490 // `(export y=sub)` would leak `y` to the parent shell.
10491 let current: HashMap<String, String> = env::vars().collect();
10492 for k in current.keys() {
10493 if !snap.env_vars.contains_key(k) {
10494 env::remove_var(k);
10495 }
10496 }
10497 for (k, v) in &snap.env_vars {
10498 if current.get(k) != Some(v) {
10499 env::set_var(k, v);
10500 }
10501 }
10502 if let Some(cwd) = snap.cwd {
10503 let _ = env::set_current_dir(&cwd);
10504 // Resync $PWD env so a parent `pwd` doesn't read
10505 // the cwd the subshell `cd`'d into.
10506 env::set_var("PWD", &cwd);
10507 }
10508 // Restore umask. zsh's `(umask 077)` doesn't leak to
10509 // parent because the subshell forks; we run in-process
10510 // so we manually reset.
10511 unsafe {
10512 libc::umask(snap.umask as libc::mode_t);
10513 }
10514 // Restore parent's traps (the subshell's own traps die
10515 // with it). zsh: `(trap "X" USR1)` doesn't leak the
10516 // USR1 trap out of the subshell. Write back to the
10517 // canonical `traps_table` (bin_trap writes there).
10518 if let Ok(mut t) = crate::ported::builtin::traps_table().lock() {
10519 *t = snap.traps;
10520 }
10521 // Restore parent's option store so `(set -e)` /
10522 // `(setopt extendedglob)` don't leak. zsh forks
10523 // subshells so child option changes die with the
10524 // child; we run in-process and must restore.
10525 crate::ported::options::opt_state_restore(snap.opts);
10526 // c:Src/exec.c — fork() means alias mutations in a
10527 // subshell die with the child. Restore parent's
10528 // alias table from snapshot. Clear current entries
10529 // then re-add parent's. Bug #209 in docs/BUGS.md.
10530 if let Ok(mut tab) = crate::ported::hashtable::aliastab_lock().write() {
10531 tab.clear();
10532 for (name, text) in snap.aliases {
10533 tab.add(crate::ported::zsh_h::alias {
10534 node: crate::ported::zsh_h::hashnode {
10535 next: None,
10536 nam: name,
10537 flags: 0,
10538 },
10539 text,
10540 inuse: 0,
10541 });
10542 }
10543 }
10544 // c:Src/exec.c::entersubsh — same fork-copy
10545 // semantics for shfunctab. Restore parent's function
10546 // table from snapshot so `(f() { ... })` definitions
10547 // inside the subshell don't leak to the parent.
10548 // Bug #208 in docs/BUGS.md.
10549 if let Ok(mut tab) = crate::ported::hashtable::shfunctab_lock().write() {
10550 tab.restore(snap.shfuncs);
10551 }
10552 // Restore the runtime dispatch tables (compiled chunks
10553 // + source). Without these, a subshell-defined
10554 // override leaves its bytecode in place even after
10555 // shfunctab is restored — `g` after the subshell would
10556 // still run the override.
10557 exec.functions_compiled = snap.functions_compiled;
10558 exec.function_source = snap.function_source;
10559 // c:Src/exec.c::entersubsh — restore parent's
10560 // modulestab so a subshell `(zmodload zsh/X)` doesn't
10561 // leak to the parent. Bug #210 in docs/BUGS.md.
10562 // Restore via per-module flag write since the
10563 // snapshot is `(name → flags)` only.
10564 if let Ok(mut t) = crate::ported::module::MODULESTAB.lock() {
10565 for (name, saved_flags) in &snap.modules {
10566 if let Some(m) = t.modules.get_mut(name) {
10567 m.node.flags = *saved_flags;
10568 }
10569 }
10570 }
10571 // c:Src/exec.c::entersubsh — restore parent's THINGYTAB
10572 // so a subshell's `zle -N w f` / `zle -D w` doesn't
10573 // affect the parent's widget registry. Bug #453.
10574 if let Ok(mut t) = crate::ported::zle::zle_thingy::thingytab().lock() {
10575 *t = snap.thingytab;
10576 }
10577 // Same for KEYMAPNAMTAB. Bug #454.
10578 if let Ok(mut t) = crate::ported::zle::zle_keymap::keymapnamtab().lock() {
10579 *t = snap.keymapnamtab;
10580 }
10581 // c:Src/exec.c entersubsh fork semantics — restore the
10582 // parent's user-range fd table. A bare `exec >file` /
10583 // `exec N>&-` inside `(...)` died with the C child;
10584 // the in-process subshell must undo it here. Flush
10585 // Rust's stdout buffer FIRST so bytes the subshell
10586 // printed drain to the SUBSHELL's fd 1, not the
10587 // restored parent fd.
10588 {
10589 use std::io::Write;
10590 let _ = std::io::stdout().flush();
10591 }
10592 for (fd, saved) in snap.saved_fds {
10593 unsafe {
10594 if saved >= 0 {
10595 libc::dup2(saved, fd);
10596 libc::close(saved);
10597 } else {
10598 // fd was closed at entry; close whatever
10599 // the subshell opened on that slot.
10600 libc::close(fd);
10601 }
10602 }
10603 }
10604 }
10605 });
10606 // Decrement SUBSHELL_DEPTH. If a deferred subshell exit
10607 // landed inside (EXIT_PENDING set with depth > 0), promote
10608 // the deferred status into the subshell's exit status now
10609 // that we're at the boundary, then clear so the parent
10610 // continues. Matches C zsh's "subshell-exit-via-fork"
10611 // boundary where the child's process::exit(N) becomes
10612 // $WAITSTATUS / $? in the parent.
10613 crate::ported::builtin::SUBSHELL_DEPTH.fetch_sub(1, std::sync::atomic::Ordering::Relaxed);
10614 // c:Src/exec.c — drain the signal queue against the now-
10615 // restored parent trap table. Pairs with the
10616 // queue_signals() call at the end of subshell_begin.
10617 // Any `kill $$` from inside the subshell is processed
10618 // here against OUTER's trap, matching C zsh's
10619 // signal-delivery-to-parent semantics. Bug #450.
10620 crate::ported::signals_h::unqueue_signals();
10621 // c:Src/exec.c — a `( … )` subshell is a FORK in C: an errflag
10622 // abort inside the child ends the child with its lastval as
10623 // the exit status, and the flag dies with the child process.
10624 // The parent's $? picks up the status and the parent's lists
10625 // keep running. zsh 5.9: `(readonly r=1; r=2); echo "after
10626 // $?"` prints `after 1`. zshrs runs the subshell in-process,
10627 // so mirror the fork isolation by clearing ERRFLAG_ERROR at
10628 // the subshell boundary — exec.last_status() already carries
10629 // the child's lastval (synced by ERREXIT_CHECK trigger 4).
10630 //
10631 // ERRFLAG_HARD must die at this boundary too: `${u:?msg}` sets
10632 // errflag |= ERRFLAG_HARD (c:Src/subst.c:3344) and then, in a
10633 // C forked subshell, `_exit(1)` (c:3353) — the parent never
10634 // sees ANY errflag bit. A leaked HARD bit here made every
10635 // subsequent zerr() take the silent arm (c:Src/utils.c:175-177
10636 // `if (errflag || noerrs) { errflag |= ERRFLAG_ERROR; return; }`),
10637 // so the next eval/source's parse silently "failed" and the
10638 // D04 harness shell wedged after chunk 10's
10639 // `(print ${unset1:?exiting1})`.
10640 crate::ported::utils::errflag.fetch_and(
10641 !(crate::ported::zsh_h::ERRFLAG_ERROR | crate::ported::zsh_h::ERRFLAG_HARD),
10642 std::sync::atomic::Ordering::Relaxed,
10643 );
10644 let exit_pending =
10645 crate::ported::builtin::EXIT_PENDING.load(std::sync::atomic::Ordering::Relaxed);
10646 if exit_pending != 0 {
10647 // c:Src/builtin.c — `exit N` masks N to 8 bits because
10648 // POSIX _exit takes the low byte as status. `(exit 256)`
10649 // and `(exit 0)` are indistinguishable to the parent;
10650 // `(exit 257)` exits with 1. Without the mask zshrs's
10651 // in-process subshell propagated the full i32 (256) into
10652 // the parent's $?, diverging from zsh.
10653 let raw = crate::ported::builtin::EXIT_VAL.load(std::sync::atomic::Ordering::Relaxed);
10654 let val = raw & 0xFF;
10655 with_executor(|exec| exec.set_last_status(val));
10656 crate::ported::builtin::EXIT_PENDING.store(0, std::sync::atomic::Ordering::Relaxed);
10657 crate::ported::builtin::RETFLAG.store(0, std::sync::atomic::Ordering::Relaxed);
10658 crate::ported::builtin::BREAKS.store(0, std::sync::atomic::Ordering::Relaxed);
10659 // Set the post-subshell-exit guard. The next GET_VAR
10660 // sync_status path consults this to skip its
10661 // vm.last_status→LASTVAL sync (which would overwrite the
10662 // deferred-exit status we just set with stale vm state
10663 // since SubshellEnd doesn't propagate status into the
10664 // VM). Cleared as soon as the next sync_status sees it.
10665 SUBSHELL_EXIT_STATUS_PENDING.with(|c| c.set(true));
10666 // Return the deferred-exit status so the VM updates its
10667 // own `last_status`. Otherwise run_chunk's post-script
10668 // `set_last_status(vm.last_status)` would clobber LASTVAL
10669 // back to the stale pre-subshell value.
10670 return Some(val);
10671 }
10672 None
10673 }
10674
10675 fn redirect(&mut self, fd: u8, op: u8, target: &str) {
10676 // Apply a redirection at the OS level for the next command/builtin.
10677 // The host tracks saved fds in a per-executor stack so a future
10678 // `with_redirects_end` can restore. For now, this is a thin wrapper
10679 // that performs the dup2; pairing with explicit save/restore is
10680 // delivered by `with_redirects_begin/end`.
