zsh/vm_helper.rs
1//! Shell executor state for zshrs.
2//!
3//! ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
4//! !!! LAST-RESORT FILE — NOT FOR NEW LOGIC !!!
5//!
6//! This file holds the `ShellExecutor` runtime state struct + VM-adjacent
7//! helpers. It is **not** the place to add zsh logic — every line here that
8//! does real shell work is a tax we pay because zshrs uses fusevm bytecode
9//! instead of C zsh's wordcode walker.
10//!
11//! **Before adding code to this file, STOP and ask:**
12//!
13//! 1. Does the C source have a fn that does this? (Check `src/zsh/Src/*.c`)
14//! → Port it into `src/ported/<file>.rs` with line-by-line citations.
15//! Then call the canonical fn from here.
16//!
17//! 2. Does `src/ported/` already have a port?
18//! → Call it directly. Don't reimplement.
19//!
20//! 3. Is this purely a Rust-only state-struct accessor (getter/setter on
21//! ShellExecutor fields, VM init plumbing, executor-context guards)?
22//! → OK to put it here. Mark it `WARNING: RUST-ONLY HELPER` per memory
23//! `feedback_rust_only_helpers_need_warning`.
24//!
25//! **NEVER:** reinvent paramsubst/expansion/glob/typeset/redirect/scope
26//! management here. Every one of those has a canonical port in `src/ported/`.
27//! When a bridge-side fn grows past ~30 lines of shell logic, that's a
28//! signal the work belongs in `src/ported/` — port it, don't inline.
29//!
30//! This file should be SHRINKING over time. Every PR that adds lines here
31//! should justify it; every PR that moves lines OUT to `src/ported/` is
32//! aligned with the project direction.
33//!
34//! See also: memory `feedback_no_shortcuts_in_porting`, `feedback_true_port_pattern`,
35//! `feedback_no_shellexecutor_in_ported` (the inverse direction).
36//! ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
37//!
38//! **Not a port of Src/exec.c.** C zsh runs compiled programs on the native
39//! **wordcode walker** in `Src/exec.c` (`execlist` / `execpline` / `execcmd`).
40//! zshrs uses fusevm bytecode instead; the bridge lives in `src/fusevm_bridge.rs`.
41//! This file holds:
42//! - `ShellExecutor` — the runtime state struct that the VM and
43//! every ported builtin/utility threads through
44//! - VM-adjacent helpers that read/write that state
45//!
46//! Path-wise this file lives at the crate root (`src/vm_helper`) rather
47//! than in `src/ported/` because nothing here corresponds 1:1 to a
48//! `Src/*.c` source file. `crate::ported::exec` is kept as a
49//! re-export alias so existing call-sites continue to compile.
50
51use crate::compsys::cache::CompsysCache;
52use crate::compsys::CompInitResult;
53use crate::history::HistoryEngine;
54use crate::options::ZSH_OPTIONS_SET;
55use crate::ported::builtin::{BREAKS, CONTFLAG};
56use crate::ported::math::mathevali;
57use crate::ported::modules::parameter::*;
58use crate::ported::subst::singsub;
59
60thread_local! {
61 /// Eval-recursion depth counter — no C counterpart by design.
62 ///
63 /// !!! WARNING: RUST-ONLY BACKSTOP — reproduces C behaviour, no C fn !!!
64 ///
65 /// zsh bounds runaway `eval` recursion via its job table: every eval'd
66 /// list runs through `execpline`, which grabs a job slot per pipeline
67 /// (`initjob`), and the table caps at `MAX_MAXJOBS` → `zerr("job table
68 /// full or recursion limit exceeded")` (Src/jobs.c:1878-1884). The fusevm
69 /// runtime that executes eval bodies allocates no job per pipeline, and
70 /// nested evals push no funcstack frame (INEVAL suppression, matching zsh's
71 /// `if (!ineval)` at Src/builtin.c:6164), so neither the job table nor the
72 /// FUNCNEST/FUNCSTACK depth reflects eval nesting — leaving eval recursion
73 /// unbounded until the (256 MB but finite) main-thread stack overflows →
74 /// uncatchable SIGBUS. This counter is the Rust proxy for zsh's count of
75 /// concurrently-held job slots: `builtin.rs::eval` bumps it around each
76 /// eval body and refuses to recurse at the same `MAX_MAXJOBS` ceiling.
77 /// Lives here (not src/ported/) because it is an architectural Rust-only
78 /// backstop with no 1:1 C symbol.
79 pub static EVAL_RECURSION_DEPTH: std::cell::Cell<usize> = const { std::cell::Cell::new(0) };
80 /// Set by `execute_cmdarg` for the next compile_script_isolated only —
81 /// the `-c` string itself, not an `eval` or trap it runs.
82 ///
83 /// !!! WARNING: RUST-ONLY FLAG — C PASSES `exiting` TO execode !!!
84 static CMDARG_EXITING: std::cell::Cell<bool> = const { std::cell::Cell::new(false) };
85}
86
87/// O(1) single-key probe against the canonical hashed storage — the
88/// fast-path companion to subst.rs `assoc_get` for order-independent
89/// single-key reads. Returns `None` when `name` (post-nameref) isn't
90/// an assoc; otherwise `Some((key_present, value))` under ONE lock,
91/// with no whole-map clone and no zsh-bucket-order rebuild (that
92/// reorder is only observable to whole-map enumeration). Hot: shell
93/// loops reading `${assoc[$k]}` per iteration were O(n²) through
94/// assoc_get — zpwr expandstats over 42k records took 43s vs zsh's ~1s.
95/// Lives here (not src/ported/) because it has no C counterpart — C's
96/// getarg IS the single-key path.
97/// Classify an assoc subscript as an EXACT-key lookup and return the
98/// key: plain text (no leading flag group) passes through; a leading
99/// `(e…)`/`(E…)` group (c:Src/params.c:1449 — literal-key flag) is
100/// stripped. Search groups ((r)/(i)/(k)/…) return `None` — they need
101/// the full getarg walk. Companion gate for [`assoc_key_hit`]'s O(1)
102/// fast paths.
103/// !!! WARNING: RUST-ONLY HELPER — body of c:Src/init.c:180-215, lifted out
104/// of loop() so the script-file event loop (`run_events_per_command`) runs
105/// the same code; src/ported/ may not gain functions. !!!
106///
107/// The `toplevel` preexec branch of loop(): when a `preexec` function or
108/// `preexec_functions` exists, call the hook with `$1` (the history line, or
109/// "" when the history ring does not hold the current line — always the
110/// case for a script), `$2` (getjobtext) and `$3` (getpermtext), then clear
111/// ERRFLAG_ERROR (c:214). `event_src` is the event's source text captured
112/// around parse_event; zshrs's parse_event returns the fusevm AST, which the
113/// text renderers cannot walk, so the source is compiled to the wordcode
114/// Eprog C would already hold. That compile is silent and leaves errflag as
115/// it found it; an event it cannot take gets empty `$2`/`$3`.
116pub fn run_preexec_hook(event_src: Option<&str>) {
117 use crate::ported::zsh_h::{interact, isset, INTERACTIVECOMMENTS, SHINSTDIN};
118 // Native p10k engine command timer (src/extensions/p10k): stamp the
119 // command start for command_execution_time, with or without a user hook.
120 crate::p10k::note_exec_start(event_src);
121 let preexec_fn = crate::ported::utils::getshfunc("preexec"); // c:181
122 let preexec_hook = crate::ported::params::paramtab()
123 .read()
124 .ok()
125 .and_then(|t| {
126 t.get(&format!("preexec{}", crate::ported::zsh_h::HOOK_SUFFIX))
127 .map(|_| ())
128 }); // c:182 realparamtab->getnode2(realparamtab, "preexec" HOOK_SUFFIX)
129 if preexec_fn.is_none() && preexec_hook.is_none() {
130 return;
131 }
132 let mut args: Vec<String> = Vec::new(); // c:191 newlinklist()
133 args.push("preexec".to_string()); // c:192
134 {
135 // c:195 — `if (hist_ring && curline.histnum == curhist)`
136 let hr = crate::ported::hist::hist_ring.lock().unwrap();
137 let cl = crate::ported::hist::curline.lock().unwrap();
138 let same = !hr.is_empty()
139 && cl.as_ref().map(|c| c.histnum).unwrap_or(0)
140 == crate::ported::hist::curhist.load(Ordering::SeqCst);
141 if same {
142 args.push(hr.first().map(|h| h.node.nam.clone()).unwrap_or_default()); // c:196
143 } else {
144 args.push(String::new()); // c:198
145 }
146 }
147 let event_prog: Option<crate::ported::zsh_h::eprog> = event_src.and_then(|src| {
148 let noerrs = crate::ported::utils::noerrs_lock();
149 let saved_noerrs = std::mem::replace(&mut *noerrs.lock().unwrap(), 1);
150 let saved_errflag = errflag.load(Ordering::SeqCst);
151 // The event was lexed with `strin == 0`, where a `#` starts a comment
152 // only under INTERACTIVECOMMENTS in an interactive stdin shell
153 // (c:Src/lex.c:678-681). parse_string lexes under `strin`, which would
154 // turn comments on; `nocomments` restores that decision, the way
155 // getoutput does around its own parse_string (c:Src/exec.c:4720-4723).
156 let onc = crate::ported::lex::LEX_NOCOMMENTS.with(|c| c.get());
157 crate::ported::lex::LEX_NOCOMMENTS.with(|c| {
158 c.set(interact() && isset(SHINSTDIN) && !isset(INTERACTIVECOMMENTS))
159 });
160 let p = crate::ported::exec::parse_string(src, 0);
161 crate::ported::lex::LEX_NOCOMMENTS.with(|c| c.set(onc));
162 errflag.store(saved_errflag, Ordering::SeqCst);
163 *noerrs.lock().unwrap() = saved_noerrs;
164 p
165 });
166 // c:210 — addlinknode(args, dupstring(getjobtext(prog, NULL)))
167 // c:211 — addlinknode(args, cmdstr = getpermtext(prog, NULL, 0))
168 let (job_text, cmdstr) = match event_prog {
169 Some(p) => (
170 crate::ported::text::getjobtext(Box::new(p.clone()), None), // c:210
171 crate::ported::text::getpermtext(Box::new(p), None, 0), // c:211
172 ),
173 None => (String::new(), String::new()),
174 };
175 args.push(crate::ported::mem::dupstring(&job_text));
176 args.push(cmdstr.clone());
177 crate::ported::utils::callhookfunc("preexec", Some(&args), 1, std::ptr::null_mut()); // c:202
178 crate::ported::mem::zsfree(cmdstr); // c:205
179 errflag.fetch_and(!ERRFLAG_ERROR, Ordering::SeqCst); // c:214
180}
181
182pub fn exact_assoc_sub_key(sub: &str) -> Option<&str> {
183 match sub.strip_prefix('(') {
184 None => Some(sub),
185 Some(rest) => {
186 let close = rest.find(')')?;
187 let grp = &rest[..close];
188 if !grp.is_empty() && grp.chars().all(|c| c == 'e' || c == 'E') {
189 Some(&rest[close + 1..])
190 } else {
191 None
192 }
193 }
194 }
195}
196
197pub fn assoc_key_hit(name: &str, key: &str) -> Option<(bool, Option<String>)> {
198 let resolved = match crate::ported::params::resolve_nameref_name(name, None) {
199 crate::ported::params::nameref_resolution::Target { name: t_, .. } => t_,
200 _ => name.to_string(),
201 };
202 // c:Src/params.c:1090-1115 createparam — a `local NAME` / `typeset
203 // NAME` replaces a special's paramtab node with a plain one, so the
204 // name is no longer a hash. Answering here would strand the read on
205 // the O(1) assoc fast path and never reach paramsubst's scalar
206 // subscript arm (`${options[1]}` came back empty for a local).
207 //
208 // …but "no longer the magic row" is NOT the same as "no longer a
209 // hash": a `local -A commands` shadow owns a table of its own and the
210 // key read must be served FROM IT (c:Src/params.c:2270 / :1597-1606).
211 // See `magic_special_shadowed_by_nonhash`, which every sibling
212 // dispatch site shares so the qualification cannot drift.
213 if magic_special_shadowed_by_nonhash(resolved.as_str()) {
214 return None;
215 }
216 // c:Src/Zle/complete.c:1272/1411 — `compstate[nmatches]` is a LIVE gsu
217 // integer (`get_nmatches` = `permmatches(0) ? 0 : nmatches`), not stored
218 // data, so the hashed store never held it and every shell-side read
219 // returned the EMPTY string. `_parameters` (`local -i nm=$compstate[
220 // nmatches]` … `(( compstate[nmatches] > nm ))`) therefore always
221 // reported "added nothing" and returned 1 — `unset <TAB>` offered 197
222 // names against zsh's 496 — and the same idiom in `_alternative`,
223 // `_describe` and `_arguments` mis-fired the same way.
224 // The same applies to the other NINE gsu-backed rows
225 // (c:complete.c:1261-1300): list_lines, list_max, unambiguous,
226 // unambiguous_cursor, unambiguous_positions, insert_positions, vared,
227 // all_quotes, ignored. Only `nmatches` was served live here, so
228 // `$compstate[list_lines]` and friends read empty from shell code
229 // where zsh reports a value.
230 if resolved == "compstate" && crate::ported::zle::compcore::LIVE_COMPSTATE_KEYS.contains(&key) {
231 return Some((
232 true,
233 Some(
234 crate::ported::zle::compcore::get_compstate_str(key).unwrap_or_else(|| {
235 if key == "nmatches" {
236 "0".to_string()
237 } else {
238 String::new()
239 }
240 }),
241 ),
242 ));
243 }
244 crate::ported::params::paramtab_hashed_storage()
245 .lock()
246 .ok()
247 .and_then(|s| {
248 s.get(resolved.as_str())
249 .map(|m| (m.contains_key(key), m.get(key).cloned()))
250 })
251}
252use crate::ported::utils::{errflag, ERRFLAG_ERROR};
253use crate::ported::zsh_h::PM_UNDEFINED;
254use crate::ported::zsh_h::WC_SIMPLE;
255use crate::ported::zsh_h::{options, MAX_OPS};
256use crate::ported::zsh_h::{PM_ARRAY, PM_HASHED, PM_INTEGER, PM_READONLY};
257use parking_lot::Mutex;
258use std::collections::HashSet;
259use std::ffi::CStr;
260use std::ffi::CString;
261use std::ffi::OsStr;
262use std::fs;
263use std::io::Read;
264use std::os::unix::ffi::OsStrExt;
265use std::os::unix::fs::FileTypeExt;
266use std::os::unix::fs::PermissionsExt;
267use std::os::unix::io::FromRawFd;
268use std::sync::atomic::AtomicI32;
269use std::sync::atomic::Ordering;
270use std::time::{SystemTime, UNIX_EPOCH};
271use walkdir::WalkDir;
272
273// Backward-compat re-exports for free ported recently relocated to their
274// canonical-C-file Rust modules. Existing call-sites in this file (and
275// elsewhere) still reference these unqualified.
276#[allow(unused_imports)]
277pub(crate) use crate::func_body_fmt::FuncBodyFmt;
278#[allow(unused_imports)]
279pub(crate) use crate::ported::hist::bufferwords as bufferwords_z_tuple;
280#[allow(unused_imports)]
281pub(crate) use crate::ported::math::{parse_assign, parse_compound, parse_pre_inc};
282#[allow(unused_imports)]
283pub use crate::ported::params::convbase as format_int_in_base;
284pub use crate::ported::params::convbase_underscore;
285// `getarrvalue` is already re-exported by `pub use crate::ported::params::*`
286// below; an explicit `pub(crate) use` here only shadowed that public export.
287#[allow(unused_imports)]
288pub(crate) use crate::ported::utils::base64_decode;
289#[allow(unused_imports)]
290pub(crate) use crate::ported::utils::{ispwd, printprompt4, quotedzputs};
291
292pub(crate) use crate::intercepts::intercept_matches;
293/// AOP advice type — before, after, or around.
294pub use crate::intercepts::{AdviceKind, Intercept};
295
296/// Result from background compinit thread.
297pub use crate::compinit_bg::CompInitBgResult;
298use std::io::Write;
299use std::sync::LazyLock;
300
301/// State snapshot for plugin delta computation.
302pub(crate) use crate::plugin_cache::PluginSnapshot;
303
304/// Cached compiled regexes for hot paths
305pub(crate) static REGEX_CACHE: LazyLock<Mutex<HashMap<String, Regex>>> =
306 LazyLock::new(|| Mutex::new(HashMap::with_capacity(64)));
307
308// fusevm VM bridge (extension; not a port of Src/exec.c) lives in
309// src/fusevm_bridge.rs. Re-exports below let the rest of the codebase
310// reference symbols as `crate::ported::exec::X`.
311pub(crate) use crate::fusevm_bridge::ExecutorContext;
312pub use crate::fusevm_bridge::*;
313
314/// `ZSH_VERSION` / `ZSH_PATCHLEVEL` / `ZSH_VERSION_DATE` consts
315/// generated by `build.rs` from `src/zsh/Config/version.mk`. Use
316/// `zsh_version::ZSH_VERSION` etc. at call sites so version bumps
317/// pick up automatically.
318pub mod zsh_version {
319 include!(concat!(env!("OUT_DIR"), "/zsh_version.rs"));
320}
321
322/// Match an intercept pattern against a command name or full command string.
323/// Supports: exact match, glob ("git *", "_*", "*"), or "all".
324
325/// O(1) builtin-name lookup set derived from the canonical
326/// `BUILTINS` table (`src/ported/builtin.rs:122`, the 1:1 port of
327/// `static struct builtin builtins[]` at `Src/builtin.c:40-137`).
328/// The set is the union of: (1) names from `BUILTINS` (walked at first
329/// access), (2) zshrs daemon-side builtins from `ZSHRS_BUILTIN_NAMES`.
330/// Both arms run once at static init, so the set cannot drift from the
331/// canonical table.
332pub(crate) static BUILTIN_NAMES: LazyLock<HashSet<String>> = LazyLock::new(|| {
333 let mut s: HashSet<String> = HashSet::new();
334 // Walk the canonical `BUILTINS` table — the 1:1 port of
335 // `static struct builtin builtins[]` at `Src/builtin.c:40-137`
336 // (ported at `src/ported/builtin.rs:122`). Every name in there is
337 // a real zsh builtin; the set stays in sync as new ports land.
338 for b in crate::ported::builtin::BUILTINS.iter() {
339 s.insert(b.node.nam.clone());
340 }
341 // Daemon-side (zshrs-specific extensions).
342 for &n in crate::daemon::builtins::ZSHRS_BUILTIN_NAMES.iter() {
343 s.insert(n.to_string());
344 }
345 s
346});
347
348use crate::exec_jobs::{JobState, JobTable};
349use crate::parse::{Redirect, RedirectOp, ShellCommand, ShellWord, VarModifier, ZshParamFlag};
350use indexmap::IndexMap;
351use std::collections::HashMap;
352use std::env;
353use std::fs::{File, OpenOptions};
354use std::io;
355use std::path::{Path, PathBuf};
356use std::process::{Child, Command, Stdio};
357
358// Re-exports for call-sites that reference `crate::ported::exec::<Name>`.
359pub use crate::bash_complete::CompSpec;
360pub use crate::ported::builtin::AutoloadFlags;
361pub use crate::ported::modules::zutil::zstyle_entry;
362
363/// One inline-assignment scope (`X=foo Y=bar cmd`).
364///
365/// `saved` holds `(name, prev_var, prev_env)` for each name the
366/// PREFIX assignments touched, so END_INLINE_ENV can put the shell
367/// var and the process env back.
368///
369/// `recording` is what keeps the frame from swallowing assignments
370/// the *command itself* performs. zsh's `addvars()` (Src/exec.c:4142)
371/// walks only the parsed WC_ASSIGN chain, and `save_params`
372/// (Src/exec.c:4410) snapshots only those names; once the command
373/// runs, the save list is closed. The bytecode emits the prefix
374/// assignments between BEGIN_INLINE_ENV and SEAL_INLINE_ENV, and
375/// SEAL clears this flag, so `X=y . file` no longer records (and
376/// then reverts) every global the sourced file assigns.
377pub struct InlineEnvFrame {
378 /// Per-name pre-assignment state, one entry per prefix assignment.
379 pub saved: Vec<SavedInlineParam>,
380 /// True only while the prefix assignments are being executed.
381 pub recording: bool,
382 /// C's `ADDVAR_EXPORT` bit for this command (c:Src/exec.c:37), decided
383 /// where C decides it: `if (is_shfunc) flags |= ADDVAR_EXPORT`
384 /// (c:Src/exec.c:4142-4143), under the comment "Export this if the
385 /// command is a shell function, but not if it's a builtin"
386 /// (c:4138-4140). `addvars` reads it to run the prefix assignment with
387 /// `opts[ALLEXPORT]` forced on (c:2646-2651), which is what makes
388 /// `X=y shellfn` leave `X` marked exported for the duration of the
389 /// call while `X=y builtin` does not.
390 pub export: bool,
391}
392
393/// One parameter a prefix assignment displaced — the Rust twin of the
394/// `tpm` that C's `save_params` builds with
395/// `copyparam(tpm, pm, 0)` (c:Src/exec.c:4491-4493).
396pub struct SavedInlineParam {
397 /// The assigned name (C's `s`, c:Src/exec.c:4475).
398 pub name: String,
399 /// The WHOLE pre-assignment `Param` — type flags, base, width, level
400 /// and the value union, with the value read back through the type's
401 /// GSU getfn exactly as `copyparam` does (c:Src/params.c:1273-1291).
402 /// `None` when the name did not exist, which is C's remove-only arm
403 /// at c:Src/exec.c:4507-4508.
404 pub pm: Option<crate::ported::zsh_h::param>,
405 /// !!! RUST-ONLY FIELD — THE SECOND HALF OF c:Src/params.c:1289 !!!
406 /// C's `copyparam` copies an association's pairs into `tpm->u.hash`,
407 /// so they travel inside the `Param` above. zshrs keeps them in the
408 /// name-keyed `paramtab_hashed_storage` instead, OUTSIDE the `Param`
409 /// (the structural mismatch `stdunsetfn` documents at
410 /// params.rs:11044), so the pairs must be carried — and restored —
411 /// alongside the node rather than within it.
412 pub hashed: Option<IndexMap<String, String>>,
413 /// Pre-assignment `environ` entry for the name.
414 pub prev_env: Option<String>,
415}
416
417impl InlineEnvFrame {
418 /// New frame, open for recording until SEAL_INLINE_ENV runs.
419 pub fn new() -> Self {
420 Self {
421 saved: Vec::new(),
422 recording: true,
423 // c:4141 `int flags = 0;` — BEGIN_INLINE_ENV sets the
424 // ADDVAR_EXPORT bit once it has resolved the command word.
425 export: false,
426 }
427 }
428
429 /// Put every displaced parameter back — the frame's half of
430 /// `restore_params` (c:Src/exec.c:4519), run when the prefixed
431 /// command returns.
432 ///
433 /// Saves are pushed as the prefix assignments execute, so a name
434 /// assigned twice (`X=1 X=2 cmd`) has its ORIGINAL in the first
435 /// entry; walking in reverse makes that the last write, which is
436 /// what C's single pre-`addvars` snapshot achieves.
437 pub fn restore(self) {
438 for saved in self.saved.into_iter().rev() {
439 // c:4525-4531 (remove the temporary parameter) followed by
440 // c:4533-4570 (put the saved one back). C's `remove_p` carries
441 // the name in BOTH arms of `save_params` (c:4504 runs whether or
442 // not a `tpm` was made), so the name is always in the remove
443 // list and the restore list holds the snapshot only if there was
444 // one.
445 crate::ported::exec::restore_params(
446 saved.pm.into_iter().collect(),
447 vec![saved.name.clone()],
448 );
449 // !!! RUST-ONLY — see `SavedInlineParam::hashed` !!!
450 // c:4561's `tpm->gsu.h->setfn(tpm, pm->u.hash)` in zshrs terms.
451 // Unconditional: when nothing was saved the row must be gone, or
452 // the pairs the command itself wrote would outlive the frame.
453 if let Ok(mut store) = crate::ported::params::paramtab_hashed_storage().lock() {
454 match saved.hashed {
455 Some(pairs) => {
456 store.insert(saved.name.clone(), pairs);
457 }
458 None => {
459 store.remove(saved.name.as_str());
460 }
461 }
462 }
463 // c:4568-4569 — `if ((pm->node.flags & PM_EXPORTED) && ...)
464 // addenv(pm, s);`. zshrs's prefix-assignment path publishes the
465 // value with `zputenv` on the way in, so the way out is the
466 // matching restore of the pre-assignment `environ` entry.
467 match saved.prev_env {
468 Some(v) => std::env::set_var(&saved.name, &v),
469 None => std::env::remove_var(&saved.name),
470 }
471 }
472 }
473}
474
475impl Default for InlineEnvFrame {
476 fn default() -> Self {
477 Self::new()
478 }
479}
480
481/// Snapshot of subshell-isolated state. Captured at `(` entry, restored at
482/// `)` exit. zsh subshell semantics: assignments inside `(…)` don't leak to
483/// the outer scope — and that includes `export`. zsh forks a child for the
484/// subshell so the child's env::set_var dies with the child; without a fork
485/// (zshrs runs subshells in-process for perf), we snapshot+restore the OS
486/// env table around the subshell. Otherwise `(export y=v)` would leak `y`
487/// to the parent shell, breaking every script that uses a subshell to
488/// scope an env override.
489/// Snapshot of mutable executor state across a subshell
490/// boundary.
491/// Port of the `entersubsh()` save/restore Src/exec.c does at
492/// line 1084 — captures everything that must be replaced when a
493/// `(...)` group fires.
494pub struct SubshellSnapshot {
495 /// Snapshot of `paramtab` (the C-canonical parameter store) at
496 /// subshell entry. Step 1 of the unification mirrors writes to
497 /// paramtab, so subshell-scoped assignments now show up there
498 /// too — without this snapshot, restoring only `variables` /
499 /// `arrays` / `assoc_arrays` leaks the subshell's writes to the
500 /// parent via paramtab (e.g. `x=outer; (x=inner); echo $x` returned
501 /// `inner` because paramsubst reads through paramtab).
502 /// Same node storage as the live table (`Src/params.c:854`
503 /// `newparamtable(151, "paramtab")`) so restoring a snapshot restores
504 /// C's bucket-walk order too, not just the name→value mapping.
505 pub paramtab: crate::ported::hashtable::hashtable_nodes<crate::ported::zsh_h::Param>,
506 /// `paramtab_hashed_storage` field.
507 pub paramtab_hashed_storage: crate::cow_map::CowHashMap<String, IndexMap<String, String>>,
508 /// `positional_params` field.
509 pub positional_params: Vec<String>,
510 /// Values of the special parameters whose backing store is a process
511 /// GLOBAL rather than the parameter table — `Src/params.c`'s `char *ifs`
512 /// (IFS), `wordchars`, `home`, `histsiz`, … — each reached through a GSU
513 /// getfn/setfn pair (the dispatch list at params.rs:12548).
514 ///
515 /// The `paramtab` snapshot above restores the param NODE, but the node only
516 /// carries the GSU pair; the value itself lives in the global, which a
517 /// paramtab restore doesn't touch. C forks for `(...)`, so a child's writes
518 /// to those globals die with it. zshrs runs subshells in-process, so
519 /// `(IFS=,; :)` left the PARENT's IFS as `,` — and every later word-split
520 /// in the parent silently used it. Same fork-copy reasoning as `opts`
521 /// and `tables`.
522 pub special_globals: Vec<(String, String)>,
523 /// Flock fds (`Src/utils.c:2111` `addlockfd`) live at subshell
524 /// entry. `zsystem flock FILE` keeps the fd open for the life of
525 /// the shell; under C's forked `(...)` the child's fd — and hence
526 /// the lock — dies when the subshell exits. zshrs runs subshells
527 /// in-process, so the lock outlived the subshell and every later
528 /// `zsystem flock` on that file (from a real forked background job)
529 /// blocked forever. Recorded here so `subshell_end` can close the
530 /// fds the subshell itself opened.
531 pub flock_fds: Vec<i32>,
532 /// `loops` / `breaks` / `contflag` at subshell entry
533 /// (c:Src/loop.c, c:Src/builtin.c bin_break). C forks for `(...)`,
534 /// so a `break` executed inside dies with the child and the parent's
535 /// loop runs on: `for i in 1 2; do (break); print after; done` prints
536 /// `after` twice. zshrs runs subshells in-process, so the three
537 /// counters have to be restored by hand at the boundary.
538 /// Fourth slot: `ancestor_loops` (c:Src/loop.c:36), which entry moves
539 /// `loops` into (c:Src/exec.c:1263-1264, 55061).
540 pub loop_flags: (i32, i32, i32, i32),
541 /// `cmdoutval` (c:Src/exec.c:225) at subshell entry. A process global
542 /// the forked child owns a private copy of: `x=$(exit 3); (y=1); exec
543 /// 3>f` still reports 3 in zsh although the subshell's assignment reset
544 /// its own copy (c:2796). Restored by hand like `loop_flags`.
545 pub cmdoutval: i32,
546 /// `stopmsg` (c:Src/builtin.c) at subshell entry. Another process
547 /// global the child owns a copy of: a nofork `${ exit N }` in a
548 /// subshell sets it (c:Src/subst.c:2100) and the child then exits, so
549 /// in C it never reaches the parent. Leaked here it made every later
550 /// `exit` in a function or nofork skip its deferral (c:5884).
551 pub stopmsg: i32,
552 /// Parent's traps at subshell entry. zsh's `(trap "echo X" EXIT;
553 /// true)` runs the trap when the subshell exits — BEFORE the parent
554 /// continues. Without this snapshot, the trap inherited from parent
555 /// would fire, OR a trap set inside the subshell would leak to the
556 /// parent's process exit. Restored on subshell_end after the
557 /// subshell's own EXIT trap (if any) has fired. Stores a snapshot
558 /// of `crate::ported::builtin::traps_table()` (canonical).
559 pub traps: HashMap<String, String>,
560 /// Parent's shell options at subshell entry. `(set -e)` /
561 /// `(setopt extendedglob)` mustn't leak; zsh forks the subshell
562 /// so child options die with the child. We run in-process, so we
563 /// must restore the option store on subshell_end.
564 pub opts: HashMap<String, bool>,
565 /// The fork-copied hash tables — `aliastab`, `sufaliastab`, … — at
566 /// subshell entry (see `crate::ported::exec::SubshForkCopy`). zsh forks
567 /// for `(...)` so `(alias x=y)` inside a subshell dies with the child
568 /// and doesn't leak to the parent. Bug #209 in docs/BUGS.md. The
569 /// tables are restored whole — node flags (ALIAS_GLOBAL / DISABLED)
570 /// and bucket order included — which the `(name, text, flags)` list
571 /// this replaced got only partly right, and it never covered
572 /// `sufaliastab` at all: `(alias -s x=y)` leaked.
573 pub tables: crate::ported::exec::SubshForkCopy,
574 /// Parent's shell-function table at subshell entry. C zsh's
575 /// `entersubsh` (`Src/exec.c`) forks before running the
576 /// subshell body so `(f() { ... })` defining a function dies
577 /// with the child and never leaks to the parent. zshrs runs
578 /// subshells in-process, so we must clone `shfunctab` on entry
579 /// and restore on exit. Bug #208 in docs/BUGS.md. Stored as a
580 /// clone of the whole `shfunc_table` — the bucket layout is part
581 /// of the state, because `${(k)functions}` / `compadd -k functions`
582 /// emit C's bucket-walk order verbatim
583 /// (`Src/Modules/parameter.c:480-481`); rebuilding the table from
584 /// an unordered map on restore would reshuffle that order after
585 /// every `( … )` / `$( … )`.
586 pub shfuncs: std::sync::Arc<crate::ported::hashtable::shfunc_table>,
587 /// Parent's compiled-function chunks at subshell entry. Companion
588 /// to `shfuncs` above — `ShellExecutor.functions_compiled` is the
589 /// runtime dispatch table that `Op::CallFunction` reads through;
590 /// without restoring it, a subshell `(g() { override; })` leaves
591 /// the override bytecode chunk in place so the parent's
592 /// `g` call still runs the override after `subshell_end`
593 /// restored shfunctab. Bug #208 in docs/BUGS.md.
594 ///
595 /// Copy-on-write (`crate::cow_map`): taking this snapshot is a
596 /// refcount bump, and the deep copy happens only if the subshell
597 /// body actually compiles or redefines a function.
598 pub functions_compiled: crate::cow_map::CowHashMap<String, fusevm::Chunk>,
599 /// Parent's function source map at subshell entry. Companion to
600 /// `functions_compiled` so `typeset -f` / `whence` show the
601 /// parent's source after subshell exit, not the subshell's
602 /// overridden body. Bug #208 in docs/BUGS.md.
603 ///
604 /// Copy-on-write, for the same reason as `functions_compiled`.
605 pub function_source: crate::cow_map::CowHashMap<String, String>,
606 /// Parent's THINGYTAB (ZLE widget registry) at subshell entry.
607 /// zsh forks for `(...)` so `zle -N w f` / `zle -D w` inside the
608 /// subshell flip widget bindings only in the child; when the
609 /// child exits the parent's widget table is untouched. zshrs runs
610 /// subshells in-process so a subshell's `zle -D w` would
611 /// otherwise unbind the parent's widget. Bug #453 in docs/BUGS.md.
612 pub thingytab: HashMap<String, crate::ported::zle::zle_thingy::Thingy>,
613 /// Parent's KEYMAPNAMTAB (named keymap registry) at subshell
614 /// entry. Same fork-copy semantics as THINGYTAB — a subshell's
615 /// `bindkey -N km` / `bindkey -D km` mutates only the child's
616 /// keymap registry in C zsh. Bug #454 in docs/BUGS.md.
617 pub keymapnamtab:
618 crate::ported::hashtable::hashtable_nodes<crate::ported::zle::zle_keymap::KeymapName>,
619 /// Parent's `$!` (clone::lastpid) at subshell entry. C zsh forks
620 /// for `(...)`, so a background job started INSIDE the subshell
621 /// sets the child's `lastpid` only — `( : & ); echo $!` prints 0
622 /// in zsh. zshrs runs subshells in-process, so restore on end.
623 pub lastpid: i32,
624 /// Job-control state at subshell entry: (JOBTAB clone, CURJOB,
625 /// PREVJOB, MAXJOB, THISJOB). C zsh forks for `(...)` so any
626 /// `disown` / `wait` / new `&` job inside the subshell mutates the
627 /// CHILD's copy of jobtab and dies with it (Src/exec.c::entersubsh
628 /// fork semantics); the parent's table is untouched. zshrs's
629 /// in-process subshell must snapshot/restore to match — without
630 /// this, `sleep 1 & (disown); jobs` shows an empty table where
631 /// zsh still lists the job. Bug #462.
632 pub jobtab: Vec<crate::ported::zsh_h::job>,
633 /// `curjob` at subshell entry (Src/jobs.c:75 global).
634 pub curjob: i32,
635 /// `prevjob` at subshell entry (Src/jobs.c:80 global).
636 pub prevjob: i32,
637 /// `maxjob` at subshell entry (Src/jobs.c:71 global).
638 pub maxjob: usize,
639 /// `thisjob` at subshell entry (Src/jobs.c:77 global).
640 pub thisjob: i32,
641 /// `oldjobtab` / `oldmaxjob` at subshell entry (Src/jobs.c:101/104).
642 /// The subshell's `clearjobtab` (c:1808-1825) saves the parent's jobs
643 /// there; in C that copy lives in the forked child only, so the
644 /// parent's own `oldjobtab` must come back unchanged or its `jobs`,
645 /// `$jobstates` and `disown` would read the dead subshell's snapshot.
646 pub oldjobtab: Vec<crate::ported::zsh_h::job>,
647 pub oldmaxjob: usize,
648 /// The descriptors 0-9 a bare `exec` redirection in the body moves
649 /// are saved here on first touch and put back when this is dropped
650 /// (see `crate::ported::exec::SubshFdFrame`). C zsh forks for `(...)`
651 /// so the child's `exec >file` / `exec 3<&-` dies with it; without
652 /// this, `(exec >t.log; ...); cat t.log` left the PARENT's fd 1
653 /// pointing at t.log and `cat` looped forever copying the file into
654 /// itself.
655 pub fd_frame: crate::ported::exec::SubshFdFrame,
656 /// `redirect_scope_stack` depth at entry. A body aborted by an error
657 /// skips its `WithRedirectsEnd`; C's forked child takes those fds with
658 /// it, so subshell_end unwinds to here.
659 pub redir_depth: usize,
660 /// Pipeline job slots held at entry (`exec_jobs::execpline_slots_mark`).
661 /// Slots the body takes come from the cleared child table, so
662 /// subshell_end drops their records along with that table.
663 pub slots_mark: usize,
664 /// `sigtrapped[]` at subshell entry (Src/signals.c:39). C's
665 /// `entersubsh` clears per-signal trap STATE via `unsettrap(sig)`
666 /// (c:Src/exec.c:1088-1092), which zeroes both the body and the
667 /// sigtrapped flags. zshrs cleared only the body table, so the flags
668 /// desynced: a subshell that dropped a trap body still reported the
669 /// signal as trapped. Snapshot the whole vector so subshell_end can
670 /// restore the parent's exact state (including an inherited
671 /// ZSIG_IGNORED on SIGQUIT).
672 pub sigtrapped: Vec<i32>,
673 /// `subsh` at subshell entry (Src/exec.c:160 global). C's
674 /// `entersubsh` sets `subsh = 1` for a real (non-ESUB_FAKE)
675 /// subshell at c:Src/exec.c:1192-1193, and the forked child
676 /// carries it for the whole body. PRINT_EXIT_VALUE reads it
677 /// (c:4309 `&& !subsh`), which is why zsh prints nothing for
678 /// `setopt printexitvalue; (false)` while still reporting a
679 /// bare `false`. zshrs runs `( … )` in-process, so the flag has
680 /// to be set on entry and restored by hand on End.
681 pub subsh: i32,
682 /// The libc locale on entry, as `setlocale(LC_ALL, NULL)` reports it.
683 /// `LC_ALL=C` / `LANG=…` / `LC_CTYPE=…` inside the body reach libc via
684 /// `setlang` / `lcsetfn` (c:Src/params.c:4811, c:4877), which in C
685 /// changes only the forked child. The paramtab restore puts the
686 /// parameters back without re-running their setfns, so the parent kept
687 /// the child's locale: `() { (LC_ALL=C; true) }` left `${(q+)…}` quoting
688 /// multibyte text as raw bytes for the rest of the script.
689 pub locale: Option<std::ffi::CString>,
690}
691
692/// !!! WARNING: RUST-ONLY !!! C forks `( … )` and `$( … )`, so a locale the
693/// body sets dies with the child. zshrs runs both in-process; this records
694/// the parent's libc locale on entry for `subsh_locale_restore`.
695pub(crate) fn subsh_locale_save() -> Option<std::ffi::CString> {
696 let cur = unsafe { libc::setlocale(libc::LC_ALL, std::ptr::null()) };
697 (!cur.is_null()).then(|| unsafe { std::ffi::CStr::from_ptr(cur) }.to_owned())
698}
699
700/// !!! WARNING: RUST-ONLY !!! Put back the locale `subsh_locale_save`
701/// recorded, if the body changed it, and redo what the locale setfns do after
702/// `setlocale` (c:Src/params.c:4841 `clear_mbstate()`, c:4842 `inittyptab()`).
703/// Call it after the paramtab restore: inittyptab reads $IFS. The typtab is
704/// rebuilt only when the locale differs, so an ordinary subshell does not
705/// rewrite the table concurrent lexers read.
706pub(crate) fn subsh_locale_restore(saved: Option<&std::ffi::CStr>) {
707 let Some(loc) = saved else { return };
708 let cur = unsafe { libc::setlocale(libc::LC_ALL, std::ptr::null()) };
709 if !cur.is_null() && unsafe { std::ffi::CStr::from_ptr(cur) } == loc {
710 return;
711 }
712 unsafe { libc::setlocale(libc::LC_ALL, loc.as_ptr()) };
713 crate::ported::params::clear_mbstate();
714 crate::ported::utils::inittyptab();
715}
716
717#[allow(unused_imports)]
718pub(crate) use crate::ported::pattern::{
719 extract_numeric_ranges, numeric_range_contains, numeric_ranges_to_star,
720};
721
722/// Top-level shell executor state.
723/// Fork-equivalent event counter — incremented on every external
724/// spawn and in-process subshell entry. C zsh's `time` keyword
725/// reports only for JOBS (forked work, Src/jobs.c printtime via the
726/// job table); zshrs runs builtins/braces/functions in-process with
727/// no job, so the TIME_SUBLIST handler compares this counter across
728/// the timed body to decide whether to emit the report.
729pub static FORK_EVENTS: std::sync::atomic::AtomicU64 = std::sync::atomic::AtomicU64::new(0);
730
731/// Port of the file-static globals + `Estate` chain Src/exec.c
732/// uses — `execlist()` (line 1349) drives every list, with
733/// `execpline()` (line 1668), `execpline2()` (line 1991),
734/// `execsimple()` (line 1290), and the per-`WC_*` `execfuncs[]`
735/// table (line 268) feeding off it. The Rust port collapses
736/// everything into one `ShellExecutor` so we don't need
737/// thread-local globals.
738pub struct ShellExecutor {
739 /// Mirrors C zsh's file-static `scriptname` (Src/init.c). Used by
740 /// PS4's `%N` and the `scriptname:line: …` prefix on error
741 /// messages. Inside a function, MUTATES to the function name
742 /// (Src/exec.c:5903 `scriptname = dupstring(name)`). Init sets
743 /// this in `-c` mode to the binary basename per init.c:479; when
744 /// sourcing a file via `source`/`bin_dot`, it becomes the
745 /// resolved file path; otherwise it falls back through `$0` →
746 /// `$ZSH_ARGZERO`.
747 pub scriptname: Option<String>,
748 /// Mirrors C zsh's `scriptfilename` global (Src/init.c). Tracks
749 /// the FILE BEING READ (vs scriptname which tracks the active
750 /// function name during a call). Used by PS4's `%x` and certain
751 /// error-message prefixes that want the file location, NOT the
752 /// function name.