10681 with_executor(|exec| exec.host_apply_redirect(fd, op, target));
10682 }
10683
10684 fn with_redirects_begin(&mut self, count: u8) {
10685 with_executor(|exec| exec.host_redirect_scope_begin(count));
10686 }
10687
10688 fn regex_match(&mut self, s: &str, regex: &str) -> bool {
10689 // c:Src/Modules/regex.c:54 `zcond_regex_match` — POSIX ERE
10690 // matching + populate `$MATCH` / `$MBEGIN` / `$MEND` /
10691 // `$match[]` / `$mbegin[]` / `$mend[]` (or `$BASH_REMATCH`
10692 // under BASHREMATCH). Direct delegation to the canonical
10693 // port at src/ported/modules/regex.rs:58.
10694 //
10695 // The bridge passthru path delivers TOKEN-form bytes here
10696 // (Inbrack \u{91}, Outbrack \u{92}, Star \u{87}, Quest
10697 // \u{86}, etc.) since the lexer tokenizes regex meta chars
10698 // inside `[[ ]]`. The host regex engine expects ASCII, so
10699 // untokenize the pattern (and subject, for safety) once at
10700 // this boundary. zsh C reaches its POSIX-ERE engine through
10701 // the same untokenize path inside zcond_regex_match.
10702 let s_clean = crate::lex::untokenize(s);
10703 let regex_clean = crate::lex::untokenize(regex);
10704 crate::ported::modules::regex::zcond_regex_match(
10705 &[s_clean.as_str(), regex_clean.as_str()],
10706 crate::ported::modules::regex::ZREGEX_EXTENDED,
10707 ) != 0
10708 }
10709
10710 fn with_redirects_end(&mut self) {
10711 with_executor(|exec| exec.host_redirect_scope_end());
10712 // c:Src/exec.c:5172 — if any redirect in this scope failed
10713 // (noclobber-blocked, ENOENT for read, etc.), the command's
10714 // exit status is forced to 1 regardless of what the (still-
10715 // executed) command's own exit was. C zsh prevents the
10716 // command from running at all when a redirect fails; the
10717 // Rust port still runs it (sinking output to /dev/null in
10718 // the noclobber arm at host_apply_redirect:5481) and then
10719 // overrides $? here. Same observable effect for the common
10720 // pattern `echo x > existing-file` under noclobber.
10721 let failed = with_executor(|exec| {
10722 let f = exec.redirect_failed;
10723 exec.redirect_failed = false;
10724 f
10725 });
10726 if failed {
10727 with_executor(|exec| exec.set_last_status(1));
10728 }
10729 }
10730
10731 fn heredoc(&mut self, content: &str) {
10732 // C `Src/exec.c:4641` — `parsestr(&buf)` runs parameter +
10733 // command substitution on the heredoc body. The lexer's
10734 // quoted-delimiter detection (`<<'EOF'`) routes through the
10735 // `Op::HereDoc` path in `compile_zsh.rs` which short-circuits
10736 // before reaching here; unquoted forms route through the
10737 // BUILTIN_EXPAND_TEXT mode-4 emit path that calls singsub.
10738 // This handler covers the verbatim/quoted case.
10739 with_executor(|exec| exec.host_set_pending_stdin(content.to_string()));
10740 }
10741
10742 fn herestring(&mut self, content: &str) {
10743 // Shell semantics: herestring appends a newline. `<<<` body
10744 // substitution (`Src/exec.c:4655 getherestr` calls
10745 // `quotesubst` + `untokenize`) lands here verbatim; the
10746 // upstream compiler routes through `Op::HereString` after
10747 // BUILTIN_EXPAND_TEXT for the substitution pass, so callers
10748 // of `host.herestring` see the already-expanded form.
10749 let mut s = content.to_string();
10750 s.push('\n');
10751 with_executor(|exec| exec.host_set_pending_stdin(s));
10752 }
10753
10754 fn exec(&mut self, args: Vec<String>) -> i32 {
10755 // c:Src/exec.c getproc + Src/jobs.c deletefilelist — close
10756 // any `>(cmd)` write ends owned by this command once it
10757 // finishes (drops on every return path below).
10758 let _psub_fds = PsubFdGuard;
10759 // c:Src/subst.c paramsubst — when `${var:?msg}` or `${var?msg}`
10760 // triggered the "parameter null or not set" error, errflag
10761 // is raised and zsh aborts the simple command without
10762 // attempting exec. The expansion may have produced empty
10763 // argv[0] which falls into the c:?/permission-denied path
10764 // below, masking the real diagnostic with a spurious
10765 // "permission denied:" line and rc=126 instead of rc=1.
10766 // Honour errflag here so the script ends with the
10767 // paramsubst error as the sole diagnostic. Bug #86.
10768 //
10769 // c:Src/exec.c — C's execlist loop clears ERRFLAG_ERROR
10770 // between sublists when the error came from a NOMATCH-style
10771 // command failure (glob no-match, etc.) so subsequent
10772 // sublists run. zshrs's vm dispatch handles this at the
10773 // post-command-boundary HERE: if THIS command has its
10774 // `current_command_glob_failed` cell set (meaning the glob
10775 // NOMATCH happened during this command's argv prep), surface
10776 // status 1 and clear BOTH the cell AND ERRFLAG_ERROR so the
10777 // NEXT exec call sees a clean state. The errflag from
10778 // genuine script-fatal errors (parse, redirect, paramsubst
10779 // `${:?msg}`) does NOT come paired with glob_failed, so
10780 // those still short-circuit + propagate.
10781 consume_tilde_globsubst_carrier();
10782 let glob_failed = with_executor(|exec| {
10783 let f = exec.current_command_glob_failed.get();
10784 exec.current_command_glob_failed.set(false);
10785 f
10786 });
10787 if glob_failed {
10788 crate::ported::utils::errflag.fetch_and(
10789 !crate::ported::zsh_h::ERRFLAG_ERROR,
10790 std::sync::atomic::Ordering::Relaxed,
10791 );
10792 with_executor(|exec| exec.set_last_status(1));
10793 return 1;
10794 }
10795 // c:Src/subst.c:505-507 — CSH_NULL_GLOB external-path
10796 // boundary: command skipped with `no match` but the NEXT
10797 // sublist runs (zsh -fc 'setopt cshnullglob; ls *nope*;
10798 // print after' prints the error then `after` — verified
10799 // zsh 5.9.1), so clear ERRFLAG like the glob_failed arm.
10800 if consume_badcshglob() {
10801 crate::ported::utils::errflag.fetch_and(
10802 !crate::ported::zsh_h::ERRFLAG_ERROR,
10803 std::sync::atomic::Ordering::Relaxed,
10804 );
10805 with_executor(|exec| exec.set_last_status(1));
10806 return 1;
10807 }
10808 if (crate::ported::utils::errflag.load(std::sync::atomic::Ordering::SeqCst)
10809 & crate::ported::zsh_h::ERRFLAG_ERROR)
10810 != 0
10811 {
10812 return 1;
10813 }
10814 // c:Src/exec.c — two distinct empty-command cases:
10815 //
10816 // 1. args=[""] — an explicit empty-string command word
10817 // (`""`, `"\$unset"`, `\$'\$x'`). zsh attempts exec(2)
10818 // on the empty path → EACCES → "permission denied", \$?
10819 // = 126.
10820 //
10821 // 2. args=[] — the WORD LIST is empty (unquoted \$(\$cmd)
10822 // that produced empty, or an unquoted unset \$var that
10823 // elided). zsh: no exec is attempted; \$? becomes the
10824 // last cmd-subst's exit status (the inner sub-VM
10825 // already set last_status), and the line completes
10826 // silently. Critically NOT 126.
10827 if args.is_empty() {
10828 // c:Src/exec.c — empty word list passes through to a
10829 // no-op; preserve whatever the inner cmd-subst's exit
10830 // is. Return last_status so the caller's SetStatus
10831 // round-trips correctly.
10832 return with_executor(|exec| exec.last_status());
10833 }
10834 if args[0].is_empty() {
10835 let script_name =
10836 crate::ported::utils::scriptname_get().unwrap_or_else(|| "zshrs".to_string());
10837 let lineno: u64 = with_executor(|exec| {
10838 exec.scalar("LINENO")
10839 .and_then(|s| s.parse::<u64>().ok())
10840 .unwrap_or(1)
10841 });
10842 eprintln!("{}:{}: permission denied: ", script_name, lineno);
10843 return 126;
10844 }
10845 // c:Src/exec.c — when any redirect in the current scope
10846 // failed (e.g. noclobber blocked a `>` overwrite), zsh
10847 // refuses to execute the command and exits with status 1.
10848 // The Rust port still applied the command (writing to the
10849 // /dev/null sink installed by host_apply_redirect's
10850 // noclobber arm), but the success status overwrote the
10851 // intended `1`. Short-circuit here so the exec returns 1
10852 // without running the body.
10853 let redir_failed = with_executor(|exec| {
10854 let f = exec.redirect_failed;
10855 exec.redirect_failed = false;
10856 f
10857 });
10858 if redir_failed {
10859 return 1;
10860 }
10861 // Track `$_` as the last argument of the last command (zsh /
10862 // bash convention). Empty arglists leave it untouched.
10863 if let Some(last) = args.last() {
10864 with_executor(|exec| {
10865 exec.set_scalar("_".to_string(), last.clone());
10866 });
10867 }
10868 // Route external command spawning through `executor.execute_external`
10869 // so intercepts (AOP before/after/around), command_hash lookups,
10870 // pre/postexec hooks, and zsh-specific fork-then-exec all apply.
10871 // Without this override, fusevm's default `host.exec` calls
10872 // `Command::new` directly, bypassing zshrs's dispatch logic.
10873 let status = with_executor(|exec| exec.host_exec_external(&args));
10874 // c:Src/jobs.c:1748 waitonejob (no-procs else-branch). zshrs's
10875 // exec model routes external commands through host_exec_external
10876 // (which already waitpid'd in-line); the canonical waitonejob
10877 // expects a Job to derive lastval, but here we already know
10878 // it. Synthesize a procs-less job so waitonejob's no-procs
10879 // branch fires the `pipestats[0]=lastval; numpipestats=1;`
10880 // update via the canonical port.
10881 crate::ported::builtin::LASTVAL.store(status, std::sync::atomic::Ordering::Relaxed);
10882 let mut synth = crate::ported::zsh_h::job::default();
10883 crate::ported::jobs::waitonejob(&mut synth);
10884 status
10885 }
10886
10887 fn cmd_subst(&mut self, sub: &fusevm::Chunk) -> String {
10888 // Run the sub-chunk on a nested VM with the same host wired up,
10889 // capturing stdout. The current executor remains active via the
10890 // thread-local — the nested VM uses CallBuiltin to dispatch shell
10891 // ops back through `with_executor`.
10892 let (read_end, write_end) = match os_pipe::pipe() {
10893 Ok(p) => p,
10894 Err(_) => return String::new(),
10895 };
10896 let saved_stdout = unsafe { libc::dup(libc::STDOUT_FILENO) };
10897 if saved_stdout < 0 {
10898 return String::new();
10899 }
10900 let saved_stderr = unsafe { libc::dup(libc::STDERR_FILENO) };
10901 let write_fd = AsRawFd::as_raw_fd(&write_end);
10902 unsafe {
10903 libc::dup2(write_fd, libc::STDOUT_FILENO);
10904 }
10905 drop(write_end);
10906
10907 // c:Bug #56 — publish the saved outer fds so a trap firing
10908 // during the nested VM run can route its body output to the
10909 // PARENT's stdout instead of the cmdsub's pipe-bound fd 1.
10910 // zsh's forked cmdsub gets this for free (trap runs in the
10911 // parent process whose fd 1 is untouched). zshrs's
10912 // in-process cmdsub needs this thread-local stack so the
10913 // trap dispatcher can find the right destination fd.
10914 CMDSUBST_OUTER_FDS.with(|s| s.borrow_mut().push((saved_stdout, saved_stderr)));
10915
10916 // Nested scope for `>(cmd)` fd ownership — commands inside
10917 // the cmdsub must not drain the enclosing command's pending
10918 // psub fds (see PSUB_SCOPE_DEPTH).
10919 let _psub_scope = PsubScope::enter();
10920
10921 // c:Src/exec.c:1161 — forked cmdsub child runs entersubsh()
10922 // which does `zsh_subshell++`; in-process equivalent.
10923 let _subshell_bump = CmdSubstSubshellBump::enter();
10924
10925 crate::fusevm_disasm::maybe_print_stdout("host.cmd_subst", sub);
10926 let mut vm = fusevm::VM::new(sub.clone());
10927 register_builtins(&mut vm);
10928 vm.set_shell_host(Box::new(ZshrsHost));
10929 let _ = vm.run();
10930 let cmd_status = vm.last_status;
10931
10932 CMDSUBST_OUTER_FDS.with(|s| {
10933 s.borrow_mut().pop();
10934 });
10935
10936 unsafe {
10937 libc::dup2(saved_stdout, libc::STDOUT_FILENO);
10938 libc::close(saved_stdout);
10939 if saved_stderr >= 0 {
10940 libc::close(saved_stderr);
10941 }
10942 }
10943
10944 // Inner cmd's status not propagated for the same reason as
10945 // run_command_substitution — see GAPS.md.
10946 let _ = cmd_status;
10947
10948 let mut buf = String::new();
10949 let mut reader = read_end;
10950 let _ = reader.read_to_string(&mut buf);
10951 // Strip trailing newlines (POSIX command substitution semantics)
10952 while buf.ends_with('\n') {
10953 buf.pop();
10954 }
10955 buf
10956 }
10957
10958 fn call_function(&mut self, name: &str, args: Vec<String>) -> Option<i32> {
10959 // c:Src/exec.c — when the command word is empty (e.g. `""`
10960 // or `"$nonexistent"`), zsh attempts the exec(2) which
10961 // returns EACCES ("permission denied") and exits 126. The
10962 // Rust port silently treated empty as a no-op (status 0).
10963 // Match zsh by emitting the diagnostic and returning 126.
10964 if name.is_empty() {
10965 let script_name =
10966 crate::ported::utils::scriptname_get().unwrap_or_else(|| "zshrs".to_string());
10967 let lineno: u64 = with_executor(|exec| {
10968 exec.scalar("LINENO")
10969 .and_then(|s| s.parse::<u64>().ok())
10970 .unwrap_or(1)
10971 });
10972 eprintln!("{}:{}: permission denied: ", script_name, lineno);
10973 with_executor(|exec| exec.set_last_status(126));
10974 return Some(126);
10975 }
10976 // c:Src/exec.c — redirect failure in this scope means the
10977 // command should NOT run. The Host::exec path already has
10978 // this gate (at fn exec above); call_function takes external
10979 // commands like `cat <&3` through a different code path, so
10980 // gate here too. Without this, bad-fd redirects produced
10981 // the diagnostic but the external command still ran, so $?
10982 // came out from the command's natural exit instead of the
10983 // forced 1.
10984 let redir_failed = with_executor(|exec| {
10985 let f = exec.redirect_failed;
10986 exec.redirect_failed = false;
10987 f
10988 });
10989 if redir_failed {
10990 with_executor(|exec| exec.set_last_status(1));
10991 return Some(1);
10992 }
10993 // ACTUALLY A ZSH FUNCTION: zmv/zcp/zln/zcalc are zsh autoload
10994 // functions, NOT builtins. zshrs ships fast native impls, but they
10995 // must behave like the zsh functions — command-not-found until
10996 // `autoload -Uz <name>` creates a function entry. When autoloaded we
10997 // run the native impl here (short-circuiting the fpath source, which
10998 // can hang zshrs's parser on zsh-specific syntax); when NOT autoloaded
10999 // we fall through (return None → resolution ends in command-not-found),
11000 // matching `zsh -f; zmv` → "command not found: zmv".
11001 if matches!(name, "zmv" | "zcp" | "zln" | "zcalc")
11002 && !with_executor(|exec| exec.function_exists(name))
11003 {
11004 return None;
11005 }
11006 match name {
11007 "zmv" => {
11008 return Some(crate::extensions::ext_builtins::zmv(&args, "mv"));
11009 }
11010 "zcp" => {
11011 return Some(crate::extensions::ext_builtins::zmv(&args, "cp"));
11012 }
11013 "zln" => {
11014 return Some(crate::extensions::ext_builtins::zmv(&args, "ln"));
11015 }
11016 "zcalc" => {
11017 return Some(crate::extensions::ext_builtins::zcalc(&args));
11018 }
11019 // ztest framework (src/extensions/ztest.rs — port of
11020 // ../strykelang's unit-test framework). All zassert_*/
11021 // ztest_* names route through the single try_dispatch
11022 // helper so adding/removing assertions only touches
11023 // ztest.rs.
11024 n if crate::extensions::ztest::try_dispatch_known(n) => {
11025 let status = with_executor(|exec| {
11026 crate::extensions::ztest::try_dispatch(exec, n, &args).unwrap_or(1)
11027 });
11028 return Some(status);
11029 }
11030 // Daemon-managed z* builtins — thin IPC wrappers. Short-circuit BEFORE
11031 // the function-lookup path so a missing daemon doesn't fall through to
11032 // "command not found". The name list is owned by the daemon crate
11033 // (zshrs_daemon::builtins::ZSHRS_BUILTIN_NAMES); routing through
11034 // try_dispatch keeps this site zero-touch as new z* builtins land.
11035 n if crate::daemon::builtins::is_zshrs_builtin(n) => {
11036 let argv: Vec<String> = std::iter::once(name.to_string()).chain(args).collect();
11037 return Some(crate::daemon::builtins::try_dispatch(n, &argv).unwrap_or(1));
11038 }
11039 _ => {}
11040 }
11041
11042 // c:Src/exec.c:3050-3068 — module-provided builtins (registered
11043 // via each module's `bintab` and folded into the canonical
11044 // `builtintab` by `createbuiltintable`) must dispatch BEFORE
11045 // PATH lookup. fusevm's `shell_builtins::builtin_id` doesn't
11046 // know about per-module entries like `log`
11047 // (Src/Modules/watch.c:693) — they reach call_function as
11048 // CallFunction ops. Consult the merged builtintab here so
11049 // `log` runs the canonical `bin_log` instead of falling
11050 // through to `/usr/bin/log` on macOS. Bug #72 in docs/BUGS.md.
11051 //
11052 // User-defined functions still take precedence over builtins
11053 // (zsh's `alias → function → builtin → external` resolution
11054 // order, c:Src/exec.c:3038-3068). Check `functions_compiled`
11055 // first so a user `log() { ... }` shadows the module bin_log.
11056 // c:Src/exec.c — shfunctab->getnode (the DISABLED-filtering
11057 // accessor) returns NULL for entries flipped to DISABLED via
11058 // `disable -f NAME`. functions_compiled holds the body
11059 // independently of the DISABLED flag, so check shfunctab first
11060 // and mask the lookup when the entry is disabled. Bug #221
11061 // in docs/BUGS.md.
11062 let user_fn_disabled = crate::ported::hashtable::shfunctab_lock()
11063 .read()
11064 .ok()
11065 .and_then(|t| {
11066 let entry = t.get_including_disabled(name)?;
11067 Some((entry.node.flags as u32 & crate::ported::zsh_h::DISABLED as u32) != 0)
11068 })
11069 .unwrap_or(false);
11070 let has_user_fn =
11071 !user_fn_disabled && with_executor(|exec| exec.functions_compiled.contains_key(name));
11072 if !has_user_fn {
11073 // c:Src/exec.c:3056 — `builtintab->getnode(builtintab,
11074 // cmdarg)` returns NULL for DISABLED entries, falling
11075 // execcmd through to PATH lookup. Mirror by gating the
11076 // bn_in_tab match on the BUILTINS_DISABLED set. Bug #106
11077 // in docs/BUGS.md.