753 ///
754 /// At -c-mode init, scriptname == scriptfilename == "zsh"
755 /// (Src/init.c:479). When entering a function, ONLY scriptname
756 /// updates (exec.c:5903); scriptfilename stays at the outer
757 /// file path, so `%x` inside a function still shows the file
758 /// the function was called from.
759 pub scriptfilename: Option<String>,
760 /// Stack of subshell-state snapshots. Each `(…)` subshell pushes a copy
761 /// of variables/arrays/assoc_arrays at entry and pops/restores at exit.
762 /// Without this, `(x=inner; …); echo $x` shows `inner` instead of the
763 /// outer-scope value.
764 pub subshell_snapshots: Vec<SubshellSnapshot>,
765 /// Stack of inline-assignment scopes — `X=foo Y=bar cmd` pushes
766 /// a frame at the start, the assigns run inside it, and `cmd`
767 /// returns into END_INLINE_ENV which restores both shell-vars
768 /// and process-env to the pre-frame state. Each frame holds
769 /// `(name, prev_var, prev_env)` per assigned name. zsh's
770 /// equivalent is the parser-level "addvar" list executed under
771 /// `addvars()` (Src/exec.c) right before the command exec.
772 pub inline_env_stack: Vec<InlineEnvFrame>,
773 /// Set by `expand_glob`'s no-match arm when `nomatch` is on (zsh
774 /// default) — instructs the simple-command dispatcher to skip
775 /// executing the current command, set last_status=1, and continue
776 /// to the next command in the script. zsh's bin_simple uses the
777 /// errflag global for the same role: error printed, command
778 /// suppressed, script continues. Without this we were calling
779 /// `process::exit(1)` deep inside expand_glob, killing the whole
780 /// shell on any unmatched glob even with multi-statement input.
781 /// `Cell` because the no-match site only has a `&self` borrow.
782 pub current_command_glob_failed: std::cell::Cell<bool>,
783 /// `jobs` field.
784 pub jobs: JobTable,
785 /// `fpath` field.
786 pub fpath: Vec<PathBuf>,
787 /// SQLite history index, opened on FIRST USE rather than at startup.
788 ///
789 /// Only `--doctor`, the cache report and `dbview history` read it, and
790 /// none of them runs on a script or `-c` invocation — but opening it
791 /// eagerly cost every `zshrs script.zsh` an sqlite open plus
792 /// `init_schema` (0.40 ms of a 1.75 ms startup, measured with
793 /// `ZSHRS_STARTUP_TRACE`). A pre-filled `None` is how a pool worker
794 /// declares it must never open one.
795 pub history: std::cell::OnceCell<Option<HistoryEngine>>,
796 pub(crate) process_sub_counter: u32,
797 pub completions: HashMap<String, CompSpec>, // command -> completion spec
798 pub zstyles: Vec<zstyle_entry>, // zstyle configurations
799 /// Current function scope depth for `local` tracking.
800 pub local_scope_depth: usize,
801 /// Last arg of the currently-running command, deferred into `$_`
802 /// when the next command dispatches. zsh: `$_` reflects the LAST
803 /// command's last arg, so `echo hi; echo $_` prints `hi` (not the
804 /// `_` arg of `echo $_` itself). Promoted in `pop_args` and
805 /// `host.exec` before the command's args are read.
806 pub pending_underscore: Option<String>,
807 /// True while expanding inside a double-quoted context. Set by
808 /// `BUILTIN_EXPAND_TEXT` mode 1 around `expand_string` calls.
809 /// Used by parameter-flag application to suppress array-only flags
810 /// (`(o)`/`(O)`/`(n)`/`(i)`/`(M)`/`(u)`) — zsh's behaviour: those
811 /// flags only fire in array context.
812 pub in_dq_context: u32,
813 /// True (>0) while expanding the RHS of a scalar assignment.
814 /// Direct port of zsh's `PREFORK_SINGLE` bit set by
815 /// Src/exec.c::addvars line 2546 (`prefork(vl, isstr ?
816 /// (PREFORK_SINGLE|PREFORK_ASSIGN) : PREFORK_ASSIGN, ...)`).
817 /// Subst_port's paramsubst reads this via `ssub` and suppresses
818 /// `(f)` / `(s:STR:)` / `(0)` / `(z)` split flags per
819 /// Src/subst.c:1759 + 3902, so `y="${(f)x}"` preserves x's
820 /// original separator (newlines) instead of re-joining with
821 /// IFS-first-char (space).
822 pub in_scalar_assign: u32,
823 /// `profiling_enabled` field.
824 pub profiling_enabled: bool,
825 // compsys - completion system cache
826 /// `compsys_cache` field.
827 /// SQLite mirror, opened on FIRST USE via [`ShellExecutor::compsys_cache`].
828 ///
829 /// It is a dbview/FTS mirror for inspection — the authoritative completion
830 /// cache is the rkyv shards — so nothing on a normal command path touches
831 /// it. Opening it in the constructor still cost every shell three file
832 /// opens (`compsys.db`, `-wal`, `-shm`) plus WAL setup, including
833 /// `zshrs -f -c exit`, which cannot consult it at all.
834 pub compsys_cache: std::cell::OnceCell<Option<CompsysCache>>,
835 // Background compinit — receiver for async fpath scan result
836 /// `compinit_pending` field.
837 pub compinit_pending: Option<(
838 std::sync::mpsc::Receiver<CompInitBgResult>,
839 std::time::Instant,
840 )>,
841 // Plugin source cache — stores side effects of source/. in SQLite
842 /// `plugin_cache` field.
843 pub plugin_cache: Option<crate::plugin_cache::PluginCache>,
844 // cdreplay - deferred compdef calls for zinit turbo mode
845 /// `deferred_compdefs` field.
846 pub deferred_compdefs: Vec<Vec<String>>,
847 // Control flow signals
848 pub returning: Option<i32>, // Set by return builtin, cleared after function returns
849 /// zsh compatibility mode - use .zcompdump, fpath scanning, etc.
850 /// Also serves as the `--zsh` parity-test flag: caches off, daemon
851 /// off, plugin_cache replay off so every `source` re-runs the file
852 /// fresh per Src/builtin.c:6080-6123 bin_dot semantics.
853 pub zsh_compat: bool,
854 /// bash compatibility mode (`--bash`). Same parity-mode semantics
855 /// as `zsh_compat` (caches/daemon/replay off) plus bash-specific
856 /// behavior tweaks where bash 5.x diverges from zsh — e.g.
857 /// `BASH_VERSION` / `BASH_REMATCH` exposed, `[[ =~ ]]` populates
858 /// match indices the bash way, mapfile/readarray as builtins.
859 pub bash_compat: bool,
860 /// POSIX sh strict mode — no SQLite, no worker pool, no zsh extensions
861 pub posix_mode: bool,
862 /// Worker thread pool for background tasks (compinit, process subs, etc.)
863 pub worker_pool: std::sync::Arc<crate::worker::WorkerPool>,
864 /// AOP intercept table: command/function name → advice chain.
865 /// Glob patterns supported (e.g. "git *", "*").
866 pub intercepts: Vec<Intercept>,
867 /// Async job handles: id → receiver for (status, stdout)
868 pub async_jobs: HashMap<u32, crossbeam_channel::Receiver<(i32, String)>>,
869 /// Next async job ID
870 pub next_async_id: u32,
871 /// Per-scope saved-fd stacks for `Op::WithRedirectsBegin/End`. Each entry
872 /// is a Vec of (fd, saved_dup_fd) pairs taken from `dup(fd)` before the
873 /// redirect was applied; `with_redirects_end` `dup2`s them back and closes.
874 pub redirect_scope_stack: Vec<Vec<(i32, i32)>>,
875 /// Per-scope MULTIOS tee state. Each entry is `(pipe_write_fd,
876 /// JoinHandle)`: the pipe write-end currently dup2'd onto the
877 /// command's fd, and the splitter thread that reads from the
878 /// pipe read-end and writes to every collected target. Closed
879 /// + joined by `host_redirect_scope_end` BEFORE the saved fds
880 /// are restored so the splitter drains every byte the body
881 /// wrote into the pipe. Bug #36 in docs/BUGS.md.
882 pub multios_scope_stack: Vec<Vec<(i32, std::thread::JoinHandle<()>)>>,
883 /// True while applying a bare `exec`'s redirect list (`exec 1>&-`,
884 /// `exec 2>/dev/null` — no command words). `host_apply_redirect`
885 /// then skips pushing the saved fd into the enclosing scope so the
886 /// fd change survives group/command teardown.
887 /// c:Src/exec.c:3978-3986 — nullexec==1: "we specifically *don't*
888 /// restore the original fd's before returning"; C's per-execcmd
889 /// `save[]` means exec's redirs never enter the enclosing group's
890 /// save list either. Toggled by `BUILTIN_EXEC_PERM_REDIRS`.
891 pub exec_redirs_permanent: bool,
892 /// Set in a forked pipeline-stage child right after its stdout is
893 /// dup2'd onto the pipe write-end. Consumed by the FIRST
894 /// `host_redirect_scope_begin` (the stage command's own redirect
895 /// list) into `pipe_output_scope`.
896 /// c:Src/exec.c:3722-3724 — `addfd(forked, save, mfds, 1, output,
897 /// 1, NULL)`: the pipe occupies mfds[1] in the SAME execcmd that
898 /// processes the stage command's redirect list.
899 pub pipe_output_pending: bool,
900 /// Index into `redirect_scope_stack` of the scope whose redirect
901 /// list shares an execcmd with the pipeline output on fd 1. A
902 /// write-side redirect of fd 1 applied at exactly this scope depth
903 /// MULTIOS-splits (tees) instead of replacing — c:Src/exec.c:
904 /// 2447-2480 addfd "split the stream". Cleared when that scope ends.
905 pub pipe_output_scope: Option<usize>,
906 /// Set by `host_apply_redirect` when a redirect target couldn't be
907 /// opened (permission denied, no such directory, etc). The next
908 /// builtin/command checks this at entry and short-circuits with
909 /// status 1 instead of running. Mirrors zsh's "command skip" on
910 /// redirect failure.
911 pub redirect_failed: bool,
912 /// Compiled function bodies — name → fusevm::Chunk. Populated by
913 /// `BUILTIN_REGISTER_FUNCTION` (from `FunctionDef` lowering) and lazily by
914 /// `ZshrsHost::call_function` when only an AST exists in `self.functions`
915 /// (autoloaded, sourced, etc.). `Op::CallFunction` dispatches through here.
916 ///
917 /// COPY-ON-WRITE. `$( … )` and `( … )` snapshot this whole map for
918 /// subshell isolation (Bug #208), so a plain `HashMap` made entering
919 /// a substitution cost one `Chunk` clone per compiled function. It
920 /// is a `CowHashMap`, so the snapshot shares and the first write
921 /// inside the body splits. Reads and writes are unchanged —
922 /// `Deref`/`DerefMut` keep every call site as it was.
923 pub functions_compiled: crate::cow_map::CowHashMap<String, fusevm::Chunk>,
924 /// Canonical source text for functions. Populated by autoload paths (the
925 /// raw file/cache body), runtime FuncDef compile (the parsed source span),
926 /// and `unfunction` removal. Used by introspection (`whence`, `which`,
927 /// `typeset -f`) instead of reconstructing from a ShellCommand AST. When a
928 /// function is in `functions_compiled` but not here, introspection falls
929 /// back to `text::getpermtext(self.functions[name])`.
930 ///
931 /// Copy-on-write for the same reason as `functions_compiled`: a
932 /// subshell snapshots it whole, and the body usually never writes.
933 pub function_source: crate::cow_map::CowHashMap<String, String>,
934 /// `first_body_line - 1` per compiled function — matches inner
935 /// `ZshCompiler::lineno_offset` / zsh `funcstack->flineno` combined with
936 /// relative `$LINENO` for Src/prompt.c:909 `%I`.
937 pub function_line_base: HashMap<String, i64>,
938 /// `scriptfilename` when `BUILTIN_REGISTER_COMPILED_FN` ran — `%x` inside
939 /// a function (prompt.c:931-934) reads `funcstack->filename`.
940 pub function_def_file: HashMap<String, Option<String>>,
941 /// Innermost-last stack of active compiled-call frames for prompt `%I` / `%x`.
942 pub prompt_funcstack: Vec<(String, i64, Option<String>)>,
943 /// Scalar→(array, sep) tie table set up by `typeset -T VAR var [SEP]`.
944 /// Array→(scalar, sep) reverse-tie table. Used by BUILTIN_SET_ARRAY to
945 /// join the array elements with `sep` and mirror to the scalar side.
946 pub tied_array_to_scalar: HashMap<String, (String, String)>,
947
948 // ── ztest framework counters (extensions/ztest.rs) ──────────────────
949 //
950 // Mirrors strykelang's per-VMHelper test counters
951 // (strykelang/builtins.rs:22292-22308 + builtins.rs::test_pass/_fail/_skip,
952 // builtins.rs::test_pass_count etc.). Each `zassert_*` builtin bumps
953 // the per-block counter; `ztest_run` rolls per-block into _total and
954 // resets the per-block side, so a single test file with multiple
955 // `ztest_run` calls can reuse the counters. The worker-pool runner in
956 // src/extensions/ztest.rs reads pass_total+pass_count and
957 // fail_total+fail_count after `execute_script` returns for the cumulative
958 // numbers (strykelang/cli_runners.rs:115-118). `ztest_run_failed` is a
959 // sticky bool the runner reads so a test that asserts but then exits
960 // 0 still flags as failed. `ztest_suppress_stdout` matches
961 // VMHelper::suppress_stdout — the runner sets it inside the forked
962 // grandchild so the per-test stderr capture stays clean.
963 /// Per-block pass count (reset by `ztest_run`).
964 pub ztest_pass_count: std::sync::atomic::AtomicUsize,
965 /// Per-block fail count (reset by `ztest_run`).
966 pub ztest_fail_count: std::sync::atomic::AtomicUsize,
967 /// Per-block skip count (reset by `ztest_run`).
968 pub ztest_skip_count: std::sync::atomic::AtomicUsize,
969 /// Cumulative pass total across the run.
970 pub ztest_pass_total: std::sync::atomic::AtomicUsize,
971 /// Cumulative fail total across the run.
972 pub ztest_fail_total: std::sync::atomic::AtomicUsize,
973 /// Cumulative skip total across the run.
974 pub ztest_skip_total: std::sync::atomic::AtomicUsize,
975 /// Sticky failure flag — set by any `ztest_run` that observed fails;
976 /// the CLI runner reads this so a test that asserts then exits 0
977 /// still counts as a failed file.
978 pub ztest_run_failed: std::sync::atomic::AtomicBool,
979 /// Suppress per-assertion `✓`/`✗` lines on stderr. Set by the worker
980 /// runner inside the forked child when it has already redirected
981 /// fd 2 to a tmp file (we still want the lines, but only after the
982 /// runner re-emits them under print_lock to avoid line-tearing).
983 pub ztest_suppress_stdout: bool,
984}
985
986/// Context-isolated nested parse — the AST-path bridge for C's
987/// `parse_string` (`Src/exec.c:283`). zshrs executes via the ZshProgram
988/// AST + `compile_zsh`, not wordcode, so this can't live in `src/ported/`
989/// (C's `parse_string` returns `Eprog`, and the build.rs port-gate rejects
990/// any non-C-named fn there). It's the bridge that wraps the AST
991/// `parse_init`+`parse` in the SAME isolation `parse_string` provides.
992///
993/// Why this exists: a runtime parse — command-substitution body,
994/// process-substitution argv — must not clobber the outer
995/// `loop()`/`parse_event` reader's live input when it interleaves parsing
996/// with execution (faithful single-event mode).
997///
998/// The load-bearing piece is `strinbeg(0)`/`strinend()` (c:290/298). It
999/// sets the `strin` flag so that when the nested lexer drains `cmd_str`,
1000/// `ingetc` returns EOF (input.rs:391) instead of falling through to
1001/// `inputline()`, which would STEAL the outer reader's next SHIN line —
1002/// e.g. `echo A` / `v=$(echo hi)` / `echo B` had the cmd-subst swallow
1003/// `echo B` off stdin, and the outer loop then hit EOF after one command.
1004/// `strinbeg` also runs `hbegin`/`lexinit`, so it must execute on isolated
1005/// history+lexer state — hence the surrounding `zcontext_save`/`restore`
1006/// (c:288/300), which saves+restores `tok`/`tokstr`/`lexbuf`/`isnewlin`/
1007/// `incmdpos`/heredocs/`lexstop`/`toklineno`/history.
1008///
1009/// Two zshrs-specific globals `zcontext` doesn't cover are saved here too:
1010/// the lexer-input window (`LEX_INPUT`/`LEX_POS`/`LEX_UNGET_BUF`) that
1011/// `lex_init` overwrites with `cmd_str`, and the line counter `LEX_LINENO`
1012/// (C saves `oldlineno` explicitly at parse_string c:291/295).
1013///
1014/// NOTE: using `inpush` (the literal C input stack) instead does NOT work
1015/// in zshrs's hybrid input model — the outer piped reader pulls from SHIN
1016/// via `inputline`, and pushing/popping the `inbuf` stack severs that
1017/// continuation. The `LEX_INPUT` window + `strin` flag is the working
1018/// equivalent.
1019pub(crate) fn parse_isolated(input: &str) -> crate::parse::ZshProgram {
1020 parse_isolated_at(input, None)
1021}
1022
1023/// `parse_isolated` with C's `reset_lineno` argument inverted: `Some(n)` is
1024/// `parse_string(s, 0)` (c:Src/exec.c:291-293), which parses from the line the
1025/// shell is executing, `n`, instead of from line 1. A parse error in a body
1026/// parsed at run time (`` `…` `` in getoutput c:4778) then carries that line,
1027/// or none at all when it is 0 (c:Src/utils.c:301).
1028pub(crate) fn parse_isolated_at(input: &str, lineno: Option<u64>) -> crate::parse::ZshProgram {
1029 use crate::ported::lex::{
1030 tok, LEXERR, LEX_INPUT, LEX_LINENO, LEX_POS, LEX_UNGET_BUF, LEX_FILE_WINDOW_STRIN,
1031 };
1032
1033 // Inline Rust FFI: rewrite every `rust { ... }` block into a
1034 // `__rust_compile '<base64>' <line>` command before it reaches the lexer.
1035 // This is the shared source-string chokepoint for `-c`, script files, and
1036 // nested (command/process-substitution) parses. The `.contains("rust")`
1037 // gate keeps the common case (no FFI block) allocation-free — the vast
1038 // majority of nested parses never mention `rust`.
1039 let ffi_desugared = input
1040 .contains("rust")
1041 .then(|| crate::rust_ffi::desugar(input));
1042 let input: &str = ffi_desugared.as_deref().unwrap_or(input);
1043
1044 crate::ported::context::zcontext_save(); // c:288
1045 // Save the zshrs-specific lexer window + line counter that lex_init
1046 // overwrites but zcontext doesn't cover.
1047 let saved_input = LEX_INPUT.with_borrow(|s| s.clone());
1048 let saved_pos = LEX_POS.get();
1049 let saved_unget = LEX_UNGET_BUF.with_borrow(|b| b.clone());
1050 let saved_lineno = LEX_LINENO.get(); // c:291 oldlineno
1051 // The nested parse installs its own window; `lex_init` marks it a
1052 // string unit. Put the outer window's kind back with the window.
1053 let saved_file_window = LEX_FILE_WINDOW_STRIN.get();
1054 // input.rs `lexstop` is the input-side half of C's single `lexstop`;
1055 // draining the nested LEX_INPUT sets it true and zcontext only covers
1056 // the lex.rs half (LEX_LEXSTOP). Restore it so the outer reader isn't
1057 // left at EOF.
1058 let saved_in_lexstop = crate::ported::input::lexstop.with(|c| c.get());
1059
1060 // c:Src/exec.c:294 `inpush(s, INP_LINENO, NULL)` — the nested parse
1061 // gets its own input frame, so whatever the OUTER reader still has
1062 // queued is parked on the input stack until the inpop below. The text
1063 // itself is lexed from LEX_INPUT (parse_init), so the pushed frame is
1064 // empty: once that window drains, ingetc hits the end of THIS frame
1065 // and returns EOF. Without the frame it fell through to the outer
1066 // buffer — the rest of a multi-line command line, still unread while
1067 // `preexec` ran an `eval` or `$(…)` — and swallowed it.
1068 crate::ported::input::inpush("", crate::ported::zsh_h::INP_LINENO, None);
1069 crate::ported::hist::strinbeg(0); // c:290 — strin++ → drained nested input EOFs (no SHIN steal)
1070 crate::ported::parse::parse_init(input); // install cmd_str as LEX_INPUT (lex_init), LEX_LINENO=1
1071 if let Some(n) = lineno {
1072 LEX_LINENO.set(n); // c:292 `if (reset_lineno)` not taken: keep the running line
1073 }
1074 let program = crate::ported::parse::parse(); // c:294 (AST analog of par_list)
1075
1076 // Capture parse failure BEFORE the restores wipe the signals.
1077 let parse_err = (errflag.load(Ordering::Relaxed) & ERRFLAG_ERROR) != 0 || tok() == LEXERR;
1078 if tok() == LEXERR && crate::ported::builtin::LASTVAL.load(Ordering::Relaxed) == 0 {
1079 crate::ported::builtin::LASTVAL.store(1, Ordering::Relaxed); // c:296-297
1080 }
1081
1082 crate::ported::hist::strinend(); // c:298 — strin--
1083 // c:Src/exec.c:304 `inpop()` — pops the pushed frame AND its continuations. An
1084 // alias expanded at the end of the body sits on top as an
1085 // INP_ALIAS|INP_CONT frame (c:Src/input.c:695), so this is also where its
1086 // `inuse` is cleared (c:Src/input.c:773): left behind, the alias stayed
1087 // `inuse` and every later `$(m)` / `m` reported "command not found".
1088 // Popping the frame pushed above restores the outer reader's queue.
1089 crate::ported::input::inpop();
1090 // Restore the zshrs window, then the token/parse/history state.
1091 LEX_INPUT.with_borrow_mut(|s| *s = saved_input);
1092 LEX_POS.set(saved_pos);
1093 LEX_UNGET_BUF.with_borrow_mut(|b| *b = saved_unget);
1094 LEX_LINENO.set(saved_lineno); // c:295
1095 LEX_FILE_WINDOW_STRIN.set(saved_file_window);
1096 crate::ported::input::lexstop.with(|c| c.set(saved_in_lexstop));
1097 crate::ported::context::zcontext_restore(); // c:300
1098 // zcontext_restore → parse_context_restore clears ERRFLAG_ERROR
1099 // (parse.c:354); re-raise so callers gating on the bit still see it.
1100 if parse_err {
1101 errflag.fetch_or(ERRFLAG_ERROR, Ordering::Relaxed);
1102 }
1103 program
1104}
1105
1106/// Build the `scriptname[:lineno]` prefix zsh puts on an execution error.
1107///
1108/// c:Src/utils.c:301 — `zerrmsg` prints the line number ONLY when it is
1109/// non-zero: `if ((unset(SHINSTDIN) || locallevel) && lineno) fprintf(file,
1110/// "%lld: ", lineno);`. The command-not-found / no-such-file / permission-denied
1111/// sites below emit DIRECTLY rather than through `zerr` (deliberately — see
1112/// their comments, routing through zerr would set errflag and abort a script
1113/// that zsh continues), but they hand-rolled `"{}:{}"` and so printed a bare
1114/// `:0` inside a one-line function where zsh prints no line number at all:
1115/// `f(){ nosuchcmd }; f` gave `f:0: command not found:` vs zsh's `f: command
1116/// not found:`. Mirrors the C condition exactly. Bug #1070.
1117fn zerr_prefix(sn: &str) -> String {
1118 let lineno = crate::ported::lex::lineno();
1119 let ll = crate::ported::params::locallevel.load(std::sync::atomic::Ordering::Relaxed);
1120 if (crate::ported::zsh_h::unset(crate::ported::zsh_h::SHINSTDIN) || ll != 0) && lineno != 0 {
1121 format!("{}:{}", sn, lineno)
1122 } else {
1123 sn.to_string()
1124 }
1125}
1126
1127thread_local! {
1128 /// `(function name, its source file)` while that function's AUTOLOAD body
1129 /// is being run. Empty at every other moment.
1130 pub static AUTOLOAD_DEF_FILE: std::cell::RefCell<Vec<(String, String)>> =
1131 const { std::cell::RefCell::new(Vec::new()) };
1132}
1133
1134/// Scope guard for [`AUTOLOAD_DEF_FILE`].
1135pub struct AutoloadFileGuard(bool);
1136
1137impl AutoloadFileGuard {
1138 fn enter(name: &str) -> Self {
1139 match crate::ported::hashtable::getshfuncfile(name) {
1140 Some(f) => {
1141 AUTOLOAD_DEF_FILE.with(|s| s.borrow_mut().push((name.to_string(), f)));
1142 Self(true)
1143 }
1144 None => Self(false),
1145 }
1146 }
1147}
1148
1149impl Drop for AutoloadFileGuard {
1150 fn drop(&mut self) {
1151 if self.0 {
1152 AUTOLOAD_DEF_FILE.with(|s| {
1153 s.borrow_mut().pop();
1154 });
1155 }
1156 }
1157}
1158
1159/// The file `name` is being autoloaded from, if that is happening right now.
1160pub fn autoload_def_file(name: &str) -> Option<String> {
1161 AUTOLOAD_DEF_FILE.with(|s| {
1162 s.borrow()
1163 .iter()
1164 .rev()
1165 .find(|(n, _)| n == name)
1166 .map(|(_, f)| f.clone())
1167 })
1168}
1169
1170/// !!! WARNING: RUST-ONLY HELPER — NO DIRECT C COUNTERPART !!!
1171///
1172/// C's `zexecve` (Src/exec.c:504-643) performs the whole `#!` recovery
1173/// in place: it is only ever reached in the already-forked child that is
1174/// about to BECOME the command, so at c:566/571/581/585/627 it simply
1175/// calls `execve()` a second time and never returns. zshrs reaches the
1176/// same decision from a second call site — the `std::process::Command`
1177/// spawn in `vm_helper::execute_external_bg` — which hands argv to the
1178/// kernel from the PARENT process and therefore cannot "re-exec in
1179/// place". This helper is c:534-634 verbatim with each of those five
1180/// `execve(prog, argv)` calls replaced by `Ok((prog, argv))`, so both
1181/// call sites share one implementation of the shebang rules.
1182///
1183/// `Err(eno)` is what C's `return eno` (c:643) would hand back — either
1184/// the original `eno` or the `errno` from a failed open/read (c:632/634).
1185#[allow(non_snake_case)]
1186pub fn zexecve_recover(pth: &str, argv: &[String], eno: i32) -> Result<(String, Vec<String>), i32> {
1187 if eno == libc::ENOEXEC || eno == libc::ENOENT {
1188 // c:534
1189 let cpth = match std::ffi::CString::new(pth) {
1190 Ok(c) => c,
1191 Err(_) => return Err(libc::ENOENT),
1192 };
1193 let fd = unsafe { libc::open(cpth.as_ptr(), libc::O_RDONLY | libc::O_NOCTTY) }; // c:538
1194 if fd < 0 {
1195 // c:633-634 — `} else eno = errno;` then fall through to `return eno`.
1196 return Err(std::io::Error::last_os_error()
1197 .raw_os_error()
1198 .unwrap_or(libc::ENOENT));
1199 }
1200 let mut buf = vec![0u8; crate::ported::exec::POUNDBANGLIMIT + 1]; // c:541
1201 let ct = unsafe {
1202 libc::read(
1203 fd,
1204 buf.as_mut_ptr() as *mut libc::c_void,
1205 crate::ported::exec::POUNDBANGLIMIT as libc::size_t,
1206 )
1207 }; // c:542
1208 unsafe {
1209 libc::close(fd);
1210 } // c:543
1211 if ct >= 0 {
1212 // c:544
1213 let ct = ct as usize;
1214 if ct >= 2 && buf[0] == b'#' && buf[1] == b'!' {
1215 // c:545
1216 let mut t0 = 0;
1217 while t0 < ct && buf[t0] != b'\n' {
1218 t0 += 1;
1219 } // c:546-548
1220 if t0 == ct {
1221 // c:549
1222 // c:550 `zerr(...)`. C is inside the forked child that is
1223 // about to `_exit`, so the errflag `zerr` raises is
1224 // irrelevant there. This runs in the PARENT, where a raised
1225 // errflag aborts the enclosing script — `( if
1226 // bad-interp-cmd; then exit 0; else exit 1; fi )` returned
1227 // 127 instead of running its else branch. `zwarn`
1228 // (utils.rs:260) emits the identical text without the flag.
1229 crate::ported::utils::zwarn(&format!(
1230 // c:550
1231 "{}: bad interpreter: {}: {}",
1232 pth,
1233 String::from_utf8_lossy(&buf[2..t0.min(ct)]),
1234 std::io::Error::from_raw_os_error(eno)
1235 ));
1236 } else {
1237 // c:552
1238 while t0 > 0 && (buf[t0] == b' ' || buf[t0] == b'\t' || buf[t0] == b'\n') {
1239 buf[t0] = 0;
1240 t0 -= 1;
1241 } // c:553-554
1242 let mut ptr_lo: usize = 2;
1243 while ptr_lo < buf.len() && buf[ptr_lo] == b' ' {
1244 ptr_lo += 1;
1245 } // c:555
1246 let ptr2_lo = ptr_lo;
1247 let mut ptr_hi = ptr2_lo;
1248 while ptr_hi < buf.len() && buf[ptr_hi] != 0 && buf[ptr_hi] != b' ' {
1249 ptr_hi += 1;
1250 } // c:556
1251 let interp_str = String::from_utf8_lossy(&buf[ptr2_lo..ptr_hi]).into_owned();
1252 if eno == libc::ENOENT {
1253 // c:557 — pathprog rewrite path.
1254 let pprog = if !interp_str.starts_with('/') {
1255 // c:561
1256 crate::ported::utils::pathprog(&interp_str)
1257 .map(|p| p.display().to_string())
1258 } else {
1259 None
1260 };
1261 if let Some(pprog) = pprog {
1262 // c:562
1263 let mut argv_new: Vec<String> = Vec::with_capacity(argv.len() + 2);
1264 argv_new.push(interp_str.clone()); // c:564
1265 if ptr_hi >= buf.len() || buf[ptr_hi] == 0 {
1266 argv_new.push(pth.to_string());
1267 } else {
1268 // c:567
1269 let mut rest_lo = ptr_hi + 1;
1270 while rest_lo < buf.len() && buf[rest_lo] == b' ' {
1271 rest_lo += 1;
1272 }
1273 let mut rest_hi = rest_lo;
1274 while rest_hi < buf.len() && buf[rest_hi] != 0 {
1275 rest_hi += 1;
1276 }
1277 let arg_str =
1278 String::from_utf8_lossy(&buf[rest_lo..rest_hi]).into_owned();
1279 argv_new.push(arg_str);
1280 argv_new.push(pth.to_string());
1281 }
1282 for orig in argv.iter().skip(1) {
1283 argv_new.push(orig.clone());
1284 }
1285 // c:565/c:570 — `winch_unblock(); execve(...)`.
1286 // Deliberately NOT ported here. In C that unblock is
1287 // the last statement before `execve` in the FORKED
1288 // CHILD, so all it decides is the mask the new
1289 // program starts with; the parent never sees it.
1290 // This resolver runs in the PARENT (see the c:550
1291 // note above) and RETURNS, so calling it dropped the
1292 // shell's standing `winch_block()`
1293 // (c:Src/init.c:1458) with nothing here to re-arm
1294 // it. The next re-arm is whenever `raw_getbyte`
1295 // returns from its read (zle_main.rs:507/654/762),
1296 // so the leak spans the REST OF THE CURRENT COMMAND,
1297 // not the session — and a completer that execs a
1298 // shebang-less script therefore finishes its
1299 // completion with the handler live. That is the one
1300 // state c:1458 exists to prevent: `adjustwinsize(1)`
1301 // inside a widget makes `calclist` count display
1302 // strings against a width they were not built at,
1303 // and repaints over the prompt row.
1304 //
1305 // The new image still gets SIGWINCH unblocked: the
1306 // child-side `zexecve` unblocks at c:527 (ported at
1307 // src/ported/exec.rs:4215), and the spawn-based
1308 // callers of this function go through
1309 // `std::process::Command`, whose child resets the
1310 // signal mask to empty before `execvp`.
1311 return Ok((pprog, argv_new)); // c:566/c:571
1312 }
1313 crate::ported::utils::zwarn(&format!(
1314 // c:574 — `zerr`; see the c:550 note above for why
1315 // this is `zwarn` in the parent-side port.
1316 "{}: bad interpreter: {}: {}",
1317 pth,
1318 interp_str,
1319 std::io::Error::from_raw_os_error(eno)
1320 ));
1321 } else if ptr_hi < buf.len() && buf[ptr_hi] != 0 {
1322 // c:576
1323 let mut rest_lo = ptr_hi + 1;
1324 while rest_lo < buf.len() && buf[rest_lo] == b' ' {
1325 rest_lo += 1;
1326 }
1327 let mut rest_hi = rest_lo;
1328 while rest_hi < buf.len() && buf[rest_hi] != 0 {
1329 rest_hi += 1;
1330 }
1331 let arg_str = String::from_utf8_lossy(&buf[rest_lo..rest_hi]).into_owned();
1332 let mut argv_new: Vec<String> =
1333 vec![interp_str.clone(), arg_str, pth.to_string()];
1334 for orig in argv.iter().skip(1) {
1335 argv_new.push(orig.clone());
1336 }
1337 // c:580 — `winch_unblock()` omitted: child-side in C,
1338 // parent-side here. See the c:565/c:570 note above.
1339 return Ok((interp_str, argv_new)); // c:581
1340 } else {
1341 // c:582
1342 let mut argv_new: Vec<String> = vec![interp_str.clone(), pth.to_string()];
1343 for orig in argv.iter().skip(1) {
1344 argv_new.push(orig.clone());
1345 }
1346 // c:584 — `winch_unblock()` omitted: child-side in C,
1347 // parent-side here. See the c:565/c:570 note above.
1348 return Ok((interp_str, argv_new)); // c:585
1349 }
1350 }
1351 } else if eno == libc::ENOEXEC {
1352 // c:588 — binary-safety + /bin/sh fallback.
1353 let nul_pos = buf[..ct].iter().position(|&b| b == 0); // c:597
1354 let isbinary = match nul_pos {
1355 None => false, // c:598
1356 Some(npos) => {
1357 let mut has_letter = false;
1358 let mut binary = true;
1359 for &b in &buf[..npos] {
1360 // c:602-609
1361 if (b as char).is_ascii_lowercase() || b == b'$' || b == b'`' {
1362 has_letter = true;
1363 }
1364 if has_letter && b == b'\n' {
1365 binary = false; // c:606
1366 break;
1367 }
1368 }
1369 binary
1370 }
1371 };
1372 if !isbinary {
1373 // c:611
1374 let mut argv_new: Vec<String> = Vec::with_capacity(argv.len() + 2);
1375 argv_new.push("sh".to_string()); // c:625
1376 // c:544 — `*argv = pth;`: by now argv[0] IS the path
1377 // being executed, not the command word, so both the
1378 // `-`/`+` guard and the operand handed to sh are `pth`.
1379 // Testing the word let `PATH=-dir tstcmd` hand sh
1380 // `-dir/tstcmd` with no `-` guard (sh: `-d: invalid
1381 // option`), and ran a bare-word script as `sh tstcmd`.
1382 if pth.starts_with('-') || pth.starts_with('+') {
1383 argv_new.push("-".to_string()); // c:623
1384 }
1385 argv_new.push(pth.to_string()); // c:544
1386 for orig in argv.iter().skip(1) {
1387 argv_new.push(orig.clone());
1388 }
1389 // c:626 — `winch_unblock()` omitted: child-side in C,
1390 // parent-side here. See the c:565/c:570 note above.
1391 return Ok(("/bin/sh".to_string(), argv_new)); // c:627
1392 }
1393 }
1394 }
1395 }
1396 Err(eno) // c:643
1397}
1398
1399
1400/// !!! WARNING: RUST-ONLY HELPER — the non-forking form of `execute()`'s
1401/// search (Src/exec.c:791-895) !!!
1402///
1403/// C runs `execute()` in the forked child, where every `zexecve` either
1404/// replaces the process or returns an errno and the walk moves on to the
1405/// next candidate. zshrs spawns from the parent, so each `zexecve` here is a
1406/// `posix_spawn(2)` attempt (`zexecve_spawn`) and the first one that starts a
1407/// process ends the walk with its pid. The candidates, their order and the
1408/// `isgooderr` bookkeeping are C's:
1409///
1410/// * c:791-801 — an `arg0` holding a `/` is exec'd as spelled; unless it is
1411/// relative, not `./…`/`../…`, and PATH_DIRS is set, a failure is final.
1412/// * c:803-819 — `command -p`: the default-path hit alone.
1413/// * c:822-870 — the `cmdnamtab` candidate, then c:877-895
1414/// `execute_skip_exec:`, the full `$path` walk (empty and `.` entries
1415/// mean `arg0` itself).
1416///
1417/// `Err(eno)` is C's `eno` at c:871: non-zero means `zerr("%e: %s")`, zero
1418/// means nothing usable was found (`command not found`).
1419pub(crate) fn execute_spawn(
1420 arg0: &str,
1421 argv: &[String],
1422 defpath: Option<&str>,
1423 hashed: Option<&str>,
1424 drop_argv0: bool,
1425 in_place: bool,
1426) -> Result<libc::pid_t, i32> {
1427 use crate::ported::exec::isgooderr;
1428 // c:791-801
1429 if let Some(slash) = arg0.find('/') {
1430 let lerrno = match zexecve_spawn(arg0, argv, drop_argv0, in_place) {
1431 Ok(pid) => return Ok(pid),
1432 Err(e) => e,
1433 }; // c:793
1434 let dotted = arg0.starts_with('.')
1435 && (slash == 1 || (arg0.starts_with("..") && slash == 2));
1436 if slash == 0 || crate::ported::zsh_h::unset(crate::ported::zsh_h::PATHDIRS) || dotted {
1437 return Err(lerrno); // c:794-798
1438 }
1439 }
1440 let mut eno = 0;
1441 let mut attempt = |pth: &str, dir: &str| -> Option<libc::pid_t> {
1442 match zexecve_spawn(pth, argv, drop_argv0, in_place) {
1443 Ok(pid) => Some(pid),
1444 Err(ee) => {
1445 if isgooderr(ee, dir) {
1446 eno = ee;
1447 }
1448 None
1449 }
1450 }
1451 };
1452 let dir_of = |p: &str| -> String {
1453 // c:817-818 / c:866-867 — `if ((dptr = strrchr(nn, '/'))) *dptr =
1454 // '\0'; isgooderr(ee, *nn ? nn : "/")`.
1455 match p.rfind('/') {
1456 Some(0) | None => "/".to_string(),
1457 Some(i) => p[..i].to_string(),
1458 }
1459 };
1460 if let Some(pbuf) = defpath {
1461 // c:815 — `ee = zexecve(pbuf, argv, newenvp);`
1462 if let Some(pid) = attempt(pbuf, &dir_of(pbuf)) {
1463 return Ok(pid);
1464 }
1465 return Err(eno);
1466 }
1467 if let Some(nn) = hashed {
1468 // c:863 — `ee = zexecve(nn, argv, newenvp);`
1469 if let Some(pid) = attempt(nn, &dir_of(nn)) {
1470 return Ok(pid);
1471 }
1472 }
1473 // c:877 `execute_skip_exec:` — `for (pp = path; *pp; pp++)`.
1474 let path = crate::ported::params::getaparam("path").unwrap_or_default();
1475 for pp in &path {
1476 let pth = if pp.is_empty() || pp == "." {
1477 arg0.to_string() // c:879-880
1478 } else {
1479 format!("{}/{}", pp, arg0) // c:884-888
1480 };
1481 if pth.len() >= libc::PATH_MAX as usize {
1482 continue; // c:887-888
1483 }
1484 if let Some(pid) = attempt(&pth, pp) {
1485 return Ok(pid); // c:890 `ee = zexecve(buf, argv, newenvp);`
1486 }
1487 }
1488 Err(eno)
1489}
1490
1491/// !!! WARNING: RUST-ONLY HELPER — `zexecve()` (Src/exec.c:504-643) from the
1492/// parent via `posix_spawn(2)` !!!
1493///
1494/// Same steps as the forked-child port (`crate::ported::exec::zexecve`), with
1495/// each `execve` a `posix_spawn` of the SAME path — never `posix_spawnp`,
1496/// whose libc-internal `/bin/sh` fallback on ENOEXEC runs a shebang-less
1497/// script without C's `-`/`+` operand guard (c:611-624) and so hands
1498/// `-dir/tstcmd` to sh as an option. The kernel's refusal comes back here
1499/// and `zexecve_recover` makes C's decision.
1500///
1501/// The environment is the child's `environ` as C leaves it right before
1502/// `execve`, built as a copy so the parent's own array is untouched (C does
1503/// these writes in the child that is about to vanish):
1504///
1505/// * c:514-520 — `zputenv("_=<pth>")`: an existing `_` entry is replaced in
1506/// place, otherwise the entry is appended.
1507/// * c:765-768 (`execute`) — `unsetenv("ARGV0")` when ARGV0 supplied argv[0].
1508///
1509/// The array keeps `environ`'s order: `std::process::Command` rebuilds any
1510/// environment it modifies from a sorted map, so externals saw the variables
1511/// alphabetically instead of in the order the shell added them.
1512fn zexecve_spawn(pth: &str, argv: &[String], drop_argv0: bool, in_place: bool) -> Result<libc::pid_t, i32> {
1513 use std::ffi::CString;
1514 use std::os::unix::ffi::OsStrExt as _;
1515 // c:514-520 — `_=pth`, made absolute against `pwd`.
1516 let underscore = if pth.starts_with('/') {
1517 pth.to_string() // c:516-517
1518 } else {
1519 format!("{}/{}", crate::ported::params::getsparam("PWD").unwrap_or_default(), pth) // c:519
1520 };
1521 let mut underscore_entry = b"_=".to_vec();
1522 underscore_entry.extend_from_slice(&crate::ported::utils::unmetafy_str(&underscore));
1523 let mut envp: Vec<CString> = Vec::new();
1524 let mut have_underscore = false;
1525 for (k, v) in std::env::vars_os() {
1526 let k = k.as_bytes();
1527 if drop_argv0 && k == b"ARGV0" {
1528 continue; // c:768 `unsetenv("ARGV0")`
1529 }
1530 let entry = if k == b"_" {
1531 have_underscore = true;
1532 underscore_entry.clone() // c:520 — replaced in place
1533 } else {
1534 let mut e = k.to_vec();
1535 e.push(b'=');
1536 e.extend_from_slice(v.as_bytes());
1537 e
1538 };
1539 if let Ok(c) = CString::new(entry) {
1540 envp.push(c);
1541 }
1542 }
1543 if !have_underscore {
1544 if let Ok(c) = CString::new(underscore_entry) {
1545 envp.push(c); // c:520 — appended
1546 }
1547 }
1548 // c:528 — `execve(pth, argv, newenvp);`
1549 let eno = match posix_spawn_argv(pth, argv, &envp, in_place) {
1550 Ok(pid) => return Ok(pid),
1551 Err(eno) => eno,
1552 };
1553 // c:534-634 — the kernel refused the file: the `#!` line's interpreter,
1554 // or `/bin/sh` for a shebang-less script.
1555 match zexecve_recover(pth, argv, eno) {
1556 // c:566/571/581/585/627 — the second `execve`; when it fails too, C
1557 // falls through to c:643 `return eno` with the ORIGINAL errno.
1558 Ok((prog, newargv)) => posix_spawn_argv(&prog, &newargv, &envp, in_place).map_err(|_| eno),
1559 Err(e) => Err(e), // c:632/634/643
1560 }
1561}
1562
1563/// !!! WARNING: RUST-ONLY HELPER — the dispositions `entersubsh` resets in the
1564/// child C forks for an external, applied at the spawn instead !!!
1565///
1566/// c:Src/exec.c:1185-1193 —
1567/// ```c
1568/// if (interact) {
1569/// signal_default(SIGTERM);
1570/// if (!(sigtrapped[SIGINT] & ZSIG_IGNORED))
1571/// signal_default(SIGINT);
1572/// if (!(sigtrapped[SIGPIPE]))
1573/// signal_default(SIGPIPE);
1574/// }
1575/// if (!(sigtrapped[SIGQUIT] & ZSIG_IGNORED))
1576/// signal_default(SIGQUIT);
1577/// ```
1578/// The shell ignores SIGQUIT always and SIGTERM when interactive
1579/// (c:Src/init.c:1448/1463), and SIG_IGN survives `execve`, so without this a
1580/// spawned `sh -c 'kill -QUIT $$'` survived its own signal. SIGPIPE is reset
1581/// unless a `trap '' PIPE` ignores it: C never ignores it otherwise, but the
1582/// Rust runtime does. The job-control stops (c:1174-1183) stay as they are: a foreground
1583/// wait here does not watch for a stopped child, so a ^Z'd child would hang
1584/// the shell.
1585fn entersubsh_sigdef() -> Vec<i32> {
1586 use crate::ported::zsh_h::{isset, INTERACTIVE, ZSIG_IGNORED};
1587 let trapped = |sig: i32| -> i32 {
1588 crate::ported::signals::sigtrapped
1589 .lock()
1590 .ok()
1591 .and_then(|t| t.get(sig as usize).copied())
1592 .unwrap_or(0)
1593 };
1594 let mut sigs = Vec::with_capacity(4);
1595 if trapped(libc::SIGPIPE) & ZSIG_IGNORED == 0 {
1596 sigs.push(libc::SIGPIPE); // c:1189-1190, and the Rust runtime's ignore
1597 }
1598 if isset(INTERACTIVE) {
1599 sigs.push(libc::SIGTERM); // c:1186
1600 if trapped(libc::SIGINT) & ZSIG_IGNORED == 0 {
1601 sigs.push(libc::SIGINT); // c:1187-1188
1602 }
1603 }
1604 if trapped(libc::SIGQUIT) & ZSIG_IGNORED == 0 {
1605 sigs.push(libc::SIGQUIT); // c:1192-1193
1606 }
1607 sigs
1608}
1609
1610/// !!! WARNING: RUST-ONLY HELPER — the signal mask the child C forks for an
1611/// external carries into `execve`, computed in the parent instead !!!
1612///
1613/// The child inherits the shell's mask minus what C unblocks on the way to
1614/// the exec:
1615///
1616/// * c:Src/exec.c:1162-1166 (`entersubsh`) — while `intrap`, every trapped,
1617/// non-ignored signal is unblocked: the trap dispatcher blocks the signal
1618/// being handled, and a command run from the trap body must not inherit it.
1619/// * c:Src/exec.c:751 (`execute`) — `child_unblock()` right before the
1620/// `zexecve` walk, so SIGCHLD is never blocked in an external.
1621///
1622/// Without this a command run from a `trap '...' USR2` handler started with
1623/// SIGUSR2 blocked, and every external started with SIGCHLD blocked.
1624fn execute_sigmask() -> libc::sigset_t {
1625 use crate::ported::signals_h::SIGCOUNT;
1626 use crate::ported::zsh_h::ZSIG_IGNORED;
1627 unsafe {
1628 let mut mask: libc::sigset_t = std::mem::zeroed();
1629 libc::sigemptyset(&mut mask);
1630 libc::pthread_sigmask(libc::SIG_BLOCK, std::ptr::null(), &mut mask);
1631 if crate::ported::signals::intrap.load(std::sync::atomic::Ordering::Relaxed) != 0 {
1632 // c:1163-1166 — `for (sig = 1; sig <= SIGCOUNT; sig++) if
1633 // (sigtrapped[sig] && sigtrapped[sig] != ZSIG_IGNORED)
1634 // signal_unblock(signal_mask(sig));`
1635 let trapped = crate::ported::signals::sigtrapped
1636 .lock()
1637 .map(|t| t.clone())
1638 .unwrap_or_default();
1639 for sig in 1..=SIGCOUNT {
1640 let st = trapped.get(sig as usize).copied().unwrap_or(0);
1641 if st != 0 && st != ZSIG_IGNORED {
1642 libc::sigdelset(&mut mask, sig);
1643 }
1644 }
1645 }
1646 libc::sigdelset(&mut mask, libc::SIGCHLD); // c:751 `child_unblock();`
1647 mask
1648 }
1649}
1650
1651/// !!! WARNING: RUST-ONLY HELPER — the `posix_spawn(2)` call standing in for
1652/// `execve(2)` in `zexecve_spawn` !!!
1653///
1654/// argv and path are unmetafied first, as C's `zexecve` does at c:510-513
1655/// (`$'\xff'` reaches the child as the raw byte). The child starts with
1656/// `execute_sigmask`'s mask and `entersubsh_sigdef`'s dispositions.
1657///
1658/// `in_place` is c:Src/exec.c:4369 in a child already forked for the command
1659/// (`last1 = forked = 1`, c:3063): `execve(2)` replaces this process instead
1660/// of spawning, and returns only on failure.
1661fn posix_spawn_argv(
1662 pth: &str,
1663 argv: &[String],
1664 envp: &[std::ffi::CString],
1665 in_place: bool,
1666) -> Result<libc::pid_t, i32> {
1667 use std::ffi::CString;
1668 let cstr = |s: &str| {
1669 let mut b = crate::ported::utils::unmetafy_str(s);
1670 if let Some(n) = b.iter().position(|c| *c == 0) {
1671 b.truncate(n); // a C string ends at its first NUL
1672 }
1673 CString::new(b).unwrap_or_default()
1674 };
1675 let cpth = cstr(pth);
1676 let cargs: Vec<CString> = argv.iter().map(|a| cstr(a)).collect();
1677 let mut argv_ptrs: Vec<*mut libc::c_char> = cargs.iter().map(|c| c.as_ptr() as *mut _).collect();
1678 argv_ptrs.push(std::ptr::null_mut());
1679 let mut envp_ptrs: Vec<*mut libc::c_char> = envp.iter().map(|c| c.as_ptr() as *mut _).collect();
1680 envp_ptrs.push(std::ptr::null_mut());
1681 let sigdef = entersubsh_sigdef();
1682 let sigmask = execute_sigmask();
1683 unsafe {
1684 if in_place {
1685 for &sig in &sigdef {
1686 libc::signal(sig, libc::SIG_DFL);
1687 }
1688 libc::pthread_sigmask(libc::SIG_SETMASK, &sigmask, std::ptr::null_mut());
1689 libc::execve(cpth.as_ptr(), argv_ptrs.as_ptr() as *const *const _, envp_ptrs.as_ptr() as *const *const _);
1690 return Err(std::io::Error::last_os_error().raw_os_error().unwrap_or(libc::ENOEXEC));
1691 }
1692 let mut attrs: libc::posix_spawnattr_t = std::mem::zeroed();
1693 let rc = libc::posix_spawnattr_init(&mut attrs);
1694 if rc != 0 {
1695 return Err(rc);
1696 }
1697 let mut dfl: libc::sigset_t = std::mem::zeroed();
1698 libc::sigemptyset(&mut dfl);
1699 for &sig in &sigdef {
1700 libc::sigaddset(&mut dfl, sig);
1701 }
1702 libc::posix_spawnattr_setsigdefault(&mut attrs, &dfl);
1703 libc::posix_spawnattr_setsigmask(&mut attrs, &sigmask);
1704 libc::posix_spawnattr_setflags(
1705 &mut attrs,
1706 (libc::POSIX_SPAWN_SETSIGDEF | libc::POSIX_SPAWN_SETSIGMASK) as _,
1707 );
1708 let mut pid: libc::pid_t = 0;
1709 let rc = libc::posix_spawn(
1710 &mut pid,
1711 cpth.as_ptr(),
1712 std::ptr::null(),
1713 &attrs,
1714 argv_ptrs.as_ptr(),
1715 envp_ptrs.as_ptr(),
1716 );
1717 libc::posix_spawnattr_destroy(&mut attrs);
1718 if rc != 0 {
1719 return Err(rc);
1720 }
1721 Ok(pid)
1722 }
1723}
1724
1725impl ShellExecutor {
1726 /// Set a scalar parameter via the canonical `paramtab`
1727 /// (`Src/params.c:3350 setsparam`). The single store.
1728 pub fn set_scalar(&mut self, name: String, value: String) {
1729 setsparam(&name, &value); // c:params.c:3350
1730 }
1731
1732 /// Read positional parameters from canonical `PPARAMS`
1733 /// `Mutex<Vec<String>>` (Src/init.c:pparams). The single store.
1734 pub fn pparams(&self) -> Vec<String> {
1735 crate::ported::builtin::PPARAMS
1736 .lock()
1737 .map(|p| p.clone())
1738 .unwrap_or_default()
1739 }
1740
1741 /// Write positional parameters to canonical `PPARAMS`.
1742 pub fn set_pparams(&mut self, params: Vec<String>) {
1743 if let Ok(mut p) = crate::ported::builtin::PPARAMS.lock() {
1744 *p = params;
1745 }
1746 }
1747
1748 /// Read PM_* type flags from the paramtab Param entry. Used by
1749 /// SET_VAR / `+=` arms (case-fold, integer-add, readonly guard).
1750 /// Returns 0 when the name isn't in paramtab. Mirrors the C
1751 /// source's direct `pm->node.flags & PM_INTEGER` checks.
1752 pub fn param_flags(&self, name: &str) -> i32 {
1753 paramtab()
1754 .read()
1755 .ok()
1756 .and_then(|t| t.get(name).map(|p| p.node.flags))
1757 .unwrap_or(0)
1758 }
1759
1760 /// `readonly` / `typeset -r` / read-only-by-design (LINENO, PPID,
1761 /// $$, $?, $!, ...) — match user-side rejection in C's
1762 /// assignstrvalue at `Src/params.c:2699-2703` which gates on
1763 /// `pm->node.flags & PM_READONLY` where the IPDEF4 family declares
1764 /// `PM_READONLY_SPECIAL = PM_SPECIAL | PM_READONLY | PM_RO_BY_DESIGN`
1765 /// (all three bits set together), which `init_partab_params` now
1766 /// stamps in full. The PM_RO_BY_DESIGN arm below is therefore no
1767 /// longer the IPDEF4 rows' only read-only marker — it remains
1768 /// because `private` params (c:Src/Modules/param_private.c:174)
1769 /// carry PM_RO_BY_DESIGN WITHOUT PM_READONLY and need the
1770 /// scope-gated test. Bug #418-family / test_lineno_intrinsic_readonly.
1771 pub fn is_readonly_param(&self, name: &str) -> bool {
1772 // c:Src/params.c:2235 fetchvalue looks the name up through
1773 // realparamtab->getnode, i.e. getprivatenode (Src/Modules/
1774 // param_private.c:678): a private of an enclosing scope with nothing
1775 // beneath it is ABSENT, and createparam reports the write.
1776 let (flags, pm_level) = crate::ported::params::paramtab()
1777 .read()
1778 .ok()
1779 .and_then(|t| {
1780 t.get(name).and_then(|p| {
1781 let vis = crate::ported::modules::param_private::getprivatenode(&**p as *const _);
1782 (!vis.is_null()).then(|| unsafe { ((*vis).node.flags as u32, (*vis).level) })
1783 })
1784 })
1785 .unwrap_or((0, 0));
1786 // c:Src/params.c:2264-2265 — assignsparam reaches its c:3216 PM_READONLY
1787 // test only through fetchvalue, which reports a node that is PM_UNSET
1788 // without PM_DECLARED as ABSENT; createparam then decides (for an
1789 // outer-scope private, hidden by scopeprivate's PM_UNSET|PM_READONLY,
1790 // that is "can't modify read-only parameter"). So such a node is not
1791 // a read-only rejection here; assignsparam reports it.
1792 if (flags & crate::ported::zsh_h::PM_UNSET) != 0 && (flags & crate::ported::zsh_h::PM_DECLARED) == 0 {
1793 return false;
1794 }
1795 // c:Src/params.c assignsparam — a real PM_READONLY param always
1796 // rejects writes.
1797 if (flags & PM_READONLY) != 0 {
1798 return true;
1799 }
1800 if (flags & crate::ported::zsh_h::PM_RO_BY_DESIGN) != 0 {
1801 // c:Src/Modules/param_private.c pps_setfn (c:300-307) — a
1802 // PRIVATE param (PM_RO_BY_DESIGN + PM_REMOVABLE) is NOT blanket
1803 // read-only: a write is permitted iff it is in the SAME scope
1804 // (`locallevel == pm->level`, e.g. `() { private p=1; p=2 }`
1805 // → 2) or above the wrap level (`locallevel >
1806 // private_wraplevel`). A deeper nested-scope write is rejected
1807 // (setfn_error) — that is how a nested fn writing an OUTER
1808 // function's private still errors and aborts. zshrs never
1809 // wires the private GSU, so the level gate is enforced here.
1810 if (flags & crate::ported::zsh_h::PM_REMOVABLE) != 0 {
1811 let ll = crate::ported::params::locallevel.load(Ordering::Relaxed);
1812 let wrap = crate::ported::modules::param_private::private_wraplevel
1813 .load(Ordering::Relaxed);
1814 return !(ll == pm_level || ll > wrap); // c:304 (negated: blocked)
1815 }
1816 // Non-removable PM_RO_BY_DESIGN = IPDEF4-family special
1817 // (LINENO/$?/$$…). These now also carry PM_READONLY and so
1818 // return true from the branch above; this arm stays as the
1819 // c:Src/zsh.h:1923 "readonly by design" fallback for any row
1820 // reached before `init_partab_params` has stamped the flag.
1821 return true;
1822 }
1823 false
1824 }
1825
1826 /// Most-recent-command exit status. Reads canonical
1827 /// `builtin::LASTVAL` AtomicI32 (`Src/builtin.c:6443`).
1828 pub fn last_status(&self) -> i32 {
1829 crate::ported::builtin::LASTVAL.load(Ordering::Relaxed)
1830 }
1831
1832 /// Write the most-recent-command exit status. The canonical
1833 /// store is `builtin::LASTVAL`; this is the single setter.
1834 /// Used everywhere `$?` / `%?` / errexit / ZERR trap read.
1835 pub fn set_last_status(&mut self, status: i32) {
1836 crate::ported::builtin::LASTVAL.store(status, Ordering::Relaxed);
1837 }
1838
1839 /// Set an indexed array parameter via canonical paramtab
1840 /// (`assignaparam`, `Src/params.c:3357`). The single store.
1841 ///
1842 /// Callers mean "make NAME this array" — `typeset -a h=(…)` over an
1843 /// existing assoc among them, whose type conversion C does in
1844 /// typeset_single before it assigns. `setaparam` now keeps an existing
1845 /// assoc hashed (C's c:3341-3342), so go one level down to the
1846 /// type-resetting store with the same `ASSPM_WARN` flag setaparam used.
1847 pub fn set_array(&mut self, name: String, value: Vec<String>) {
1848 assignaparam(&name, value, crate::ported::zsh_h::ASSPM_WARN); // c:params.c:3766
1849 }
1850
1851 /// Set an associative array parameter via canonical
1852 /// `sethparam` (`Src/params.c:3602`). The single store.
1853 pub fn set_assoc(&mut self, name: String, value: IndexMap<String, String>) {
1854 let mut flat: Vec<String> = Vec::with_capacity(value.len() * 2);
1855 for (k, v) in &value {
1856 flat.push(k.clone());
1857 flat.push(v.clone());
1858 }
1859 sethparam(&name, flat); // c:params.c:3602
1860 }
1861
1862 /// Read a scalar parameter. Mirrors C `getsparam` at
1863 /// `Src/params.c:3076` — reads through paramtab, falls back to
1864 /// special-var hooks and env.
1865 pub fn scalar(&self, name: &str) -> Option<String> {
1866 getsparam(name)
1867 }
1868
1869 /// Read an array parameter via canonical `getaparam`
1870 /// (`Src/params.c:3101`).
1871 pub fn array(&self, name: &str) -> Option<Vec<String>> {
1872 getaparam(name)
1873 }
1874
1875 /// Read an associative array parameter from canonical
1876 /// `paramtab_hashed_storage`. Mirrors C `gethparam` at
1877 /// `Src/params.c:3115` — returns the typed `IndexMap`.
1878 pub fn assoc(&self, name: &str) -> Option<IndexMap<String, String>> {
1879 // c:Src/params.c:570-575 — nameref deref before the read.
1880 let resolved = match crate::ported::params::resolve_nameref_name(name, None) {
1881 // c:2264-2267 + c:2297 — a reference to an ELEMENT leaves its
1882 // `[sub]` unparsed for a bracks-0 getvalue, so it is no hash.
1883 crate::ported::params::nameref_resolution::Target { subscript: Some(_), .. } => {
1884 return Default::default()
1885 }
1886 crate::ported::params::nameref_resolution::Target { name: t_, .. } => t_,
1887 _ => name.to_string(),
1888 };
1889 // The live param's TYPE is authoritative — paramtab_hashed_storage is
1890 // keyed by NAME ONLY (no scope), so a local ARRAY that shadows a special
1891 // assoc (`local -a options` / `local -a commands` over the hidden
1892 // `options`/`commands` specials) leaves a stale (emptied) hashed_storage
1893 // entry behind. Without this guard `exec.assoc("options")` returned
1894 // Some(empty), so a subsequent bare `options=(-a -b -c)` routed through
1895 // sethparam → "bad set of key/value pairs for associative array"
1896 // (odd count), breaking `_sqlite`'s `local -a options; options=(…)`
1897 // (separate statements — the one-statement `local -a commands=(…)` form
1898 // that openssl uses is typed correctly by bin_typeset). Consult the
1899 // current param: if it exists and is NOT PM_HASHED, it's an array/scalar
1900 // shadow and the stale assoc storage must not be seen.
1901 if let Some(flags) = crate::ported::params::paramtab()
1902 .read()
1903 .ok()
1904 .and_then(|t| t.get(resolved.as_str()).map(|p| p.node.flags as u32))
1905 {
1906 if (flags & PM_HASHED) == 0 {
1907 return None;
1908 }
1909 }
1910 paramtab_hashed_storage()
1911 .lock()
1912 .ok()
1913 .and_then(|m| m.get(resolved.as_str()).cloned())
1914 }
1915
1916 /// Test whether a scalar parameter exists in paramtab.
1917 /// Mirrors the C `paramtab->getnode(name) != NULL` check.
1918 pub fn has_scalar(&self, name: &str) -> bool {
1919 getsparam(name).is_some()
1920 }
1921
1922 /// Test whether an array parameter exists in paramtab. Mirrors
1923 /// `getaparam(name).is_some()` (PM_ARRAY + populated `u_arr`, with
1924 /// digit-first-name rejection and nameref deref) WITHOUT cloning the
1925 /// backing vector — `getaparam` returns an owned `Vec<String>`, so a
1926 /// bare existence probe on a large array copied every element. Hot in
1927 /// the subscript-store dispatch (`a[i]=v` in a loop), so keep it a
1928 /// flag read.
1929 pub fn has_array(&self, name: &str) -> bool {
1930 if name.starts_with(|c: char| c.is_ascii_digit()) {
1931 return false;
1932 }
1933 let resolved = match crate::ported::params::resolve_nameref_name(name, None) {
1934 crate::ported::params::nameref_resolution::Target { name: t_, .. } => t_,
1935 _ => name.to_string(),
1936 };
1937 crate::ported::params::paramtab()
1938 .read()
1939 .ok()
1940 .and_then(|t| {
1941 t.get(resolved.as_str()).map(|p| {
1942 (p.node.flags as u32 & crate::ported::zsh_h::PM_ARRAY) != 0 && p.u_arr.is_some()
1943 })
1944 })
1945 .unwrap_or(false)
1946 }
1947
1948 /// Test whether an associative array parameter exists. Reads
1949 /// canonical `paramtab_hashed_storage` (Src/params.c hashed
1950 /// PM_HASHED slot).
1951 pub fn has_assoc(&self, name: &str) -> bool {
1952 // c:Src/params.c:570-575 — nameref deref before the read.
1953 let resolved = match crate::ported::params::resolve_nameref_name(name, None) {
1954 // c:2264-2267 + c:2297 — a reference to an ELEMENT leaves its
1955 // `[sub]` unparsed for a bracks-0 getvalue, so it is no hash.
1956 crate::ported::params::nameref_resolution::Target { subscript: Some(_), .. } => {
1957 return Default::default()
1958 }
1959 crate::ported::params::nameref_resolution::Target { name: t_, .. } => t_,
1960 _ => name.to_string(),
1961 };
1962 // Live-param type is authoritative (see `assoc` above): a non-PM_HASHED
1963 // shadow (e.g. `local -a options`) hides the stale name-keyed
1964 // hashed_storage entry.
1965 if let Some(flags) = crate::ported::params::paramtab()
1966 .read()
1967 .ok()
1968 .and_then(|t| t.get(resolved.as_str()).map(|p| p.node.flags as u32))
1969 {
1970 if (flags & PM_HASHED) == 0 {
1971 return false;
1972 }
1973 }
1974 paramtab_hashed_storage()
1975 .lock()
1976 .ok()
1977 .map(|m| m.contains_key(resolved.as_str()))
1978 .unwrap_or(false)
1979 }
1980
1981 /// Unset an associative array parameter via canonical
1982 /// `unsetparam` (Src/params.c:3819) — PM_READONLY rejection,
1983 /// stdunsetfn dispatch, env clear. Also clears the zshrs-side
1984 /// `paramtab_hashed_storage` parallel IndexMap shadow.
1985 pub fn unset_assoc(&mut self, name: &str) {
1986 unsetparam(name);
1987 let _ = paramtab_hashed_storage()
1988 .lock()
1989 .ok()
1990 .as_deref_mut()
1991 .map(|m| m.remove(name));
1992 }
1993
1994 /// Set a regular alias. Writes canonical aliastab with
1995 /// ALIAS_GLOBAL bit cleared.
1996 pub fn set_alias(&mut self, name: String, value: String) {
1997 if let Ok(mut tab) = crate::ported::hashtable::aliastab_lock().write() {
1998 tab.add(crate::ported::hashtable::createaliasnode(&name, &value, 0));
1999 }
2000 }
2001
2002 /// Set a global alias (`alias -g`). Writes canonical aliastab
2003 /// with ALIAS_GLOBAL bit set.
2004 pub fn set_global_alias(&mut self, name: String, value: String) {
2005 if let Ok(mut tab) = crate::ported::hashtable::aliastab_lock().write() {
2006 tab.add(crate::ported::hashtable::createaliasnode(
2007 &name,
2008 &value,
2009 crate::ported::zsh_h::ALIAS_GLOBAL as u32,
2010 ));
2011 }
2012 }
2013
2014 /// Set a suffix alias (`alias -s ext=cmd`). Writes canonical
2015 /// sufaliastab with ALIAS_SUFFIX node flag — mirrors C
2016 /// Src/builtin.c:4480-4481 (`flags1 |= ALIAS_SUFFIX; ht =
2017 /// sufaliastab;`) → c:4527 (`createaliasnode(value, flags1)`).
2018 /// Without ALIAS_SUFFIX in node.flags, `${saliases[k]}` /
2019 /// `${(k)saliases}` introspection (parameter.c:1953/2018) fails
2020 /// because both paths strict-equality-match flags == ALIAS_SUFFIX.
2021 pub fn set_suffix_alias(&mut self, name: String, value: String) {
2022 if let Ok(mut tab) = crate::ported::hashtable::sufaliastab_lock().write() {
2023 tab.add(crate::ported::hashtable::createaliasnode(
2024 &name,
2025 &value,
2026 crate::ported::zsh_h::ALIAS_SUFFIX as u32,
2027 ));
2028 }
2029 }
2030
2031 /// Snapshot the alias map as a sorted `Vec<(name, value)>`,
2032 /// only entries WITHOUT the ALIAS_GLOBAL flag (regular aliases).
2033 pub fn alias_entries(&self) -> Vec<(String, String)> {
2034 if let Ok(tab) = crate::ported::hashtable::aliastab_lock().read() {
2035 tab.iter_sorted()
2036 .into_iter()
2037 .filter(|(_, a)| (a.node.flags & crate::ported::zsh_h::ALIAS_GLOBAL as i32) == 0)
2038 .map(|(k, a)| (k.clone(), a.text.clone()))
2039 .collect()
2040 } else {
2041 Vec::new()
2042 }
2043 }
2044
2045 /// Snapshot the global-alias entries (ALIAS_GLOBAL flag set).
2046 pub fn global_alias_entries(&self) -> Vec<(String, String)> {
2047 if let Ok(tab) = crate::ported::hashtable::aliastab_lock().read() {
2048 tab.iter_sorted()
2049 .into_iter()
2050 .filter(|(_, a)| (a.node.flags & crate::ported::zsh_h::ALIAS_GLOBAL as i32) != 0)
2051 .map(|(k, a)| (k.clone(), a.text.clone()))
2052 .collect()
2053 } else {
2054 Vec::new()
2055 }
2056 }
2057
2058 /// Snapshot the suffix-alias entries.
2059 pub fn suffix_alias_entries(&self) -> Vec<(String, String)> {
2060 if let Ok(tab) = crate::ported::hashtable::sufaliastab_lock().read() {
2061 tab.iter_sorted()
2062 .into_iter()
2063 .map(|(k, a)| (k.clone(), a.text.clone()))
2064 .collect()
2065 } else {
2066 Vec::new()
2067 }
2068 }
2069
2070 /// Unset an array parameter via canonical `unsetparam`
2071 /// (Src/params.c:3819). Routes through the C-faithful port
2072 /// that runs PM_NAMEREF skip + PM_READONLY rejection via
2073 /// unsetparam_pm + stdunsetfn dispatch + pm.old scope restore.
2074 /// Inline `tab.remove(name)` skipped all four.
2075 pub fn unset_array(&mut self, name: &str) {
2076 unsetparam(name);
2077 }
2078
2079 /// Unset a scalar parameter via canonical `unsetparam`. Same
2080 /// C-faithful path as `unset_array`; the C `unsetparam` itself
2081 /// is type-agnostic and dispatches through PM_TYPE inside.
2082 pub fn unset_scalar(&mut self, name: &str) {
2083 unsetparam(name);
2084 }
2085 /// Lightweight executor for a POOL WORKER THREAD. Unlike [`new`] (a full
2086 /// session bootstrap that re-derives PWD, imports the environment, seeds
2087 /// `OPTS_LIVE`, and writes ~30 default params into the GLOBAL param table),
2088 /// this constructs ONLY the per-executor struct fields and touches NO
2089 /// global state. A worker shares the already-populated, `RwLock`-synchronized
2090 /// globals (params / functions / options); re-seeding them here would clobber
2091 /// the live main session's values (IFS, OPTIND, `$_`, user options, …).
2092 ///
2093 /// The worker pool is shared (Arc) — a worker never spins up its own pool.
2094 /// Per-worker SQLite caches (compsys / plugin) and the history engine are
2095 /// left `None`: a worker runs short compute bodies, not interactive editing.
2096 ///
2097 /// Phase 1 of the in-process thread-execution model (replaces the
2098 /// subprocess-forking parallel builtins). The caller runs the body under
2099 /// `ExecutorContext::enter(&mut wex)` so the VM's thread_local executor
2100 /// resolves on the worker thread; param writes flow to the shared globals.
2101 pub fn new_worker(pool: std::sync::Arc<crate::worker::WorkerPool>) -> Self {
2102 // fpath from the inherited env, same as new() — pure read, no global write.
2103 let raw_fpath: Vec<PathBuf> = env::var("FPATH")
2104 .unwrap_or_default()
2105 .split(':')
2106 .filter(|s| !s.is_empty())
2107 .map(PathBuf::from)
2108 .collect();
2109 // c:Src/init.c:1132-1143 seeds `<prefix>/share/zsh/<ver>/functions` at
2110 // a FIXED position, and every `#compdef` claim is resolved against
2111 // `$fpath` ORDER. The bundled tree is that tree's replacement — a
2112 // byte-identical superset of it — so it belongs AT ITS INDEX, not at
2113 // either end.
2114 let host_at = raw_fpath.iter().position(|d| is_host_zsh_function_tree(d));
2115 let mut fpath: Vec<PathBuf> = raw_fpath
2116 .into_iter()
2117 .filter(|d| !is_host_zsh_function_tree(d))
2118 .collect();
2119 if fpath.is_empty() {
2120 fpath = default_fpath(); // c:Src/init.c:1143
2121 }
2122 // Same directory as the main constructor, and LAST for the same
2123 // reason, but never written from a pool worker -- materialisation
2124 // is the session's job.
2125 if let Some(d) = crate::bundled_functions::functions_dir() {
2126 if d.is_dir() && !fpath.contains(&d) {
2127 // Insert AT the host tree's inherited index, not at either
2128 // end. `host_at` is that index in the raw FPATH; the filter
2129 // above removed only host-tree entries, all of which are at or
2130 // after it, so the index still addresses the same slot.
2131 //
2132 // Both ends are wrong, and each was tried. LAST (the previous
2133 // behaviour) let `zsh-more-completions`' appended dirs
2134 // out-rank the distribution for 343 files / 729 commands —
2135 // the opposite of real zsh, where the tree sits at index 24
2136 // and those plugins at 40-49, and the opposite of what
2137 // `zsh-more-completions.plugin.zsh` itself asks for: it
2138 // PREPENDS `override_src` and APPENDS `src`/`more_src*`/
2139 // `man_src`. FIRST shadowed everything the user curated,
2140 // `override_src` included (that is the `ls -<TAB>` regression
2141 // recorded in this comment's earlier revision). The host
2142 // tree's own index is after `override_src` and before the
2143 // appended plugin dirs, which is exactly where zsh has it.
2144 //
2145 // Measured on a 60-cell corpus of contested commands:
2146 // 1 PASS -> 57 PASS, zero regressions.
2147 match host_at {
2148 Some(i) if i <= fpath.len() => fpath.insert(i, d),
2149 _ => fpath.push(d),
2150 }
2151 }
2152 }
2153 Self {
2154 scriptname: Some("zsh".to_string()),
2155 scriptfilename: Some("zsh".to_string()),
2156 subshell_snapshots: Vec::new(),
2157 inline_env_stack: Vec::new(),
2158 current_command_glob_failed: std::cell::Cell::new(false),
2159 jobs: JobTable::new(),
2160 fpath,
2161 // Worker: no interactive history engine, ever — a pre-filled
2162 // cell so `history()` cannot open one from a pool thread.
2163 history: std::cell::OnceCell::from(None),
2164 completions: HashMap::new(),
2165 process_sub_counter: 0,
2166 zstyles: Vec::new(),
2167 local_scope_depth: 0,
2168 pending_underscore: None,
2169 in_dq_context: 0,
2170 in_scalar_assign: 0,
2171 profiling_enabled: false,
2172 compsys_cache: std::cell::OnceCell::from(None), // worker: no per-thread SQLite mirror
2173 compinit_pending: None,
2174 plugin_cache: None, // worker: no per-thread plugin cache
2175 deferred_compdefs: Vec::new(),
2176 returning: None,
2177 zsh_compat: false,
2178 bash_compat: false,
2179 posix_mode: false,
2180 worker_pool: pool, // SHARED — never spawn a nested pool
2181 intercepts: Vec::new(),
2182 async_jobs: HashMap::new(),
2183 next_async_id: 1,
2184 redirect_scope_stack: Vec::new(),
2185 multios_scope_stack: Vec::new(),
2186 exec_redirs_permanent: false,
2187 pipe_output_pending: false,
2188 pipe_output_scope: None,
2189 redirect_failed: false,
2190 functions_compiled: crate::cow_map::CowHashMap::new(),
2191 function_source: crate::cow_map::CowHashMap::new(),
2192 function_line_base: HashMap::new(),
2193 function_def_file: HashMap::new(),
2194 prompt_funcstack: Vec::new(),
2195 tied_array_to_scalar: HashMap::new(),
2196 ztest_pass_count: std::sync::atomic::AtomicUsize::new(0),
2197 ztest_fail_count: std::sync::atomic::AtomicUsize::new(0),
2198 ztest_skip_count: std::sync::atomic::AtomicUsize::new(0),
2199 ztest_pass_total: std::sync::atomic::AtomicUsize::new(0),
2200 ztest_fail_total: std::sync::atomic::AtomicUsize::new(0),
2201 ztest_skip_total: std::sync::atomic::AtomicUsize::new(0),
2202 ztest_run_failed: std::sync::atomic::AtomicBool::new(false),
2203 ztest_suppress_stdout: false,
2204 }
2205 }
2206
2207 /// `new` — see implementation.
2208 pub fn new() -> Self {
2209 tracing::debug!("ShellExecutor::new() initializing");
2210
2211 // zsh's man/info pages ride along with the function tree: same
2212 // reason (zshrs is not installed under a zsh prefix), same
2213 // one-stamp-read steady-state cost. This runs FIRST because it
2214 // edits MANPATH/INFOPATH in the OS environment, and the env is
2215 // read into paramtab further down -- publishing after that point
2216 // left an inherited $INFOPATH untouched in the shell while the
2217 // process environment carried the new value.
2218 crate::bundled_docs::install_and_publish();
2219
2220 // c:Src/init.c:1236-1259 — setupvals' pwd/oldpwd init, ported
2221 // here because the bin entry skips setupvals (see the
2222 // init_bltinmods note below). The validated value lands in the
2223 // live OS env: the bin entry's `$PWD` carrier (the analog of
2224 // C's `pwd` global — see the subshell-snapshot comment at
2225 // fusevm_bridge.rs `cwd:` field). set_pwd_env() pours it into
2226 // paramtab after the env-import loop, same order as C
2227 // (params.c:955).
2228 //
2229 // c:1242-1245 — "Try a cheap test to see if we can initialize
2230 // `PWD' from `HOME'." EMULATE_ZSH reads the `home` global,
2231 // which setupvals derives from getpwuid(getuid())->pw_dir
2232 // (c:1222-1225), falling back to "/" (c:1230-1232).
2233 let home = unsafe {
2234 let pw = libc::getpwuid(libc::getuid());
2235 if pw.is_null() {
2236 None
2237 } else {
2238 Some(
2239 std::ffi::CStr::from_ptr((*pw).pw_dir)
2240 .to_string_lossy()
2241 .into_owned(),
2242 )
2243 }
2244 }
2245 .unwrap_or_else(|| "/".to_string()); // c:1230-1232 EMULATE_ZSH home = "/"
2246 // ispwd (src/zsh/Src/utils.c:809-829): a candidate is honored
2247 // only when it (a) is absolute, (b) stat's to the same
2248 // dev+inode as ".", and (c) has no `.`/`..` components.
2249 // Without this chain, a child that inherits $PWD from a parent
2250 // run in a different directory (cargo test setting
2251 // current_dir(tempdir) while leaking PWD=/project/root) treats
2252 // the stale PWD as the logical-path base, so `cd sub` resolves
2253 // against the wrong directory.
2254 let pwd_val = if ispwd(&home) {
2255 home // c:1245-1246 — pwd = ztrdup(ptr) [HOME]
2256 } else if let Some(p) = env::var("PWD")
2257 .ok()
2258 .filter(|p| p.len() < libc::PATH_MAX as usize && ispwd(p))
2259 {
2260 p // c:1247-1249 — pwd = ztrdup(getenv("PWD"))
2261 } else {
2262 crate::ported::compat::zgetcwd() // c:1250-1252 — pwd = zgetcwd()
2263 };
2264 // c:1255-1259 — oldpwd = getenv("OLDPWD") ?: ztrdup(pwd).
2265 let oldpwd_val = env::var("OLDPWD").unwrap_or_else(|_| pwd_val.clone()); // c:1257
2266 // Both stay out of the environment until set_pwd_env below: in C they
2267 // are the `pwd`/`oldpwd` globals, and only set_pwd_env's addenv
2268 // (c:Src/builtin.c:821-826) exports them, after HOME, LOGNAME and
2269 // SHLVL (c:Src/params.c:960-977). Publishing them here put a
2270 // non-inherited PWD/OLDPWD ahead of those in every child's env.
2271
2272 // Initialize fpath from FPATH env var, or from the compiled-in
2273 // defaults when it is absent (c:Src/init.c:1132-1143).
2274 let raw_fpath: Vec<PathBuf> = env::var("FPATH")
2275 .unwrap_or_default()
2276 .split(':')
2277 .filter(|s| !s.is_empty())
2278 .map(PathBuf::from)
2279 .collect();
2280 // c:Src/init.c:1132-1143 seeds `<prefix>/share/zsh/<ver>/functions` at
2281 // a FIXED position, and every `#compdef` claim is resolved against
2282 // `$fpath` ORDER. The bundled tree is that tree's replacement — a
2283 // byte-identical superset of it — so it belongs AT ITS INDEX, not at
2284 // either end.
2285 let host_at = raw_fpath.iter().position(|d| is_host_zsh_function_tree(d));
2286 let mut fpath: Vec<PathBuf> = raw_fpath
2287 .into_iter()
2288 .filter(|d| !is_host_zsh_function_tree(d))
2289 .collect();
2290 if fpath.is_empty() {
2291 fpath = default_fpath();
2292 }
2293 // zshrs ships zsh's function tree with the binary; its directory
2294 // goes LAST, so it supplies only what nothing else on fpath does.
2295 //
2296 // It was first at one point, mirroring where zsh's own
2297 // <prefix>/share/zsh/<ver>/functions sits. That is wrong here: zsh
2298 // puts its tree ahead of nothing the user curated, because the
2299 // user's own directories are prepended to it, whereas an inherited
2300 // FPATH arrives ALREADY assembled. Leading it shadowed every
2301 // curated completion of the same name -- 242 of
2302 // zsh-more-completions' files on this author's setup, `_ls` among
2303 // them, which is how `ls -<TAB>` started diverging from zsh.
2304 //
2305 // Being last costs nothing: the only tree that could out-rank the
2306 // bundle is a host zsh's own, and `is_host_zsh_function_tree`
2307 // already removes that from fpath entirely. Materialised here (not
2308 // only when FPATH is absent) so an inherited FPATH from a shell
2309 // that lacks these still resolves is-at-least/colors/add-zsh-hook.
2310 if let Some(d) = crate::bundled_functions::functions_dir() {
2311 let _ = crate::bundled_functions::ensure_installed();
2312 if d.is_dir() && !fpath.contains(&d) {
2313 // Insert AT the host tree's inherited index, not at either
2314 // end. `host_at` is that index in the raw FPATH; the filter
2315 // above removed only host-tree entries, all of which are at or
2316 // after it, so the index still addresses the same slot.
2317 //
2318 // Both ends are wrong, and each was tried. LAST (the previous
2319 // behaviour) let `zsh-more-completions`' appended dirs
2320 // out-rank the distribution for 343 files / 729 commands —
2321 // the opposite of real zsh, where the tree sits at index 24
2322 // and those plugins at 40-49, and the opposite of what
2323 // `zsh-more-completions.plugin.zsh` itself asks for: it
2324 // PREPENDS `override_src` and APPENDS `src`/`more_src*`/
2325 // `man_src`. FIRST shadowed everything the user curated,
2326 // `override_src` included (that is the `ls -<TAB>` regression
2327 // recorded in this comment's earlier revision). The host
2328 // tree's own index is after `override_src` and before the
2329 // appended plugin dirs, which is exactly where zsh has it.
2330 //
2331 // Measured on a 60-cell corpus of contested commands:
2332 // 1 PASS -> 57 PASS, zero regressions.
2333 match host_at {
2334 Some(i) if i <= fpath.len() => fpath.insert(i, d),
2335 _ => fpath.push(d),
2336 }
2337 }
2338 }
2339
2340 let history = std::cell::OnceCell::new();
2341
2342 // Seed canonical OPTS_LIVE with defaults BEFORE any setsparam
2343 // call. assignstrvalue early-returns when `unset(EXECOPT)`
2344 // (c:2701 guard); without the option table populated, EXECOPT
2345 // reads false and every paramtab write below is a silent no-op.