11078 let disabled = crate::ported::builtin::BUILTINS_DISABLED
11079 .lock()
11080 .map(|s| s.contains(name))
11081 .unwrap_or(false);
11082 let bn_in_tab =
11083 !disabled && crate::ported::builtin::createbuiltintable().contains_key(name);
11084 if bn_in_tab {
11085 return Some(dispatch_builtin_raw(name, args));
11086 }
11087 }
11088
11089 // c:Src/lex.c — alias expansion is a LEXER-TIME pass, not a
11090 // run-time lookup. zsh parses the whole `-c` argument (or
11091 // script) before executing, so aliases defined in the same
11092 // parse unit don't apply to commands parsed earlier. Only at
11093 // an INTERACTIVE prompt does each line parse separately with
11094 // the latest aliastab visible.
11095 //
11096 // Gate the run-time alias-rewrite path on `interactive` so
11097 // `alias hi='echo hello'; hi` in `-c` mode falls through to
11098 // "command not found" (matching zsh) while interactive REPL
11099 // input still re-parses with the live aliastab.
11100 let interactive = crate::ported::zsh_h::isset(crate::ported::zsh_h::INTERACTIVE);
11101 let already_expanding = if interactive {
11102 crate::ported::hashtable::aliastab_lock()
11103 .read()
11104 .ok()
11105 .and_then(|tab| tab.get(name).map(|a| a.inuse != 0))
11106 .unwrap_or(false)
11107 } else {
11108 true // suppress lookup entirely in non-interactive mode
11109 };
11110 let alias_body = if already_expanding {
11111 None
11112 } else {
11113 with_executor(|exec| exec.alias(name))
11114 };
11115 if let Some(body) = alias_body {
11116 let combined = if args.is_empty() {
11117 body
11118 } else {
11119 let quoted: Vec<String> = args
11120 .iter()
11121 .map(|a| {
11122 let escaped = a.replace('\'', "'\\''");
11123 format!("'{}'", escaped)
11124 })
11125 .collect();
11126 format!("{} {}", body, quoted.join(" "))
11127 };
11128 // Bump inuse → run → clear, matching C's lexer behavior.
11129 if let Ok(mut tab) = crate::ported::hashtable::aliastab_lock().write() {
11130 if let Some(a) = tab.get_mut(name) {
11131 a.inuse += 1;
11132 }
11133 }
11134 let status = with_executor(|exec| exec.execute_script(&combined).unwrap_or(1));
11135 if let Ok(mut tab) = crate::ported::hashtable::aliastab_lock().write() {
11136 if let Some(a) = tab.get_mut(name) {
11137 a.inuse = (a.inuse - 1).max(0);
11138 }
11139 }
11140 return Some(status);
11141 }
11142
11143 // $_ pre-body bump and pending-underscore tracking are
11144 // ZshrsHost-only concerns (prompt rendering). Apply BEFORE
11145 // delegating to dispatch_function_call so the body sees the
11146 // bumped value.
11147 //
11148 // c:Src/exec.c:3491 — `setunderscore((args && nonempty(args))
11149 // ? ((char *) getdata(lastnode(args))) : "")`. C sets $_ to
11150 // the LAST node of the WHOLE args list (which includes argv[0]
11151 // == the function name). So for a no-arg `f`, $_ becomes "f"
11152 // inside the function body. The Rust port at the CallFunction
11153 // op-handler receives `args` WITHOUT the command name
11154 // (compile_zsh.rs:1571 only pushes simple.words[1..]). The
11155 // last() fallback `|| fn_name.clone()` already covers the
11156 // no-arg case, but `exec.set_scalar("_", ...)` writes paramtab
11157 // — the canonical `$_` read goes through `underscoregetfn`
11158 // (params.rs:7836) which reads the `zunderscore` Mutex.
11159 // setsparam("_") doesn't update that mutex, so the body's
11160 // `${_}` returned empty. Bug #279 in docs/BUGS.md. Mirror the
11161 // C `setunderscore` by writing via `set_zunderscore` directly.
11162 let fn_name = name.to_string();
11163 with_executor(|exec| {
11164 let dollar_underscore = args.last().cloned().unwrap_or_else(|| fn_name.clone());
11165 exec.set_scalar("_".to_string(), dollar_underscore.clone());
11166 crate::ported::params::set_zunderscore(std::slice::from_ref(&dollar_underscore));
11167 exec.pending_underscore = Some(dollar_underscore);
11168 });
11169
11170 // Delegate the actual function dispatch to the canonical
11171 // `dispatch_function_call` (which itself wraps the canonical
11172 // `doshfunc` port from `Src/exec.c:5823`). Single doshfunc
11173 // call-site keeps scope-mgmt invariants in one place.
11174 let status = with_executor(|exec| exec.dispatch_function_call(&fn_name, &args));
11175
11176 // Anonymous functions (`() { … } args`, compiled by
11177 // parse_anon_funcdef as `_zshrs_anon_N` / `_zshrs_anon_kw_N`)
11178 // execute exactly ONCE and must not persist. zsh runs the body and
11179 // frees the function, so `${functions}` / `typeset -f` never show
11180 // it. Remove every trace right after the single invocation —
11181 // AFTER `status` is captured, so the body's exit code is preserved
11182 // ($? — calling `unfunction` here would reset it to 0 instead).
11183 // Without this, real plugins that use `() { … }` (fzf-tab, zinit,
11184 // p10k, …) leaked dozens of `_zshrs_anon_N` into `$functions`,
11185 // diverging from zsh's function table on every such config.
11186 if fn_name.starts_with("_zshrs_anon_") {
11187 // `${functions}` / `typeset -f` enumerate the canonical
11188 // `shfunctab` (via scanpmfunctions); the bytecode call path
11189 // also keeps the body in the executor's compiled-fn maps. Clear
11190 // BOTH so no trace of the one-shot anon survives.
11191 crate::ported::hashtable::removeshfuncnode(&fn_name);
11192 with_executor(|exec| {
11193 exec.functions_compiled.remove(&fn_name);
11194 exec.function_source.remove(&fn_name);
11195 exec.function_line_base.remove(&fn_name);
11196 exec.function_def_file.remove(&fn_name);
11197 });
11198 }
11199
11200 // $_ post-body — last call-arg or function name. Mirrors the
11201 // C `setunderscore` invocation after the body returns.
11202 with_executor(|exec| {
11203 let last_call_arg = args.last().cloned().unwrap_or_else(|| fn_name.clone());
11204 exec.set_scalar("_".to_string(), last_call_arg.clone());
11205 exec.pending_underscore = Some(last_call_arg);
11206 });
11207
11208 status
11209 }
11210}
11211
11212// ───────────────────────────────────────────────────────────────────────────
11213/// Render a failed-redirect open error the way C's `zerrmsg` `%e` format
11214/// code does (Src/utils.c): `strerror(errno)` with the first character
11215/// lowercased, except `EIO` (kept capitalized) and `EINTR` (→ "interrupt").
11216/// C's redirect open failures call `zwarn("%e: %s", errno, fname)`
11217/// (Src/exec.c:3741); zshrs's `zwarning` takes a pre-built string, so the
11218/// `%e` part is built here. Replaces the prior hardcoded `ErrorKind` match
11219/// that fell back to a generic "redirect failed" for `EROFS`/`EACCES`/etc.
11220fn redir_errno_msg(err: &std::io::Error) -> String {
11221 let errno = match err.raw_os_error() {
11222 Some(n) if n != 0 => n,
11223 _ => return "redirect failed".to_string(),
11224 };
11225 if errno == libc::EINTR {
11226 return "interrupt".to_string(); // c:zerrmsg %e — EINTR special-case
11227 }
11228 let cptr = unsafe { libc::strerror(errno) };
11229 if cptr.is_null() {
11230 return "redirect failed".to_string();
11231 }
11232 let msg = unsafe { std::ffi::CStr::from_ptr(cptr) }.to_string_lossy();
11233 if errno == libc::EIO {
11234 return msg.into_owned(); // c:zerrmsg %e — EIO keeps capitalization
11235 }
11236 // c:zerrmsg %e — `fputc(tulower(errmsg[0])); fputs(errmsg + 1)`.
11237 let mut chars = msg.chars();
11238 match chars.next() {
11239 Some(first) => first.to_lowercase().collect::<String>() + chars.as_str(),
11240 None => "redirect failed".to_string(),
11241 }
11242}
11243
11244// Host-routed shell ops: ShellExecutor methods invoked by ZshrsHost from the
11245// fusevm VM. Not a port of Src/exec.c (see file-level docs above) — they're
11246// the bridge between fusevm opcodes and ShellExecutor state.
11247// ───────────────────────────────────────────────────────────────────────────
11248impl ShellExecutor {
11249 // ─── Host-routed shell ops (called by ZshrsHost from fusevm) ────────────
11250
11251 /// Apply a single redirection. The current scope's saved-fd vec gets a
11252 /// dup of the original fd so it can be restored by `host_redirect_scope_end`.
11253 /// `op_byte` matches `fusevm::op::redirect_op::*`.
11254 /// Apply a file-open result to a redirect fd; on error, emit
11255 /// zsh-format diagnostic, set redirect_failed, sink fd to /dev/null.
11256 /// Shared between WRITE/APPEND/READ/CLOBBER arms in
11257 /// host_apply_redirect to keep the error-handling identical.