2346 if opt_state_len() == 0 {
2347 for (k, v) in Self::default_options() {
2348 opt_state_set(&k, v);
2349 }
2350 }
2351
2352 // c:Src/params.c:838-847 — `for (ip = special_params; ip->node.nam;
2353 // ip++) paramtab->addnode(paramtab, ztrdup(ip->node.nam), ip);`
2354 // The specials go into paramtab FIRST, ahead of every non-special
2355 // seed below, because creation ORDER is observable: a new key is
2356 // front-inserted into its bucket chain (c:Src/hashtable.c:214-215)
2357 // and `${(k)parameters}` prints that chain walk verbatim
2358 // (c:Src/hashtable.c:420-434). Seeding NULLCMD / FUNCNEST / PS1 /
2359 // … before this loop put them AHEAD of the specials they follow in
2360 // the C table (`#` at c:304 vs NULLCMD at c:378; UID at c:312 vs
2361 // FUNCNEST at c:366), which is exactly where zshrs's parameter
2362 // order diverged from zsh's. With the table seeded first, those
2363 // later `setsparam`/`setiparam` calls hit an existing node and
2364 // replace it IN PLACE (c:187-203 `replacing:`), keeping C's slot.
2365 // c:Src/params.c:384-394 — IPDEF8/IPDEF9 macros stamp
2366 // `PM_SCALAR|PM_SPECIAL` (IPDEF8 for `PATH`/`FPATH`/etc.) and
2367 // `PM_ARRAY|PM_SPECIAL|PM_DONTIMPORT` (IPDEF9 for `path`/
2368 // `fpath`/etc.) on every entry in the createparamtable table.
2369 // setsparam/setaparam above create plain PM_SCALAR/PM_ARRAY
2370 // entries; this loop applies the PM_SPECIAL + PM_TIED bits
2371 // (plus the IPDEF9 PM_DONTIMPORT bit on the array side) so
2372 // `${(t)PATH}` reads `scalar-tied-export-special` and
2373 // `${(t)path}` reads `array-tied-special`.
2374 //
2375 // Walks the `special_params` table (params.rs:464+) which is
2376 // the Rust port of the C IPDEF list. For each entry: OR the
2377 // declared pm_flags onto the existing paramtab entry. The
2378 // tied-pair entries (PM_TIED) also need PM_SPECIAL OR'd in
2379 // since the IPDEF8/IPDEF9 macros add PM_SPECIAL implicitly;
2380 // the table declares only the per-entry-distinct flags.
2381 let stamp_special_params = || {
2382 use crate::ported::params::{paramtab, special_params};
2383 use crate::ported::zsh_h::{PM_ARRAY, PM_DONTIMPORT, PM_SCALAR, PM_SPECIAL, PM_TIED};
2384 if let Ok(mut tab) = paramtab().write() {
2385 // Stamp PM_SPECIAL onto every entry the special_params
2386 // table declares. For tied scalars (PATH/FPATH/etc),
2387 // also walks `tied_name` to apply IPDEF9-flag bits
2388 // (PM_ARRAY|PM_SPECIAL|PM_DONTIMPORT|PM_TIED) onto the
2389 // partner array entry (path/fpath/etc) — those array
2390 // names aren't in the special_params table directly
2391 // but C zsh's createparamtable emits IPDEF9 rows for
2392 // them at Src/params.c:425-432.
2393 use crate::ported::zsh_h::{hashnode, param, PM_DONTIMPORT as PM_DI, PM_UNSET};
2394 // c:Src/params.c:838-847 — the first section is always
2395 // loaded; past it, sh/ksh emulation loads the untied scalar
2396 // `special_params_sh` rows INSTEAD of the zsh-only tail
2397 // (array-tied pairs, lowercase aliases, pipestatus):
2398 // if (EMULATION(EMULATE_SH|EMULATE_KSH))
2399 // for (ip = special_params_sh; ...) addnode(...);
2400 // else
2401 // while ((++ip)->node.nam) addnode(...);
2402 // `zsh --emulate sh -c 'typeset -p PATH'` therefore prints
2403 // `export PATH=…`, never `export -T PATH path=( … )`.
2404 let head = &special_params[..crate::ported::params::SPECIAL_PARAMS_ZSH_START];
2405 let tail: &[crate::ported::params::special_paramdef] = if crate::ported::zsh_h::EMULATION(
2406 crate::ported::zsh_h::EMULATE_SH | crate::ported::zsh_h::EMULATE_KSH,
2407 ) {
2408 crate::ported::params::special_params_sh // c:841-843
2409 } else {
2410 &special_params[crate::ported::params::SPECIAL_PARAMS_ZSH_START..] // c:845-847
2411 };
2412 for entry in head.iter().chain(tail.iter()) {
2413 // !!! DASH-FAITHFUL GATE (no C counterpart) !!! dash has no
2414 // $RANDOM / $SECONDS: they are ordinary variables there.
2415 if crate::dash_mode::dash_faithful()
2416 && matches!(entry.name, "RANDOM" | "SECONDS")
2417 {
2418 continue;
2419 }
2420 // c:384/394 IPDEF8/9 — `D|PM_SCALAR|PM_SPECIAL` or
2421 // `D|PM_ARRAY|PM_SPECIAL|PM_DONTIMPORT`.
2422 //
2423 // Mask `entry.pm_flags` to the attribute bits that
2424 // may be OR'd onto an existing Param.
2425 //
2426 // PM_READONLY IS included. C declares it on 16 rows
2427 // via `PM_READONLY_SPECIAL` (c:Src/zsh.h:1925 —
2428 // `PM_SPECIAL|PM_READONLY|PM_RO_BY_DESIGN`): the
2429 // IPDEF1 pair `#`/`TTYIDLE` (c:304,314), IPDEF2 `-`
2430 // (c:318), the IPDEF4 block `!`/`$`/`?`/`HISTCMD`/
2431 // `LINENO`/`PPID`/`ZSH_SUBSHELL` (c:351-358) plus
2432 // `status` (c:424), IPDEF9 `*`/`@` (c:392-393) and
2433 // `zsh_eval_context` (c:438), and IPDEF8
2434 // `ZSH_EVAL_CONTEXT` (c:408). `special_params`
2435 // (params.rs:477+) declares exactly those 16 and no
2436 // others, so the bit lands on precisely C's set.
2437 //
2438 // An earlier revision stripped PM_READONLY here to
2439 // keep internal-runtime writes from tripping
2440 // `assignstrvalue`'s guard. Nearly all of those
2441 // writers already mutate the paramtab node in place
2442 // — exactly as C writes the backing C global behind
2443 // a no-op GSU (c:Src/params.c:351, IPDEF4 uses
2444 // `varint_readonly_gsu` =
2445 // `{intvargetfn, nullintsetfn, stdunsetfn}`). The
2446 // in-place sites are `fusevm_bridge.rs:242,261,13045`
2447 // (ZSH_SUBSHELL bump on `u_val`) and
2448 // `fusevm_bridge.rs:12590,12594,12714,12718` +
2449 // `exec.rs:7651,7655` (zsh_eval_context /
2450 // ZSH_EVAL_CONTEXT push+pop). None call
2451 // `setsparam`/`setiparam`.
2452 //
2453 // The ONE writer that did route through `setsparam`
2454 // was `endparamscope`'s deferred scope-pop restore
2455 // (`params.rs`, the `None =>` arm of the `deferred`
2456 // loop): with no GSU wired it name-routes the
2457 // restore and so re-entered the guard, emitting a
2458 // spurious `read-only variable: NAME` while
2459 // unwinding a scope that had shadowed one. C cannot
2460 // reach the guard there because c:5915-5933 calls
2461 // the setfn directly. That arm now drops the bit for
2462 // the duration of the restore, matching C.
2463 //
2464 // With that handled, restoring the flag costs the
2465 // runtime nothing while making `typeset X=v`,
2466 // `readonly X=v` and `X+=v` reject the way
2467 // c:Src/params.c:3216 does, and making
2468 // `paramtypestr` (c:Src/Modules/parameter.c:75-76)
2469 // emit the `-readonly` component that
2470 // `${parameters[X]}` is read for.
2471 //
2472 // PM_UNSET is included: lookup_special_var arms for
2473 // TRY_BLOCK_ERROR / TRY_BLOCK_INTERRUPT (and other
2474 // PM_UNSET entries with sentinel defaults) check
2475 // this bit to decide between "stored value" vs
2476 // "uninitialized → return -1 sentinel". The flag
2477 // gets cleared by assignstrvalue at c:3660 on any
2478 // write, so it correctly tracks "ever assigned".
2479 // Bug #143 in docs/BUGS.md.
2480 let safe_pm_flags = entry.pm_flags
2481 & (PM_TIED | PM_DI | PM_UNSET | crate::ported::zsh_h::PM_READONLY);
2482 // c:Src/params.c — IPDEF macros set PM_TYPE bits
2483 // (PM_INTEGER for IPDEF5/6, PM_ARRAY for IPDEF9,
2484 // PM_HASHED for IPDEF-hash) along with PM_SPECIAL.
2485 // zshrs's previous init only ORed PM_SPECIAL +
2486 // tied/di/unset/readonly — never the type bit. If
2487 // setsparam ran BEFORE init_partab_params (it does
2488 // for OPTIND/SHLVL at vm_helper.rs:874/878), the
2489 // param entry stayed PM_SCALAR and `typeset -p
2490 // OPTIND` emitted `typeset OPTIND=1` instead of
2491 // zsh's `typeset -i10 OPTIND=1`. OR the pm_type
2492 // into the bits so the type attribute lands.
2493 let mut bits = safe_pm_flags | PM_SPECIAL | entry.pm_type;
2494 // c:Src/zsh.h:1925 — `PM_READONLY_SPECIAL` is the
2495 // three-bit set `PM_SPECIAL|PM_READONLY|
2496 // PM_RO_BY_DESIGN`. `special_params` stores only
2497 // PM_READONLY per row (the other two are implied by
2498 // the IPDEF macro), so complete the triple here:
2499 // PM_SPECIAL is already OR'd into `bits` above, and
2500 // this adds the PM_RO_BY_DESIGN companion that
2501 // distinguishes a by-design readonly special from a
2502 // user `readonly` (c:Src/zsh.h:1923).
2503 if (entry.pm_flags & crate::ported::zsh_h::PM_READONLY) != 0 {
2504 bits |= crate::ported::zsh_h::PM_RO_BY_DESIGN;
2505 }
2506 if entry.pm_type == PM_ARRAY {
2507 bits |= PM_DI;
2508 }
2509 let _ = PM_SCALAR;
2510 let _ = PM_DONTIMPORT;
2511 if let Some(pm) = tab.get_mut(entry.name) {
2512 let was_integer =
2513 (pm.node.flags as u32 & crate::ported::zsh_h::PM_INTEGER) != 0;
2514 pm.node.flags |= bits as i32;
2515 // c:Src/params.c:344 IPDEF4 / c:353 IPDEF5 — the
2516 // C struct literal initialises the `base` field
2517 // to 10 for every PM_INTEGER special. zshrs's
2518 // initial paramtab seeding doesn't carry that
2519 // through (the special_paramdef table has no
2520 // `base` field). Set the default here so
2521 // `printparamnode`'s PMTF_USE_BASE arm at
2522 // params.rs:9341 emits "10" between
2523 // `integer` and the name (`integer 10 readonly
2524 // !=0`). Bug #297 in docs/BUGS.md.
2525 if entry.pm_type == crate::ported::zsh_h::PM_INTEGER && pm.base == 0 {
2526 pm.base = 10;
2527 }
2528 // When OR-ing PM_INTEGER onto a param that
2529 // was previously PM_SCALAR (i.e. setsparam ran
2530 // BEFORE init_partab_params, storing the value
2531 // in u_str), parse the u_str into u_val so the
2532 // integer getter reads the correct value. C
2533 // zsh's setsparam-equivalent path detects the
2534 // pm's PM_TYPE first and routes through
2535 // intsetfn, but zshrs's setsparam at the bin
2536 // entry point predates init_partab_params, so
2537 // it lands as PM_SCALAR storage that the
2538 // type-flip needs to migrate.
2539 if !was_integer
2540 && entry.pm_type == crate::ported::zsh_h::PM_INTEGER
2541 && pm.u_val == 0
2542 {
2543 if let Some(ref s) = pm.u_str {
2544 pm.u_val = s.parse::<i64>().unwrap_or(0);
2545 pm.u_str = None;
2546 }
2547 }
2548 // c:Src/zsh.h IPDEF8/IPDEF9 — the third macro
2549 // arg is the tied partner name; mapped into
2550 // `pm->ename` so `typeset -p` can find the
2551 // peer for the PM_TIED swap. Bug #410.
2552 if let Some(peer) = entry.tied_name {
2553 pm.ename = Some(peer.to_string());
2554 }
2555 } else {
2556 // Param hasn't been created yet (e.g. PATH gets
2557 // imported lazily via the env fallback in
2558 // getsparam at params.rs:4104; array specials
2559 // like `pipestatus` / `funcstack` / `dirstack`
2560 // / `zsh_scheduled_events` aren't pre-populated).
2561 // Seed an empty placeholder carrying the
2562 // canonical flag set so subsequent setsparam /
2563 // `(t)X` / `${+X}` observers see the IPDEF
2564 // attribute bits AND `${+X}` returns 1.
2565 let u_arr = if entry.pm_type == PM_ARRAY {
2566 Some(Vec::new())
2567 } else {
2568 None
2569 };
2570 let pm: crate::ported::zsh_h::Param = Box::new(param {
2571 node: hashnode {
2572 next: None,
2573 nam: entry.name.to_string(),
2574 flags: (entry.pm_type as i32) | bits as i32,
2575 },
2576 u_data: 0,
2577 u_tied: None,
2578 u_arr,
2579 u_str: None,
2580 u_val: 0,
2581 u_dval: 0.0,
2582 u_hash: None,
2583 gsu_s: None,
2584 gsu_i: None,
2585 gsu_f: None,
2586 gsu_a: None,
2587 gsu_h: None,
2588 // c:Src/params.c:344 IPDEF4 / c:353 IPDEF5 —
2589 // PM_INTEGER specials default base=10.
2590 base: if entry.pm_type == crate::ported::zsh_h::PM_INTEGER {
2591 10
2592 } else {
2593 0
2594 },
2595 width: 0,
2596 env: None,
2597 // c:Src/zsh.h IPDEF8/IPDEF9 — tied partner
2598 // name. Bug #410.
2599 ename: entry.tied_name.map(|s| s.to_string()),
2600 old: None,
2601 level: 0,
2602 });
2603 tab.insert(entry.name.to_string(), pm);
2604 }
2605 // Tied partner side. The previous loop body ORed
2606 // PM_ARRAY|PM_SPECIAL|PM_DONTIMPORT|PM_TIED onto the
2607 // partner indiscriminately, but for a SCALAR ↔
2608 // ARRAY tied pair (PATH ↔ path, FIGNORE ↔ fignore),
2609 // that incorrectly stamped PM_ARRAY onto the scalar
2610 // partner (FIGNORE, PATH, FPATH, MAILPATH, MANPATH,
2611 // PSVAR, CDPATH, MODULE_PATH). Result: `(t)PATH`
2612 // returned `array-tied-export-special` instead of
2613 // `scalar-tied-export-special`.
2614 //
2615 // Both partners are already listed in `special_params`
2616 // (the scalar at the IPDEF8 block, the array at the
2617 // IPDEF9 block past the sentinel), so each gets its
2618 // own pass through this loop and ends up with the
2619 // correct flags. No cross-stamping needed.
2620 let _ = entry.tied_name;
2621 }
2622 }
2623 };
2624 // c:Src/init.c:1277 — `inittyptab(); /* initialize the ztypes table */`
2625 // runs inside setupvals BEFORE `createparamtable()` (c:Src/init.c:1286).
2626 // This executor is the fusevm runtime's createparamtable entry point,
2627 // and the seeding below reaches `isident()` (WORDCHARS, …), which is
2628 // typtab-driven — with a zeroed typtab every name fails IIDENT and the
2629 // seed aborts with "not an identifier: WORDCHARS".
2630 crate::ported::utils::inittyptab(); // c:1277
2631 crate::startup_trace::mark("inittyptab");
2632 stamp_special_params(); // c:838-847 — create in C's order
2633 // Standard zsh scalar param defaults — direct port of
2634 // `createparamtable` (Src/params.c:817-988) + the `setupvals`
2635 // tail. Writes through canonical `setsparam` (Src/params.c:3350).
2636 //
2637 // c:params.c:972-973 — ZSH_VERSION / ZSH_PATCHLEVEL.
2638 // `zsh_version::ZSH_VERSION` (emitted by build.rs from the
2639 // vendored `Config/version.mk`) is the development snapshot
2640 // tag `5.9.0.3-test`; shipped zsh binaries report the clean
2641 // release form (`5.9`). Bug #73 in docs/BUGS.md — cross-shell
2642 // scripts that gate on `[[ $ZSH_VERSION = 5.9 ]]` or split on
2643 // `.` expecting MAJOR.MINOR break on the `-test` suffix.
2644 //
2645 // Use the cleaned `patchlevel::ZSH_VERSION` here ("5.9") and
2646 // surface the full snapshot tag as `$ZSHRS_VERSION` for
2647 // zshrs-specific identity checks.
2648 // ZSH_VERSION / ZSH_PATCHLEVEL / ZSHRS_VERSION / ZSH_NAME /
2649 // ZSH_ARGZERO are NOT seeded here: C creates them at the END of
2650 // `createparamtable` (c:970-973, after the environ import) and
2651 // ZSH_NAME at `Src/init.c:1364` (setupvals, later still). They
2652 // are seeded at those C positions further down, because a name
2653 // created before the import lands in a different chain slot —
2654 // `ZSH_NAME` seeded here came out BEHIND every same-bucket
2655 // environment variable in `${(k)parameters}` instead of ahead
2656 // of them (c:Src/hashtable.c:214-215 front-insert).
2657 setsparam("WORDCHARS", "*?_-.[]~=/&;!#$%^(){}<>");
2658 // SHLVL is NOT seeded here. c:Src/params.c:948-951 increments it
2659 // AFTER the environ-import loop, so the +1 lives at the end of that
2660 // loop below — see the `c:948-951` block. Doing it here instead meant
2661 // parsing the raw env string with `parse::<i32>()`, which mis-read
2662 // every non-decimal form C accepts via zstrtol_underscore
2663 // (SHLVL=0x10 must give 17, 010 → 9, 1_0 → 11, 9abc → 10, abc → 1).
2664 // POSIX/zsh default IFS: space + tab + newline + NUL.
2665 setsparam("IFS", " \t\n\0");
2666 // POSIX getopts: OPTIND starts at 1.
2667 setsparam("OPTIND", "1");
2668 // Note: OPTERR is NOT pre-initialised. zsh leaves it unset
2669 // even after `getopts` calls (verified: `getopts ":a" opt -a`
2670 // does not set it). It's a user-writable variable that
2671 // starts unset. Bug #150 in docs/BUGS.md.
2672 // zsh wipes inherited `$_` (unlike bash).
2673 setsparam("_", "");
2674 // c:params.c:5064 — histchars derives from bangchar+hatchar+
2675 // hashchar (defaults `!`, `^`, `#`). At init the special
2676 // entry may not exist yet — fall back to the literal default.
2677 let histchars_val = paramtab()
2678 .read()
2679 .ok()
2680 .and_then(|t| {
2681 t.get("histchars")
2682 .or_else(|| t.get("HISTCHARS"))
2683 .map(|pm| histcharsgetfn(pm))
2684 })
2685 .unwrap_or_else(|| "!^#".to_string());
2686 setsparam("histchars", &histchars_val);
2687
2688 // c:Src/params.c:870-871 — `setsparam("TIMEFMT", ...)` etc.
2689 // Seed TIMEFMT explicitly so `${(k)parameters}` lists it
2690 // (the createparamtable() ported in ported::params isn't
2691 // invoked from this bin entry — its setsparam calls don't
2692 // run, so TIMEFMT only existed via the lookup_special_var
2693 // fallback, which scanpmparameters can't see).
2694 setsparam("TIMEFMT", crate::ported::zsh_system_h::DEFAULT_TIMEFMT);
2695 // c:Src/params.c:892 — `setsparam("TMPPREFIX",
2696 // ztrdup_metafy(DEFAULT_TMPPREFIX));`, the line immediately
2697 // before the TIMEFMT seed above. `DEFAULT_TMPPREFIX` is
2698 // "/tmp/zsh" (c:configure.ac:3030 → config.h). Same reason as
2699 // TIMEFMT: createparamtable() is not reached from this bin
2700 // entry, so without this seed `$TMPPREFIX` existed only when
2701 // the environment happened to export it — every scrubbed-env
2702 // launch (cron, launchd/systemd unit, container entrypoint,
2703 // `env -i`) left it unset and every temp-file path derived
2704 // from it fell back per-call-site.
2705 //
2706 // C seeds unconditionally BEFORE the import loop (c:870 vs
2707 // c:893+) and the import then overwrites via assignsparam, so
2708 // an exported $TMPPREFIX still wins. zshrs's import only
2709 // rewrites an entry that is still PM_UNSET, so the env value is
2710 // resolved HERE instead — same end state, and the node is
2711 // created at C's position in the bucket chain. Skipping the
2712 // seed when the environment had TMPPREFIX (the previous shape)
2713 // deferred creation into the import loop, which put TMPPREFIX
2714 // behind every environment variable that hashes to its bucket.
2715 //
2716 // The lookup reads the process-entry environ snapshot, the same
2717 // source the import loop below walks (see the `environ` static
2718 // in ported::params for why the live environment is not it).
2719 let env_at_entry = |name: &str| -> Option<String> {
2720 crate::ported::params::environ
2721 .get()
2722 .and_then(|v| {
2723 v.iter()
2724 .find(|(k, _)| k == name)
2725 .map(|(_, val)| val.clone())
2726 })
2727 .or_else(|| std::env::var(name).ok())
2728 };
2729 setsparam(
2730 "TMPPREFIX",
2731 env_at_entry("TMPPREFIX")
2732 .as_deref()
2733 .unwrap_or(crate::ported::config_h::DEFAULT_TMPPREFIX),
2734 ); // c:870
2735 // c:Src/init.c:1214-1215 — `nullcmd = ztrdup("cat");
2736 // readnullcmd = ztrdup(DEFAULT_READNULLCMD);`. Real paramtab
2737 // seeds (NOT read-time fallbacks) so `unset NULLCMD` truly
2738 // unsets — the bare-redirect "redirection with no command"
2739 // diagnostic depends on getsparam returning None afterwards.
2740 // c:config.h:48 DEFAULT_READNULLCMD "more" — the parity
2741 // floor agrees: scrubbed-env Homebrew zsh 5.9.1 -fc reports
2742 // READNULLCMD=more (probed; the previous macOS arm's "less"
2743 // guess came from the USER's env exporting READNULLCMD=less
2744 // — zpwr sets it). This block runs AFTER the env import, so
2745 // these are DEFAULT seeds only: an env-imported value must
2746 // win (C seeds before the import loop, c:854-885 vs c:893+).
2747 if getsparam("NULLCMD").map_or(true, |v| v.is_empty()) {
2748 setsparam("NULLCMD", "cat");
2749 }
2750 if getsparam("READNULLCMD").map_or(true, |v| v.is_empty()) {
2751 setsparam("READNULLCMD", crate::ported::config_h::DEFAULT_READNULLCMD);
2752 }
2753 // c:Src/params.c:873-876 — `gethostname(hostnam, 256);
2754 // setsparam("HOST", ztrdup_metafy(hostnam));`
2755 // Seeded HERE, before the import loop, exactly like C; it used
2756 // to run at the very end of this constructor, which put HOST
2757 // ahead of same-bucket specials (PROMPT) that C creates first.
2758 // The env value is resolved up front for the same reason as
2759 // TMPPREFIX above (C's import would overwrite it).
2760 let mut host_buf = [0u8; 256];
2761 let host_rc = unsafe { libc::gethostname(host_buf.as_mut_ptr() as *mut libc::c_char, 256) }; // c:874
2762 let hostname = if host_rc == 0 {
2763 std::ffi::CStr::from_bytes_until_nul(&host_buf)
2764 .ok()
2765 .and_then(|c| c.to_str().ok())
2766 .unwrap_or("")
2767 .to_string()
2768 } else {
2769 String::new()
2770 };
2771 setsparam("HOST", env_at_entry("HOST").as_deref().unwrap_or(&hostname)); // c:875
2772 // c:Src/params.c:878-882 — `setsparam("LOGNAME", (str = getlogin())
2773 // && *str ? ztrdup_metafy(str) : ztrdup(cached_username));`
2774 // Also pre-import in C (c:878 vs c:893+); creating it during the
2775 // import instead put LOGNAME behind the environment variables
2776 // sharing its bucket.
2777 let logname_default = {
2778 let from_getlogin = unsafe {
2779 let p = libc::getlogin(); // c:880
2780 if p.is_null() {
2781 String::new()
2782 } else {
2783 std::ffi::CStr::from_ptr(p).to_string_lossy().into_owned()
2784 }
2785 };
2786 if from_getlogin.is_empty() {
2787 crate::ported::utils::get_username() // c:882 cached_username
2788 } else {
2789 from_getlogin
2790 }
2791 };
2792 setsparam(
2793 "LOGNAME",
2794 env_at_entry("LOGNAME")
2795 .as_deref()
2796 .unwrap_or(&logname_default),
2797 ); // c:878
2798 // c:Src/init.c:1186-1193 — default prompt strings. zsh sets
2799 // PS4 to "+%N:%i> " for ZSH emulation ("+ " for KSH/SH).
2800 // Without seeding, PS4 reads empty and `set -x` output has
2801 // no prefix at all. Bug #92 in docs/BUGS.md.
2802 //
2803 // C zsh runs createparamtable's env-import loop (c:893-924)
2804 // BEFORE init.c:1186 fires, so an exported $PS4 in the parent
2805 // env wins over the default seed. zshrs's env import happens
2806 // further down in ShellExecutor::new() (at the createparamtable
2807 // call site), so getsparam() reads None here even when env has
2808 // a value, and the default would clobber the user's PS4.
2809 //
2810 // Additional wrinkle: C zsh's PROMPT / PROMPT2 / PROMPT3 /
2811 // PROMPT4 params are ALIASES for PS1..PS4 (Src/params.c:381,
2812 // 415-421 — both IPDEF7R entries bind to the same `prompt*`
2813 // global). So `export PROMPT4=...` in the parent env sets the
2814 // shared global, and `$PS4` reads the same string. The user's
2815 // interactive shell exports PROMPT4 (the form zsh's prompt
2816 // theme system uses), so when zshrs -x runs, PROMPT4 is in
2817 // env but PS4 is not. Without aliasing in the env-probe step,
2818 // zshrs seeds default PS4 and ignores the user's customised
2819 // prefix.
2820 //
2821 // Probe env::var directly for the name AND its alias; first
2822 // non-empty wins. Only fall through to the default seed when
2823 // every candidate is empty. Mirrors C zsh's behavior without
2824 // reshuffling the rest of new(). Bug: `zshrs -x` ignored the
2825 // user's custom PS4/PROMPT4 unless re-forwarded with
2826 // `PS4=$PROMPT4 zshrs -x`.
2827 // !!! DASH-FAITHFUL GATE (no C counterpart) !!! dash has no `PROMPTn` aliases
2828 // (an imported `PROMPT4` is an ordinary variable) and always starts with
2829 // `PS1='$ '` (`# ` for root), `PS2='> '`, `PS4='+ '`, interactive or not.
2830 let dash_prompts = crate::extensions::dash_mode::dash_faithful();
2831 // c:Src/init.c:1212-1213 — `ifs = EMULATION(EMULATE_KSH|EMULATE_SH) ?
2832 // ztrdup(DEFAULT_IFS_SH) : ztrdup(DEFAULT_IFS)`: the sh default has no
2833 // Meta+space tail. An inherited `IFS` still wins (importenv).
2834 if dash_prompts && std::env::var_os("IFS").is_none() {
2835 *crate::ported::params::ifs_lock().lock().unwrap() =
2836 Some(crate::ported::zsh_h::DEFAULT_IFS_SH.to_string());
2837 }
2838 let seed_prompt = |name: &str, alias: Option<&str>, default: &str| {
2839 let alias = if dash_prompts { None } else { alias };
2840 let default = match (dash_prompts, name) {
2841 (true, "PS1") => {
2842 if crate::ported::utils::privasserted() {
2843 "# "
2844 } else {
2845 "$ "
2846 }
2847 }
2848 (true, "PS2") => "> ",
2849 (true, "PS4") => "+ ",
2850 _ => default,
2851 };
2852 let cur = crate::ported::params::getsparam(name);
2853 let have_param = cur.as_deref().map_or(false, |s| !s.is_empty());
2854 if have_param {
2855 return;
2856 }
2857 // Probe primary name first, then the C-side alias.
2858 // An EMPTY exported value counts: C's env import (c:893-924)
2859 // assigns whatever `environ` holds, empty string included, and
2860 // it runs before the c:1196 defaults, so `export PS1=` yields
2861 // an empty prompt rather than `%m%# `. Testing only for a
2862 // NON-empty value skipped that case and re-seeded the default.
2863 for candidate in std::iter::once(name).chain(alias.into_iter()) {
2864 if let Ok(env_val) = std::env::var(candidate) {
2865 setsparam(name, &env_val);
2866 return;
2867 }
2868 }
2869 setsparam(name, default);
2870 };
2871 seed_prompt("PS4", Some("PROMPT4"), "+%N:%i> ");
2872 // c:Src/init.c:1181-1190 —
2873 // if(unset(INTERACTIVE)) {
2874 // prompt = ztrdup("");
2875 // prompt2 = ztrdup("");
2876 // } else ... {
2877 // prompt = ztrdup("%m%# ");
2878 // prompt2 = ztrdup("%_> ");
2879 // }
2880 // Non-interactive shells get EMPTY primary/secondary prompts
2881 // — `zsh -fc 'typeset'` lists PS1='' — while interactive ones
2882 // get the %m%# defaults. PS3/PS4/SPROMPT are seeded
2883 // unconditionally in C (c:1191-1194). PS1 may be reset by the
2884 // prompt-theme layer; only seed when the slot is empty so any
2885 // prior theme write wins.
2886 let interactive = crate::ported::zsh_h::isset(crate::ported::zsh_h::INTERACTIVE);
2887 seed_prompt(
2888 "PS1",
2889 Some("PROMPT"),
2890 if interactive { "%m%# " } else { "" },
2891 );
2892 seed_prompt(
2893 "PS2",
2894 Some("PROMPT2"),
2895 if interactive { "%_> " } else { "" },
2896 );
2897 // c:Src/init.c:1191 — `prompt3 = ztrdup("?# ");`
2898 seed_prompt("PS3", Some("PROMPT3"), "?# ");
2899 // c:Src/init.c:1194 — `sprompt = ztrdup("zsh: correct '%R'
2900 // to '%r' [nyae]? ");` — spelling-correction prompt.
2901 seed_prompt("SPROMPT", None, "zsh: correct '%R' to '%r' [nyae]? ");
2902 // c:Src/params.c:417-422 — `PROMPT*` aliases for `PS*`.
2903 // C zsh's IPDEF7("PROMPT", &prompt), IPDEF7("PROMPT2",
2904 // &prompt2), IPDEF7("PROMPT3", &prompt3), IPDEF7("PROMPT4",
2905 // &prompt4) all point to the same C globals as the matching
2906 // IPDEF7("PS{1..4}", ...) entries — they're aliases in C,
2907 // sharing storage. zshrs's paramtab keeps them as separate
2908 // entries; mirror the alias by mirroring the value here.
2909 // Bug #274 in docs/BUGS.md (PROMPT3 was the visible report;
2910 // PROMPT/PROMPT2/PROMPT4 had the same gap silently).
2911 for (alias, source) in &[
2912 ("PROMPT", "PS1"),
2913 ("PROMPT2", "PS2"),
2914 ("PROMPT3", "PS3"),
2915 ("PROMPT4", "PS4"),
2916 ] {
2917 if dash_prompts {
2918 continue;
2919 }
2920 if crate::ported::params::getsparam(alias).map_or(true, |s| s.is_empty()) {
2921 if let Some(v) = crate::ported::params::getsparam(source) {
2922 setsparam(alias, &v);
2923 }
2924 }
2925 }
2926 // c:params.c:858-860 — standard non-special param defaults.
2927 // C uses `setiparam(...)` (PM_INTEGER) for these so
2928 // `(t)MAILCHECK` etc. report `integer`. zshrs previously
2929 // routed through `setsparam` (PM_SCALAR) — the value worked
2930 // but the type bit was wrong, breaking
2931 // `case "${(t)LISTMAX}" in *integer*)` and any path that
2932 // gates on arithmetic-typed semantics. Bug #268 in
2933 // docs/BUGS.md.
2934 crate::ported::params::setiparam("MAILCHECK", 60); // c:858
2935 crate::ported::params::setiparam("KEYTIMEOUT", 40); // c:859
2936 crate::ported::params::setiparam("LISTMAX", 100); // c:860
2937 // c:config.h:1004 — MAX_FUNCTION_DEPTH=500. Advisory cap;
2938 // dispatch_function_call enforces against this.
2939 crate::ported::params::setiparam("FUNCNEST", 500);
2940
2941 // Run setlocale(LC_ALL, "") so nl_langinfo() (used by the
2942 // `langinfo` module) returns the host's actual locale instead
2943 // of the C/POSIX default ("US-ASCII"). Direct port of zsh's
2944 // Src/init.c:1208 setlocale call. unsafe { } around libc is
2945 // standard for this exact use-case — setlocale is process-
2946 // global and must run once at startup.
2947 unsafe {
2948 libc::setlocale(libc::LC_ALL, c"".as_ptr());
2949 }
2950
2951 // c:hashtable.c:1206 createaliastables() — seeds aliastab with
2952 // the `run-help` / `which-command` defaults. Run once at shell
2953 // init so the canonical port owns the default-alias set; the
2954 // Executor's `aliases` HashMap then mirrors aliastab.
2955 crate::ported::hashtable::createaliastables();
2956 crate::startup_trace::mark("createaliastables");
2957 // Build the initial $path tied array as a local — fans out
2958 // to paramtab below; no ShellExecutor mirror anymore.
2959 let mut arrays: HashMap<String, Vec<String>> = HashMap::new();
2960 // c:Src/init.c:1125-1130 — `path` starts as the compiled-in default;
2961 // an inherited $PATH replaces it in the environ import. With no PATH
2962 // in the environment the default stands and is not exported.
2963 let path_dirs: Vec<String> = match env::var("PATH") {
2964 Ok(p) => p.split(':').map(|s| s.to_string()).collect(),
2965 Err(_) => ["/bin", "/usr/bin", "/usr/ucb", "/usr/local/bin"] // c:1126-1129
2966 .iter()
2967 .map(|s| s.to_string())
2968 .collect(),
2969 };
2970 arrays.insert("path".to_string(), path_dirs);
2971 let mut exec = Self {
2972 // c:Src/init.c:479 — `-c` mode: scriptname = scriptfilename
2973 // = ztrdup("zsh"). Both start at the literal "zsh".
2974 // dispatch_function_call overrides scriptname per c:5903;
2975 // scriptfilename stays at the outer file.
2976 scriptname: Some("zsh".to_string()),
2977 scriptfilename: Some("zsh".to_string()),
2978 subshell_snapshots: Vec::new(),
2979 inline_env_stack: Vec::new(),
2980 current_command_glob_failed: std::cell::Cell::new(false),
2981 jobs: JobTable::new(),
2982 fpath,
2983 history,
2984 completions: HashMap::new(),
2985 process_sub_counter: 0,
2986 zstyles: Vec::new(),
2987 local_scope_depth: 0,
2988 pending_underscore: None,
2989 in_dq_context: 0,
2990 in_scalar_assign: 0,
2991 profiling_enabled: false,
2992 compsys_cache: std::cell::OnceCell::new(),
2993 compinit_pending: None, // (receiver, start_time)
2994 plugin_cache: {
2995 let pc_path = crate::plugin_cache::default_cache_path();
2996 if let Some(parent) = pc_path.parent() {
2997 let _ = fs::create_dir_all(parent);
2998 }
2999 match crate::plugin_cache::PluginCache::open(&pc_path) {
3000 Ok(pc) => {
3001 let (plugins, functions) = pc.stats();
3002 tracing::info!(
3003 plugins,
3004 cached_functions = functions,
3005 path = %pc_path.display(),
3006 "plugin_cache: sqlite opened"
3007 );
3008 Some(pc)
3009 }
3010 Err(e) => {
3011 // A corrupt cache file is not a permanent condition:
3012 // `plugins.db` is derived data, rebuilt by the next
3013 // plugin scan. SQLite answers a clobbered header with
3014 // SQLITE_NOTADB ("file is not a database"), and the
3015 // old behaviour logged that and moved on — so the
3016 // cache stayed dead for every future shell too, with
3017 // nothing but a log line to say why plugin lookups
3018 // were slow. Discard the file and open once more;
3019 // this is the same "drop it and rebuild silently"
3020 // rule the shard cache already applies to a version
3021 // mismatch. A second failure keeps the old
3022 // behaviour, so a permissions problem still degrades
3023 // instead of looping.
3024 let corrupt = matches!(
3025 e.sqlite_error_code(),
3026 Some(rusqlite::ErrorCode::NotADatabase)
3027 | Some(rusqlite::ErrorCode::DatabaseCorrupt)
3028 );
3029 if corrupt {
3030 tracing::warn!(
3031 error = %e,
3032 path = %pc_path.display(),
3033 "plugin_cache: corrupt — discarding and rebuilding"
3034 );
3035 let _ = fs::remove_file(&pc_path);
3036 // The -wal/-shm side files belong to the database
3037 // that just went away; leaving them makes the
3038 // fresh open inherit a journal for a file that no
3039 // longer exists.
3040 for side in ["-wal", "-shm"] {
3041 let mut p = pc_path.clone().into_os_string();
3042 p.push(side);
3043 let _ = fs::remove_file(PathBuf::from(p));
3044 }
3045 match crate::plugin_cache::PluginCache::open(&pc_path) {
3046 Ok(pc) => {
3047 tracing::info!(
3048 path = %pc_path.display(),
3049 "plugin_cache: rebuilt after corruption"
3050 );
3051 Some(pc)
3052 }
3053 Err(e2) => {
3054 tracing::warn!(error = %e2, "plugin_cache: reopen after discard failed");
3055 None
3056 }
3057 }
3058 } else {
3059 tracing::warn!(error = %e, "plugin_cache: failed to open");
3060 None
3061 }
3062 }
3063 }
3064 },
3065 deferred_compdefs: Vec::new(),
3066 returning: None,
3067 zsh_compat: false,
3068 bash_compat: false,
3069 posix_mode: false,
3070 worker_pool: {
3071 let config = crate::config::load();
3072 let pool_size = crate::config::resolve_pool_size(&config.worker_pool);
3073 std::sync::Arc::new(crate::worker::WorkerPool::new(pool_size))
3074 },
3075 intercepts: Vec::new(),
3076 async_jobs: HashMap::new(),
3077 next_async_id: 1,
3078 redirect_scope_stack: Vec::new(),
3079 multios_scope_stack: Vec::new(),
3080 exec_redirs_permanent: false,
3081 pipe_output_pending: false,
3082 pipe_output_scope: None,
3083 redirect_failed: false,
3084 functions_compiled: crate::cow_map::CowHashMap::new(),
3085 function_source: crate::cow_map::CowHashMap::new(),
3086 function_line_base: HashMap::new(),
3087 function_def_file: HashMap::new(),
3088 prompt_funcstack: Vec::new(),
3089 tied_array_to_scalar: HashMap::new(),
3090 ztest_pass_count: std::sync::atomic::AtomicUsize::new(0),
3091 ztest_fail_count: std::sync::atomic::AtomicUsize::new(0),
3092 ztest_skip_count: std::sync::atomic::AtomicUsize::new(0),
3093 ztest_pass_total: std::sync::atomic::AtomicUsize::new(0),
3094 ztest_fail_total: std::sync::atomic::AtomicUsize::new(0),
3095 ztest_skip_total: std::sync::atomic::AtomicUsize::new(0),
3096 ztest_run_failed: std::sync::atomic::AtomicBool::new(false),
3097 ztest_suppress_stdout: false,
3098 };
3099 // Publish the session worker pool so preprompt-time async hooks
3100 // (async_precmd) can reach it without an entered executor context.
3101 crate::async_precmd::set_session_pool(std::sync::Arc::clone(&exec.worker_pool));
3102 crate::startup_trace::mark("worker pool ready");
3103 // Mirror env-derived path arrays into the `arrays` table so
3104 // user-level `fpath` / `path` array reads see the inherited
3105 // entries. zsh: `fpath+=…` should append to the inherited
3106 // 43-entry array, not replace it. Same for `path` (PATH).
3107 crate::startup_trace::mark("exec struct built");
3108 let fpath_arr: Vec<String> = exec
3109 .fpath
3110 .iter()
3111 .map(|p| p.to_string_lossy().to_string())
3112 .collect();
3113 if !fpath_arr.is_empty() {
3114 exec.set_array("fpath".to_string(), fpath_arr);
3115 }
3116 if let Ok(path) = env::var("PATH") {
3117 let path_arr: Vec<String> = path
3118 .split(':')
3119 .filter(|s| !s.is_empty())
3120 .map(String::from)
3121 .collect();
3122 if !path_arr.is_empty() {
3123 exec.set_array("path".to_string(), path_arr);
3124 }
3125 }
3126 // Register the standard tied path-family pairs so `path+=` /
3127 // `fpath+=` / etc. mirror through the array→scalar sync hook
3128 // in BUILTIN_APPEND_ARRAY (and the SET_ARRAY tied path).
3129 // Direct port of the implicit ties that zsh wires up at
3130 // startup for PATH/path, FPATH/fpath, etc. Source-of-truth
3131 // for the pairs is Src/init.c's `setupvals()` PM_TIED entries.
3132 // c:Src/params.c:395-422 IPDEF8 — full PM_TIED colonarr list:
3133 // CDPATH, FIGNORE, FPATH, MAILPATH, PATH, PSVAR, MODULE_PATH,
3134 // MANPATH (ZSH_EVAL_CONTEXT is readonly-special, excluded).
3135 for (scalar, arr) in [
3136 ("PATH", "path"),
3137 ("FPATH", "fpath"),
3138 ("MANPATH", "manpath"),
3139 ("CDPATH", "cdpath"),
3140 ("MODULE_PATH", "module_path"),
3141 ("PSVAR", "psvar"),
3142 ("FIGNORE", "fignore"),
3143 ("MAILPATH", "mailpath"),
3144 ] {
3145 exec.tied_array_to_scalar
3146 .insert(arr.to_string(), (scalar.to_string(), ":".to_string()));
3147 }
3148
3149 // Pour `path` (from env PATH split) into paramtab. The IPDEF9
3150 // flag set was stamped by the c:838-847 pass above and survives
3151 // the assignment (`assignaparam` keeps PM_DONTIMPORT on a
3152 // PM_SPECIAL node — c:3374 + params.rs), so no re-stamp is
3153 // needed here.
3154 for (k, v) in &arrays {
3155 setaparam(k, v.clone()); // c:params.c:3595
3156 }
3157
3158 // c:Src/params.c:893-924 — the environment import runs AFTER the
3159 // specials table (moved above, c:838-847) and after the c:854-885
3160 // non-special seeds, exactly as `createparamtable` sequences them.
3161 //
3162 // Carries c:951's `addenv(pm, buf)` argument out of the paramtab
3163 // write-lock below: `addenv` takes that same lock itself, so the call
3164 // cannot be made while the guard is alive.
3165 let mut shlvl_env: Option<String> = None;
3166 // !!! LOCK DISCIPLINE — NO GSU CALLBACK MAY RUN UNDER THE PARAMTAB GUARD !!!
3167 //
3168 // C's import loop calls `assignsparam(..., ASSPM_ENV_IMPORT)` (c:907-908)
3169 // and holds no lock at all, so a setfn reached from there is free to read
3170 // any parameter — and several do. `intsetfn` name-dispatches `$COLUMNS` /
3171 // `$LINES` to `zlevarsetfn` (c:4226), which re-enters `adjustwinsize`,
3172 // which falls back to `getsparam("COLUMNS")` when its ioctl comes back
3173 // empty. `paramtab()` is a non-reentrant `std::sync::RwLock`, so running
3174 // that under the write guard below parked the shell on itself:
3175 //
3176 // COLUMNS=200 zshrs -f -c 'print $COLUMNS'
3177 //
3178 // never printed anything and never exited, with `sample` showing
3179 // 2665/2665 samples in `semaphore_wait_trap` down
3180 // `ShellExecutor::new → intsetfn → zlevarsetfn → adjustwinsize →
3181 // adjustlines → getsparam → RwLock::lock_contended`. That is the same
3182 // chain 5929e82b81 diagnosed on the startup path; that fix removed a
3183 // Rust-only return value from `adjustwinsize` so the ONE caller it knew
3184 // about stopped re-entering, and left the pattern in place for this one.
3185 //
3186 // So the dispatches are queued here and replayed after the guard is
3187 // released, the way `shlvl_env` and `tied_env_arrays` already are. One
3188 // queue, not one per signature, because C's order is per-variable and
3189 // must survive: `termsetfn` populates the termcap geometry that a later
3190 // `$LINES` import reads back through `adjustwinsize`.
3191 enum DeferredEnvSetfn {
3192 /// c:2774 — `v->pm->gsu.i->setfn(v->pm, ival)`.
3193 Int(String, i64),
3194 /// c:907-908 — the cached-storage scalar specials' setfn.
3195 Scalar(String, String),
3196 }
3197 let mut deferred_env_setfns: Vec<DeferredEnvSetfn> = Vec::new();
3198 {
3199 use crate::ported::params::paramtab;
3200 if let Ok(mut tab) = paramtab().write() {
3201 use crate::ported::zsh_h::{param, PM_UNSET};
3202 // c:Src/params.c:893-924 environment-import loop —
3203 // every env var gets either a fresh exported paramtab
3204 // entry OR (when the entry pre-exists from
3205 // special_params) PM_EXPORTED OR'd onto its flags.
3206 // Without this, `declare -p PATH` printed `typeset -T
3207 // PATH=''` and `declare -p USER` printed nothing at
3208 // all because USER was never in paramtab.
3209 use crate::ported::zsh_h::hashnode as _hn;
3210 use crate::ported::zsh_h::{PM_EXPORTED, PM_SCALAR};
3211 // c:Src/params.c:4329-4342 colonarrsetfn — assigning a
3212 // tied IPDEF8 scalar (MANPATH, CDPATH, MODULE_PATH, …)
3213 // colonsplit()s the value into the partner array,
3214 // preserving empty components. The env import below
3215 // bypasses the GSU setfn, so collect the tied pairs
3216 // here and pour them through setaparam after the
3217 // paramtab lock drops. PATH→path / FPATH→fpath are
3218 // seeded earlier (vm_helper ~1160-1199) and skipped.
3219 let mut tied_env_arrays: Vec<(String, Vec<String>)> = Vec::new();
3220 // c:Src/params.c:893 — walk the process-entry environ
3221 // snapshot, not the live env (frameworks can mutate it
3222 // before init — see params.rs `environ` static).
3223 let environ_vars: Vec<(String, String)> = crate::ported::params::environ
3224 .get()
3225 .cloned()
3226 .unwrap_or_else(|| std::env::vars().collect());
3227 for (env_name, env_value) in environ_vars {
3228 if env_name.is_empty() || env_name.contains('[') {
3229 continue;
3230 }
3231 if env_name.as_bytes()[0].is_ascii_digit() {
3232 continue;
3233 }
3234 if !crate::ported::params::isident(&env_name) {
3235 continue;
3236 }
3237 if let Some(pm) = tab.get_mut(&env_name) {
3238 // c:Src/params.c:902-906 — the import loop runs
3239 // `dontimport(pm->node.flags)` BEFORE doing
3240 // anything to the entry; PM_DONTIMPORT names
3241 // (`_`, IFS, GID/EGID, KEYBOARD_HACK — the
3242 // IPDEF7/IPDEF2 rows, c:796-800) are skipped
3243 // ENTIRELY: no PM_EXPORTED stamp, no value
3244 // seed. zshrs previously OR'd PM_EXPORTED
3245 // first, so an inherited env `_` made the
3246 // special `_` exported and it leaked into
3247 // `typeset +x -r` / `export -p` listings where
3248 // zsh shows nothing.
3249 if (pm.node.flags as u32 & crate::ported::zsh_h::PM_DONTIMPORT) != 0 {
3250 continue; // c:905 `continue;`
3251 }
3252 pm.node.flags |= PM_EXPORTED as i32;
3253 // c:Src/params.c:2769-2776 — assignstrvalue's
3254 // PM_INTEGER arm, which C's env import reaches via
3255 // `assignsparam(..., ASSPM_ENV_IMPORT)` (c:907-908;
3256 // assignsparam forwards its `flags` verbatim at
3257 // c:params.c assignstrvalue(v, val, flags)):
3258 // if (flags & ASSPM_ENV_IMPORT) {
3259 // char *ptr;
3260 // ival = zstrtol_underscore(val, &ptr, 0, 1);
3261 // } else
3262 // ival = mathevali(val);
3263 // v->pm->gsu.i->setfn(v->pm, ival);
3264 // An integer param keeps its value in `u.val`, NOT in
3265 // the scalar slot, so the `pm.u_str = env_value` seed
3266 // below stored the digits somewhere no integer reader
3267 // ever looks and EVERY pre-existing PM_INTEGER param
3268 // silently ignored the environment: COLUMNS/LINES
3269 // (IPDEF5, c:355-356) read back 0, while HISTSIZE,
3270 // SAVEHIST, LISTMAX, MAILCHECK, KEYTIMEOUT and
3271 // FUNCNEST kept their built-in defaults — i.e.
3272 // `HISTSIZE=5000 zshrs -c ...` was a no-op.
3273 //
3274 // Base 0 + underscore=1 is not incidental: it is what
3275 // makes `COLUMNS=0x10` 16, `COLUMNS=0b101` 5,
3276 // `COLUMNS=010` 8 (octal — c:utils.c:2452-2461 takes
3277 // the leading `0` then falls to `base = 8`),
3278 // `COLUMNS=1_0` 10, and a trailing-garbage value like
3279 // `9abc` a silent 9. mathevali would instead ERROR on
3280 // `9abc`, which is precisely why C splits the two
3281 // paths — importing a hostile environment must not
3282 // abort the shell (upstream 546203a770, "33276: safer
3283 // import of numerical variables from environment").
3284 if (pm.node.flags as u32 & crate::ported::zsh_h::PM_INTEGER) != 0 {
3285 let (ival, _) =
3286 crate::ported::utils::zstrtol_underscore(&env_value, 0, true); // c:2773
3287 // c:3660 — any assignstrvalue write clears PM_UNSET.
3288 pm.node.flags &= !(PM_UNSET as i32);
3289 // c:2774 — `v->pm->gsu.i->setfn(v->pm, ival)`.
3290 // intsetfn is this port's stand-in for the gsu_i
3291 // vtable: it name-dispatches the specials whose
3292 // setter has side effects (SECONDS, RANDOM,
3293 // HISTSIZE, …) and writes u.val otherwise. Queued,
3294 // not called: see the lock-discipline note above.
3295 //
3296 // The plain storage half of the setfn still lands
3297 // HERE, synchronously, because C's write is
3298 // synchronous and code further down this same guard
3299 // reads it back: the c:948-951 SHLVL block computes
3300 // `++shlvl` from the value this loop just imported,
3301 // and the C comment on it ("the increment must
3302 // observe the imported value") is load-bearing.
3303 // Deferring the storage too made `SHLVL=5 zshrs -f
3304 // -c 'print $SHLVL'` answer 1 instead of 6. The
3305 // queued dispatch then re-applies it along with the
3306 // side effects only it can produce.
3307 pm.u_val = ival; // c:2774 (the `pm->u.val = x` half)
3308 deferred_env_setfns
3309 .push(DeferredEnvSetfn::Int(env_name.clone(), ival)); // c:2774
3310 pm.env = Some(format!("{env_name}={env_value}"));
3311 continue;
3312 }
3313 // c:Src/params.c:893-924 — C's env-import calls
3314 // `assignsparam(..., ASSPM_ENV_IMPORT)` which
3315 // routes through the param's GSU setfn. For
3316 // SPECIAL scalars with cached storage (HOME,
3317 // USERNAME, TERM, WORDCHARS, TERMINFO,
3318 // TERMINFO_DIRS, KEYBOARD_HACK, histchars) the
3319 // setfn writes to a separate `*_lock` global
3320 // (e.g. home_lock). Just OR'ing PM_EXPORTED
3321 // leaves those globals empty, so `$HOME` reads
3322 // back "" even though HOME is in env. Mirror
3323 // C by copying the env value into pm.u_str and
3324 // (for cached specials) the matching global.
3325 // Only seed cached state when the param was
3326 // still marked PM_UNSET — i.e. nothing has set
3327 // it yet. ShellExecutor::new's earlier init
3328 // block (vm_helper line 837+) already ran
3329 // setsparam for a few names (ZSH_ARGZERO,
3330 // WORDCHARS, SHLVL with the +1 increment, IFS,
3331 // OPTIND, …); those calls clear PM_UNSET so we
3332 // must not overwrite them with the raw env
3333 // value here. The PM_UNSET-still-set case is
3334 // the "C zsh would have called
3335 // assignsparam(...,ASSPM_ENV_IMPORT) and ours
3336 // didn't yet" gap that bug #599 (HOME=` `) and
3337 // %~ prompt expansion need.
3338 let still_unset =
3339 (pm.node.flags as u32 & crate::ported::zsh_h::PM_UNSET) != 0;
3340 if still_unset {
3341 pm.u_str = Some(env_value.clone());
3342 pm.env = Some(format!("{}={}", env_name, env_value));
3343 // c:Src/params.c:3660 — `assignstrvalue`
3344 // clears PM_UNSET on any write. HOME / TERM
3345 // / TERMINFO / TERMINFO_DIRS / WORDCHARS
3346 // start life with PM_UNSET in
3347 // `special_params` (params.rs SPECIAL_PARAMS
3348 // table) so `lookup_special_var` skips the
3349 // getfn for uninitialized specials; env
3350 // import is the canonical "now it's set"
3351 // event, so clear the bit.
3352 pm.node.flags &= !(PM_UNSET as i32);
3353 // Cached-state specials: route through
3354 // the matching setfn so the global cache
3355 // (home_lock / wordchars_lock / etc.)
3356 // reflects the env value. Queued, not
3357 // called: see the lock-discipline note
3358 // above. `termsetfn` in particular reaches
3359 // the terminal setup, which is exactly the
3360 // kind of callee that reads parameters.
3361 if matches!(
3362 env_name.as_str(),
3363 "HOME"
3364 | "USERNAME"
3365 | "TERM"
3366 | "WORDCHARS"
3367 | "TERMINFO"
3368 | "TERMINFO_DIRS"
3369 ) {
3370 deferred_env_setfns.push(DeferredEnvSetfn::Scalar(
3371 env_name.clone(),
3372 env_value.clone(),
3373 )); // c:907-908
3374 }
3375 }
3376 // c:Src/params.c:907-908 — env import always
3377 // assigns through the GSU setfn; for tied
3378 // IPDEF8 scalars that is colonarrsetfn
3379 // (c:4329-4342), which colonsplit()s the value
3380 // into the partner array, empties preserved.
3381 // Not gated on still_unset: C re-assigns on
3382 // import regardless.
3383 if (pm.node.flags as u32 & crate::ported::zsh_h::PM_TIED) != 0 {
3384 if let Some(ref peer) = pm.ename {
3385 if peer != "path" && peer != "fpath" {
3386 tied_env_arrays.push((
3387 peer.clone(),
3388 env_value.split(':').map(String::from).collect(), // c:4339 colonsplit
3389 ));
3390 }
3391 }
3392 }
3393 } else {
3394 // Fresh entry — PM_SCALAR + PM_EXPORTED, value
3395 // taken from env. Mirrors C zsh's c:907-908
3396 // `assignsparam(..., ASSPM_ENV_IMPORT)` for
3397 // names not already in the special table.
3398 let pm: crate::ported::zsh_h::Param = Box::new(param {
3399 node: _hn {
3400 next: None,
3401 nam: env_name.clone(),
3402 flags: (PM_SCALAR | PM_EXPORTED) as i32,
3403 },
3404 u_data: 0,
3405 u_tied: None,
3406 u_arr: None,
3407 u_str: Some(env_value.clone()),
3408 u_val: 0,
3409 u_dval: 0.0,
3410 u_hash: None,
3411 gsu_s: None,
3412 gsu_i: None,
3413 gsu_f: None,
3414 gsu_a: None,
3415 gsu_h: None,
3416 base: 0,
3417 width: 0,
3418 env: Some(format!("{}={}", env_name, env_value)),
3419 ename: None,
3420 old: None,
3421 level: 0,
3422 });
3423 tab.insert(env_name, pm);
3424 }
3425 }
3426 // Apply the collected tied-pair splits after the env
3427 // walk. setaparam (the canonical store) needs the
3428 // same paramtab write lock held here, so write u_arr
3429 // directly on the peer entry — array reads route
3430 // through paramtab so this is the single store.
3431 for (peer, parts) in tied_env_arrays {
3432 if let Some(apm) = tab.get_mut(peer.as_str()) {
3433 apm.u_arr = Some(parts); // c:4339 — `*dptr = colonsplit(x, …)`
3434 apm.node.flags &= !(crate::ported::zsh_h::PM_UNSET as i32);
3435 }
3436 }
3437 // c:Src/params.c:948-951 — runs AFTER the import loop:
3438 // pm = (Param) paramtab->getnode(paramtab, "SHLVL");
3439 // sprintf(buf, "%d", (int)++shlvl);
3440 // /* shlvl value in environment needs updating unconditionally */
3441 // addenv(pm, buf);
3442 // SHLVL is `IPDEF5("SHLVL", &shlvl, varinteger_gsu)` (c:358), so
3443 // the loop above has already parsed any inherited value into it
3444 // with zstrtol_underscore; C then increments THAT, in place.
3445 // Ordering is the whole point: the increment must observe the
3446 // imported value, and the import must not clobber the
3447 // increment. When SHLVL is absent from the environment the
3448 // param is still 0 here, so ++ yields 1 — matching C, whose
3449 // `shlvl` global starts at 0.
3450 //
3451 // addenv also exports the INCREMENTED value, which is why a
3452 // forked child sees 6 for `SHLVL=5 zsh -fc 'printenv SHLVL; true'`.
3453 // (A bare `printenv SHLVL` shows 5 because zsh exec's the last
3454 // command in place and backs the increment out — the shell is
3455 // being replaced, not nested. That is a separate mechanism,
3456 // ported for the explicit `exec` form at
3457 // fusevm_bridge.rs's BUILTIN_EXEC.)
3458 if let Some(pm) = tab.get_mut("SHLVL") {
3459 let next = pm.u_val + 1; // c:949 `++shlvl`
3460 // Queued like the import loop's dispatches above — SHLVL's
3461 // setfn is a leaf today, but the rule is the pattern, not
3462 // the current callee list. Storage half applied here for
3463 // the same reason it is up there: c:951's `addenv` and
3464 // everything after this guard must see the incremented
3465 // value.
3466 pm.u_val = next; // c:949 (the `*pm->u.valptr = x` half)
3467 deferred_env_setfns.push(DeferredEnvSetfn::Int("SHLVL".to_string(), next)); // c:949
3468 pm.node.flags &= !(PM_UNSET as i32);
3469 // c:951 — the argument of the unconditional `addenv`.
3470 // C renders it with `sprintf(buf, "%d", (int)shlvl)`
3471 // AFTER the increment, so it is the incremented value.
3472 shlvl_env = Some(next.to_string()); // c:950
3473 }
3474 }
3475 }
3476 // ---- paramtab guard released; replay the queued dispatches ----
3477 //
3478 // This is the c:907-908 / c:2774 / c:949 setfn call, moved to where no
3479 // lock is held, in the order C makes it. Each one runs against a
3480 // detached node — which is what C's pointer-to-the-live-node amounts to
3481 // once the lock is out of the way — and the integer arm publishes the
3482 // union slot the setfn wrote back onto the table afterwards. The scalar
3483 // specials need no publish: their setfns write a cached global
3484 // (`home_lock`, `wordchars_lock`, …) and ignore `pm`, matching C's
3485 // `UNUSED(Param pm)`. A node the setfn removed outright is simply
3486 // skipped.
3487 for deferred in deferred_env_setfns {
3488 match deferred {
3489 DeferredEnvSetfn::Int(name, ival) => {
3490 let staged = crate::ported::params::paramtab()
3491 .read()
3492 .ok()
3493 .and_then(|tab| tab.get(name.as_str()).map(|pm| (**pm).clone()));
3494 let Some(mut staged) = staged else { continue };
3495 crate::ported::params::intsetfn(&mut staged, ival); // c:2774
3496 if let Ok(mut tab) = crate::ported::params::paramtab().write() {
3497 if let Some(pm) = tab.get_mut(name.as_str()) {
3498 pm.u_val = staged.u_val;
3499 }
3500 }
3501 }
3502 DeferredEnvSetfn::Scalar(name, value) => {
3503 let staged = crate::ported::params::paramtab()
3504 .read()
3505 .ok()
3506 .and_then(|tab| tab.get(name.as_str()).map(|pm| (**pm).clone()));
3507 let Some(mut staged) = staged else { continue };
3508 match name.as_str() {
3509 "HOME" => crate::ported::params::homesetfn(&mut staged, value),
3510 "USERNAME" => crate::ported::params::usernamesetfn(&mut staged, value),
3511 "TERM" => crate::ported::params::termsetfn(&mut staged, value),
3512 "WORDCHARS" => crate::ported::params::wordcharssetfn(&mut staged, value),
3513 "TERMINFO" => crate::ported::params::terminfosetfn(&mut staged, value),
3514 "TERMINFO_DIRS" => {
3515 crate::ported::params::terminfodirssetfn(&mut staged, value)
3516 }
3517 _ => {}
3518 }
3519 }
3520 }
3521 }
3522 // c:Src/params.c:960-965 — HOME wiring, which C runs right
3523 // after the environment-import loop:
3524 // pm = (Param) realparamtab->getnode2(realparamtab, "HOME");
3525 // if (EMULATION(EMULATE_ZSH))
3526 // {
3527 // pm->node.flags &= ~PM_UNSET;
3528 // if (!(pm->node.flags & PM_EXPORTED))
3529 // addenv(pm, home);
3530 // } else if (!home)
3531 // pm->node.flags |= PM_UNSET;
3532 // `home` itself was synthesised from the password database
3533 // back in setupvals (c:Src/init.c:1237-1250) BEFORE the import
3534 // loop, so an inherited $HOME wins: the import calls
3535 // `homesetfn` and overwrites the synthesised value.
3536 //
3537 // zshrs reaches neither of those C sites from this bin entry,
3538 // so `$HOME` was whatever the environment supplied and nothing
3539 // else — a scrubbed launch (cron, launchd/systemd unit,
3540 // container entrypoint, `env -i`) got no $HOME at all, and
3541 // every `~`, rc-file path and cache path derived from it
3542 // silently resolved to "" (`~/x` expanded to `/x`).
3543 //
3544 // Ordering is preserved by only synthesising when the import
3545 // produced nothing: `var_os` is the exact "was it in the
3546 // environment" test C's loop keys off, so an explicitly empty
3547 // `HOME=` still stays empty (reference binary: `env -i … HOME=
3548 // zsh -f -c 'print -r -- "[${HOME-UNSET}]"'` prints `[]`).
3549 //
3550 // `home` itself comes from c:Src/init.c:1237-1250, inlined here
3551 // because C has no function to port — it is straight-line code
3552 // inside `setupvals()`:
3553 // #ifdef USE_GETPWUID
3554 // if ((pswd = getpwuid(cached_uid))) {
3555 // if (EMULATION(EMULATE_ZSH))
3556 // home = ztrdup_metafy(pswd->pw_dir);
3557 // cached_username = ztrdup_metafy(pswd->pw_name);
3558 // }
3559 // else
3560 // #endif /* USE_GETPWUID */
3561 // {
3562 // if (EMULATION(EMULATE_ZSH))
3563 // home = ztrdup("/");
3564 // cached_username = ztrdup("");
3565 // }
3566 // Both arms are guarded on EMULATE_ZSH: under sh/ksh emulation
3567 // the C global stays NULL and `$HOME` can only come from the
3568 // environment. Reference binary agrees — `env -i TERM=dumb
3569 // PATH=/usr/bin:/bin zsh --emulate sh -f -c 'echo
3570 // "HOME=${HOME-UNSET}"'` prints `HOME=UNSET`, while the same
3571 // command without `--emulate sh` prints the password-database
3572 // home. `cached_username` is seeded separately (see the
3573 // getlogin() block below), so only the `home` half is here.
3574 if std::env::var_os("HOME").is_none()
3575 && crate::ported::zsh_h::EMULATION(crate::ported::zsh_h::EMULATE_ZSH)
3576 {
3577 // c:Src/init.c:1239 — `getpwuid(cached_uid)`, cached_uid
3578 // being `getuid()` from c:1235.
3579 let pswd = unsafe { libc::getpwuid(libc::getuid()) };
3580 let pw_dir = if pswd.is_null() {
3581 std::ptr::null()
3582 } else {
3583 unsafe { (*pswd).pw_dir }
3584 };
3585 let h = if pw_dir.is_null() {
3586 // c:1248 — password lookup failed: `home = ztrdup("/")`.
3587 // A NULL `pw_dir` on a present entry is the same
3588 // "no usable home" case.
3589 "/".to_string()
3590 } else {
3591 // c:1241 — `home = ztrdup_metafy(pswd->pw_dir)`.
3592 crate::ported::utils::metafy(
3593 &unsafe { std::ffi::CStr::from_ptr(pw_dir) }.to_string_lossy(),
3594 )
3595 };
3596 // Routes through `homesetfn`, so the `home` global and the
3597 // paramtab entry agree (c:Src/params.c:5118).
3598 crate::ported::params::setsparam("HOME", &h);
3599 // c:964-965 — the param is not PM_EXPORTED (it was not
3600 // imported), so C addenv's it. The reference binary
3601 // confirms the synthesised value reaches children:
3602 // `env -i TERM=dumb PATH=/usr/bin:/bin zsh -f -c
3603 // '/usr/bin/env'` lists `HOME=/Users/…`.
3604 crate::ported::params::addenv("HOME", &h);
3605 }
3606
3607 // c:Src/params.c:968-970 — `pm = realparamtab->getnode2(realparamtab,
3608 // "LOGNAME"); if (!(pm->node.flags & PM_EXPORTED)) addenv(pm, pm->u.str);`
3609 //
3610 // Unconditional in C, right after the HOME block above, and the same
3611 // shape: the param was SEEDED (c:878-882, from `getlogin()`) rather
3612 // than imported, so it carries no PM_EXPORTED and C exports it here.
3613 // The faithful port of this lives in `createparamtable`
3614 // (ported/params.rs, c:968-970) but that function is not on this
3615 // entry path — `ShellExecutor::new` is — so the seeded value never
3616 // reached the environment. Measured under a scrubbed env:
3617 //
3618 // env -i HOME=$HOME TERM=xterm PATH=... <shell> -f -c '${(t)LOGNAME}'
3619 // zsh : scalar-export zshrs: scalar
3620 //
3621 // and a child of zshrs saw no LOGNAME at all where a child of zsh did.
3622 // Like the SHLVL case above, C exports a value the import loop did not
3623 // supply; unlike it, there is no paired exec-time adjustment.
3624 if let Some(v) = crate::ported::params::getsparam("LOGNAME") {
3625 crate::ported::params::addenv("LOGNAME", &v); // c:970
3626 }
3627
3628 // c:Src/params.c:951 — `addenv(pm, buf)`. Unconditional: the C comment
3629 // on the line above it is "shlvl value in environment needs updating
3630 // unconditionally". `addenv` both zputenv's the string and stamps
3631 // `pm->node.flags |= PM_EXPORTED` (c:Src/params.c:5482-5484), so this
3632 // one call is what makes `${(t)SHLVL}` read `integer-export-special`
3633 // AND what puts the incremented number in a forked child's
3634 // environment.
3635 //
3636 // Omitting it left $SHLVL unable to count nesting at all — measured
3637 // with each shell nesting itself twice, the inner shells FORKED so
3638 // that no exec-in-place decrement applies:
3639 // zsh L1=1 L2=2 L3=3
3640 // zshrs L1=1 L2=1 L3=1
3641 // because every nested zshrs re-read the grandparent's stale
3642 // environment entry instead of the parent's incremented one. It also
3643 // left the parameter unexported under a scrubbed environment
3644 // (`env -i … -f -c '${(t)SHLVL}'`: zsh `integer-export-special`,
3645 // zshrs `integer-special`), which a completion listing sees directly:
3646 // `env <TAB>` runs `_parameters -g "*export*"` and zsh offered SHLVL
3647 // there where zshrs did not.
3648 // It follows the HOME and LOGNAME exports, as in C (c:960-974), so a
3649 // child sees them in zsh's environment order.
3650 if let Some(v) = shlvl_env {
3651 crate::ported::params::addenv("SHLVL", &v); // c:951
3652 }
3653
3654 // SHLVL's paired exec-time half — c:Src/exec.c:
3655 // 4332-4336, "for either implicit or explicit exec, decrease $SHLVL as
3656 // we're now done as a shell", guarded by `!subsh && !forked` — is
3657 // ported for the EXPLICIT `exec cmd` form at fusevm_bridge.rs's
3658 // BUILTIN_EXEC, the one place fusevm replaces its own process.
3659 //
3660 // Still open: C also execs IN PLACE for the LAST command of a `-c`
3661 // script, and fusevm always forks there (`zsh -fc 'echo $$; /bin/sh -c
3662 // "echo \$\$"'` reports one pid on zsh, two on zshrs). So
3663 // `SHLVL=5 <shell> -fc '/usr/bin/env'` shows SHLVL=5 on zsh (exec'd,
3664 // decremented) and SHLVL=6 on zshrs (forked, so C's own rule says do
3665 // not decrement). That divergence belongs to the unported implicit
3666 // exec-in-place, not to the SHLVL wiring: fixing it means teaching
3667 // fusevm C's `do_exec` for the final command, and until then the
3668 // forked answer is the one C would give for a forked command.
3669 // c:Src/init.c:1907-1909 — `SHTTY = -1; init_io(cmd); setupvals(...)`.
3670 // zsh_main runs those three in that order, and this constructor stands
3671 // in for setupvals's param setup on the drivers that never reach
3672 // zsh_main: `zshrs -c CODE` dispatches at bins/zshrs.rs:1716 (after
3673 // --zsh/-f are stripped) straight into ShellExecutor::new + exit. So
3674 // init_io has to happen here, or SHTTY is still -1 and the winsize
3675 // probe below silently no-ops (adjustwinsize early-returns at
3676 // c:1900-1901). setupvals's own adjustwinsize(0) covers the zsh_main
3677 // path; init_io is idempotent (c:615-618 closes and reopens SHTTY), so
3678 // that path just re-establishes it a moment later.
3679 crate::ported::init::init_io(None); // c:1908
3680 crate::startup_trace::mark("exec: init_io");
3681
3682 // c:Src/init.c:1274-1276 — `adjustwinsize(0)`, the first thing after
3683 // createparamtable (c:1270). Probes the tty via TIOCGWINSZ and
3684 // publishes the geometry to $COLUMNS/$LINES.
3685 //
3686 // Ordering is C's, and load-bearing in both directions: it must FOLLOW
3687 // init_io (which sets SHTTY) and it must FOLLOW the environ import
3688 // above, because the tty geometry OVERRIDES an inherited COLUMNS — a
3689 // 97-column terminal reports 97 even when COLUMNS=10 was exported in.
3690 // (C gets that via the c:1906-1907 "Signal missed while a job owned the
3691 // tty?" promotion of from=0 to from=1, which makes adjustcolumns take
3692 // the signalled path and overwrite zterm_columns with ws_col.)
3693 //
3694 // With no terminal at all, SHTTY stays -1 and the imported value (or 0)
3695 // survives untouched — which is what C's zterm_columns does there, and
3696 // why a piped `COLUMNS=20 zshrs -c` still reports 20. Note SHTTY does
3697 // NOT require stdin/stdout to be a tty: init_io's last resort is
3698 // `open("/dev/tty")` (c:667-670), so a piped-but-still-attached shell
3699 // gets the real width, matching `zsh -fc 'print $COLUMNS | cat'`.
3700 let _ = crate::ported::utils::adjustwinsize(0); // c:1276
3701 crate::startup_trace::mark("exec: adjustwinsize");
3702
3703 // c:Src/params.c:955 — `set_pwd_env();` runs AFTER the environ
3704 // import loop, overwriting the imported $PWD/$OLDPWD paramtab
3705 // entries with the ispwd()-validated values computed above
3706 // (c:Src/init.c:1242-1259). Without this, a stale inherited
3707 // $PWD (env-import snapshot taken at process entry) survives
3708 // in paramtab even though the live env was corrected.
3709 // The live environment is the carrier set_pwd_env reads for C's
3710 // `pwd`/`oldpwd` globals (see the c:Src/init.c:1242-1259 block above).
3711 env::set_var("PWD", &pwd_val);
3712 env::set_var("OLDPWD", &oldpwd_val);
3713 crate::dash_mode::set_oldpwd_global(&oldpwd_val); // c:1255-1259
3714 crate::ported::builtin::set_pwd_env();
3715 crate::startup_trace::mark("exec: set_pwd_env");
3716
3717 // c:Src/params.c:975-992 — host/arch identification params:
3718 // CPUTYPE / MACHTYPE / OSTYPE / VENDOR. C zsh reads from
3719 // compile-time `#define`s (set by ./configure) for MACHTYPE /
3720 // OSTYPE / VENDOR, and from uname().machine at runtime for
3721 // CPUTYPE.
3722 //
3723 // Rust port: probe uname() at startup for CPUTYPE, and use
3724 // const strings parameterized by build-target for the
3725 // others. Match homebrew zsh's values where possible.
3726 let mut uname_buf: libc::utsname = unsafe { std::mem::zeroed() };
3727 let _ = unsafe { libc::uname(&mut uname_buf) };
3728 let to_str = |b: &[libc::c_char]| -> String {
3729 // c-string → owned String, truncated at first NUL.
3730 let bytes: Vec<u8> = b
3731 .iter()
3732 .take_while(|&&c| c != 0)
3733 .map(|&c| c as u8)
3734 .collect();
3735 String::from_utf8_lossy(&bytes).into_owned()
3736 };
3737 let cputype = to_str(&uname_buf.machine);
3738 crate::ported::params::setsparam("CPUTYPE", &cputype); // c:961
3739 // OSTYPE: configure's `$host_os`, resolved on the build host and
3740 // frozen into config.h — C never re-derives it from uname() at
3741 // startup. Deriving it here made the two writers disagree, so the
3742 // same binary answered `darwin25.5.0` under -c and `darwin23.6.0`
3743 // under -i. Single source of truth: config_h::OSTYPE, exactly as
3744 // MACHTYPE below.
3745 crate::ported::params::setsparam("OSTYPE", crate::ported::config_h::OSTYPE); // c:990
3746 // MACHTYPE: configure's `$host_cpu`, i.e. the config.guess
3747 // canonical arch name — NOT uname's `machine`. The two differ
3748 // on Apple Silicon (uname says `arm64`, config.guess says
3749 // `aarch64`), and zsh reports the latter. Single source of
3750 // truth: config_h::MACHTYPE (= build target arch).
3751 crate::ported::params::setsparam("MACHTYPE", crate::ported::config_h::MACHTYPE); // c:967
3752 // VENDOR: configure's `$host_vendor`. Deriving it from uname's
3753 // `sysname` here was a second, non-C writer that disagreed with
3754 // `config_h::VENDOR` off Darwin: config.guess emits `pc` for x86_64
3755 // Linux (config.guess:1222) and `unknown` for aarch64 Linux
3756 // (config.guess:1009), a distinction `sysname` cannot make. Single
3757 // source of truth, exactly as OSTYPE/MACHTYPE above.
3758 crate::ported::params::setsparam("VENDOR", crate::ported::config_h::VENDOR); // c:992
3759
3760 // c:Src/init.c:963 — `setsparam("TTY", ttyname(0) ?: "")`, which
3761 // C reaches at c:969 in the createparamtable tail. Even a
3762 // non-interactive -fc shell creates the param.
3763 let tty_str = unsafe {
3764 let p = libc::ttyname(0);
3765 if p.is_null() {
3766 String::new()
3767 } else {
3768 std::ffi::CStr::from_ptr(p)
3769 .to_str()
3770 .unwrap_or("")
3771 .to_string()
3772 }
3773 };
3774 crate::ported::params::setsparam("TTY", &tty_str); // c:969
3775 // c:Src/params.c:971 — `setsparam("ZSH_ARGZERO", ztrdup(posixzero))`:
3776 // the kernel-supplied argv[0] of THIS binary, in --zsh parity mode
3777 // too. The bin entrypoint overrides this with the script path for
3778 // -c / runscript invocations. (A previous revision probed the
3779 // system zsh install path and reported THAT as ZSH_ARGZERO for
3780 // byte-parity — faking the shell's identity. Parity tests that
3781 // compare the value must normalize the machine-specific binary
3782 // path in the test row instead.)
3783 let argzero_default = env::args().next().unwrap_or_else(|| "zsh".to_string());
3784 crate::ported::params::setsparam("ZSH_ARGZERO", &argzero_default); // c:971
3785 // c:Src/params.c:972 — ZSH_VERSION. `zsh_version::ZSH_VERSION`
3786 // (emitted by build.rs from the vendored `Config/version.mk`) is
3787 // the development snapshot tag `5.9.0.3-test`; shipped zsh
3788 // binaries report the clean release form (`5.9`). Bug #73 in
3789 // docs/BUGS.md — cross-shell scripts that gate on
3790 // `[[ $ZSH_VERSION = 5.9 ]]` or split on `.` expecting
3791 // MAJOR.MINOR break on the `-test` suffix. Use the cleaned
3792 // `patchlevel::ZSH_VERSION` here ("5.9") and surface the full
3793 // snapshot tag as `$ZSHRS_VERSION` for zshrs identity checks.
3794 crate::ported::params::setsparam("ZSH_VERSION", crate::ported::patchlevel::ZSH_VERSION); // c:972
3795 // c:Src/params.c:973 + Src/patchlevel.h — `ZSH_PATCHLEVEL` is a
3796 // git-describe-style identifier (`zsh-MAJOR.MINOR-N-gHASH`) of
3797 // the upstream commit zshrs targets. `build.rs` emits "unknown"
3798 // because the vendored zsh tarball ships no CUSTOM_PATCHLEVEL
3799 // define; use the canonical const in `patchlevel.rs` instead.
3800 // Bug #90 in docs/BUGS.md — scripts that fingerprint by
3801 // $ZSH_PATCHLEVEL fell to the wildcard arm under "unknown".
3802 crate::ported::params::setsparam(
3803 "ZSH_PATCHLEVEL",
3804 crate::ported::patchlevel::ZSH_PATCHLEVEL,
3805 ); // c:973
3806 // Skip ZSHRS_VERSION whenever the zsh-compatible namespace must
3807 // stay free of zshrs-original names, so `${(k)parameters}`
3808 // doesn't carry a name zsh doesn't ship — same predicate and
3809 // reasoning as the guard in `ported::params::createparamtable`.
3810 // `hide_ext_builtins()` is `--zsh` OR `ZSHRS_HIDE_EXT_BUILTINS`
3811 // (the parity harnesses' knob). Scripts can still detect zshrs
3812 // via `$ZSH_VERSION`, which carries a `-test` suffix.
3813 if !crate::ext_builtins::hide_ext_builtins() {
3814 crate::ported::params::setsparam(
3815 "ZSHRS_VERSION",
3816 crate::ported::patchlevel::ZSHRS_VERSION,
3817 );
3818 }
3819 // `$HELPDIR` points at the bundled help tree that
3820 // `bundled_docs::install_and_publish` (top of this constructor)
3821 // materialised, so `run-help` works on a host with no zsh install.
3822 // It is seeded HERE, after the environment-import loop above, for
3823 // two reasons: the table has to exist to hold it, and an exported
3824 // `HELPDIR` the user supplied is already imported by now, so the
3825 // seed's own guard leaves that value (and its PM_EXPORTED) alone.
3826 //
3827 // The parameter is NOT exported. `run-help` is a shell function
3828 // that reads `${HELPDIR:-…}` out of the shell's own table, so the
3829 // environment entry this used to create bought nothing and leaked
3830 // the name into every child process — `env <TAB>`, which lists
3831 // exported names, offered a name zsh does not have.
3832 crate::bundled_docs::seed_helpdir_param();
3833 // c:Src/params.c:974-979 — `setaparam("signals", …)`.
3834 {
3835 use crate::ported::signals_h::SIGS;
3836 // c:signames.c sigs[] (generated) — index 0 is "EXIT",
3837 // entries 1..=SIGCOUNT are in PLATFORM SIGNAL-NUMBER
3838 // order, tail is "ZERR", "DEBUG" (zsh.h SIGZERR/SIGDEBUG).
3839 // SIGS is declared in Linux textual order, so sort by the
3840 // libc number to reproduce the generated table's order on
3841 // every platform. Same construction as params.rs — keep
3842 // in sync.
3843 let mut by_num: Vec<(&str, i32)> = SIGS.to_vec();
3844 by_num.sort_by_key(|&(_, n)| n);
3845 let mut signals_arr: Vec<String> = Vec::with_capacity(by_num.len() + 3);
3846 signals_arr.push("EXIT".to_string()); // c:sigs[0]
3847 signals_arr.extend(by_num.iter().map(|(n, _)| n.to_string()));
3848 // c:Src/params.c:1018-1021 — the real-time names follow.
3849 #[cfg(target_os = "linux")]
3850 signals_arr.extend(
3851 (libc::SIGRTMIN()..=libc::SIGRTMAX()).map(crate::ported::signals::rtsigname),
3852 );
3853 signals_arr.push("ZERR".to_string()); // c:sigs tail
3854 signals_arr.push("DEBUG".to_string()); // c:sigs tail
3855 crate::ported::params::setaparam("signals", signals_arr); // c:974
3856 }
3857 // c:Src/init.c:1364 — `setsparam("ZSH_NAME", ztrdup(zsh_name))`,
3858 // which setupvals runs AFTER createparamtable, so the node lands
3859 // ahead of the imported environment in its bucket chain.
3860 crate::ported::params::setsparam("ZSH_NAME", "zsh"); // c:Src/init.c:1364
3861 crate::startup_trace::mark("exec: base params set");
3862 // LOGNAME is seeded pre-import now (c:878) — see the block by the
3863 // TMPPREFIX/HOST seeds above.
3864 //
3865 // DO NOT setsparam("USERNAME", ...) anywhere in init. `$USERNAME`
3866 // is a special parameter whose SETTER (`usernamesetfn` in
3867 // params.rs) performs setgid(2) + setuid(2) to actually change
3868 // the effective user — a deliberate upstream zsh feature for
3869 // `USERNAME=other-user cmd`. Calling it at init seeds the value
3870 // AND tries to change uid/gid; when the resolved pwd's pw_uid
3871 // differs from `getuid()` (sudo launches, macOS Keychain-helper
3872 // inherited env, container entry points, etc.) the setgid call
3873 // fails with EPERM and emits `zsh:1: failed to change group ID:
3874 // Operation not permitted`. Upstream seeds `$USERNAME` via the
3875 // GETTER path (`usernamegetfn` reads through `cached_username`
3876 // populated by `inittyptab` → `get_username`), no setter call.
3877
3878 // c:Src/init.c:1176 — `module_path = mkarray(MODULE_DIR)`.
3879 // The canonical init lives in `init::setupvals` (port of
3880 // `Src/init.c:setupvals`); the bin entry skips setupvals (per
3881 // the init_bltinmods comment above), so call the lightweight
3882 // module_path bootstrap exposed by init.rs from here. This
3883 // mirrors the HOST gethostname seeding pattern above:
3884 // duplicated init that should collapse into a full setupvals
3885
3886 // c:Src/init.c:1945 init_bltinmods — runs right after setupvals
3887 // (c:1942), i.e. after createparamtable's import, so the module
3888 // autoload stubs (`WATCH`, `watch`, …) are created HERE. The bin
3889 // entry skips zsh_main → init_bltinmods, so run it from
3890 // ShellExecutor::new for the same effect. Bug #270.