11258 fn redir_open_or_fail(
11259 fd: i32,
11260 result: std::io::Result<fs::File>,
11261 target: &str,
11262 redirect_failed: &mut bool,
11263 ) -> bool {
11264 match result {
11265 Ok(file) => {
11266 let new_fd = file.into_raw_fd();
11267 unsafe {
11268 // When the target fd was already closed (e.g. `exec 0<&-;
11269 // cmd < file`), open() returns the lowest free fd, which is
11270 // `fd` itself. Then `dup2(fd, fd)` is a no-op: closing new_fd
11271 // would CLOSE the fd we just opened, AND — since Rust's
11272 // File::open sets O_CLOEXEC and a no-op dup2 does NOT clear
11273 // it — an exec'd child would lose the descriptor. So in the
11274 // reuse case, keep the fd and clear its close-on-exec flag;
11275 // otherwise dup2 (which clears cloexec on the copy) + close.
11276 if new_fd != fd {
11277 libc::dup2(new_fd, fd);
11278 libc::close(new_fd);
11279 } else {
11280 libc::fcntl(fd, libc::F_SETFD, 0);
11281 }
11282 }
11283 true
11284 }
11285 Err(e) => {
11286 // c:Src/exec.c:3741 — zwarn("%e: %s", errno, fname) with the
11287 // real lineno prefix; redir_errno_msg builds the `%e` errno
11288 // message for all errnos (not just the few hardcoded before).
11289 let msg = redir_errno_msg(&e);
11290 crate::ported::utils::zwarn(&format!("{}: {}", msg, target));
11291 *redirect_failed = true;
11292 // The /dev/null sink keeps a failed scoped redirect
11293 // from leaking the aborted command's output to the
11294 // wrong fd until scope-end restores it. For a bare
11295 // `exec` redirect (permanent, no scope restore) C
11296 // leaves the fd UNTOUCHED — execerr() aborts the
11297 // statement and the original fd 1 keeps flowing
11298 // (A04redirect: `exec >./nonexistent/x` then `echo
11299 // output` still prints). c:Src/exec.c:3735-3742.
11300 let permanent = with_executor(|exec| exec.exec_redirs_permanent);
11301 if !permanent {
11302 if let Ok(devnull) = fs::OpenOptions::new()
11303 .read(true)
11304 .write(true)
11305 .open("/dev/null")
11306 {
11307 let new_fd = devnull.into_raw_fd();
11308 unsafe {
11309 if new_fd != fd {
11310 libc::dup2(new_fd, fd);
11311 libc::close(new_fd);
11312 } else {
11313 libc::fcntl(fd, libc::F_SETFD, 0);
11314 }
11315 }
11316 }
11317 }
11318 false
11319 }
11320 }
11321 }
11322 /// `host_apply_redirect` — see implementation.
11323 pub fn host_apply_redirect(&mut self, fd: u8, op_byte: u8, target: &str) {
11324 // `&>` / `&>>` always target both fd 1 and fd 2 regardless of the
11325 // fd byte the parser supplied (the lexer's tokfd clamp makes the
11326 // raw value unreliable for these forms).
11327 let fd: i32 = if matches!(op_byte, r::WRITE_BOTH | r::APPEND_BOTH) {
11328 1
11329 } else {
11330 fd as i32
11331 };
11332 // c:Src/exec.c — for DUP_READ / DUP_WRITE forms (<&N / >&N),
11333 // validate the source fd is open BEFORE the save-and-dup
11334 // dance below. The save's `dup(fd)` reclaims the lowest free
11335 // fd, which on closed-fd reuse would let dup2(src=N, …)
11336 // succeed against the freshly-claimed slot — masking the
11337 // user's "bad file descriptor" error. Check src_fd first.
11338 if matches!(op_byte, r::DUP_READ | r::DUP_WRITE) {
11339 let n_check = target.trim_start_matches('&');
11340 if n_check != "-" {
11341 if let Ok(src_fd) = n_check.parse::<i32>() {
11342 if unsafe { libc::fcntl(src_fd, libc::F_GETFD) } == -1 {
11343 // c:Src/exec.c — zwarn with real lineno prefix.
11344 crate::ported::utils::zwarn(&format!("{}: bad file descriptor", src_fd));
11345 self.set_last_status(1);
11346 self.redirect_failed = true;
11347 return;
11348 }
11349 }
11350 }
11351 }
11352 // c:Src/exec.c:3978-3986 — bare `exec` redirects (nullexec==1)
11353 // skip the save entirely: "we specifically *don't* restore the
11354 // original fd's". C's save[] is per-execcmd, so exec's redirs
11355 // never enter an enclosing group's save list either; pushing
11356 // into `redirect_scope_stack.last_mut()` here (the enclosing
11357 // group's scope) made `{ exec 1>&-; … } 2>/dev/null` restore
11358 // stdout at group end — diverging from zsh, which keeps fd 1
11359 // closed for the rest of the script.
11360 if !self.exec_redirs_permanent {
11361 let saved = unsafe { libc::dup(fd) };
11362 if saved >= 0 {
11363 if let Some(top) = self.redirect_scope_stack.last_mut() {
11364 top.push((fd, saved));
11365 } else {
11366 // No scope — leave saved fd open and let the next scope
11367 // reclaim it. (Caller without a scope leaks the dup; this
11368 // matches `WithRedirects` parser construction always wrapping.)
11369 unsafe { libc::close(saved) };
11370 }
11371 }
11372 // For `&>` / `&>>` also save fd 2 so the scope restores it after
11373 // the body. Otherwise stderr stays redirected past the command.
11374 if matches!(op_byte, r::WRITE_BOTH | r::APPEND_BOTH) {
11375 let saved2 = unsafe { libc::dup(2) };
11376 if saved2 >= 0 {
11377 if let Some(top) = self.redirect_scope_stack.last_mut() {
11378 top.push((2, saved2));
11379 } else {
11380 unsafe { libc::close(saved2) };
11381 }
11382 }
11383 }
11384 }
11385 // c:Src/exec.c:3722-3724 + 2447-2480 — MULTIOS split when this
11386 // command's stdout IS the pipeline output. C registers the pipe
11387 // in mfds[1] (`addfd(forked, save, mfds, 1, output, 1, NULL)`)
11388 // BEFORE walking the explicit redirect list, so a write-side
11389 // redirect of fd 1 finds mfds[1] occupied and, with MULTIOS
11390 // set, "split[s] the stream": fd 1 becomes the write end of an
11391 // internal pipe whose reader tees every chunk to BOTH the
11392 // pipeline pipe and the new target. That is why
11393 // `{ echo a; echo b >&2; } 3>&1 1>&2 2>&3 3>&- | cat` sends
11394 // `a` to the pipe (via the tee) AND to stderr — plain dup2
11395 // replacement loses the pipe stream. The scope-depth gate
11396 // mirrors mfds being per-execcmd: only the redirect list
11397 // attached to the stage's own command joins the pipe; nested
11398 // commands inside the body (`{ echo a > f; } | cat`) get a
11399 // fresh "mfds" and replace as usual.
11400 if fd == 1
11401 && self
11402 .pipe_output_scope
11403 .is_some_and(|d| d + 1 == self.redirect_scope_stack.len())
11404 && crate::ported::options::opt_state_get("multios").unwrap_or(true)
11405 {
11406 // Resolve the new write target exactly as the plain arms
11407 // below would, but as a raw fd for the tee.
11408 let new_target_fd: i32 = match op_byte {
11409 r::DUP_WRITE => {
11410 // Numeric `>&N` only; `-` (close) and `p` (coproc)
11411 // fall through to the plain arms.
11412 target
11413 .trim_start_matches('&')
11414 .parse::<i32>()
11415 .map(|src| unsafe { libc::dup(src) })
11416 .unwrap_or(-1)
11417 }
11418 r::WRITE | r::CLOBBER => fs::File::create(target)
11419 .map(|f| f.into_raw_fd())
11420 .unwrap_or(-1),
11421 r::APPEND => fs::OpenOptions::new()
11422 .create(true)
11423 .append(true)
11424 .open(target)
11425 .map(|f| f.into_raw_fd())
11426 .unwrap_or(-1),
11427 _ => -1,
11428 };
11429 if new_target_fd >= 0 {
11430 let pipe_dup = unsafe { libc::dup(1) };
11431 match (pipe_dup >= 0).then(os_pipe::pipe) {
11432 Some(Ok((read_end, write_end))) => {
11433 // Splitter: same read-loop shape as
11434 // BUILTIN_MULTIOS_REDIRECT, with one ordering
11435 // refinement. C's tee is a forked process
11436 // (closemn → teeproc) whose wakeup latency lets
11437 // the stage's DIRECT pipe writes land first —
11438 // observed zsh output for `{ echo a; echo b >&2; }
11439 // 3>&1 1>&2 2>&3 3>&- | cat` is `b` then `a`,
11440 // 15/15 runs. A Rust thread wakes faster than
11441 // the debug-build VM dispatches the next echo,
11442 // inverting the order. Emulate the C timing
11443 // observably: stream to the NEW target (file /
11444 // stderr dup) immediately, but defer the
11445 // pipe-bound copy until EOF (or a 64KB cap so a
11446 // long-running stream still flows instead of
11447 // growing memory unboundedly).
11448 let write_now = |tfd: i32, data: &[u8]| {
11449 let mut off = 0;
11450 while off < data.len() {
11451 let w = unsafe {
11452 libc::write(
11453 tfd,
11454 data[off..].as_ptr() as *const libc::c_void,
11455 data.len() - off,
11456 )
11457 };
11458 if w <= 0 {
11459 break;
11460 }
11461 off += w as usize;
11462 }
11463 };
11464 let handle = std::thread::spawn(move || {
11465 let mut rd = read_end;
11466 let mut buf = [0u8; 8192];
11467 let mut pipe_pending: Vec<u8> = Vec::new();
11468 loop {
11469 match std::io::Read::read(&mut rd, &mut buf) {
11470 Ok(0) | Err(_) => break,
11471 Ok(n) => {
11472 write_now(new_target_fd, &buf[..n]);
11473 pipe_pending.extend_from_slice(&buf[..n]);
11474 if pipe_pending.len() >= 65536 {
11475 write_now(pipe_dup, &pipe_pending);
11476 pipe_pending.clear();
11477 }
11478 }
11479 }
11480 }
11481 write_now(pipe_dup, &pipe_pending);
11482 unsafe {
11483 libc::close(pipe_dup);
11484 libc::close(new_target_fd);
11485 }
11486 });
11487 let write_raw = AsRawFd::as_raw_fd(&write_end);
11488 unsafe { libc::dup2(write_raw, 1) };
11489 drop(write_end);
11490 // Scope-end closes this dup (the last writer once
11491 // the saved fd 1 is restored) → EOF → join.