3891 // c:Src/init.c:1085 — `(void)addhookdefs(NULL, zshhooks, ...)`, a
3892 // setupvals step. The bin entry skips setupvals, so without this
3893 // hooktab stayed empty: `zmodload zsh/nearcolor`'s boot_
3894 // (nearcolor.c:199) found no "get_color_attr" hookdef to attach
3895 // getnearestcolor to, and `%F{#hex}` never mapped to the palette.
3896 // Shared once-guard with setupvals (see init::ZSHHOOKS_ADDED).
3897 crate::ported::init::ZSHHOOKS_ADDED.call_once(|| {
3898 let base = crate::ported::init::zshhooks.load(std::sync::atomic::Ordering::SeqCst);
3899 let _ = crate::ported::module::addhookdefs(std::ptr::null(), base, 4);
3900 });
3901 crate::ported::init::init_bltinmods(); // c:Src/init.c:1945
3902 crate::startup_trace::mark("exec: init_bltinmods");
3903
3904 // Populate paramtab with PM_SPECIAL Params for every PARTAB /
3905 // PARTAB_ARRAY magic-assoc name. Mirrors what C's zsh/parameter
3906 // module boot_ → handlefeatures chain does — which happens when
3907 // the module LOADS, after init_bltinmods planted its autoload
3908 // stubs, and `addparamdef` unsets the stub before creating the
3909 // real param (c:Src/module.c addparamdef → unsetparam_pm +
3910 // createparam), so these names take a FRESH chain slot ahead of
3911 // the stubs. Running this before init_bltinmods put `usergroups`
3912 // and friends behind `WATCH` in `${(k)parameters}`.
3913 init_partab_params(); // c:Src/Modules/parameter.c:2341 boot_/enables_ chain
3914 crate::startup_trace::mark("exec: init_partab_params");
3915
3916 // HOST is seeded pre-import now (c:875) — see the block next to
3917 // the TMPPREFIX/LOGNAME seeds above.
3918 // bash startup delta: bash defines TERM itself when the
3919 // environment does not carry one, and exports it. zsh leaves
3920 // TERM unset in that case, so `zshrs --bash` inherited zsh's
3921 // behavior and diverged from the reference shell:
3922 //
3923 // $ env -u TERM /bin/bash -c 'printf "%s\n" "${TERM+set}"'
3924 // set
3925 // $ env -u TERM /bin/bash -c 'echo "$TERM"'
3926 // dumb
3927 // $ env -u TERM /bin/zsh -f -c 'printf "%s\n" "${TERM+set}"'
3928 // (empty — zsh leaves it unset)
3929 //
3930 // Same on bash 3.2.57 (macOS /bin/bash) and 5.3.15, so it is
3931 // not a version artifact. Only the bare `--bash` drop-in takes
3932 // it: `--bash --zsh` asks for zsh-STYLE emulation, where zsh's
3933 // leave-it-unset behavior is the correct answer. Guarded on the
3934 // environment so an inherited TERM always wins.
3935 if crate::extensions::dash_mode::bash_mode() && std::env::var_os("TERM").is_none() {
3936 crate::ported::params::setsparam("TERM", "dumb");
3937 // bash exports it (`declare -x TERM` shows up in `export -p`);
3938 // addenv stamps PM_EXPORTED and pushes it into the child env.
3939 crate::ported::params::addenv("TERM", "dumb");
3940 }
3941
3942 // c:Src/init.c:479 — `-c` mode: scriptname = scriptfilename
3943 // = ztrdup("zsh"). Both globals start as the literal "zsh"
3944 // (not the binary path) so PS4's %x / %N print "zsh" not
3945 // "/path/to/zshrs" at the top level. Function dispatch
3946 // overrides scriptname per c:5903; scriptfilename stays.
3947 crate::startup_trace::mark("exec: pre-scriptname");
3948 crate::ported::utils::set_scriptname(Some("zsh".to_string()));
3949 // c:Src/init.c:470-479 — `scriptname = scriptfilename =
3950 // ztrdup("zsh")` sits INSIDE the `-c` branch of the option parse.
3951 // An interactive shell (or one running a script file) leaves
3952 // `scriptfilename` NULL, and exec.c:5383 copies it onto every
3953 // Shfunc it defines — which is why zsh reports an EMPTY
3954 // `$functions_source[f]` for a function typed at the prompt.
3955 // Stamping "zsh" unconditionally made zshrs answer "zsh" there.
3956 let dash_c = std::env::args_os()
3957 .skip(1)
3958 .map(|a| a.to_string_lossy().into_owned())
3959 .any(|a| a.starts_with('-') && !a.starts_with("--") && a.contains('c'));
3960 if dash_c {
3961 crate::ported::utils::set_scriptfilename(Some("zsh".to_string())); // c:479
3962 }
3963
3964 // call once that port is complete.
3965 crate::startup_trace::mark("exec: pre-module_path_init");
3966 crate::ported::init::module_path_init();
3967
3968 // Publish `fpath` LAST, once every special has been seeded.
3969 //
3970 // The earlier `set_array("fpath", …)` above lands before the tied
3971 // FPATH/fpath special is created, and creating it resets the node
3972 // to empty -- so an INTERACTIVE shell reached its first prompt with
3973 // typeset -aT FPATH fpath=( )
3974 // while `$FPATH` still read the full four-entry string. `path`
3975 // hid the bug: `PATH` is always in the environ, so the env import
3976 // refills it, whereas zsh never exports `FPATH` and nothing
3977 // refilled that one.
3978 //
3979 // The damage was not a missing default: a `.zshrc` that does
3980 // `fpath=( mydir $fpath )` -- the standard idiom -- appended to
3981 // NOTHING, so the shell ended up with only what the rc file added
3982 // and lost the bundled tree entirely.
3983 //
3984 // `-c` was unaffected, which is why every non-interactive probe of
3985 // this passed.
3986 let fpath_final: Vec<String> = exec
3987 .fpath
3988 .iter()
3989 .map(|p| p.to_string_lossy().to_string())
3990 .collect();
3991 if !fpath_final.is_empty() {
3992 // The ENVIRONMENT has to carry the final value too, not just
3993 // paramtab. `setupvals` re-seeds the tied specials and the
3994 // split that refills `fpath` reads the FPATH string; with an
3995 // inherited FPATH that string was still the one this process
3996 // started with, so the bundled tree -- appended to the param
3997 // only -- vanished the moment a shell was interactive AND had
3998 // FPATH set. `-c` never re-seeds, which is why it kept showing
3999 // the correct list.
4000 unsafe { std::env::set_var("FPATH", fpath_final.join(":")) };
4001 exec.set_array("fpath".to_string(), fpath_final);
4002 }
4003 crate::startup_trace::mark("new() end");
4004
4005 exec
4006 }
4007
4008 /// The SQLite history index, opened on first call.
4009 ///
4010 /// Startup no longer pays for it: only the three diagnostic readers
4011 /// (`--doctor`, the cache report, `dbview history`) reach this, and a
4012 /// pool worker's cell is pre-filled with `None` so it stays closed
4013 /// there.
4014 pub fn history(&self) -> Option<&HistoryEngine> {
4015 self.history
4016 .get_or_init(|| HistoryEngine::new().ok())
4017 .as_ref()
4018 }
4019
4020 /// Execute a script file with bytecode caching — skips lex+parse+compile on cache hit.
4021 /// Bytecode is stored in rkyv keyed by (path, mtime).
4022 pub fn execute_script_file(&mut self, file_path: &str) -> Result<i32, String> {
4023 let path = Path::new(file_path);
4024 let abs_path = path
4025 .canonicalize()
4026 .unwrap_or_else(|_| path.to_path_buf())
4027 .to_string_lossy()
4028 .to_string();
4029
4030 // Try bytecode cache first — rkyv shard at ~/.zshrs/scripts.rkyv.
4031 // The cache validates path + mtime + zshrs binary mtime; on any
4032 // miss we fall through to lex/parse/compile. Cached path uses
4033 // `run_chunk` (the shared VM-execution helper); script-eval
4034 // path delegates to `execute_script_zsh_pipeline` so the
4035 // full parse/compile/cache-save/run flow stays in one place.
4036 if let Some(bc_blob) = crate::script_cache::try_load_bytes(path) {
4037 if let Ok(chunk) = bincode::deserialize::<fusevm::Chunk>(&bc_blob) {
4038 if !chunk.ops.is_empty() {
4039 tracing::trace!(
4040 path = %abs_path,
4041 ops = chunk.ops.len(),
4042 "execute_script_file: bytecode cache hit"
4043 );
4044 return self.run_chunk(chunk, &format!("execute_script_file:cache:{abs_path}"));
4045 }
4046 }
4047 }
4048
4049 // Cache miss.
4050 // c:Src/init.c:1394-1395 — `SHIN = movefd(open(funmeta, …))`, then
4051 // zsh_main's `loop(1,0)` (c:1963) reads the script ONE EVENT AT A
4052 // TIME: `lexinit(); parse_event(ENDINPUT); … execode(prog, …)`
4053 // (c:155-220). Every complete command runs before the next is
4054 // lexed, so an `alias` a line installs is in force for the next
4055 // line, and a syntax error on line N leaves lines 1..N-1 already
4056 // executed. A whole-file compile lexed every line with the state
4057 // the file started with and ran nothing when any line failed to
4058 // parse. The event loop is the one `source` uses (c:1626-1627
4059 // `loop(0, 0)`): with `interact` unset, c:234's
4060 // `(!interact || sourcelevel) && errflag` break is the same test.
4061 // c:Src/init.c:1566 / Src/input.c — the file is read as RAW BYTES and
4062 // metafied (Src/utils.c:4856), never rejected for non-UTF-8.
4063 let content = crate::script_bytes::read_script_file(file_path)
4064 .map_err(|e| format!("{}: {}", file_path, e))?;
4065 let mut events: Vec<crate::parse::ZshList> = Vec::new();
4066 // c:Src/init.c:1963 — zsh_main runs the script through `loop(1, 0)`:
4067 // toplevel.
4068 let mut status = self.run_events_per_command(&content, Some(&mut events), true)?;
4069 // c:Src/init.c:1969-1974 — `if (tok == LEXERR || errexit) { if
4070 // (!lastval) lastval = 1; stopmsg = 1; zexit(lastval, …); }`: a
4071 // parse error, or an error abort in a non-interactive shell, exits
4072 // non-zero even when the last command that ran succeeded.
4073 if (errflag.load(Ordering::Relaxed) & ERRFLAG_ERROR) != 0 {
4074 if status == 0 {
4075 status = 1; // c:1971-1972
4076 self.set_last_status(status);
4077 }
4078 // c:Src/builtin.c:6006 zexit — `errflag = 0;` before the EXIT trap
4079 // runs, so the trap body is not itself aborted by the error.
4080 errflag.fetch_and(!ERRFLAG_ERROR, Ordering::Relaxed);
4081 }
4082 let status = self.fire_script_exit_hooks(status)?;
4083
4084 // Best-effort cache save — failures don't block execution. The
4085 // cached program is the events exactly as the loop lexed them, so
4086 // a hit replays what this run parsed; `events` is empty unless the
4087 // loop drained the whole file without an error.
4088 if !events.is_empty() {
4089 let program = crate::parse::ZshProgram { lists: events };
4090 let compiler = crate::compile_zsh::ZshCompiler::new();
4091 let chunk = compiler.compile(&program);
4092 if let Ok(blob) = bincode::serialize(&chunk) {
4093 let _ = crate::script_cache::try_save_bytes(path, &blob);
4094 tracing::trace!(
4095 path = %abs_path,
4096 bytes = blob.len(),
4097 "execute_script_file: bytecode cached"
4098 );
4099 }
4100 }
4101
4102 Ok(status)
4103 }
4104
4105 /// Run a compiled `fusevm::Chunk` to completion inside this
4106 /// executor's context. Shared by `execute_script_zsh_pipeline`,
4107 /// `execute_script_file`'s bytecode-cache hit path, and the
4108 /// function-dispatch body_runner. Centralises the VM setup so
4109 /// `register_builtins` and `ExecutorContext::enter` invariants
4110 /// stay in lockstep.
4111 fn run_chunk(&mut self, chunk: fusevm::Chunk, label: &str) -> Result<i32, String> {
4112 if chunk.ops.is_empty() {
4113 return Ok(self.last_status());
4114 }
4115 crate::fusevm_disasm::maybe_print_stdout(label, &chunk);
4116 let mut vm = crate::vm_pool::acquire(chunk);
4117 // Seed vm.last_status with the executor's current LASTVAL so
4118 // sub-VMs (EXIT trap bodies, eval, source) see the inherited
4119 // `$?` from the caller's last command — matching C zsh where
4120 // lastval is a process global. Without this, the new VM
4121 // started at 0 and BUILTIN_GET_VAR's sync_status would write
4122 // 0 back into LASTVAL on the first `$?` read.
4123 vm.last_status = self.last_status();
4124 let _ctx = ExecutorContext::enter(self);
4125 // c:Src/loop.c — `loops` is bracketed by the C interpreter's own
4126 // recursion, so a `return` or an errflag abort out of a loop
4127 // unwinds it for free. A compiled chunk instead jumps straight to
4128 // its end, skipping the loop's `loops--`. Restoring the count the
4129 // chunk started with makes that structurally impossible to leak:
4130 // whatever loops this chunk opened are closed when it finishes.
4131 let loops_entry = crate::ported::builtin::LOOPS.load(Ordering::Relaxed);
4132 // c:Src/exec.c:4364 `done:` → `fixfds(save)` — every execcmd puts
4133 // back the fds its redirections replaced, whether it ran or was
4134 // abandoned on errflag. A chunk abandoned mid-command (`eval 'echo >
4135 // nx(N)'`: the target's expansion raises errflag after the scope
4136 // opened) jumped past its WithRedirectsEnd, so the caller kept
4137 // writing into the failed redirection's sink. Same unwind the
4138 // function-call path does (`unwind_redirect_scopes_to`).
4139 let redir_depth = self.redirect_scope_stack.len();
4140 // Same for the job slots of pipelines the chunk was running when it
4141 // was abandoned (see exec_jobs::execpline_slot_open).
4142 let slots_entry = crate::exec_jobs::execpline_slots_mark();
4143 let result = vm.run();
4144 crate::ported::builtin::LOOPS.store(loops_entry, Ordering::Relaxed);
4145 crate::exec_jobs::execpline_slots_release_to(slots_entry);
4146 self.unwind_redirect_scopes_to(redir_depth);
4147 match result {
4148 fusevm::VMResult::Ok(_) | fusevm::VMResult::Halted => {
4149 self.set_last_status(vm.last_status);
4150 }
4151 fusevm::VMResult::Error(e) => return Err(format!("VM error: {}", e)),
4152 }
4153 Ok(self.last_status())
4154 }
4155
4156 /// Execute via the lex+parse free ported + ZshCompiler pipeline.
4157 /// This is the only execution path; `execute_script` delegates here.
4158 /// Parse + compile `script` in an isolated lexer context, without
4159 /// running it.
4160 ///
4161 /// Split out of [`ShellExecutor::execute_script_zsh_pipeline`] so the
4162 /// autoload loader can get its hands on the compiled chunk: that chunk
4163 /// is what lands in `~/.zshrs/autoloads.rkyv`, so the next process can
4164 /// install the same function without re-parsing the definition file.
4165 fn compile_script_isolated(&mut self, script: &str) -> Result<fusevm::Chunk, String> {
4166 self.compile_script_isolated_at(script, None)
4167 }
4168
4169 /// `compile_script_isolated` with C's `parse_string(s, reset_lineno)`
4170 /// argument: `Some(n)` is `parse_string(s, 0)` (c:Src/exec.c:291-300),
4171 /// the body parsed from the running line `n` (see `parse_isolated_at`).
4172 fn compile_script_isolated_at(
4173 &mut self,
4174 script: &str,
4175 lineno: Option<u64>,
4176 ) -> Result<fusevm::Chunk, String> {
4177 // Skip history expansion for non-interactive script execution
4178 // (`zsh -c '…'`, internal eval, sourced files). zsh's `!`
4179 // history sub only fires on the REPL command line, never on
4180 // a pre-parsed script body. The interactive REPL has its
4181 // own dedicated path that calls expand_history before
4182 // dispatching here.
4183 // Save & clear errflag around the parse so a fresh syntax
4184 // error is distinguishable from one already in flight. Mirrors
4185 // Src/init.c loop()'s pre-parse `errflag &= ~ERRFLAG_ERROR;`.
4186 let saved_errflag = errflag.load(Ordering::Relaxed);
4187 errflag.fetch_and(!ERRFLAG_ERROR, Ordering::Relaxed);
4188 // Context-isolated parse (c:Src/exec.c:283 parse_string). eval /
4189 // source / autoload-register / trap bodies all reach here and run
4190 // DURING execution; on the faithful single-event loop()/parse_event
4191 // reader, a bare parse_init/lex_init would steal the outer's next
4192 // SHIN line into this nested program (e.g. `eval "x=5"` swallowed the
4193 // following `echo $x` off stdin). parse_isolated sets `strin` so the
4194 // string drains to EOF; execution below stays in the current shell.
4195 let program = parse_isolated_at(script, lineno);
4196 let parse_failed = (errflag.load(Ordering::Relaxed) & ERRFLAG_ERROR) != 0;
4197 errflag.store(saved_errflag, Ordering::Relaxed);
4198 if parse_failed {
4199 // c:Src/init.c — when the parser fires `zerr(...)`, the C
4200 // shell's `loop()` body skips the eval pass and continues;
4201 // there's no second "parse error" diagnostic. The Rust
4202 // binary's call sites print `zshrs: <e>` on Err, doubling
4203 // up on the message the parser already emitted via zerr.
4204 // Use a `__SILENCED__` sentinel that the binary's
4205 // execute_script wrapper recognizes as "already reported,
4206 // exit silently". Bug #142 in docs/BUGS.md (double-print
4207 // half).
4208 // !!! DASH-FAITHFUL GATE (no C counterpart) !!! a dash syntax error is
4209 // `sh_error` -> `exraise(EXERROR)`: it ends a non-interactive shell with
4210 // status 2, wherever the text came from (`-c`, `eval`, `.`, a script),
4211 // instead of zsh's "report, status 1, carry on".
4212 if crate::extensions::dash_mode::dash_faithful()
4213 && !crate::ported::zsh_h::isset(crate::ported::zsh_h::INTERACTIVE)
4214 {
4215 let st = crate::extensions::dash_mode::fatal_error_status().unwrap_or(2);
4216 crate::ported::builtin::LASTVAL.store(st, Ordering::Relaxed);
4217 crate::ported::builtin::EXIT_VAL.store(st, Ordering::Relaxed);
4218 crate::ported::builtin::EXIT_PENDING.store(1, Ordering::Relaxed);
4219 }
4220 return Err("__SILENCED__".to_string());
4221 }
4222
4223 let mut compiler = crate::compile_zsh::ZshCompiler::new();
4224 compiler.nested_lineno_base = lineno; // c:Src/exec.c:297 reset_lineno == 0
4225 if CMDARG_EXITING.with(|c| c.replace(false)) {
4226 // c:Src/init.c:1568 — `execstring(cmd, 0, 1, "cmdarg")`: the `-c`
4227 // string runs as an EXITING list, so its last command is the
4228 // fake exec's candidate (BUILTIN_EXEC_FORKED_SIMPLE mode 2).
4229 compiler.forked_tail_list = crate::compile_zsh::program_exiting_tail(&program);
4230 compiler.exiting_tail_mode = 2;
4231 }
4232 Ok(compiler.compile(&program))
4233 }
4234
4235 /// `execstring(cmd, 0, 1, "cmdarg")` (c:Src/init.c:1568) — run the `-c`
4236 /// string as the shell's last act: [`Self::execute_script`] with the
4237 /// string compiled as an exiting list (see CMDARG_EXITING).
4238 pub fn execute_cmdarg(&mut self, script: &str) -> Result<i32, String> {
4239 CMDARG_EXITING.with(|c| c.set(true));
4240 let chunk = self.compile_script_isolated(script);
4241 CMDARG_EXITING.with(|c| c.set(false));
4242 let status = self.run_chunk(chunk?, "cmdarg")?;
4243 // c:Src/init.c:1572 — `zexit(lastval, ZEXIT_NORMAL)` follows the
4244 // `-c` string, and c:Src/builtin.c:6012-6015 clears the error state
4245 // before the EXIT trap (`errflag = intrap = 0; if
4246 // (sigtrapped[SIGEXIT]) dotrap(SIGEXIT);`): dotrapargs refuses to
4247 // run anything while errflag is set (c:Src/signals.c:1101), so a
4248 // script that ended on an error abort (`ro=2` on a readonly, `${x?}`,
4249 // `$((1/0))`) would otherwise skip its EXIT trap. The flag is put
4250 // back afterwards because the caller still reads it to pick the
4251 // exit status of the dash-family drop-ins.
4252 let saved_errflag = errflag.load(Ordering::Relaxed);
4253 errflag.store(0, Ordering::Relaxed); // c:6012
4254 let result = self.fire_script_exit_hooks(status);
4255 errflag.fetch_or(saved_errflag, Ordering::Relaxed);
4256 result
4257 }
4258
4259 /// Run an already-compiled top-level chunk, then fire the end-of-script
4260 /// hooks (`EXIT` trap, `TRAPEXIT`, `zshexit` + `zshexit_functions`) the
4261 /// script pipeline owes them.
4262 fn run_chunk_with_exit_hooks(
4263 &mut self,
4264 chunk: fusevm::Chunk,
4265 label: &str,
4266 ) -> Result<i32, String> {
4267 let status = self.run_chunk(chunk, label)?;
4268 self.fire_script_exit_hooks(status)
4269 }
4270
4271 /// The end-of-script hooks alone (`EXIT` trap, `TRAPEXIT`, `zshexit` +
4272 /// `zshexit_functions`), for a script that already ran; `status` is the
4273 /// script's `$?` and is what the call returns.
4274 fn fire_script_exit_hooks(&mut self, status: i32) -> Result<i32, String> {
4275 // Fire EXIT trap if set. Two storage paths:
4276 // (a) `trap 'cmd' EXIT` writes the body text into
4277 // `traps_table` via bin_trap (Src/builtin.c) — fire
4278 // directly via execute_script.
4279 // (b) `TRAPEXIT() { ... }` function-named form goes
4280 // through settrap(SIGEXIT, None, ZSIG_FUNC) at
4281 // funcdef time (fusevm_bridge.rs BUILTIN_REGISTER_COMPILED_FN
4282 // arm) and lives in shfunctab + sigtrapped — fire
4283 // via dotrap(SIGEXIT) which dispatches the named
4284 // shfunc. Bug #157 in docs/BUGS.md.
4285 // Remove the trap from `traps_table` first to prevent
4286 // infinite recursion of `(a)`; `(b)`'s sigtrapped flag
4287 // is cleared by dotrap's own intrap guard.
4288 let exit_body = crate::ported::builtin::traps_table()
4289 .lock()
4290 .ok()
4291 .and_then(|mut t| t.remove("EXIT"));
4292 if let Some(action) = exit_body {
4293 tracing::debug!("firing EXIT trap (new pipeline)");
4294 // c:Src/signals.c — the EXIT trap body sees $? at the
4295 // value the script left off (so `trap 'echo $?' EXIT;
4296 // (exit 7)` prints 7), but the SHELL's final exit code
4297 // is still the pre-trap value (running `echo` inside
4298 // the trap doesn't reset the script's exit status).
4299 // Preserve `status` and re-apply it after the trap
4300 // body returns.
4301 //
4302 // c:Src/signals.c:1123/1236 — `intrap++` … `intrap--` bracket a
4303 // trap body, and while intrap the EXIT, DEBUG and ZERR traps are
4304 // suppressed (c:1112-1119, the guard in dotrap). This path runs
4305 // the EXIT body through the script pipeline rather than dotrap,
4306 // so nothing raised intrap and the body's own failure re-entered
4307 // the ERR trap: `trap 'print err' ERR; trap 'true; false' EXIT`
4308 // printed err where zsh prints nothing.
4309 //
4310 // A counter, not a flag, and paired with dotrap's SELECTIVE
4311 // guard: signals zsh does deliver from inside a trap body (e.g.
4312 // `trap 'kill -USR1 $$' EXIT`) must still dispatch.
4313 crate::ported::signals::intrap.fetch_add(1, Ordering::SeqCst); // c:1123
4314 // c:Src/signals.c:1170 — `execode((Eprog)sigfn, 1, 0, "trap")`:
4315 // the EXIT body sees "trap" in zsh_eval_context (c:Src/exec.c:1251).
4316 let trap_ctx = crate::ported::exec::EvalContextFrame::push("trap");
4317 let _ = self.execute_script_zsh_pipeline(&action);
4318 drop(trap_ctx);
4319 crate::ported::signals::intrap.fetch_sub(1, Ordering::SeqCst); // c:1236
4320 self.set_last_status(status);
4321 }
4322 // c:Src/signals.c::dotrap(SIGEXIT) — fire TRAPEXIT() shfunc
4323 // if installed via the function-name path. The TRAPEXIT()
4324 // form goes through settrap(SIGEXIT, None, ZSIG_FUNC) at
4325 // funcdef time (sets sigtrapped[SIGEXIT] |= ZSIG_FUNC).
4326 // Dispatching from here AFTER run_chunk returns means we're
4327 // outside the VM context — dotrap can't safely re-enter
4328 // via dispatch_function_call (which uses with_executor).
4329 // Route through execute_script_zsh_pipeline which sets up
4330 // a fresh VM context — invoke the function by name.
4331 let trapped = crate::ported::signals::sigtrapped
4332 .lock()
4333 .ok()
4334 .and_then(|g| g.get(crate::signals_h::SIGEXIT as usize).copied())
4335 .unwrap_or(0);
4336 // c:Src/signals.c:1112-1119 — `if (intrap) { switch (sig) { case
4337 // SIGEXIT: … return; } }`, and c:Src/signals.c:892 `if (!intrap &&
4338 // …)` in endtrapscope. An EXIT trap never fires from inside another
4339 // trap body. This site is a Rust-only end-of-pipeline hook (every
4340 // `eval` / `source` / trap body runs its own pipeline and reaches
4341 // here), and it dispatches TRAPEXIT by NAME without going through
4342 // dotrap — so nothing consulted `intrap` and nothing cleared
4343 // sigtrapped. Once `endtrapscope` started restoring a saved
4344 // ZSIG_FUNC EXIT trap (the c:929-931 arm), the TRAPEXIT body's own
4345 // nested pipeline re-entered this hook with the flag still set and
4346 // recursed without bound:
4347 // f() { eval 'TRAPEXIT() { echo T; }' }; f
4348 // The `intrap++ … intrap--` bracket is the same one the string-form
4349 // branch above already carries (c:1123 / c:1236).
4350 // c:Src/signals.c:744-752 — `sigtrapped[sig] |= (locallevel <<
4351 // ZSIG_SHIFT)`: a trap installed inside a function carries its
4352 // scope's locallevel, and `endtrapscope` (c:892-903/945-956) is what
4353 // fires THAT one, at the scope exit. Only an untagged (locallevel 0)
4354 // EXIT trap belongs to the shell-exit path this hook stands in for.
4355 // Without the test, `f() { TRAPEXIT() { echo T } }; f` fired twice —
4356 // once from f's endtrapscope and once more from the pipeline hook.
4357 let exit_trap_locallevel = trapped >> crate::ported::zsh_h::ZSIG_SHIFT;
4358 if (trapped & crate::ported::zsh_h::ZSIG_FUNC as i32) != 0
4359 && exit_trap_locallevel == 0
4360 && crate::ported::signals::intrap.load(Ordering::SeqCst) == 0
4361 {
4362 // The TRAP<SIG> function is stored in shfunctab as
4363 // "TRAPEXIT"; calling it by name re-enters
4364 // execute_script_zsh_pipeline with a fresh VM context.
4365 crate::ported::signals::intrap.fetch_add(1, Ordering::SeqCst); // c:1123
4366 let _ = self.execute_script_zsh_pipeline("TRAPEXIT");
4367 crate::ported::signals::intrap.fetch_sub(1, Ordering::SeqCst); // c:1236
4368 }
4369 // c:Src/init.c::zexit — `callhookfunc("zshexit", NULL, 1, NULL)`.
4370 // Fire the `zshexit` shfunc + walk `zshexit_functions` array.
4371 // Routed through execute_script_zsh_pipeline calls because
4372 // we're outside the VM context here (post-run_chunk). Iterate
4373 // the array directly + call zshexit by name. Bug #215 in
4374 // docs/BUGS.md.
4375 //
4376 // Re-entry guard: each call to execute_script_zsh_pipeline
4377 // (whether top-level script or the named-fn dispatch below)
4378 // hits this code at its tail. Without a guard, the zshexit
4379 // hook recurses infinitely (calls itself at end via this
4380 // path). Use a thread-local depth counter and skip the
4381 // dispatch when depth > 0.
4382 thread_local! {
4383 static ZSHEXIT_HOOK_DEPTH: std::cell::Cell<u32> = const {
4384 std::cell::Cell::new(0)
4385 };
4386 }
4387 let hook_depth = ZSHEXIT_HOOK_DEPTH.with(|c| c.get());
4388 if hook_depth == 0 {
4389 ZSHEXIT_HOOK_DEPTH.with(|c| c.set(hook_depth + 1));
4390 if crate::ported::hashtable::shfunctab_lock()
4391 .read()
4392 .ok()
4393 .map(|t| t.contains_key("zshexit"))
4394 .unwrap_or(false)
4395 {
4396 let _ = self.execute_script_zsh_pipeline("zshexit");
4397 }
4398 let exit_arr = crate::ported::params::paramtab()
4399 .read()
4400 .ok()
4401 .and_then(|t| t.get("zshexit_functions").and_then(|p| p.u_arr.clone()))
4402 .unwrap_or_default();
4403 for fn_name in exit_arr {
4404 let exists = crate::ported::hashtable::shfunctab_lock()
4405 .read()
4406 .ok()
4407 .map(|t| t.contains_key(&fn_name))
4408 .unwrap_or(false);
4409 if exists {
4410 let _ = self.execute_script_zsh_pipeline(&fn_name);
4411 }
4412 }
4413 ZSHEXIT_HOOK_DEPTH.with(|c| c.set(hook_depth));
4414 }
4415 // Preserve script status; trap body shouldn't override it.
4416 self.set_last_status(status);
4417
4418 let _ = status;
4419 Ok(self.last_status())
4420 }
4421 /// zshrs's script entry: lex + parse + compile + run, then the
4422 /// end-of-script hooks. `eval`, `source`, trap bodies and autoload
4423 /// registration all funnel through here.
4424 pub fn execute_script_zsh_pipeline(&mut self, script: &str) -> Result<i32, String> {
4425 let chunk = self.compile_script_isolated(script)?;
4426 self.run_chunk_with_exit_hooks(chunk, "execute_script_zsh_pipeline")
4427 }
4428
4429 /// Run a source string the way C's `execstring` does (c:Src/exec.c:1228
4430 /// → `execode`, c:1245): parse, execute, return the status. No `EXIT`
4431 /// trap, `TRAPEXIT` or `zshexit` hook runs here. In C those belong to
4432 /// `zexit` (c:Src/builtin.c:6037-6043) and to the end of a function's
4433 /// trap scope (`endtrapscope`, c:Src/signals.c:880), never to the end of
4434 /// an `eval`, a trap body or a `sched` command. Firing them there ran a
4435 /// script's EXIT trap in the middle of the script:
4436 /// `trap 'print T' EXIT; eval :; print after` printed T before `after`.
4437 pub fn execute_string_without_exit_hooks(&mut self, script: &str) -> Result<i32, String> {
4438 let chunk = self.compile_script_isolated(script)?;
4439 self.run_chunk(chunk, "execstring")
4440 }
4441
4442 /// `execode(parse_string(cmdarg, 0), 1, 0, "cmdsubst")` — a nofork
4443 /// substitution body (c:Src/subst.c:2044-2046) parsed from the running
4444 /// line, so its statements carry the outer line numbers and leave
4445 /// `lineno` on them, as C's do.
4446 pub fn execute_string_at_lineno(&mut self, script: &str, lineno: u64) -> Result<i32, String> {
4447 let chunk = self.compile_script_isolated_at(script, Some(lineno))?;
4448 self.run_chunk(chunk, "execstring")
4449 }
4450
4451 /// Run the TEXT that `getpermtext` reconstructed from an already-compiled
4452 /// `.zwc` program.
4453 ///
4454 /// c:Src/init.c:1618-1622 — the compiled arm of `source()` is
4455 /// `execode(prog, 1, 0, "filecode")`. The wordcode runs as it stands and
4456 /// NOTHING is lexed; a `.zwc` is quote-resolved once, at `zcompile` time.
4457 ///
4458 /// !!! WARNING: RUST-ONLY HELPER !!!
4459 /// zshrs has no execute-the-wordcode path — it deparses the program back
4460 /// to source (`getpermtext`) and lexes it again — so the round trip is
4461 /// lossless only while the lexer reads quotes the way the deparse writes
4462 /// them. `untokenize` (c:Src/exec.c:2134) renders EVERY quote null through
4463 /// `ztokens[Snull - Pound]`, and that entry is a bare single quote
4464 /// (c:Src/lex.c:38), so a closing null followed by an opening one comes
4465 /// back out as two adjacent quotes. Under RCQUOTES the lexer reads that
4466 /// pair inside a quoted word as one LITERAL quote (c:Src/lex.c:1328)
4467 /// instead of as two delimiters, so the openshift-aliases plugin's
4468 /// `alias opodr='oc …=''{…}'''` — which `zcompile` resolved with no
4469 /// literal quotes at all — re-lexed with two of them. The deparse
4470 /// spelling is by construction the DEFAULT-option spelling, so the option
4471 /// is cleared for the compile to restore C's "not lexed at all" property.
4472 ///
4473 /// It is cleared for the COMPILE ONLY. A `.zwc` that does `setopt
4474 /// rcquotes` (zsh-expand's plugin entry does, at its line 39) must still
4475 /// set the option for real, and that setting must outlive the source — so
4476 /// the previous value is restored before the chunk RUNS, not after. The
4477 /// same split applies to alias expansion: a function or `eval` body the
4478 /// program runs is lexed at RUNTIME and must see the live alias table.
4479 pub fn execute_zwc_program(&mut self, script: &str) -> Result<i32, String> {
4480 let chunk = {
4481 let _relex = ZwcRelexGuard::enter();
4482 self.compile_script_isolated(script)
4483 };
4484 self.run_chunk_with_exit_hooks(chunk?, "execute_zwc_program")
4485 }
4486
4487 /// Run `script` the way C runs a PLAIN sourced file: parse ONE event,
4488 /// execute it, parse the next — so lexer-time state that one line
4489 /// establishes is in force when the next line is lexed.
4490 ///
4491 /// c:Src/init.c:1618-1641 — `source()` has two arms. A file that was
4492 /// already compiled (`try_source_file` found a `.zwc`) runs whole, as one
4493 /// program: `execode(prog, 1, 0, "filecode")` (c:1621). A plain file runs
4494 /// through the per-command loop: `/* loop through the file to be sourced
4495 /// */ switch (loop(0, 0))` (c:1626-1627), whose body is `lexinit();
4496 /// parse_event(ENDINPUT); … execode(prog, 0, 0, "file")` (c:155-220).
4497 /// This is the second arm.
4498 ///
4499 /// The difference is observable whenever a line changes something the
4500 /// LEXER consults, because a whole-file compile lexes every line with the
4501 /// state the file STARTED with:
4502 ///
4503 /// ```text
4504 /// alias greet='print -r -- hello' # takes effect at execution time
4505 /// greet # …but this line is lexed after it
4506 /// ```
4507 ///
4508 /// Same for `setopt rcquotes` (c:Src/lex.c:1326), `unsetopt aliases`, and
4509 /// a syntax error late in the file (C has already run the good lines).
4510 ///
4511 /// **Re-entrancy.** Every nested context — `$(source f)`, `` `source f` ``,
4512 /// `eval "source f"`, a pipe stage, a `( … )` subshell, a `source` inside
4513 /// a sourced file — reaches here through the normal builtin path, so this
4514 /// must be safe to enter while an outer instance of itself is parked
4515 /// mid-file. Two properties make it so, and both are deliberate:
4516 ///
4517 /// * It never touches the shell's INPUT STACK. C's `source` points
4518 /// `SHIN` at the file (c:1584) and lets `loop`'s `ingetc` pull from
4519 /// it; doing that here would fight the outer reader for the one
4520 /// global. Instead the file body is installed as the lexer's own
4521 /// `LEX_INPUT` window under `strinbeg` — the exact parking
4522 /// [`parse_isolated`] uses for a command-substitution body — and the
4523 /// outer window is saved on the Rust stack and restored on the way
4524 /// out. Nesting is then just stack discipline.
4525 /// * It never dispatches through `execode`. `execode`
4526 /// (`src/ported/exec.rs`) runs its program on the installed SESSION
4527 /// executor, which is the right one only for the top-level REPL; from
4528 /// inside a command substitution the live executor is the sub-VM that
4529 /// owns the capture. Each event is compiled and run here on `self` —
4530 /// the same executor `execute_script` would have used — via
4531 /// [`Self::run_chunk`]. `$(source f)` therefore captures exactly what
4532 /// `$(…)` captures from any other builtin.
4533 ///
4534 /// Returns the file's `$?`. `Err` only for a VM error, as
4535 /// [`Self::run_chunk`] reports it.
4536 pub fn execute_script_per_command(&mut self, script: &str) -> Result<i32, String> {
4537 // c:Src/init.c:1626-1627 — source() runs `loop(0, 0)`: not toplevel.
4538 self.run_events_per_command(script, None, false)
4539 }
4540
4541 /// The `loop()` body shared by a sourced file and a script file.
4542 ///
4543 /// `events`, when given, receives every executed event's lists, and is
4544 /// CLEARED unless the loop drained the input to a clean `ENDINPUT` — so a
4545 /// caller holding a non-empty collection after the call has the exact
4546 /// program the lexer produced, one event at a time, for the whole file.
4547 ///
4548 /// `toplevel` is loop()'s parameter of the same name: true only for the
4549 /// script file zsh_main runs (c:Src/init.c:1963 `loop(1, 0)`), false for
4550 /// `source` / `.` (c:1626-1627 `loop(0, 0)`). It gates the preexec hook
4551 /// (c:180).
4552 fn run_events_per_command(
4553 &mut self,
4554 script: &str,
4555 mut events: Option<&mut Vec<crate::parse::ZshList>>,
4556 toplevel: bool,
4557 ) -> Result<i32, String> {
4558 use crate::ported::lex::{
4559 tok, ENDINPUT, LEXERR, LEX_FILE_WINDOW_STRIN, LEX_INPUT, LEX_LINENO, LEX_POS,
4560 LEX_UNGET_BUF,
4561 };
4562
4563 // Inline Rust FFI blocks are rewritten before the lexer sees them,
4564 // as on every other source-string entry (see `parse_isolated`).
4565 let ffi_desugared = script
4566 .contains("rust")
4567 .then(|| crate::rust_ffi::desugar(script));
4568 let script: &str = ffi_desugared.as_deref().unwrap_or(script);
4569
4570 // c:Src/init.c:121 — `if (!toplevel) zcontext_save();`, plus the
4571 // zshrs-only lexer window that `zcontext` doesn't cover (identical
4572 // list to `parse_isolated`, which parks it for the same reason).
4573 crate::ported::context::zcontext_save(); // c:121
4574 let saved_input = LEX_INPUT.with_borrow(|s| s.clone());
4575 let saved_pos = LEX_POS.get();
4576 let saved_unget = LEX_UNGET_BUF.with_borrow(|b| b.clone());
4577 let saved_lineno = LEX_LINENO.get();
4578 let saved_in_lexstop = crate::ported::input::lexstop.with(|c| c.get());
4579 let saved_file_window = LEX_FILE_WINDOW_STRIN.get();
4580
4581 // `strin` makes a drained window report EOF instead of falling
4582 // through to `inputline()` and STEALING the outer reader's next line
4583 // (input.rs:391). Without it a sourced file's last event swallows the
4584 // caller's next stdin line.
4585 crate::ported::hist::strinbeg(0);
4586 // c:1588 — `lineno = 1;`. `lex_init` (called by `parse_init`)
4587 // installs the body as the window and resets the line counter.
4588 crate::ported::parse::parse_init(script);
4589 // C reads this file through `inputline` (c:Src/input.c:366) with
4590 // `strin == 0`: ONE LINE per buffer, which is what makes each line
4591 // its own event (c:Src/lex.c:310 / c:Src/parse.c:657), and every
4592 // newline counted (c:Src/input.c:330). zshrs installs the body as
4593 // one window under one `strinbeg`, so the file kind is declared
4594 // here, tagged with THIS strinbeg's depth — a nested string push
4595 // inside the file is deeper and stays a string.
4596 LEX_FILE_WINDOW_STRIN.set(crate::ported::input::strin.with(|s| s.get()));
4597
4598 // c:116 — `int err, non_empty = 0;`
4599 let mut non_empty = false;
4600 let mut vm_error: Option<String> = None;
4601 let mut drained = false;
4602
4603 loop {
4604 // c:155 — `lexinit();` Resets `tok` and BOTH `lexstop` copies;
4605 // it does NOT move the window, so the next event resumes where
4606 // the last one stopped.
4607 crate::ported::lex::lexinit(); // c:155
4608 // Event source for preexec's `$2`/`$3` — see run_preexec_hook.
4609 let event_mark = toplevel.then(crate::funcdef_capture::body_mark_begin);
4610 let prog = crate::ported::parse::parse_event(ENDINPUT); // c:156
4611 let event_src = event_mark.and_then(crate::funcdef_capture::event_text);
4612 let Some(prog) = prog else {
4613 // c:159-174 — no event this pass. Break on clean EOF or on a
4614 // parse error (`!toplevel` makes C's LEXERR arm
4615 // unconditional here); a bare separator just goes round
4616 // again, which is C's `continue` at c:174.
4617 let tok_v = tok(); // c:159
4618 let errflag_v = errflag.load(Ordering::Relaxed);
4619 if (tok_v == ENDINPUT && errflag_v == 0) || tok_v == LEXERR {
4620 // c:159-162
4621 drained = tok_v == ENDINPUT;
4622 if tok_v == LEXERR
4623 && crate::ported::builtin::LASTVAL.load(Ordering::Relaxed) == 0
4624 {
4625 crate::ported::builtin::LASTVAL.store(1, Ordering::Relaxed); // c:173
4626 }
4627 // !!! DASH-FAITHFUL GATE (no C counterpart) !!! a syntax error in a
4628 // sourced file is `exraise(EXERROR)`: status 2 and the shell ends.
4629 if tok_v == LEXERR
4630 && crate::extensions::dash_mode::dash_faithful()
4631 && !crate::ported::zsh_h::isset(crate::ported::zsh_h::INTERACTIVE)
4632 {
4633 std::process::exit(
4634 crate::extensions::dash_mode::fatal_error_status().unwrap_or(2),
4635 );
4636 }
4637 break;
4638 }
4639 if tok_v == ENDINPUT || errflag_v != 0 {
4640 // Drained (or aborted) with a flag set — nothing left to
4641 // read, so looping again would spin.
4642 break;
4643 }
4644 continue; // c:174
4645 };
4646 non_empty = true; // c:179
4647 if toplevel {
4648 // callhookfunc dispatches through the live executor, which
4649 // is only installed while a chunk runs (see run_chunk).
4650 let _ctx = ExecutorContext::enter(self);
4651 run_preexec_hook(event_src.as_deref()); // c:180-215
4652 }
4653
4654 // c:220 — `execode(prog, 0, 0, "file")`. The eval-context entry
4655 // C's `execode` pushes for this arm is already on the stack:
4656 // `bin_dot` pushes it once around the whole file (builtin.rs), so
4657 // pushing it per event would report `file:file:file:…`.
4658 //
4659 // `LEX_LINENO` is saved across execution for the same reason
4660 // `loop()` does it (init.rs, c:Src/exec.c:1376/1640): each
4661 // statement's `SET_LINENO` overwrites the counter while it runs,
4662 // and the NEXT event must be lexed from the line the file is
4663 // really on.
4664 let chunk = crate::compile_zsh::ZshCompiler::new().compile(&prog);
4665 let saved_lex_lineno = LEX_LINENO.get();
4666 let run = self.run_chunk(chunk, "source");
4667 LEX_LINENO.set(saved_lex_lineno);
4668 if let Some(ev) = events.as_deref_mut() {
4669 ev.extend(prog.lists);
4670 }
4671 if let Err(e) = run {
4672 vm_error = Some(e);
4673 break;
4674 }
4675
4676 // c:234 — `if (((!interact || sourcelevel) && errflag) || retflag)
4677 // break;`. A sourced file always runs with `sourcelevel` bumped
4678 // (bin_dot, c:1606), so the errflag half is unconditional here.
4679 if (errflag.load(Ordering::Relaxed) & ERRFLAG_ERROR) != 0
4680 || crate::ported::builtin::RETFLAG.load(Ordering::Relaxed) != 0
4681 {
4682 // c:Src/init.c:1960-1968 — after loop(1,0) breaks on a
4683 // runtime error, zsh_main re-enters it unless
4684 // `errflag && !interact && !isset(CONTINUEONERROR)`; the
4685 // re-entered loop's hbegin clears ERRFLAG_ERROR
4686 // (c:Src/hist.c:1115) before the next event. Only the script
4687 // file (toplevel) has that outer loop: `source` breaks
4688 // (c:234 `sourcelevel`), and a parse error never gets here
4689 // (the LEXERR arm above breaks first).
4690 if toplevel
4691 && crate::ported::builtin::RETFLAG.load(Ordering::Relaxed) == 0
4692 && crate::ported::builtin::EXIT_PENDING.load(Ordering::Relaxed) == 0
4693 && !crate::ported::zsh_h::isset(crate::ported::zsh_h::INTERACTIVE)
4694 && crate::ported::zsh_h::isset(crate::ported::zsh_h::CONTINUEONERROR)
4695 {
4696 errflag.fetch_and(!ERRFLAG_ERROR, Ordering::Relaxed); // c:Src/hist.c:1115
4697 continue;
4698 }
4699 break; // c:235
4700 }
4701 // C's `exit` inside a sourced file calls `zexit` → `realexit()`
4702 // and the process is gone, so C never reaches the next event.
4703 // zshrs defers the exit (EXIT_PENDING plus a jump to chunk end)
4704 // and lets the caller unwind, so the loop has to stop on its own.
4705 if crate::ported::builtin::EXIT_PENDING.load(Ordering::Relaxed) != 0 {
4706 break;
4707 }
4708 }
4709
4710 // c:245 — `err = errflag;` is read BEFORE the context is restored,
4711 // because `zcontext_restore` → `parse_context_restore` ends with
4712 // `errflag &= ~ERRFLAG_ERROR` (c:Src/parse.c:354). C carries the
4713 // answer out as the `LOOP_ERROR` return value; zshrs's `bin_dot`
4714 // reads the flag itself (its c:1623-1624 + c:1663 block), so the bit
4715 // is put back after the restore instead.
4716 let err = (errflag.load(Ordering::Relaxed) & ERRFLAG_ERROR) != 0; // c:245
4717 if let Some(ev) = events {
4718 if !drained || err || vm_error.is_some() {
4719 ev.clear();
4720 }
4721 }
4722
4723 // c:246-249 — leave the loop's context exactly as it was found.
4724 crate::ported::hist::strinend();
4725 LEX_INPUT.with_borrow_mut(|s| *s = saved_input);
4726 LEX_POS.set(saved_pos);
4727 LEX_UNGET_BUF.with_borrow_mut(|b| *b = saved_unget);
4728 LEX_LINENO.set(saved_lineno);
4729 crate::ported::input::lexstop.with(|c| c.set(saved_in_lexstop));
4730 LEX_FILE_WINDOW_STRIN.set(saved_file_window);
4731 crate::ported::context::zcontext_restore(); // c:247
4732 if err {
4733 errflag.fetch_or(ERRFLAG_ERROR, Ordering::Relaxed); // see the c:245 note
4734 }
4735
4736 if let Some(e) = vm_error {
4737 return Err(e);
4738 }
4739 // c:1633-1636 — `case LOOP_EMPTY: /* Empty code resets status */
4740 // lastval = 0;`. `source /dev/null` (or a comments-only file) clears
4741 // `$?` rather than leaving the caller's.
4742 if !non_empty {
4743 self.set_last_status(0); // c:1635
4744 }
4745 Ok(self.last_status())
4746 }
4747
4748 /// Install an autoloaded function by running its definition program,
4749 /// reusing the rkyv-cached chunk when the cache can PROVE the chunk
4750 /// was compiled from this same definition text by this same binary.
4751 ///
4752 /// `registered` is what `autoload_register_source` produced: either
4753 /// `name() { <file body> }` or, for a file that already contains the
4754 /// definition, the body verbatim. Running it installs the function;
4755 /// the compiled chunk for it is exactly what the cache stores, so a hit
4756 /// skips lex+parse+compile of the whole file. For `_git` that is 424 KB
4757 /// of shell — the dominant cost of the first `git <tab>`.
4758 ///
4759 /// Two conditions gate caching, because outside them the chunk is not a
4760 /// function of the definition text alone:
4761 /// * ksh-style autoload (`KSHAUTOLOAD` / `PM_KSHSTORED`) runs the file
4762 /// at top level instead of wrapping it, so the same bytes produce a
4763 /// different program depending on a runtime option;
4764 /// * without `PM_UNALIASED` (`autoload` without `-U`) the body is
4765 /// parsed WITH alias expansion, so the chunk depends on the alias
4766 /// table too. Every compsys / plugin autoload uses `-Uz`.
4767 ///
4768 /// A hit that runs without defining `name` is treated as a corrupt
4769 /// entry, not as a failed load: the entry is dropped and the real
4770 /// source compiled. Installing a function is the one thing this
4771 /// function exists to do, so "it ran and the function is not there"
4772 /// is a fact the loader can check for itself rather than leaving the
4773 /// caller to report `function not defined by file` for what is
4774 /// actually a bad cache line.
4775 ///
4776 /// `from_wordcode` says the text is a `.zwc` deparse rather than file
4777 /// bytes. C never re-lexes such a body at all — `shf->funcdef =
4778 /// stripkshdef(prog, …)` (c:Src/exec.c:5753-5755) installs the dump's own
4779 /// wordcode and the ksh arm runs it with `execode(prog, 1, 0,
4780 /// "evalautofunc")` (c:5795) — so the compile here, which is zshrs's
4781 /// stand-in for that install, runs under [`ZwcRelexGuard`]. The guard
4782 /// covers the COMPILE only: a ksh-style body executes real user code, and
4783 /// an `eval` or function body IT reaches is lexed at runtime against the
4784 /// live alias table and the live RCQUOTES, exactly as in C.
4785 fn run_autoload_definition(
4786 &mut self,
4787 name: &str,
4788 registered: &str,
4789 ksh_style: bool,
4790 from_wordcode: bool,
4791 parse_failed: bool,
4792 ) -> Result<i32, String> {
4793 let unaliased = crate::ported::utils::getshfunc(name)
4794 .map(|f| (f.node.flags as u32 & crate::ported::zsh_h::PM_UNALIASED) != 0)
4795 .unwrap_or(false);
4796 let key = if ksh_style || !unaliased {
4797 None
4798 } else {
4799 autoload_source_key(name, registered, from_wordcode)
4800 };
4801 if let Some((dir, sha)) = key.as_ref() {
4802 if let Some(blob) = crate::autoload_cache::try_load_for_source(name, dir, sha) {
4803 match bincode::deserialize::<fusevm::Chunk>(&blob) {
4804 Ok(chunk) if !chunk.ops.is_empty() => {
4805 tracing::debug!(
4806 name,
4807 ops = chunk.ops.len(),
4808 "autoload: rkyv chunk hit, skipping parse+compile"
4809 );
4810 let status = self.run_chunk(chunk, "autoload:cached");
4811 if self.functions_compiled.contains_key(name) {
4812 return status;
4813 }
4814 // The chunk ran and `name` is still undefined, so
4815 // it is not this function's definition program
4816 // whatever the key said. Drop it and fall through
4817 // to a real compile — a wrong answer here costs
4818 // every completion on the shell.
4819 tracing::warn!(
4820 name,
4821 "autoload: cached chunk did not define the function; \
4822 dropping the entry and recompiling"
4823 );
4824 crate::autoload_cache::try_remove(name);
4825 }
4826 _ => {}
4827 }
4828 }
4829 }
4830 let chunk = {
4831 // c:Src/exec.c:5753-5755 / c:5795 — the dump's wordcode is
4832 // installed/executed as it stands, so nothing about the live
4833 // option or alias state can reach it. zshrs lexes the deparse
4834 // instead; pin the lexer to the spelling the deparse was written
4835 // in for the duration of that compile, and only that compile.
4836 let _relex = from_wordcode.then(ZwcRelexGuard::enter);
4837 // c:Src/exec.c:6264-6266 / c:5726 — C parses the file ONCE and a
4838 // failure ends the load (`return NULL`), so exactly one diagnostic
4839 // reaches the terminal. zshrs parsed the raw body in
4840 // `autoload_definition_source` and parses the WRAPPED text again
4841 // here, so a body zsh rejects printed the error twice — the second
4842 // time shifted one line by the wrap's leading `name() {`. The
4843 // first pass already reported it against the loadee's name; mute
4844 // this one. `noerrs` still sets ERRFLAG_ERROR (c:Src/utils.c:175),
4845 // which is what `compile_script_isolated` reads, so the load still
4846 // fails.
4847 let _muted = parse_failed.then(ZerrMute::enter);
4848 self.compile_script_isolated(registered)?
4849 };
4850 if let Some((dir, sha)) = key.as_ref() {
4851 match bincode::serialize(&chunk) {
4852 Ok(blob) => {
4853 if let Err(e) = crate::autoload_cache::try_save_one(name, &blob, dir, *sha) {
4854 tracing::warn!(name, error = %e, "autoload: rkyv chunk save failed");
4855 }
4856 }
4857 Err(e) => tracing::warn!(name, error = %e, "autoload: chunk serialize failed"),
4858 }
4859 }
4860 // c:Src/exec.c:5807-5816 — installing the definition is a funcdef
4861 // store (`execode(…, "evalautofunc")` for ksh style); nothing on that
4862 // path reaches the shell-exit hooks. Running the definition through
4863 // `run_chunk_with_exit_hooks` fired the EXIT trap, `TRAPEXIT` and
4864 // `zshexit` on the first call of EVERY autoloaded function (and
4865 // consumed a string EXIT trap so it never ran at exit): `trap 'echo T'
4866 // EXIT; autoload -Uz is-at-least; is-at-least 1.0` printed T first.
4867 self.run_chunk(chunk, "autoload:compiled")
4868 }
4869
4870 /// `execute_script` — see implementation.
4871 #[tracing::instrument(skip(self, script), fields(len = script.len()))]
4872 pub fn execute_script(&mut self, script: &str) -> Result<i32, String> {
4873 // lex+parse free ported + ZshCompiler is the only execution path.
4874 self.execute_script_zsh_pipeline(script)
4875 }
4876
4877 /// Run `script` with stdout AND stderr captured, returning `(exit status,
4878 /// output)` — the entry point for an embedder that owns the terminal (a
4879 /// TUI), where a stray `echo` corrupts the display.
4880 ///
4881 /// A shell cannot capture its output into an in-process buffer the way a
4882 /// single-runtime language can: a forked child writes fd 1 directly and
4883 /// knows nothing about the parent's buffers. The capture is therefore at fd
4884 /// level, and it differs from `$(…)` in the one way that matters to an
4885 /// embedder: [`Self::run_command_substitution`] runs on a sub-VM, as a
4886 /// subshell must, so a variable it sets is gone afterwards. This runs the
4887 /// script on THIS VM, so state persists across captured runs exactly as it
4888 /// does across ordinary [`Self::execute_script`] calls.
4889 ///
4890 /// The saved fds go through `movefd` to land at fd >= 10 and marked
4891 /// `FDT_INTERNAL`, per zsh's invariant that shell-internal fds never live
4892 /// below 10 — otherwise a script doing `exec 9>&-` closes the capture's own
4893 /// bookkeeping. A temp file, not a pipe, receives the output: with no
4894 /// concurrent reader, a pipe deadlocks the moment a script writes past the
4895 /// 64 KiB buffer.
4896 ///
4897 /// # Concurrency contract
4898 ///
4899 /// **While a capture is in flight, no other thread in the process may write
4900 /// fd 1 or fd 2.** POSIX has no per-thread fd table, so pointing fd 1 at the
4901 /// capture points it there for every thread at once; any byte another thread
4902 /// writes during the window lands in the returned `String` instead of on the
4903 /// terminal. The `CAPTURE_LOCK` below excludes a second *capture*, which is
4904 /// all a lock can do — a thread that never calls this function (a logger, a
4905 /// progress meter, a test harness's own reporter) is not excluded by
4906 /// anything, and its output is silently absorbed.
4907 ///
4908 /// This is not a gap that a different capture mechanism closes. C zsh dodges
4909 /// it for `$(…)` by forking: `getoutput` (`Src/exec.c:4816`) calls `zfork`
4910 /// and only the child does `redup(pipes[1], 1)` (`Src/exec.c:4837`), so the
4911 /// parent's fd 1 is never touched — but the child then runs `entersubsh`
4912 /// (`Src/exec.c:4838`) and a variable it sets is gone. Forking here would
4913 /// throw away the one property this call exists to provide (state persists
4914 /// on THIS VM across captured runs), so the cost is paid as a contract
4915 /// instead: **capture from one thread, and quiesce the rest.**
4916 pub fn execute_script_captured(&mut self, script: &str) -> (i32, String) {
4917 use std::io::{Read, Seek, SeekFrom};
4918 use std::os::unix::io::AsRawFd;
4919
4920 /// Serializes the redirect/restore window. fd 1 belongs to the process,
4921 /// not to a `ShellExecutor`, so two threads capturing at once would
4922 /// restore each other's fds mid-run and each would read back an empty
4923 /// file. An embedder that evaluates on one thread never contends here.
4924 /// It excludes another *capture* and nothing else: see the concurrency
4925 /// contract on this function for what remains the caller's problem.
4926 static CAPTURE_LOCK: std::sync::Mutex<()> = std::sync::Mutex::new(());
4927 let _guard = CAPTURE_LOCK.lock().unwrap_or_else(|e| e.into_inner());
4928
4929 let Ok(mut tmp) = tempfile::tempfile() else {
4930 // No temp file, no capture: run it anyway rather than silently
4931 // dropping the script, and report nothing captured.
4932 let status = self.execute_script(script).unwrap_or(1);
4933 return (status, String::new());
4934 };
4935
4936 // Flush Rust's buffered stdout against the REAL fd 1 before the swap,
4937 // or bytes written before this call drain into the capture instead
4938 // (the same ordering bug `run_command_substitution` documents).
4939 let _ = io::stdout().flush();
4940
4941 /// Puts fds 1 and 2 back on the way out, including when the run
4942 /// unwinds. A panic anywhere under `execute_script` would otherwise
4943 /// leave the whole PROCESS writing into a temp file that is already
4944 /// unlinked — every later write vanishes, starting with the one
4945 /// reporting the panic, which turns a localized bug into a silent one.
4946 struct RestoreFds {
4947 saved_out: i32,
4948 saved_err: i32,
4949 }
4950 impl Drop for RestoreFds {
4951 fn drop(&mut self) {
4952 let _ = io::stdout().flush();
4953 unsafe {
4954 libc::dup2(self.saved_out, libc::STDOUT_FILENO);
4955 libc::dup2(self.saved_err, libc::STDERR_FILENO);
4956 }
4957 crate::ported::utils::zclose(self.saved_out);
4958 crate::ported::utils::zclose(self.saved_err);
4959 }
4960 }
4961
4962 let saved_out = crate::ported::utils::movefd(unsafe { libc::dup(libc::STDOUT_FILENO) });
4963 let saved_err = crate::ported::utils::movefd(unsafe { libc::dup(libc::STDERR_FILENO) });
4964 unsafe {
4965 libc::dup2(tmp.as_raw_fd(), libc::STDOUT_FILENO);
4966 libc::dup2(tmp.as_raw_fd(), libc::STDERR_FILENO);
4967 }
4968 let restore = RestoreFds {
4969 saved_out,
4970 saved_err,
4971 };
4972
4973 let status = self.execute_script(script);
4974
4975 // Explicit, not end-of-scope: the temp file must be read back only
4976 // after the real fds are restored, or a diagnostic emitted while
4977 // reading would land in the very buffer being read.
4978 drop(restore);
4979
4980 let mut output = String::new();
4981 let _ = tmp.seek(SeekFrom::Start(0));
4982 let mut bytes = Vec::new();
4983 if tmp.read_to_end(&mut bytes).is_ok() {
4984 // Lossless for the same reason as the `$( cmd )` capture above:
4985 // a captured byte that is not valid UTF-8 is a byte the script
4986 // really wrote, not an encoding error to paper over.
4987 output = crate::script_bytes::decode_script_bytes(&bytes);
4988 }
4989 // Match `$(…)`: one trailing newline is an artifact of the last `echo`,
4990 // not part of the output.
4991 while output.ends_with('\n') {
4992 output.pop();
4993 }
4994
4995 (status.unwrap_or_else(|_| self.last_status()), output)
4996 }
4997
4998 /// Run an ALREADY-PARSED program (the back half of
4999 /// `execute_script_zsh_pipeline`): compile the `ZshProgram` to a
5000 /// fusevm Chunk and run it. Used by the ported `loop()` REPL
5001 /// (Src/init.c:220 `execode`), which parses via `parse_event` and
5002 /// hands the program here through the `execute_program` exec hook.
5003 /// Returns the resulting `$?` (1 on a compile/run error).
5004 pub fn execute_program(&mut self, program: &crate::parse::ZshProgram) -> i32 {
5005 let chunk = crate::compile_zsh::ZshCompiler::new().compile(program);
5006 match self.run_chunk(chunk, "loop") {
5007 Ok(status) => status,
5008 Err(_) => 1,
5009 }
5010 }
5011
5012 /// Whether `name` is a known function. Checks the compiled-functions
5013 /// table and the autoload-pending registry — `autoload foo` should
5014 /// make `whence foo`/`type foo`/`functions foo` recognize `foo` as
5015 /// a function before it's actually loaded. Doesn't trigger autoload
5016 /// itself; use `maybe_autoload` first if you need to load before
5017 /// introspecting.
5018 pub fn function_exists(&self, name: &str) -> bool {
5019 // Either compiled (already loaded) or shfunctab has an
5020 // autoload stub with PM_UNDEFINED set (pending). Matches C's
5021 // `lookupshfunc(name)` semantics at `Src/exec.c:5215`.
5022 if self.functions_compiled.contains_key(name) {
5023 return true;
5024 }
5025 crate::ported::hashtable::shfunctab_lock()
5026 .read()
5027 .ok()
5028 .map(|t| t.get(name).is_some())
5029 .unwrap_or(false)
5030 }
5031
5032 /// Sorted list of every known function name (union of compiled + source).
5033 pub fn function_names(&self) -> Vec<String> {
5034 let mut set: std::collections::BTreeSet<String> = std::collections::BTreeSet::new();
5035 for k in self.functions_compiled.keys() {
5036 set.insert(k.clone());
5037 }
5038 for k in self.function_source.keys() {
5039 set.insert(k.clone());
5040 }
5041 set.into_iter().collect()
5042 }
5043
5044 /// Dispatch a function by name. Thin passthru — autoload-materialize
5045 /// the body if needed, build a synthetic `shfunc`, and hand off to
5046 /// the canonical `doshfunc` port (`Src/exec.c:5823` →
5047 /// `src/ported/exec.rs::doshfunc`). doshfunc owns ALL scope
5048 /// management (starttrapscope/endtrapscope, startparamscope/
5049 /// endparamscope, funcdepth bump, pipestats save/restore, scriptname
5050 /// snapshot, BREAKS/CONTFLAG/LOOPS/RETFLAG snapshot+restore, `$0`
5051 /// override via FUNCTIONARGZERO, etc.). The body run itself is the
5052 /// Rust-only adaptation passed via the `body_runner` closure because
5053 /// zshrs runs function bodies through fusevm bytecode (not C zsh's
5054 /// wordcode walker via `runshfunc`).
5055 ///
5056 /// Returns `None` when the name isn't a known function so the caller
5057 /// can fall through to external dispatch.
5058 /// Body-only counterpart to [`dispatch_function_call`] — runs
5059 /// the function body WITHOUT wrapping in `doshfunc`. Used as the
5060 /// `body_runner` closure target by `src/ported/` callers that
5061 /// already wrap their own `crate::ported::exec::doshfunc(...)`
5062 /// call (so going back through `dispatch_function_call` would
5063 /// double-wrap the scope). Mirrors C's `runshfunc(prog, wrappers,
5064 /// name)` at `exec.c:6042` from doshfunc's perspective.
5065 pub fn run_function_body_only(&mut self, name: &str, args: &[String]) -> Option<i32> {
5066 // Held for the WHOLE call, not just the load: an autoloaded function is
5067 // registered TWICE — once when its file's text defines it, and again
5068 // (unchanged) when its chunk is compiled at call time — and the second
5069 // stamp would otherwise relabel it with the caller's scriptfilename.
5070 // See the AUTOLOAD_DEF_FILE consumer in fusevm_bridge.
5071 let mut _autoload_file_guard: Option<AutoloadFileGuard> = None;
5072 // Same Rust-port short-circuit as dispatch_function_call,
5073 // sans the doshfunc wrap.
5074 //
5075 // c:Src/exec.c:3760-3763 — `if (errflag) { lastval = 1; goto err; }`,
5076 // the gate execcmd_exec applies AFTER argument expansion and BEFORE it
5077 // invokes anything. A shell function whose argument list raised errflag
5078 // (`_describe -t t "${(j:,:)${(s: :)x y z}}" d` — a bad substitution)
5079 // is therefore never entered. zshrs's ordinary function path honours
5080 // that, but the compsys router short-circuits above it, so a NATIVE
5081 // PORT ran where the stock shell function did not and the completion
5082 // listed matches zsh never offered. Gate the short-circuit on the same
5083 // condition; the ordinary path below is left alone, since it already
5084 // aborts upstream.
5085 if (errflag.load(Ordering::Relaxed) & ERRFLAG_ERROR) != 0 {
5086 crate::ported::builtin::LASTVAL.store(1, Ordering::Relaxed); // c:3761
5087 return Some(1); // c:3762 goto err
5088 }
5089 if let Some(rc) = crate::compsys::router::dispatch_compsys(name, args) {
5090 // Plugin override (ABI v4) wins over the built-in Rust port.
5091 return Some(rc);
5092 }
5093 // Bug #657 gap #2 — `_regex_arguments`-generated completion functions
5094 // live in a runtime registry, not the static router table (a plain
5095 // `fn` ptr can't carry the dynamic name). Consult that registry here
5096 // so `compdef mycmd` → `_comps[cmd]=mycmd` → this call routes to the
5097 // compiled regex state machine.
5098 if let Some(rc) = crate::compsys::ported::_regex_arguments::dispatch_if_registered(name) {
5099 return Some(rc);
5100 }
5101 // c:Src/exec.c:5626 — see the twin site in
5102 // `dispatch_function_call`: a body loaded on THIS call runs one
5103 // `zsh_eval_context` frame deeper ("loadautofunc") than the
5104 // caller's "shfunc".
5105 let mut did_autoload = false;
5106 // Autoload prelude (same as dispatch_function_call's).
5107 if !self.functions_compiled.contains_key(name) {
5108 // On-demand $fpath autoload for `_`-prefixed compsys helpers that
5109 // compinit didn't register as autoload stubs — see the fuller
5110 // note in dispatch_function_call.
5111 if name.starts_with('_') && crate::ported::utils::getshfunc(name).is_none() {
5112 // c:6219 getfpfunc — gate the stub on the definition file
5113 // actually existing in $fpath, mirroring zsh's `compdef -na`
5114 // (only autoloads `_`-names present in fpath). Without this a
5115 // `_`-name with no file (e.g. a fasd completer trigger absent
5116 // from this fpath) got a phantom PM_UNDEFINED stub, and
5117 // loadautofn then leaked "function definition file not found"
5118 // to the terminal during completion. test_only=1 is a pure
5119 // probe; dump_out is preserved so .zwc-dump autoloads resolve.
5120 let mut _dir: Option<String> = None;
5121 let mut _dump = None;
5122 if crate::ported::exec::getfpfunc(name, &mut _dir, None, 1, &mut _dump).is_some() {
5123 let _ = self.execute_script_zsh_pipeline(&format!("autoload -rUz -- {name}"));
5124 }
5125 }
5126 if let Some(stub) = crate::ported::utils::getshfunc(name) {
5127 // c:Src/exec.c:5684-5704 (loadautofn) —
5128 // int noalias = noaliases;
5129 // noaliases = (shf->node.flags & PM_UNALIASED);
5130 // prog = getfpfunc(...); /* parses the file */
5131 // noaliases = noalias;
5132 // `autoload -U` records PM_UNALIASED (c:3354-3357), and its ONLY
5133 // effect is that the autoloaded body is PARSED with alias
5134 // expansion disabled. zshrs recorded the bit but never consulted
5135 // it, so a body calling `helper` picked up a caller-defined
5136 // `alias helper=...` — exactly what -U exists to prevent.
5137 let unaliased = (stub.node.flags as u32 & crate::ported::zsh_h::PM_UNALIASED) != 0;
5138 let noalias_save = crate::ported::lex::noaliases(); // c:5684
5139 crate::ported::lex::set_noaliases(unaliased); // c:5697
5140 let _restore_noaliases = NoAliasesRestore(noalias_save); // c:5704
5141 if (stub.node.flags as u32 & PM_UNDEFINED) != 0 {
5142 did_autoload = true; // c:5626 — body runs as "loadautofunc"
5143 let _sticky_load = AutoloadStickyGuard::enter(&stub);
5144 let boxed = Box::new(stub.clone());
5145 let ptr = Box::into_raw(boxed);
5146 let load_rc = crate::ported::exec::loadautofn(ptr, 0, 0, 0);
5147 unsafe {
5148 let _ = Box::from_raw(ptr);
5149 }
5150 // c:Src/exec.c:5713-5719 — `if (prog == &dummy_eprog) {
5151 // zwarn("%s: function definition file not found",
5152 // shf->node.nam); … return NULL; }`, and
5153 // c:5635-5644 execautofn: `if (!loadautofn(...)) return 1;`
5154 // A failed load is TERMINAL: C has already replaced
5155 // shf->funcdef with the mkautofn trampoline (c:3180), so
5156 // nothing of the old stub body survives to be re-run.
5157 // zshrs keeps the stub's TEXT on the shfunc node, and the
5158 // `if let Some(body)` arm below would hand that text back
5159 // to run_autoload_definition — re-executing the very
5160 // `autoload -X` that triggered this load. `cod() {
5161 // autoload -XUz }; cod` recursed until FUNCNEST and
5162 // printed the diagnostic 500 times (C04funcdef:38,39,40).
5163 if load_rc != 0 {
5164 return Some(1); // c:5719 NULL → c:5644 `return 1`
5165 }
5166 // c:5657 — preserve the fpath dir + PM_LOADDIR across the
5167 // funcdef re-register (which stamps filename="zsh"), so
5168 // `whence -v` reports the source. See the twin site below.
5169 let loaded_dir = crate::ported::utils::getshfunc(name)
5170 .and_then(|f| f.filename)
5171 .filter(|d| d != "zsh");
5172 // c:Src/exec.c:5682-5760 — C loads the body IN PLACE on the
5173 // existing Shfunc, so every stub flag except PM_UNDEFINED
5174 // survives. See the twin site below for why PM_ABSPATH_USED
5175 // in particular has to come back.
5176 let abspath_used =
5177 (stub.node.flags as u32 & crate::ported::zsh_h::PM_ABSPATH_USED) != 0;
5178 let ksh_style = autoload_is_ksh_style(name); // c:5781 (pre-registration)
5179 if let Some(body) = crate::ported::utils::getshfunc(name).and_then(|f| f.body) {
5180 _autoload_file_guard = Some(AutoloadFileGuard::enter(name));
5181 let AutoloadSource {
5182 text: registered,
5183 from_wordcode,
5184 parse_failed,
5185 } = autoload_register_source(name, &body);
5186 {
5187 // c:Src/exec.c:5735-5760 — C INSTALLS the parsed Eprog
5188 // as the function body; it executes nothing at load
5189 // time, so the global `lineno` still holds the line the
5190 // CALL was made on when doshfunc records
5191 // `funcsave->fstack.lineno = lineno` (c:6013). zshrs
5192 // installs the body by RUNNING `name() { … }` through
5193 // the pipeline, which walks the counter to the file's
5194 // last line — so the very first call of an autoloaded
5195 // function reported its caller's line as that instead:
5196 // `$functrace` read `script.zsh:1` where zsh reads
5197 // `script.zsh:4`, and inside completion `_subscript:0`
5198 // where zsh reads `_subscript:125`. Every LATER call
5199 // was already correct, because the load only happens
5200 // once.
5201 let caller_lineno = crate::ported::lex::lineno();
5202 // c:5384-5388 assigns `shf->lineno` only when a
5203 // `name() { … }` STATEMENT defines the function. An
5204 // autoload stub's Shfunc keeps the 0 it was created
5205 // with, and loadautofn replaces only `funcdef`, so zsh
5206 // reports `funcsourcetrace` as `<file>:0`. Running a
5207 // synthesized wrapper here stamps line 1 instead, so
5208 // put the stub's value back when the wrapper was ours.
5209 let synthesized = registered != body;
5210 let _ = self.run_autoload_definition(
5211 name,
5212 ®istered,
5213 ksh_style,
5214 from_wordcode,
5215 parse_failed,
5216 );
5217 crate::ported::lex::set_lineno(caller_lineno);
5218 if synthesized {
5219 // c:5384-5388 sets `shf->lineno` only where a
5220 // `name() { … }` STATEMENT defines the function; an
5221 // autoload stub keeps the 0 it was created with and
5222 // loadautofn replaces only `funcdef`, so
5223 // `funcsourcetrace` reads `<file>:0`. Executing our
5224 // synthesized wrapper records a line base of 1
5225 // instead. -1 marks "autoload-installed" so the
5226 // call-time clamp below can tell that apart from an
5227 // INLINE `f() { … }`, whose base underflows to 0 but
5228 // whose def line really is >= 1.
5229 self.function_line_base.insert(name.to_string(), -1);
5230 }
5231 }
5232 }
5233 if let Some(dir) = loaded_dir.as_deref() {
5234 restore_loaddir(name, dir, abspath_used, ksh_style);
5235 }
5236 } else if let Some(body) = stub.body.clone() {
5237 // c:Src/exec.c:5978 — the body runs under the stub's sticky
5238 // emulation; see AutoloadStickyGuard.
5239 let _sticky_load = AutoloadStickyGuard::enter(&stub);
5240 // c:Src/builtin.c:3180 (eval_autoload) — `autoload +X NAME`
5241 // loads the body EAGERLY through `loadautofn`, which sets
5242 // `body` + `filename`/PM_LOADDIR and clears PM_UNDEFINED
5243 // but leaves no compiled chunk behind. The first CALL of
5244 // such a function therefore lands in THIS arm, never the
5245 // PM_UNDEFINED arm above, so it needs the same
5246 // post-registration restore — otherwise `autoload +X
5247 // /abs/dir/NAME` lost the pair before the body ran and a
5248 // sibling `autoload -Uz SIB` inside it failed with
5249 // "function definition file not found" where zsh loads
5250 // /abs/dir/SIB. `functions[name]=body` (parameter.c
5251 // setpmfunction) reaches this arm too and simply has no
5252 // PM_LOADDIR, so the restore is skipped for it.
5253 let loaded_dir = ((stub.node.flags as u32 & crate::ported::zsh_h::PM_LOADDIR)
5254 != 0)
5255 .then(|| stub.filename.clone())
5256 .flatten();
5257 let abspath_used =
5258 (stub.node.flags as u32 & crate::ported::zsh_h::PM_ABSPATH_USED) != 0;
5259 // c:Src/Modules/parameter.c:295-315 — a body that arrived
5260 // through `functions[NAME]=BODY` is NOT a file. See
5261 // `body_came_from_param_assignment`.
5262 let param_body = body_came_from_param_assignment(stub.node.flags);
5263 let ksh_style = !param_body && autoload_is_ksh_style(name); // c:5781 (pre-registration)
5264 _autoload_file_guard = Some(AutoloadFileGuard::enter(name));
5265 // A `functions[name]=body` assignment is already a parsed
5266 // body, so it takes no file parse and cannot have reported
5267 // one (`parse_failed` is false by construction).
5268 let (registered, from_wordcode, parse_failed) = if param_body {
5269 (param_definition_source(name, &body), false, false)
5270 } else {
5271 let AutoloadSource {
5272 text,
5273 from_wordcode,
5274 parse_failed,
5275 } = autoload_register_source(name, &body);
5276 (text, from_wordcode, parse_failed)
5277 };
5278 {
5279 // c:Src/exec.c:5735-5760 — C INSTALLS the parsed Eprog
5280 // as the function body; it executes nothing at load
5281 // time, so the global `lineno` still holds the line the
5282 // CALL was made on when doshfunc records
5283 // `funcsave->fstack.lineno = lineno` (c:6013). zshrs
5284 // installs the body by RUNNING `name() { … }` through
5285 // the pipeline, which walks the counter to the file's
5286 // last line — so the very first call of an autoloaded
5287 // function reported its caller's line as that instead:
5288 // `$functrace` read `script.zsh:1` where zsh reads
5289 // `script.zsh:4`, and inside completion `_subscript:0`
5290 // where zsh reads `_subscript:125`. Every LATER call
5291 // was already correct, because the load only happens
5292 // once.
5293 let caller_lineno = crate::ported::lex::lineno();
5294 // c:5384-5388 assigns `shf->lineno` only when a
5295 // `name() { … }` STATEMENT defines the function. An
5296 // autoload stub's Shfunc keeps the 0 it was created
5297 // with, and loadautofn replaces only `funcdef`, so zsh
5298 // reports `funcsourcetrace` as `<file>:0`. Running a
5299 // synthesized wrapper here stamps line 1 instead, so
5300 // put the stub's value back when the wrapper was ours.
5301 let synthesized = registered != body;
5302 let _ = self.run_autoload_definition(
5303 name,
5304 ®istered,
5305 ksh_style,
5306 from_wordcode,
5307 parse_failed,
5308 );
5309 crate::ported::lex::set_lineno(caller_lineno);
5310 if synthesized {
5311 // c:5384-5388 sets `shf->lineno` only where a
5312 // `name() { … }` STATEMENT defines the function; an
5313 // autoload stub keeps the 0 it was created with and
5314 // loadautofn replaces only `funcdef`, so
5315 // `funcsourcetrace` reads `<file>:0`. Executing our
5316 // synthesized wrapper records a line base of 1
5317 // instead. -1 marks "autoload-installed" so the
5318 // call-time clamp below can tell that apart from an
5319 // INLINE `f() { … }`, whose base underflows to 0 but
5320 // whose def line really is >= 1.
5321 self.function_line_base.insert(name.to_string(), -1);
5322 }
5323 }
5324 if let Some(dir) = loaded_dir.as_deref() {
5325 restore_loaddir(name, dir, abspath_used, ksh_style);
5326 }
5327 }
5328 }
5329 }
5330 let chunk = self.functions_compiled.get(name).cloned()?;
5331 // c:5626 — `execode(shf->funcdef, 1, 0, "loadautofunc")`. Held
5332 // across the body run and dropped with the VM below.
5333 let _load_ctx =
5334 did_autoload.then(|| crate::ported::exec::EvalContextFrame::push("loadautofunc"));
5335 // c:Src/exec.c:5551-5557 (zsh-5.9.1, execautofn_basic) — the freshly
5336 // loaded body runs with `scriptname = shf->node.nam` and
5337 // `scriptfilename = getshfuncfile(shf)`, restored afterwards.
5338 // `doshfunc` skips its own `scriptname = name` for a PM_UNDEFINED
5339 // function (c:5835-5836), so without this every diagnostic in a body
5340 // autoloaded here kept the CALLER's name: `_complete:117:` printed as
5341 // `_main_complete:117:`. `lineno` is zeroed for the body here rather
5342 // than in `doshfunc` (see the PM_UNDEFINED note there), because the
5343 // load above must still see the caller's line.
5344 let autofn_saved = did_autoload.then(|| {
5345 let saved = (
5346 crate::ported::utils::scriptname_get(),
5347 crate::ported::utils::scriptfilename_get(),
5348 );
5349 crate::ported::utils::set_scriptname(Some(name.to_string())); // c:5553
5350 crate::ported::utils::set_scriptfilename(crate::ported::hashtable::getshfuncfile(
5351 name,
5352 )); // c:5554
5353 crate::ported::lex::set_lineno(0);
5354 saved
5355 });
5356 let seed_status = self.last_status();
5357 let _ = args; // fusevm body reads $1..$N from PPARAMS
5358 // Reuse a VM from the per-thread pool instead of building one from
5359 // scratch every call. `register_builtins` installs ~hundreds of
5360 // fn-pointer handlers into the VM's builtin_table; the table is
5361 // identical for every VM, so re-running it per function call was
5362 // pure waste (~130 profile samples in a tight call loop, the #2 hot
5363 // spot after option lookups). `VM::reset(chunk)` clears execution
5364 // state but PRESERVES builtin_table / host / JIT wiring, so a
5365 // recycled VM is call-ready without re-registration. Fresh VMs pay
5366 // the registration once. Nested calls simply check out additional
5367 // VMs; the pool grows to the max call depth. Re-entrant and
5368 // panic-safe: the VM is returned on the normal path below.
5369 // c:Src/exec.c:4364 — a `return` out of a redirected compound
5370 // command still runs `fixfds(save)`. See
5371 // `unwind_redirect_scopes_to`.
5372 let redir_depth = self.redirect_scope_stack.len();
5373 // A `return` past an open pipeline leaves its job slot held; see
5374 // exec_jobs::execpline_slots_mark.
5375 let slots_entry = crate::exec_jobs::execpline_slots_mark();
5376 let mut vm = crate::vm_pool::acquire(chunk);
5377 vm.last_status = seed_status;
5378 let _ = vm.run();
5379 let status = vm.last_status;
5380 drop(vm);
5381 crate::exec_jobs::execpline_slots_release_to(slots_entry);
5382 self.unwind_redirect_scopes_to(redir_depth);
5383 if let Some((sn, sfn)) = autofn_saved {
5384 crate::ported::utils::set_scriptname(sn); // c:5556
5385 crate::ported::utils::set_scriptfilename(sfn); // c:5557
5386 }
5387 Some(status)
5388 }
5389
5390 /// Run a shell function, growing the stack first when little is left.
5391 ///
5392 /// Each nested shell-function call costs thousands of bytes of native stack,
5393 /// and Linux gives the main thread the 8 MB `ulimit -s` (macOS honours the
5394 /// larger `-stack_size` link flag, see build.rs). FUNCNEST allows 500 levels,
5395 /// so a runaway or merely deep recursion overflowed the stack before the
5396 /// FUNCNEST guard could report it. `maybe_grow` continues on a fresh segment
5397 /// when fewer than 256 KiB remain, on the same thread, so signals, job control
5398 /// and traps behave exactly as before.
5399 pub fn dispatch_function_call(&mut self, name: &str, args: &[String]) -> Option<i32> {
5400 stacker::maybe_grow(256 * 1024, 32 * 1024 * 1024, || {
5401 self.dispatch_function_call_on_stack(name, args)
5402 })
5403 }
5404
5405 fn dispatch_function_call_on_stack(&mut self, name: &str, args: &[String]) -> Option<i32> {
5406 // Held for the WHOLE call, not just the load: an autoloaded function is
5407 // registered TWICE — once when its file's text defines it, and again
5408 // (unchanged) when its chunk is compiled at call time — and the second
5409 // stamp would otherwise relabel it with the caller's scriptfilename.