11492 let close_on_end = unsafe { libc::dup(1) };
11493 if let Some(top) = self.multios_scope_stack.last_mut() {
11494 top.push((close_on_end, handle));
11495 } else {
11496 unsafe { libc::close(close_on_end) };
11497 let _ = handle.join();
11498 }
11499 return;
11500 }
11501 _ => unsafe {
11502 // pipe()/dup failure — fall through to plain replace.
11503 if pipe_dup >= 0 {
11504 libc::close(pipe_dup);
11505 }
11506 libc::close(new_target_fd);
11507 },
11508 }
11509 }
11510 }
11511 match op_byte {
11512 r::WRITE => {
11513 // Honor `setopt noclobber`: refuse to overwrite an
11514 // existing regular file unless `>!` / `>|` (CLOBBER).
11515 // zsh internally stores the inverted-name `clobber`
11516 // (default ON); `setopt noclobber` writes
11517 // `clobber=false`. Honor both keys.
11518 //
11519 // c:Src/exec.c:2241-2245 clobber_open recover path:
11520 // after O_EXCL fails, reopen and `if (!S_ISREG(...))
11521 // return fd;` — non-regular targets (char/block-
11522 // special, FIFO, socket) bypass the noclobber check.
11523 // Bug #30 in docs/BUGS.md: this bridge-side check did
11524 // a bare `Path::exists()` and treated `/dev/null` as
11525 // a protected file, breaking `setopt no_clobber; echo
11526 // hi > /dev/null` and every `2> /dev/null` idiom.
11527 // Add a regular-file stat gate that matches the C
11528 // semantic. The canonical clobber_open at
11529 // src/ported/exec.rs:2123 already handles this; the
11530 // bridge duplicates a stripped-down version here and
11531 // must mirror the same check.
11532 let noclobber = opt_state_get("noclobber").unwrap_or(false)
11533 || !opt_state_get("clobber").unwrap_or(true);
11534 let target_meta = std::fs::metadata(target).ok();
11535 let target_is_regular_file = target_meta
11536 .as_ref()
11537 .map(|m| m.file_type().is_file())
11538 .unwrap_or(false);
11539 // c:Src/exec.c:2313 clobber_open — CLOBBER_EMPTY permits
11540 // re-using an EMPTY regular file under noclobber: `setopt
11541 // noclobber clobberempty; : >f; echo hi >f` overwrites f.
11542 // The inline bridge check ignored this and errored.
11543 let clobber_empty_ok = opt_state_get("clobberempty").unwrap_or(false)
11544 && target_meta.as_ref().map(|m| m.len() == 0).unwrap_or(false);
11545 if noclobber && target_is_regular_file && !clobber_empty_ok {
11546 eprintln!(
11547 "{}:{}: file exists: {}",
11548 shname(),
11549 crate::ported::lex::lineno(),
11550 target
11551 );
11552 self.set_last_status(1);
11553 // c:Src/exec.c — set redirect_failed so the scope-end
11554 // hook (`with_redirects_end` in this file) forces
11555 // $? to 1 regardless of the still-running command's
11556 // own exit. Without this the next command (e.g.
11557 // `echo x` writing to /dev/null below) succeeds
11558 // and overwrites the redirect-failure status,
11559 // making noclobber unobservable from $?.
11560 self.redirect_failed = true;
11561 // Sink the upcoming command's stdout to /dev/null
11562 // so we don't leak its output to the terminal.
11563 // zsh skips the command entirely; we approximate by
11564 // discarding the output (the redirect target was
11565 // the user's chosen sink, but with noclobber the
11566 // file is protected — discarding matches the
11567 // user's intent better than printing to terminal).
11568 if let Ok(file) = fs::OpenOptions::new().write(true).open("/dev/null") {
11569 let new_fd = file.into_raw_fd();
11570 unsafe {
11571 libc::dup2(new_fd, fd);
11572 libc::close(new_fd);
11573 }
11574 }
11575 return;
11576 }
11577 if !Self::redir_open_or_fail(
11578 fd,
11579 fs::File::create(target),
11580 target,
11581 &mut self.redirect_failed,
11582 ) {
11583 self.set_last_status(1);
11584 }
11585 }
11586 r::CLOBBER => {
11587 if !Self::redir_open_or_fail(
11588 fd,
11589 fs::File::create(target),
11590 target,
11591 &mut self.redirect_failed,
11592 ) {
11593 self.set_last_status(1);
11594 }
11595 }
11596 r::APPEND => {
11597 // c:Src/exec.c:3924-3927 — `>>` honors NO_CLOBBER+!APPENDCREATE
11598 // by opening O_APPEND|O_WRONLY WITHOUT O_CREAT, so missing
11599 // files yield ENOENT. zsh source:
11600 // if (!isset(CLOBBER) && !isset(APPENDCREATE) &&
11601 // !IS_CLOBBER_REDIR(fn->type))
11602 // mode = O_WRONLY|O_APPEND|O_NOCTTY;
11603 // else mode = O_WRONLY|O_APPEND|O_CREAT|O_NOCTTY;
11604 // (IS_CLOBBER_REDIR — `>>!`/`>>|` — is currently flattened
11605 // to plain APPEND at compile time in
11606 // src/extensions/compile_zsh.rs:1654-1655, so the bang/pipe
11607 // forms can't be distinguished here yet.)
11608 let noclobber = opt_state_get("noclobber").unwrap_or(false)
11609 || !opt_state_get("clobber").unwrap_or(true);
11610 let append_create = opt_state_get("appendcreate").unwrap_or(false)
11611 || opt_state_get("append_create").unwrap_or(false);
11612 let open_result = if noclobber && !append_create {
11613 fs::OpenOptions::new().append(true).open(target) // no create
11614 } else {
11615 fs::OpenOptions::new()
11616 .create(true)
11617 .append(true)
11618 .open(target)
11619 };
11620 if !Self::redir_open_or_fail(fd, open_result, target, &mut self.redirect_failed) {
11621 self.set_last_status(1);
11622 }
11623 }
11624 r::READ => {
11625 if !Self::redir_open_or_fail(
11626 fd,
11627 fs::File::open(target),
11628 target,
11629 &mut self.redirect_failed,
11630 ) {
11631 self.set_last_status(1);
11632 }
11633 }
11634 r::READ_WRITE => {
11635 if let Ok(file) = fs::OpenOptions::new()
11636 .create(true)
11637 .truncate(false) // <> opens existing-or-new without truncating
11638 .read(true)
11639 .write(true)
11640 .open(target)
11641 {
11642 let new_fd = file.into_raw_fd();
11643 unsafe {
11644 // See redir_open_or_fail: when the opened fd IS the
11645 // destination (target fd was closed), keep it and clear
11646 // O_CLOEXEC; else dup2 + close.
11647 if new_fd != fd {
11648 libc::dup2(new_fd, fd);
11649 libc::close(new_fd);
11650 } else {
11651 libc::fcntl(fd, libc::F_SETFD, 0);
11652 }
11653 }
11654 }
11655 }
11656 r::DUP_READ | r::DUP_WRITE => {
11657 // Target is a numeric fd reference like `&3`. The parser
11658 // strips the `&` prefix before we get here in some paths,
11659 // others retain it — accept both. Also support `-` for
11660 // close-fd (`<&-` / `>&-`) per POSIX. The src_fd
11661 // validity check ran above before the save-and-dup.
11662 let n = target.trim_start_matches('&');
11663 if n == "-" {
11664 unsafe { libc::close(fd) };
11665 } else if n == "p" {
11666 // c:Src/exec.c — `<&p` / `>&p` route through the
11667 // coprocin / coprocout globals. zsh's `coproc CMD`
11668 // launch publishes those fds; the canonical
11669 // bin_print / bin_read `-p` arms already consume
11670 // them. The DUP redirect form is the third
11671 // consumer: it must dup the coproc fd onto the
11672 // target slot so the next command's stdin/stdout
11673 // is wired to the running coprocess. Bug #388.
11674 let coproc_fd = if op_byte == r::DUP_READ {
11675 crate::ported::modules::clone::coprocin
11676 .load(std::sync::atomic::Ordering::Relaxed)
11677 } else {
11678 crate::ported::modules::clone::coprocout
11679 .load(std::sync::atomic::Ordering::Relaxed)
11680 };
11681 if coproc_fd < 0 {
11682 eprintln!("{}:1: no coprocess", shname());
11683 self.set_last_status(1);
11684 self.redirect_failed = true;
11685 } else {
11686 unsafe {
11687 libc::dup2(coproc_fd, fd);
11688 }
11689 }
11690 } else if let Ok(src_fd) = n.parse::<i32>() {
11691 unsafe { libc::dup2(src_fd, fd) };
11692 } else if op_byte == r::DUP_WRITE {
11693 // c:Src/glob.c:2184-2187 xpandredir — a MERGEOUT
11694 // word that expands to a non-number becomes
11695 // REDIR_ERRWRITE: `cmd >& word` opens `word` and
11696 // routes BOTH fd 1 and fd 2 there. Reached only
11697 // for dynamic words (`>&$var`); static filenames
11698 // were converted at compile time.
11699 if let Ok(file) = fs::File::create(target) {
11700 let new_fd = file.into_raw_fd();
11701 unsafe {
11702 libc::dup2(new_fd, 1);
11703 libc::dup2(new_fd, 2);
11704 libc::close(new_fd);
11705 }
11706 }
11707 } else {
11708 // c:Src/glob.c:2185 — MERGEIN non-number:
11709 // `zerr("file number expected")`.
11710 crate::ported::utils::zerr("file number expected");
11711 self.set_last_status(1);
11712 self.redirect_failed = true;
11713 }
11714 }
11715 r::WRITE_BOTH => {
11716 if let Ok(file) = fs::File::create(target) {
11717 let new_fd = file.into_raw_fd();
11718 unsafe {
11719 libc::dup2(new_fd, 1);
11720 libc::dup2(new_fd, 2);
11721 libc::close(new_fd);
11722 }
11723 }
11724 }
11725 r::APPEND_BOTH => {
11726 if let Ok(file) = fs::OpenOptions::new()
11727 .create(true)
11728 .append(true)
11729 .open(target)
11730 {
11731 let new_fd = file.into_raw_fd();
11732 unsafe {
11733 libc::dup2(new_fd, 1);
11734 libc::dup2(new_fd, 2);
11735 libc::close(new_fd);
11736 }
11737 }
11738 }
11739 _ => {}
11740 }
11741 }
11742
11743 /// Push a fresh redirect scope. `_count` is informational — the actual
11744 /// saved fds are appended by host_apply_redirect into the top scope.
11745 pub fn host_redirect_scope_begin(&mut self, _count: u8) {
11746 // c:Src/exec.c:3722-3724 — the pipeline child set
11747 // `pipe_output_pending` right after dup2'ing its stdout onto
11748 // the pipe; the FIRST redirect scope opened in that child is
11749 // the stage command's own redirect list (same execcmd as the
11750 // pipe's addfd into mfds[1]). Capture the depth so only THAT
11751 // list's fd-1 write redirects MULTIOS-join the pipe.
11752 if self.pipe_output_pending {
11753 self.pipe_output_pending = false;
11754 self.pipe_output_scope = Some(self.redirect_scope_stack.len());
11755 }
11756 self.redirect_scope_stack.push(Vec::new());
11757 self.multios_scope_stack.push(Vec::new());
11758 }
11759
11760 /// Pop the top redirect scope, restoring saved fds.
11761 pub fn host_redirect_scope_end(&mut self) {
11762 // c:Src/exec.c — restore saved fds FIRST so the multios
11763 // pipe-write end is released from `fd`, then close our
11764 // tracked close_on_end (the last surviving writer dup), then
11765 // join the splitter thread. If we closed close_on_end before
11766 // restoring saved, `fd` would still hold a pipe writer and
11767 // the thread would block forever waiting for EOF.
11768 if let Some(saved) = self.redirect_scope_stack.pop() {
11769 for (fd, saved_fd) in saved.into_iter().rev() {
11770 unsafe {
11771 libc::dup2(saved_fd, fd);
11772 libc::close(saved_fd);
11773 }
11774 }
11775 }
11776 if let Some(scope) = self.multios_scope_stack.pop() {
11777 // Close ALL tracked writer dups BEFORE joining any
11778 // thread. When one splitter holds a dup of another's
11779 // pipe write-end (two multios in one scope where a later
11780 // one duped fd 1 while an earlier splitter owned it),
11781 // joining in push order deadlocks: splitter A's EOF
11782 // waits on splitter B's writer dup, which only closes
11783 // after B's thread exits — blocked behind A's join.
11784 let mut handles = Vec::with_capacity(scope.len());
11785 for (write_fd, handle) in scope {
11786 if write_fd >= 0 {
11787 unsafe {
11788 libc::close(write_fd);
11789 }
11790 }
11791 handles.push(handle);
11792 }
11793 for handle in handles {
11794 let _ = handle.join();
11795 }
11796 }
11797 // The scope that captured the pipeline-output marker is gone;
11798 // deeper-nested future scopes must not re-match its depth.
11799 if self.pipe_output_scope == Some(self.redirect_scope_stack.len()) {
11800 self.pipe_output_scope = None;
11801 }
11802 }
11803
11804 /// Set up `content` as stdin (fd 0) for the next command.
11805 /// Used by `Op::HereDoc(idx)` and `Op::HereString`.
11806 ///
11807 /// c:Src/exec.c:4655 getherestr — C writes the body to a TEMP
11808 /// FILE (gettempfile → write_loop → close → reopen O_RDONLY →
11809 /// unlink), NOT a pipe. The previous pipe+writer-thread shape
11810 /// SIGPIPE'd the whole shell when the consumer never read the
11811 /// body (`: <<< ${(F)x/y}` — D04parameter chunk 211, flaky
11812 /// rc=141): the redirect-scope teardown closed the read end
11813 /// while the detached thread was still in write_all, and the
11814 /// shell's SIGPIPE disposition is SIG_DFL. A temp file has no
11815 /// reader/writer coupling — matching C exactly, including
11816 /// lseek-ability of fd 0, which pipes don't give.
11817 pub fn host_set_pending_stdin(&mut self, content: String) {
11818 // c:4673 — `gettempfile(NULL, 1, &s)`.
11819 let mut tmp = std::env::temp_dir();
11820 tmp.push(format!(
11821 "zshrs-herestr-{}-{:x}",
11822 std::process::id(),
11823 std::time::SystemTime::now()
11824 .duration_since(std::time::UNIX_EPOCH)
11825 .map(|d| d.as_nanos())
11826 .unwrap_or(0)
11827 ));
11828 // c:4675 — `write_loop(fd, t, len); close(fd);`
11829 if std::fs::write(&tmp, content.as_bytes()).is_err() {
11830 return; // c:4674 — tempfile failure → no redirect
11831 }
11832 // c:Src/utils.c gettempfile → mkstemp creates the temp file mode
11833 // 0600 IGNORING the umask, so the O_RDONLY reopen below always
11834 // succeeds. `std::fs::write` honors the umask, so under `umask
11835 // 0777` the file landed mode 0000 and the reopen failed with
11836 // EACCES — `cat <<<x` then read empty stdin. Force 0600 to match
11837 // mkstemp's umask-independent permissions.
11838 let _ = std::fs::set_permissions(
11839 &tmp,
11840 <std::fs::Permissions as std::os::unix::fs::PermissionsExt>::from_mode(0o600),
11841 );
11842 // c:4677 — `fd = open(s, O_RDONLY | O_NOCTTY);`
11843 let file = match std::fs::File::open(&tmp) {
11844 Ok(f) => f,
11845 Err(_) => {
11846 let _ = std::fs::remove_file(&tmp);
11847 return;
11848 }
11849 };
11850 // c:4678 — `unlink(s);` — fd stays valid, name disappears.
11851 let _ = std::fs::remove_file(&tmp);
11852 let saved = unsafe { libc::dup(libc::STDIN_FILENO) };
11853 if saved >= 0 {
11854 if let Some(top) = self.redirect_scope_stack.last_mut() {
11855 top.push((libc::STDIN_FILENO, saved));
11856 } else {
11857 unsafe { libc::close(saved) };
11858 }
11859 }
11860 let read_fd = AsRawFd::as_raw_fd(&file);
11861 unsafe { libc::dup2(read_fd, libc::STDIN_FILENO) };
11862 drop(file);
11863 }
11864
11865 /// Spawn an external command using zshrs's full dispatch logic
11866 /// (intercepts, command_hash, redirect handling). Used by
11867 /// `ZshrsHost::exec` so the bytecode VM's `Op::Exec` and
11868 /// `Op::CallFunction` external fallback get the same semantics as
11869 /// the tree-walker's `execute_external` rather than a plain
11870 /// `Command::new` shortcut. Returns the exit status.
11871 pub fn host_exec_external(&mut self, args: &[String]) -> i32 {
11872 // If a glob expansion in this command's argv triggered the
11873 // nomatch error path, suppress the actual exec and return
11874 // status 1 — mirrors zsh's command-aborted-on-glob-error
11875 // behaviour. The flag is reset BEFORE returning so the next
11876 // command starts clean.
11877 //
11878 // c:Src/glob.c:1876-1880 + Src/exec.c — NOMATCH sets
11879 // ERRFLAG_ERROR but C's execlist clears the bit per-sublist
11880 // so subsequent commands run. Symmetric with the builtin
11881 // dispatcher's clear at fusevm_bridge.rs:299 — clear it here
11882 // too at the external-command post-command-boundary.
11883 consume_tilde_globsubst_carrier();
11884 if self.current_command_glob_failed.get() {
11885 self.current_command_glob_failed.set(false);
11886 crate::ported::utils::errflag.fetch_and(
11887 !crate::ported::zsh_h::ERRFLAG_ERROR,
11888 std::sync::atomic::Ordering::Relaxed,
11889 );
11890 self.set_last_status(1);
11891 return 1;
11892 }
11893 // c:Src/subst.c:505-507 — CSH_NULL_GLOB sibling of the
11894 // NOMATCH gate above, same external-path semantics (skip
11895 // command, `no match`, clear ERRFLAG so the next sublist
11896 // runs).
11897 if consume_badcshglob() {
11898 crate::ported::utils::errflag.fetch_and(
11899 !crate::ported::zsh_h::ERRFLAG_ERROR,
11900 std::sync::atomic::Ordering::Relaxed,
11901 );
11902 self.set_last_status(1);
11903 return 1;
11904 }
11905 let Some((cmd, rest)) = args.split_first() else {
11906 return 0;
11907 };
11908 // Empty command name (e.g. result of an empty `$(false)`
11909 // command-sub being the only word) — zsh: no command runs,
11910 // exit status preserved from prior step. Was hitting the
11911 // "command not found: " path with empty name.
11912 if cmd.is_empty() && rest.is_empty() {
11913 return self.last_status();
11914 }
11915 let rest_vec: Vec<String> = rest.to_vec();
11916 // Update `$_` with the just-arriving argv so the next command
11917 // reads `_=<last_arg>`. Mirrors C zsh's writeback in
11918 // `execcmd_exec` (Src/exec.c). Per `args.last()` semantics,
11919 // when invoked as `cmd a b c`, `$_` becomes "c" — for a bare
11920 // command with no args, `$_` becomes the command name itself.
11921 crate::ported::params::set_zunderscore(args);
11922
11923 // Builtins not in fusevm's name→id table fall through to
11924 // host.exec. Catch them here before the OS-level exec attempts
11925 // to spawn a non-existent binary.
11926 match cmd.as_str() {
11927 "sched" => return dispatch_builtin("sched", rest_vec.clone()),
11928 "echotc" => return dispatch_builtin("echotc", rest_vec.clone()),
11929 "echoti" => return dispatch_builtin("echoti", rest_vec.clone()),
11930 "zpty" => return dispatch_builtin("zpty", rest_vec.clone()),
11931 "ztcp" => return dispatch_builtin("ztcp", rest_vec.clone()),
11932 "zsocket" => {
11933 // c:Src/Modules/socket.c:276 BUILTIN spec — BUILTINS["zsocket"]
11934 // optstr "ad:ltv" parsed by execbuiltin.
11935 return dispatch_builtin("zsocket", rest_vec.clone());
11936 }
11937 "private" => {
11938 // c:Src/Modules/param_private.c:217 — bin_private via
11939 // BUILTINS["private"]. The autoload require_module
11940 // (exec.c:2700-2717) fires inside
11941 // dispatch_builtin_raw, the chokepoint for all routes.
11942 return dispatch_builtin("private", rest_vec.clone());
11943 }
11944 "zformat" => return dispatch_builtin("zformat", rest_vec.clone()),
11945 "zregexparse" => return dispatch_builtin("zregexparse", rest_vec.clone()),
11946 // `unalias`/`unhash`/`unfunction` share `bin_unhash` but
11947 // each carries its own funcid (BIN_UNALIAS / BIN_UNHASH /
11948 // BIN_UNFUNCTION) — dispatch_builtin handles the BUILTINS
11949 // lookup + funcid propagation via execbuiltin.
11950 "unalias" | "unhash" | "unfunction" => {
11951 return dispatch_builtin(cmd.as_str(), rest_vec.clone());
11952 }
11953 // ACTUALLY A ZSH FUNCTION: zmv/zcp/zln/zcalc are zsh autoload
11954 // functions — implemented natively in Rust so `autoload -Uz zmv`
11955 // works without shipping the function source (and without the
11956 // fpath source hanging the parser). The `function_exists` guard
11957 // keeps them command-not-found until autoloaded, exactly like zsh;
11958 // an un-guarded arm ran them for bare `zmv`, diverging from
11959 // `zsh -f; zmv` → "command not found: zmv".
11960 "zmv" if self.function_exists("zmv") => {
11961 return crate::extensions::ext_builtins::zmv(&rest_vec, "mv")
11962 }
11963 "zcp" if self.function_exists("zcp") => {
11964 return crate::extensions::ext_builtins::zmv(&rest_vec, "cp")
11965 }
11966 "zln" if self.function_exists("zln") => {
11967 return crate::extensions::ext_builtins::zmv(&rest_vec, "ln")
11968 }
11969 "zcalc" if self.function_exists("zcalc") => {
11970 return crate::extensions::ext_builtins::zcalc(&rest_vec)
11971 }
11972 "zselect" => {
11973 // Route through canonical dispatch_builtin which goes
11974 // via execbuiltin → BUILTINS["zselect"] (zselect.c:272).
11975 return dispatch_builtin("zselect", rest_vec.clone());
11976 }
11977 "cap" => return dispatch_builtin("cap", rest_vec.clone()),
11978 "getcap" => return dispatch_builtin("getcap", rest_vec.clone()),
11979 "setcap" => return dispatch_builtin("setcap", rest_vec.clone()),
11980 "yes" => return self.builtin_yes(&rest_vec),
11981 "nl" => return self.builtin_nl(&rest_vec),
11982 "env" => return self.builtin_env(&rest_vec),
11983 "printenv" => return self.builtin_printenv(&rest_vec),
11984 "tty" => return self.builtin_tty(&rest_vec),
11985 // c:Src/Modules/files.c:806 — BUILTINS["chgrp"] with
11986 // BIN_CHGRP funcid + "hRs" optstr.
11987 "chgrp" => return dispatch_builtin("chgrp", rest_vec.clone()),
11988 "nproc" => return self.builtin_nproc(&rest_vec),
11989 "expr" => return self.builtin_expr(&rest_vec),
11990 "sha256sum" => return self.builtin_sha256sum(&rest_vec),
11991 "base64" => return self.builtin_base64(&rest_vec),
11992 "tac" => return self.builtin_tac(&rest_vec),
11993 "expand" => return self.builtin_expand(&rest_vec),
11994 "unexpand" => return self.builtin_unexpand(&rest_vec),
11995 "paste" => return self.builtin_paste(&rest_vec),
11996 "fold" => return self.builtin_fold(&rest_vec),
11997 "shuf" => return self.builtin_shuf(&rest_vec),
11998 "comm" => return self.builtin_comm(&rest_vec),
11999 "cksum" => return self.builtin_cksum(&rest_vec),
12000 "factor" => return self.builtin_factor(&rest_vec),
12001 "tsort" => return self.builtin_tsort(&rest_vec),
12002 "sum" => return self.builtin_sum(&rest_vec),
12003 "mkfifo" => return self.builtin_mkfifo(&rest_vec),
12004 "link" => return self.builtin_link(&rest_vec),
12005 "unlink" => return self.builtin_unlink(&rest_vec),
12006 "dircolors" => return self.builtin_dircolors(&rest_vec),
12007 "groups" => return self.builtin_groups(&rest_vec),
12008 "arch" => return self.builtin_arch(&rest_vec),
12009 "nice" => return self.builtin_nice(&rest_vec),
12010 "logname" => return self.builtin_logname(&rest_vec),
12011 "tput" => return self.builtin_tput(&rest_vec),
12012 "users" => return self.builtin_users(&rest_vec),
12013 // "sync" => return self.bin_sync(&rest_vec),
12014 "zbuild" => return self.builtin_zbuild(&rest_vec),
12015 // `zf_*` aliases from `zsh/files` (Src/Modules/files.c
12016 // BUILTIN table at line 816-824). The C source binds
12017 // both unprefixed (`chmod`) and prefixed (`zf_chmod`)
12018 // names to the SAME `bin_chmod` etc. handlers — the
12019 // prefixed forms exist so a script can portably reach
12020 // the builtin even when a function or alias has shadowed
12021 // the bare name. Each arm routes through the canonical
12022 // zf_* aliases route through canonical BUILTINS entries
12023 // (files.c:816-824) — execbuiltin parses each fn's optstr
12024 // automatically.
12025 "mkdir" | "zf_mkdir" | "zf_rm" | "zf_rmdir" | "zf_chmod" | "zf_chown" | "zf_chgrp"
12026 | "zf_ln" | "zf_mv" | "zf_sync"
12027 // `--zsh` parity gate: zsh -fc has zsh/files UNLOADED
12028 // — bare `mkdir` is /bin/mkdir (so `command mkdir -p`
12029 // honors the system flag set; zconvey.plugin.zsh:44
12030 // got "File exists" from the in-process bin_mkdir
12031 // that this arm intercepted) and `zf_*` names are
12032 // command-not-found 127 until `zmodload zsh/files`.
12033 // Fall through to the external/exec path in --zsh
12034 // mode; default zshrs mode keeps the anti-fork
12035 // intercept.
12036 if !crate::IS_ZSH_MODE.load(std::sync::atomic::Ordering::Relaxed) =>
12037 {
12038 return dispatch_builtin(cmd.as_str(), rest_vec.clone());
12039 }
12040 // `zstat` — port of zsh/stat module (Src/Modules/stat.c
12041 // BUILTIN("zstat", …)). Returns file metadata as
12042 // `field value` pairs / an assoc / a plus-separated
12043 // list depending on flags. zsh ALSO registers `stat`
12044 // bound to the same handler, but that name conflicts
12045 // with the system `stat(1)` binary (every script that
12046 // calls `stat -f '%Lp' …` would break). zsh resolves
12047 // this through opt-in `zmodload`; zshrs's modules are
12048 // statically linked so we keep `stat` routing to the
12049 // external command and only intercept the unambiguous
12050 // `zstat` name.
12051 "zstat" => {
12052 // Canonical bin_stat per stat.c:638 via BUILTINS["zstat"].
12053 return dispatch_builtin("zstat", rest_vec.clone());
12054 }
12055 _ => {}
12056 }
12057
12058 // AOP intercepts: when an `intercept :before/:around/:after foo` block
12059 // is registered, dynamic-command-name dispatch must consult it before
12060 // spawning. Without this, `cmd=ls; $cmd` bypasses every intercept that
12061 // a literal `ls` would trigger. The full_cmd string mirrors what the
12062 // tree-walker era passed (cmd + args joined by space) so existing
12063 // pattern matchers continue to work.
12064 if !self.intercepts.is_empty() {
12065 let full_cmd = if rest_vec.is_empty() {
12066 cmd.clone()
12067 } else {
12068 format!("{} {}", cmd, rest_vec.join(" "))
12069 };
12070 if let Some(intercept_result) = self.run_intercepts(cmd, &full_cmd, &rest_vec) {
12071 return intercept_result.unwrap_or(127);
12072 }
12073 }
12074
12075 // User-defined function lookup before OS-level exec. zsh's
12076 // dynamic-command-name dispatch (`cmd=hook1; $cmd`) checks
12077 // the function table FIRST — without this, `$f` for a
12078 // function-name `f` was always falling through to
12079 // `execute_external` and erroring "command not found".
12080 // Plugin code uses this pattern constantly:
12081 // for f in "${precmd_functions[@]}"; do "$f"; done
12082 if self.function_exists(cmd) {
12083 if let Some(status) = self.dispatch_function_call(cmd, &rest_vec) {
12084 return status;
12085 }
12086 }
12087
12088 self.execute_external(cmd, &rest_vec, &[]).unwrap_or(127)
12089 }
12090}