5410 // See the AUTOLOAD_DEF_FILE consumer in fusevm_bridge.
5411 let mut _autoload_file_guard: Option<AutoloadFileGuard> = None;
5412 // Nested scope for `>(cmd)` fd ownership — builtins running
5413 // inside the function body must not close the CALLER's
5414 // pending psub fds (`myfn >(cmd)` keeps /dev/fd/N alive for
5415 // the whole function, like C's per-job filelist). See
5416 // PSUB_SCOPE_DEPTH in fusevm_bridge.rs.
5417 let _psub_scope = crate::fusevm_bridge::PsubScope::enter();
5418 // c:Src/exec.c — `disable -f NAME` flips the DISABLED flag on
5419 // the shfunctab entry. `lookupshfunc` (which dispatch consults)
5420 // returns NULL for DISABLED entries, falling through to PATH
5421 // lookup → "command not found". zshrs keeps the compiled body
5422 // in functions_compiled independently of the flag, so check
5423 // shfunctab and short-circuit when DISABLED is set. Bug #221
5424 // in docs/BUGS.md.
5425 let is_disabled = crate::ported::hashtable::shfunctab_lock()
5426 .read()
5427 .ok()
5428 .and_then(|t| {
5429 let entry = t.get_including_disabled(name)?;
5430 Some((entry.node.flags as u32 & crate::ported::zsh_h::DISABLED as u32) != 0)
5431 })
5432 .unwrap_or(false);
5433 if is_disabled {
5434 return None;
5435 }
5436 // `_regex_arguments NAME …` (e.g. `_regex_arguments _sed_expressions …`
5437 // in `_sed`) eval-defines a real shell function NAME in zsh. This port
5438 // stores it in a runtime registry keyed by NAME (a static router fn-ptr
5439 // can't carry a dynamic name). `run_function_body_only` already consults
5440 // that registry, but `dispatch_function_call` — the path an `_arguments`
5441 // action (`:sed script:_sed_expressions`) or any by-name caller takes —
5442 // did not, so the call fell through to the autoload prelude and errored
5443 // "function definition file not found" (`sed -<TAB>`). Consult the
5444 // registry here too, before autoload. Returned directly (like
5445 // run_function_body_only) — the regex body drives compsys globals, not
5446 // function locals, so it needs no doshfunc scope wrap.
5447 if let Some(rc) = crate::compsys::ported::_regex_arguments::dispatch_if_registered(name) {
5448 return Some(rc);
5449 }
5450 // zshrs-original: `[compsys] backend = "rust"` short-circuit.
5451 // When a `_NAME` has a Rust port AND the user opted into the
5452 // rust backend, run the Rust fn directly here — but still
5453 // through the canonical doshfunc scope-management path below
5454 // (we synthesize a body_runner from the fn pointer). Router
5455 // returns None for names without a Rust port → graceful
5456 // fallback to the shfunc autoload path.
5457 //
5458 // Note: `compcore::callcompfunc` (the compsys entry hit by
5459 // Tab) wraps doshfunc itself per C `compcore.c:835`, so the
5460 // Rust _main_complete dispatch lands HERE only when called
5461 // from a non-compcore caller (e.g. a user shell script
5462 // directly invoking `_main_complete`). The doshfunc scope
5463 // wrap below applies uniformly to both.
5464 let direct_rust_fn: Option<fn(&[String]) -> i32> =
5465 crate::compsys::router::try_rust_dispatch(name);
5466 // A plugin-registered override (ABI v4, `zmodload -R`) also
5467 // intercepts natively: it supplies the body, so no shell autoload
5468 // or compiled chunk is needed — same as a built-in Rust port.
5469 let has_plugin_override = crate::extensions::plugin_host::compfn_override(name).is_some();
5470 // c:Src/exec.c:3760-3763 — the twin of the gate in
5471 // `run_function_body_only`: a native compsys port stands in for the
5472 // COMMAND here, so it must not run when the argument expansion that
5473 // built its argument list already raised errflag. Scoped to the ported
5474 // names on purpose — this function is also the shim ports use to call
5475 // ordinary shell functions (src/ported/exec.rs:8588), and those keep
5476 // whatever behaviour the executor gives them.
5477 if (direct_rust_fn.is_some() || has_plugin_override)
5478 && (errflag.load(Ordering::Relaxed) & ERRFLAG_ERROR) != 0
5479 {
5480 crate::ported::builtin::LASTVAL.store(1, Ordering::Relaxed); // c:3761
5481 return Some(1); // c:3762 goto err
5482 }
5483 // c:Src/exec.c:5626 — the body of a function loaded on THIS call
5484 // runs through `execode(shf->funcdef, 1, 0, "loadautofunc")`
5485 // (execautofn_basic), nested inside runshfunc's "shfunc" frame.
5486 // zshrs performs the load here, before `doshfunc`, so the flag
5487 // carries the fact into the body_runner that pushes the frame.
5488 let mut did_autoload = false;
5489 // Autoload prelude skipped when a Rust port OR plugin override wins
5490 // — no upstream shell function to load.
5491 if direct_rust_fn.is_none()
5492 && !has_plugin_override
5493 && !self.functions_compiled.contains_key(name)
5494 {
5495 // c:Src/exec.c:3105-3109 — `execcmd` resolves a command word as
5496 // if (!(cflags & (BINF_BUILTIN | BINF_COMMAND)) &&
5497 // (hn = shfunctab->getnode(shfunctab, cmdarg))) {
5498 // is_shfunc = 1; break; }
5499 // the identical gate repeating at c:3484-3488, then `builtintab`,
5500 // then `$PATH`, with a name that resolves nowhere ending at c:903
5501 // `zerr("command not found: %s", arg0)`. `getnode` finds a node
5502 // ALREADY IN `shfunctab` — a real definition, or the autoload STUB
5503 // that `autoload` / `compinit` put there. A file merely PRESENT in
5504 // `$fpath` is NOT a command in zsh; nothing autoloads it on demand.
5505 //
5506 // This used to probe `$fpath` with `getfpfunc` (c:Src/exec.c:6279)
5507 // for ANY `_`-prefixed name with no shfunctab entry and, on a hit,
5508 // run `autoload -rUz -- NAME` and execute the file. That gate keyed
5509 // on the FILENAME, where zsh keys on the file's TAG LINE: compinit
5510 // sh:511-525 reads `IFS=$' \t' read -rA _i_line < $_i_file` and then
5511 // `case $_i_tag in (\#compdef) compdef -na … ;; (\#autoload)
5512 // autoload -rUz "$_i_line[@]" ${_i_name} ;; esac`, so an UNTAGGED
5513 // `_`-file is registered by neither mechanism — yet zshrs ran it:
5514 // fpath=(/tmp/d); __zzz_probe
5515 // zsh zsh:1: command not found: __zzz_probe (rc 127)
5516 // zshrs RAN (rc 0)
5517 // The cost was a SWALLOWED DIAGNOSTIC: with the untagged
5518 // `__fasd_files_comp` / `__fasd_dirs_comp` in `$fpath`, a `_files`
5519 // override offering them as `_alternative` actions makes zsh report
5520 // `_alternative:71: command not found: __fasd_files_comp`, where
5521 // zshrs reported nothing at all because it executed the file.
5522 //
5523 // Nothing replaces the probe. The providers that stand in for a
5524 // stock `$fpath` file already short-circuit ABOVE this block
5525 // (`direct_rust_fn`, `has_plugin_override`), and `compinit` installs
5526 // a real PM_UNDEFINED stub for every `#compdef`/`#autoload` file it
5527 // scans — on the scan path and from a dump alike
5528 // (`register_autoload_stubs`, src/compsys/ported/compinit.rs:831-855,
5529 // called at :1137). So the `getshfunc` below IS C's
5530 // `shfunctab->getnode`, and a name it misses falls through to
5531 // builtin → `$PATH` → command-not-found, exactly as `execcmd` does.
5532 if let Some(stub) = crate::ported::utils::getshfunc(name) {
5533 // c:Src/exec.c:5684-5704 (loadautofn) — `autoload -U` records
5534 // PM_UNALIASED, whose ONLY effect is that the autoloaded body is
5535 // PARSED with alias expansion disabled:
5536 // int noalias = noaliases;
5537 // noaliases = (shf->node.flags & PM_UNALIASED);
5538 // prog = getfpfunc(...); /* parses the file */
5539 // noaliases = noalias;
5540 let unaliased = (stub.node.flags as u32 & crate::ported::zsh_h::PM_UNALIASED) != 0;
5541 let noalias_save = crate::ported::lex::noaliases(); // c:5684
5542 crate::ported::lex::set_noaliases(unaliased); // c:5697
5543 // c:5704 — restored on EVERY exit from this block, including the
5544 // early `return Some(1)` paths below.
5545 let _restore_noaliases = NoAliasesRestore(noalias_save);
5546 if (stub.node.flags as u32 & PM_UNDEFINED) != 0 {
5547 did_autoload = true; // c:5626 — body runs as "loadautofunc"
5548 let _sticky_load = AutoloadStickyGuard::enter(&stub);
5549 let boxed = Box::new(stub.clone());
5550 let ptr = Box::into_raw(boxed);
5551 let load_rc = crate::ported::exec::loadautofn(ptr, 0, 0, 0);
5552 unsafe {
5553 let _ = Box::from_raw(ptr);
5554 }
5555 // c:Src/exec.c:5713-5719 / 5635-5644 — a failed load is
5556 // TERMINAL: `loadautofn` already emitted "function
5557 // definition file not found" and C's `execautofn` returns
5558 // 1 without ever touching a body (C replaced shf->funcdef
5559 // with the mkautofn trampoline at c:3180). zshrs still has
5560 // the stub's TEXT on the shfunc node, and the `if let
5561 // Some(body)` arm below would re-run it — for an
5562 // `autoload -X` stub that means re-entering the autoload
5563 // path, which recursed to FUNCNEST and printed the
5564 // diagnostic 500 times (C04funcdef:38,39,40). The
5565 // `else if load_rc != 0` arm below stays as the (now
5566 // unreachable) faithful mirror of the same C line.
5567 if load_rc != 0 {
5568 return Some(1); // c:5719 NULL → c:5644 `return 1`
5569 }
5570 // c:Src/exec.c:5657 loadautofnsetfile — capture the fpath
5571 // directory loadautofn wrote so it can be restored (as an
5572 // absolutized path with PM_LOADDIR) after the funcdef pipeline
5573 // below clobbers `filename` to scriptfilename ("zsh"). Without
5574 // this, `whence -v <autoloaded>` printed "from zsh".
5575 let loaded_dir = crate::ported::utils::getshfunc(name)
5576 .and_then(|f| f.filename)
5577 .filter(|d| d != "zsh");
5578 // c:Src/exec.c:5682-5760 — C loads the body IN PLACE on the
5579 // existing Shfunc: `shf->node.flags &= ~PM_UNDEFINED`
5580 // (c:5751) is the only flag C clears, so PM_ABSPATH_USED —
5581 // stamped by `autoload -Uz /abs/dir/NAME`
5582 // (`add_autoload_function`, Src/builtin.c:3290-3291) —
5583 // survives the load. zshrs re-registers the body through the
5584 // funcdef pipeline, which builds a FRESH node and drops the
5585 // whole flag word; `loadautofnsetfile` below puts filename +
5586 // PM_LOADDIR back, and PM_ABSPATH_USED has to come back with
5587 // them. It is read by `add_autoload_function`'s sibling arm
5588 // (Src/builtin.c:3310-3323): when a function loaded by
5589 // absolute path autoloads a sibling with a bare name, C
5590 // inherits the CALLER's load directory — `if ((shf2 = ...
5591 // getnode2(shfunctab, calling_f)) && (shf2->node.flags &
5592 // PM_LOADDIR) && (shf2->node.flags & PM_ABSPATH_USED) && ...`
5593 // Without the restore that test never fired, so
5594 // `autoload -Uz $D/wrapper; wrapper` → `autoload -Uz sibling`
5595 // failed with "function definition file not found" (zsh runs
5596 // $D/sibling), and compsys `_`-names fell through to the Rust
5597 // port instead of the user's file.
5598 let abspath_used =
5599 (stub.node.flags as u32 & crate::ported::zsh_h::PM_ABSPATH_USED) != 0;
5600 if let Some(body) = crate::ported::utils::getshfunc(name).and_then(|f| f.body) {
5601 let ksh_style = autoload_is_ksh_style(name); // c:5781 (pre-registration)
5602 _autoload_file_guard = Some(AutoloadFileGuard::enter(name));
5603 let AutoloadSource {
5604 text: registered,
5605 from_wordcode,
5606 parse_failed,
5607 } = autoload_register_source(name, &body);
5608 // c:Src/exec.c:5739 — the ksh-autoload body runs via
5609 // `execode(prog, 1, 0, "evalautofunc")` at the function
5610 // invocation's locallevel, so a `return`/`break`/
5611 // `continue` inside the file body is CONTAINED to the
5612 // autoload call. add-zle-hook-widget's first line is
5613 // `zmodload -e zsh/zle || return 1`; when a plugin has
5614 // leaked `ksh_autoload` on (e.g. a bare `emulate sh`),
5615 // that `return` must NOT propagate out and abort the
5616 // caller's precmd/shell (zsh warns "not defined by file"
5617 // and CONTINUES). Save & restore the control-flow flags
5618 // around the body run to reinstate that boundary.
5619 {
5620 use crate::ported::builtin::{
5621 BREAKS, EXIT_PENDING, EXIT_VAL, RETFLAG, SHELL_EXITING,
5622 };
5623 use std::sync::atomic::Ordering::Relaxed;
5624 // c:Src/exec.c:5739 — `execode(prog, 1, 0,
5625 // "evalautofunc")` runs the file body as part of the
5626 // autoload invocation. add-zle-hook-widget's
5627 // `zmodload -e zsh/zle || return 1` sits at the
5628 // file's TOP LEVEL (above its anon-func wrapper); at
5629 // script scope a top-level `return` is a shell EXIT,
5630 // so running the body as a plain script aborted the
5631 // caller's precmd/shell. `return` is contained when
5632 // `locallevel || sourcelevel` (bin_return, c:5840) —
5633 // raise SOURCELEVEL (the file-source counter, which
5634 // unlike locallevel does NOT open a local scope, so
5635 // the body's global assignments still land globally)
5636 // so the top-level `return` returns from the load
5637 // instead of exiting. Save/restore the control-flow
5638 // flags so nothing leaks — matching zsh's
5639 // warn-and-continue.
5640 use crate::ported::init::sourcelevel;
5641 let saved_retflag = RETFLAG.swap(0, Relaxed);
5642 let saved_breaks = BREAKS.swap(0, Relaxed);
5643 let saved_exit_pending = EXIT_PENDING.swap(0, Relaxed);
5644 let saved_exit_val = EXIT_VAL.swap(0, Relaxed);
5645 let saved_shell_exiting = SHELL_EXITING.swap(0, Relaxed);
5646 sourcelevel.fetch_add(1, Relaxed);
5647 {
5648 // c:Src/exec.c:5739 — the ksh-autoload branch
5649 // runs the file through
5650 // `execode(prog, 1, 0, "evalautofunc")`, so
5651 // that label is on `zsh_eval_context` while
5652 // the file body executes.
5653 let _ctx =
5654 crate::ported::exec::EvalContextFrame::push("evalautofunc");
5655 {
5656 // c:Src/exec.c:5735-5760 — C INSTALLS the parsed Eprog
5657 // as the function body; it executes nothing at load
5658 // time, so the global `lineno` still holds the line the
5659 // CALL was made on when doshfunc records
5660 // `funcsave->fstack.lineno = lineno` (c:6013). zshrs
5661 // installs the body by RUNNING `name() { … }` through
5662 // the pipeline, which walks the counter to the file's
5663 // last line — so the very first call of an autoloaded
5664 // function reported its caller's line as that instead:
5665 // `$functrace` read `script.zsh:1` where zsh reads
5666 // `script.zsh:4`, and inside completion `_subscript:0`
5667 // where zsh reads `_subscript:125`. Every LATER call
5668 // was already correct, because the load only happens
5669 // once.
5670 let caller_lineno = crate::ported::lex::lineno();
5671 // c:5384-5388 assigns `shf->lineno` only when a
5672 // `name() { … }` STATEMENT defines the function. An
5673 // autoload stub's Shfunc keeps the 0 it was created
5674 // with, and loadautofn replaces only `funcdef`, so zsh
5675 // reports `funcsourcetrace` as `<file>:0`. Running a
5676 // synthesized wrapper here stamps line 1 instead, so
5677 // put the stub's value back when the wrapper was ours.
5678 let synthesized = registered != body;
5679 let _ = self.run_autoload_definition(
5680 name,
5681 ®istered,
5682 ksh_style,
5683 from_wordcode,
5684 parse_failed,
5685 );
5686 crate::ported::lex::set_lineno(caller_lineno);
5687 if synthesized {
5688 // c:5384-5388 sets `shf->lineno` only where a
5689 // `name() { … }` STATEMENT defines the function; an
5690 // autoload stub keeps the 0 it was created with and
5691 // loadautofn replaces only `funcdef`, so
5692 // `funcsourcetrace` reads `<file>:0`. Executing our
5693 // synthesized wrapper records a line base of 1
5694 // instead. -1 marks "autoload-installed" so the
5695 // call-time clamp below can tell that apart from an
5696 // INLINE `f() { … }`, whose base underflows to 0 but
5697 // whose def line really is >= 1.
5698 self.function_line_base.insert(name.to_string(), -1);
5699 }
5700 }
5701 }
5702 sourcelevel.fetch_sub(1, Relaxed);
5703 RETFLAG.store(saved_retflag, Relaxed);
5704 BREAKS.store(saved_breaks, Relaxed);
5705 EXIT_PENDING.store(saved_exit_pending, Relaxed);
5706 EXIT_VAL.store(saved_exit_val, Relaxed);
5707 SHELL_EXITING.store(saved_shell_exiting, Relaxed);
5708 }
5709 if let Some(dir) = loaded_dir.as_deref() {
5710 restore_loaddir(name, dir, abspath_used, ksh_style);
5711 }
5712 if !self.functions_compiled.contains_key(name) {
5713 // c:Src/exec.c:5742-5745 — ksh-style load ran
5714 // the file (`execode`, "evalautofunc") but it
5715 // didn't define NAME:
5716 // `zwarn("%s: function not defined by file", n);`
5717 // That diagnostic lives INSIDE C's
5718 // `ksh == 2 || (ksh == 1 && isset(KSHAUTOLOAD))`
5719 // branch (c:5726). The zsh-style branch (c:5753-
5720 // 5758) installs the parsed program as the body
5721 // outright and cannot fail this way, so zsh prints
5722 // nothing there — and when the file did not parse,
5723 // `getfpfunc` already returned NULL and the load
5724 // ended at c:5722 with only the parse error.
5725 // zshrs used to print it for every failed
5726 // registration, so `g(){ emulate ksh -c _files }`
5727 // gained a `_files: function not defined by file`
5728 // line zsh never emits. Status still propagates:
5729 // c:5644 `if (!loadautofn(...)) return 1`.
5730 if ksh_style {
5731 crate::ported::utils::zwarn(&format!(
5732 "{}: function not defined by file",
5733 name
5734 ));
5735 }
5736 return Some(1);
5737 }
5738 } else if load_rc != 0 {
5739 // c:Src/exec.c:5713-5719 / 5635-5644 —
5740 // `execautofn`'s `if (!loadautofn(...)) return 1`
5741 // propagates the loadautofn failure as the
5742 // command's exit status. zshrs's previous
5743 // path returned None here, falling through to
5744 // execute_external which emitted a SECOND
5745 // diagnostic (`command not found: NAME`) on
5746 // top of loadautofn's `function definition
5747 // file not found`. Mirror C: when load failed
5748 // AND the stub still has no body, surface
5749 // status=1 so the caller does NOT fall back
5750 // to PATH search.
5751 return Some(1);
5752 }
5753 } else if let Some(body) = stub.body.clone() {
5754 // c:Src/exec.c:5978 — the body runs under the stub's sticky
5755 // emulation; see AutoloadStickyGuard.
5756 let _sticky_load = AutoloadStickyGuard::enter(&stub);
5757 // c:Src/Modules/parameter.c::setpmfunction — function
5758 // registered via `functions[name]=body` lives in
5759 // shfunctab with `body` set but `functions_compiled`
5760 // empty (the canonical port stores the parsed eprog,
5761 // not a fusevm Chunk). Lazy-compile here by feeding
5762 // the body through the standard funcdef pipeline so
5763 // the next CallFunction op finds the chunk.
5764 //
5765 // c:Src/builtin.c:3180 (eval_autoload) — `autoload +X NAME`
5766 // reaches this arm as well: it loads the body EAGERLY via
5767 // `loadautofn`, which sets `body` + `filename`/PM_LOADDIR
5768 // and clears PM_UNDEFINED but leaves no compiled chunk, so
5769 // the first CALL never sees the PM_UNDEFINED arm above.
5770 // Restore the load directory after re-registration exactly
5771 // as that arm does — otherwise `autoload +X /abs/dir/NAME`
5772 // dropped PM_LOADDIR|PM_ABSPATH_USED before the body ran
5773 // and a sibling `autoload -Uz SIB` inside it failed with
5774 // "function definition file not found" where zsh loads
5775 // /abs/dir/SIB. The `functions[name]=body` case has no
5776 // PM_LOADDIR, so the restore is skipped for it.
5777 let loaded_dir = ((stub.node.flags as u32 & crate::ported::zsh_h::PM_LOADDIR)
5778 != 0)
5779 .then(|| stub.filename.clone())
5780 .flatten();
5781 let abspath_used =
5782 (stub.node.flags as u32 & crate::ported::zsh_h::PM_ABSPATH_USED) != 0;
5783 // c:Src/Modules/parameter.c:295-315 — a body that arrived
5784 // through `functions[NAME]=BODY` is NOT a file. See
5785 // `body_came_from_param_assignment`.
5786 let param_body = body_came_from_param_assignment(stub.node.flags);
5787 let ksh_style = !param_body && autoload_is_ksh_style(name); // c:5781 (pre-registration)
5788 _autoload_file_guard = Some(AutoloadFileGuard::enter(name));
5789 // A `functions[name]=body` assignment is already a parsed
5790 // body, so it takes no file parse and cannot have reported
5791 // one (`parse_failed` is false by construction).
5792 let (registered, from_wordcode, parse_failed) = if param_body {
5793 (param_definition_source(name, &body), false, false)
5794 } else {
5795 let AutoloadSource {
5796 text,
5797 from_wordcode,
5798 parse_failed,
5799 } = autoload_register_source(name, &body);
5800 (text, from_wordcode, parse_failed)
5801 };
5802 {
5803 // c:Src/exec.c:5735-5760 — C INSTALLS the parsed Eprog
5804 // as the function body; it executes nothing at load
5805 // time, so the global `lineno` still holds the line the
5806 // CALL was made on when doshfunc records
5807 // `funcsave->fstack.lineno = lineno` (c:6013). zshrs
5808 // installs the body by RUNNING `name() { … }` through
5809 // the pipeline, which walks the counter to the file's
5810 // last line — so the very first call of an autoloaded
5811 // function reported its caller's line as that instead:
5812 // `$functrace` read `script.zsh:1` where zsh reads
5813 // `script.zsh:4`, and inside completion `_subscript:0`
5814 // where zsh reads `_subscript:125`. Every LATER call
5815 // was already correct, because the load only happens
5816 // once.
5817 let caller_lineno = crate::ported::lex::lineno();
5818 // c:5384-5388 assigns `shf->lineno` only when a
5819 // `name() { … }` STATEMENT defines the function. An
5820 // autoload stub's Shfunc keeps the 0 it was created
5821 // with, and loadautofn replaces only `funcdef`, so zsh
5822 // reports `funcsourcetrace` as `<file>:0`. Running a
5823 // synthesized wrapper here stamps line 1 instead, so
5824 // put the stub's value back when the wrapper was ours.
5825 let synthesized = registered != body;
5826 let _ = self.run_autoload_definition(
5827 name,
5828 ®istered,
5829 ksh_style,
5830 from_wordcode,
5831 parse_failed,
5832 );
5833 crate::ported::lex::set_lineno(caller_lineno);
5834 if synthesized {
5835 // c:5384-5388 sets `shf->lineno` only where a
5836 // `name() { … }` STATEMENT defines the function; an
5837 // autoload stub keeps the 0 it was created with and
5838 // loadautofn replaces only `funcdef`, so
5839 // `funcsourcetrace` reads `<file>:0`. Executing our
5840 // synthesized wrapper records a line base of 1
5841 // instead. -1 marks "autoload-installed" so the
5842 // call-time clamp below can tell that apart from an
5843 // INLINE `f() { … }`, whose base underflows to 0 but
5844 // whose def line really is >= 1.
5845 self.function_line_base.insert(name.to_string(), -1);
5846 }
5847 }
5848 if let Some(dir) = loaded_dir.as_deref() {
5849 restore_loaddir(name, dir, abspath_used, ksh_style);
5850 }
5851 }
5852 }
5853 }
5854 // When a Rust port is registered, skip the fusevm Chunk
5855 // lookup entirely — the body_runner closure below will run
5856 // the Rust fn pointer directly. Otherwise require a compiled
5857 // chunk for the autoloaded body.
5858 let mut chunk_opt = if direct_rust_fn.is_some() || has_plugin_override {
5859 None
5860 } else {
5861 Some(self.functions_compiled.get(name).cloned()?)
5862 };
5863
5864 // c:Src/exec.c:5634 (execshfunc) — `xtrerr = stderr;`: the body traces
5865 // to the stderr the call's own redirections set up.
5866 crate::ported::utils::xtrerr.store(2, Ordering::Relaxed);
5867
5868 // zshrs-specific bookkeeping that doshfunc doesn't own:
5869 // - prompt_funcstack (PS4 trace) push/pop
5870 // - local_scope_depth FUNCNEST guard
5871 //
5872 // c:Src/exec.c::funcnest_check — C zsh allows FUNCNEST=500 by
5873 // default. zshrs's per-call stack usage is heavier (vm_helper
5874 // state, fusevm closures, parse buffers), so on the default 8MB
5875 // stack a deep recursion overflowed around depth ~80-120 and
5876 // crashed. That is now fixed at the source: the shell runs on a
5877 // 512MB-stack thread (see bins/zshrs.rs::main), which comfortably
5878 // fits FUNCNEST (500) nested heavy frames. So the effective limit
5879 // is the user's FUNCNEST (default 500), matching zsh — no premature
5880 // clamp — with a generous hard ceiling as a last-resort backstop
5881 // that stays well under the big stack's capacity. Bug #519 (the
5882 // crash) / #643 (the false-positive clamp at 80 that broke
5883 // legitimately deep recursion). The authoritative FUNCNEST error
5884 // is also enforced in doshfunc (exec.rs) on the FS_FUNC depth.
5885 const FUNCNEST_RUST_CEILING: usize = 6000;
5886 let funcnest_user: usize = self
5887 .scalar("FUNCNEST")
5888 .and_then(|s| s.parse().ok())
5889 .unwrap_or(500);
5890 let funcnest_limit = funcnest_user.min(FUNCNEST_RUST_CEILING);
5891 if self.local_scope_depth >= funcnest_limit {
5892 // c:Src/exec.c:6060-6063 —
5893 // zerr("maximum nested function level reached; increase FUNCNEST?");
5894 // lastval = 1;
5895 // goto undoshfunc;
5896 // `zerr` is what makes this FATAL: it raises errflag, so the
5897 // enclosing list stops and a non-interactive shell exits.
5898 // zsh 5.9:
5899 // zsh -fc 'FUNCNEST=2; f() { f; }; f; printf after'
5900 // prints only the diagnostic and exits 1 — no `after`. bash
5901 // agrees ("Function invocations that exceed this nesting level
5902 // cause the current command to abort", bash(1) FUNCNEST), and
5903 // its own message is likewise followed by exit 1.
5904 //
5905 // This guard printed with a bare `eprintln!` and returned 1
5906 // WITHOUT raising errflag, so the runaway recursion stopped but
5907 // the script kept running — `printf after` ran and the shell
5908 // exited 0. The ported check in exec.rs::doshfunc already does
5909 // the C trio, but this one fires first (it is the zshrs-only
5910 // stack backstop, evaluated before dispatch reaches doshfunc),
5911 // so it has to carry the same side effects.
5912 //
5913 // The message is written here rather than through `zerr`
5914 // because C's prefix is the *function* name — `scriptname` is
5915 // the running function inside doshfunc — and this guard runs
5916 // before that switch; going through zerr would print the outer
5917 // script name instead. Byte-compared against zsh 5.9.
5918 // c:Src/utils.c zerr → zerrmsg prints `scriptname:lineno: msg`
5919 // whenever `scriptname` is set (which it is here: c:5963
5920 // `scriptname = dupstring(name)` runs BEFORE the c:6060 check).
5921 // The `:lineno` half was missing, so
5922 // ( FUNCNEST=0; fn() { true; }; fn )
5923 // printed `fn: maximum …` where zsh prints `fn:4: maximum …`
5924 // (C04funcdef:46).
5925 // c:Src/utils.c:301-308 zerrmsg — the `lineno: ` part is only
5926 // printed when `lineno` is non-zero (else a single space), so a
5927 // recursion on a function's first line reads `f: maximum …`.
5928 let lineno = crate::ported::lex::lineno();
5929 let shows_lineno = (crate::ported::zsh_h::unset(crate::ported::zsh_h::SHINSTDIN)
5930 || crate::ported::params::locallevel.here() != 0)
5931 && lineno != 0;
5932 if shows_lineno {
5933 eprintln!(
5934 "{}:{}: maximum nested function level reached; increase FUNCNEST?",
5935 name, lineno
5936 );
5937 } else {
5938 eprintln!("{}: maximum nested function level reached; increase FUNCNEST?", name);
5939 }
5940 errflag.fetch_or(ERRFLAG_ERROR, Ordering::Relaxed); // c:6061 (zerr)
5941 crate::ported::builtin::LASTVAL.store(1, Ordering::Relaxed); // c:6062
5942 return Some(1);
5943 }
5944 let display_name = if name.starts_with("_zshrs_anon_") {
5945 "(anon)".to_string()
5946 } else {
5947 name.to_string()
5948 };
5949 let line_base = self.function_line_base.get(name).copied().unwrap_or(0);
5950 let def_file = self.function_def_file.get(name).cloned().flatten();
5951 self.prompt_funcstack
5952 .push((name.to_string(), line_base, def_file));
5953 self.local_scope_depth += 1;
5954
5955 // Synthetic shfunc for doshfunc — carries the name + def-file
5956 // info so funcstack push gets a proper filename. funcdef/body
5957 // stay None because the wordcode body is irrelevant on this
5958 // path (body_runner runs the fusevm Chunk directly).
5959 // c:Src/exec.c:5390-5410 — execfuncdef records the
5960 // current `scriptfilename` on the shfunc at definition
5961 // time so funcsourcetrace can show file:line of the
5962 // function's source. The function_def_file map stores
5963 // this; fall back to the live scriptfilename so dynamic
5964 // / non-`compile_funcdef`-routed definitions still get a
5965 // sensible filename. Without the fallback, the synth_shf
5966 // saw None and the funcstack push at exec.rs:5719
5967 // defaulted to an empty string, which the funcsourcetrace
5968 // getfn rendered as `:N` (or worse, picked up the
5969 // function name from a parallel field). Bug #515.
5970 // c:Src/exec.c:5620/5625 — the source file is `getshfuncfile(shf)`,
5971 // which reads the shfunc's own `filename` (authoritative: set by
5972 // execfuncdef for a normally-defined function, and by loadautofn — as
5973 // the fpath dir with PM_LOADDIR — for an AUTOLOADED one). Prefer it.
5974 // `function_def_file` is a zshrs-only side map that, for an autoloaded
5975 // function, was stamped with the OUTER `scriptfilename` ("zsh") at
5976 // compile time, not the fpath file — so it must NOT override
5977 // getshfuncfile. Consulting it second still covers functions whose
5978 // shfunc `filename` wasn't recorded (compile_funcdef-routed defs);
5979 // scriptfilename is the final fallback. Without getshfuncfile winning,
5980 // funcsourcetrace reported "zsh" for every autoloaded completer, which
5981 // broke `_git`: its first git-completion.bash search path is
5982 // `"$(dirname ${funcsourcetrace[1]%:*})"/git-completion.bash` — "zsh"
5983 // resolved to `./git-completion.bash`, found nothing, and
5984 // `. "$script"` errored (`_git:.:48: no such file or directory`).
5985 let synth_filename = crate::ported::hashtable::getshfuncfile(name)
5986 .or_else(|| self.function_def_file.get(name).cloned().flatten())
5987 .or_else(|| self.scriptfilename.clone());
5988 // c:Src/exec.c:5409 — `shf->lineno = lineno;` (def line).
5989 // `function_line_base[name]` carries compile_funcdef's
5990 // `lineno_offset = first_body_line - 1` — equals the def line
5991 // for multi-line `f() {\n body }` but underflows to 0 for
5992 // INLINE `f() { body }` (def and body share a line). zsh's
5993 // funcsourcetrace reports the def line as 1-based, so clamp
5994 // to >= 1 to handle the inline case without rebuilding
5995 // line tracking through the parser. Bug #396.
5996 let synth_lineno = {
5997 let base = self.function_line_base.get(name).copied().unwrap_or(0);
5998 if base < 0 {
5999 // Autoload-installed (see the -1 marker at the install
6000 // site): c:5384 never runs for it, so the def line is 0.
6001 0
6002 } else {
6003 std::cmp::max(1i64, base)
6004 }
6005 };
6006 // Carry the REAL function's attribute flags over from shfunctab.
6007 // `functions -t/-T/-W` store PM_TAGGED / PM_TAGGED_LOCAL /
6008 // PM_WARNNESTED on the shfunctab node (builtin.rs c:3719), and
6009 // doshfunc turns PM_TAGGED* into XTRACE for the duration of the call
6010 // (exec.c:5954-5960). Hardcoding 0 here severed that link: the flags
6011 // were parsed and stored correctly, but the synthesized shfunc handed
6012 // to doshfunc always claimed "no attributes", so `functions -t f; f`
6013 // ran silently while `setopt xtrace` (a global option, not routed
6014 // through this struct) traced normally. Bug #1058.
6015 // The shfunctab key is the REGISTRATION name, which for an
6016 // anonymous function is the generated `_zshrs_anon_*` (only the
6017 // DISPLAY name is `(anon)` — c:Src/exec.c:5492 sets
6018 // `shf->node.nam = ANONYMOUS_FUNCTION_NAME` on the same struct
6019 // that already carries `tracing_flags` from c:5437). Looking the
6020 // flags up under the display name missed every anonymous
6021 // function, so `function -T { … }` ran untraced (E02xtrace:7,9).
6022 let synth_flags = crate::ported::hashtable::shfunctab_lock()
6023 .read()
6024 .ok()
6025 .and_then(|t| {
6026 t.get(name)
6027 .or_else(|| t.get(display_name.as_str()))
6028 .map(|s| s.node.flags)
6029 })
6030 .unwrap_or(0)
6031 // PM_LOADDIR is the ONE flag that must not ride along, because it
6032 // describes the `filename` FIELD and `synth_filename` above is not
6033 // that field — it is already `getshfuncfile`'s ANSWER.
6034 // c:Src/hashtable.c:1061 `zhtricat(shf->filename, "/", shf->node.nam)`
6035 // says PM_LOADDIR means "filename holds the fpath DIRECTORY, append
6036 // the name"; doshfunc re-applies exactly that when it builds the
6037 // funcstack frame (exec.rs c:6019, the PM_LOADDIR arm). Carrying
6038 // the flag onto a node whose filename is the resolved FILE appended
6039 // `/name` a SECOND time, so an autoloaded completer's
6040 // `$funcsourcetrace[1]` read `<dir>/_git/_git:0` where zsh reads
6041 // `<dir>/_git:0`. That is not cosmetic: `_git` (git's own zsh
6042 // wrapper) locates git-completion.bash with
6043 // `"$(dirname ${funcsourcetrace[1]%:*})"/git-completion.bash`, so
6044 // the doubled component sent it to a directory that does not exist,
6045 // left `$script` empty, and `. "$script"` failed with
6046 // `_git:.:48: no such file or directory:`. This silently undid
6047 // 8d8adf9c48, whose whole purpose was to make that search succeed.
6048 // `functions_source` / `whence -v` were unaffected — they call
6049 // `getshfuncfile` against the REAL shfunctab node, which is
6050 // correct; only the synthesized frame doubled.
6051 & !(crate::ported::zsh_h::PM_LOADDIR as i32);
6052 // c:Src/exec.c:5978 — `if (sticky_emulation_differs(shfunc->sticky))`
6053 // reads the STORED per-function sticky snapshot that
6054 // `shfunc_set_sticky` (c:5402) stamped at definition time. The
6055 // synthesized shfunc hardcoded `sticky: None`, so a function
6056 // defined under `emulate sh -c '...'` never re-entered its
6057 // emulation when called (B07emulate.ztst:6,7,8,12,13,14).
6058 // Carry it over from shfunctab like `synth_flags` above.
6059 let synth_sticky = crate::ported::hashtable::shfunctab_lock()
6060 .read()
6061 .ok()
6062 .and_then(|t| {
6063 t.get(name)
6064 .or_else(|| t.get(display_name.as_str()))
6065 .and_then(|s| {
6066 s.sticky
6067 .as_deref()
6068 .map(|b| crate::ported::exec::sticky_emulation_dup(b, 0))
6069 })
6070 });
6071 let mut synth_shf = crate::ported::zsh_h::shfunc {
6072 node: crate::ported::zsh_h::hashnode {
6073 next: None,
6074 nam: display_name.clone(),
6075 flags: synth_flags,
6076 },
6077 filename: synth_filename,
6078 lineno: synth_lineno,
6079 funcdef: None,
6080 redir: None,
6081 sticky: synth_sticky,
6082 body: None,
6083 redir_text: None,
6084 };
6085 // doshargs: C convention — argv[0] = function name (for
6086 // FUNCTIONARGZERO `$0`), argv[1..] = real positional args.
6087 let mut doshargs: Vec<String> = vec![display_name.clone()];
6088 doshargs.extend(args.iter().cloned());
6089
6090 // Seed `$?` with the parent's last status — C zsh's
6091 // doshfunc inherits lastval automatically because it's a
6092 // process-global; the fusevm VM creates a fresh
6093 // `vm.last_status = 0` per call, so we mirror the inherit
6094 // explicitly. Without this, a function reading `$?` BEFORE
6095 // running any command sees 0 instead of the caller's status.
6096 let seed_status = self.last_status();
6097 let body_args: Vec<String> = args.to_vec();
6098 let name_owned = name.to_string();
6099 let body_runner = move || -> i32 {
6100 // c:Src/exec.c:5626 — `execode(shf->funcdef, 1, 0,
6101 // "loadautofunc")`. On the call that autoloaded the function,
6102 // C runs its body one `zsh_eval_context` frame deeper than
6103 // runshfunc's "shfunc", which is why zsh reports
6104 // `shfunc:loadautofunc:…` down a chain of freshly autoloaded
6105 // completers where zshrs reported a flat `shfunc:shfunc:…`.
6106 let _load_ctx =
6107 did_autoload.then(|| crate::ported::exec::EvalContextFrame::push("loadautofunc"));
6108 // Branch: plugin override (ABI v4) → built-in Rust port →
6109 // fusevm Chunk (autoloaded shell body). All run INSIDE
6110 // doshfunc's scope so prologue/epilogue applies identically.
6111 if let Some(rc) =
6112 crate::extensions::plugin_host::dispatch_compfn(&name_owned, &body_args)
6113 {
6114 return rc;
6115 }
6116 if let Some(f) = direct_rust_fn {
6117 return crate::compsys::router::run_port(&name_owned, &body_args, f);
6118 }
6119 // The function body is already this closure's own copy (taken at
6120 // dispatch, above), and fusevm's VM owns the chunk it runs. Move
6121 // it into the VM and take it back afterwards instead of copying
6122 // it again: compinit alone runs compdef ~15,000 times, and the
6123 // second deep copy of every body was the largest single cost in
6124 // its profile. `VM::run` never writes `vm.chunk`, so what comes
6125 // back is the same body. Taking it back also keeps a repeated
6126 // invocation working — C's runshfunc runs the body again when a
6127 // wrapper handler returns `cont` after running it
6128 // (Src/exec.c runshfunc: `while (wrap) { cont = handler(...);
6129 // if (!cont) return; ... } execode(prog, ...)`).
6130 let chunk = chunk_opt
6131 .take()
6132 .expect("chunk_opt must be Some when direct_rust_fn is None");
6133 crate::fusevm_disasm::maybe_print_stdout(
6134 &format!(
6135 "function:{}",
6136 body_args.first().map(|s| s.as_str()).unwrap_or("")
6137 ),
6138 &chunk,
6139 );
6140 let mut vm = crate::vm_pool::acquire(chunk);
6141 vm.last_status = seed_status;
6142 let _ = vm.run();
6143 chunk_opt = Some(std::mem::take(&mut vm.chunk));
6144 vm.last_status
6145 };
6146
6147 // Enter executor context BEFORE doshfunc so the body_runner's
6148 // VM builtins can `with_executor(...)` to reach this state.
6149 // c:Src/exec.c:5572-5585 — execshfunc swaps in a FRESH, EMPTY cmdstack
6150 // for the duration of a shell-function call and restores the caller's
6151 // afterwards:
6152 // ocs = cmdstack; ocsp = cmdsp;
6153 // cmdstack = zalloc(CMDSTACKSZ); cmdsp = 0;
6154 // doshfunc(shf, args, 0);
6155 // free(cmdstack); cmdstack = ocs; cmdsp = ocsp;
6156 // The cmdstack is what `%_` renders, so without the swap a function
6157 // body inherits the CALLER's parser context: `f(){ print -rP "[%_]" }`
6158 // printed `[cursh]` inside `{ f }`, `[then]` inside an `if`, `[for]`
6159 // inside a loop and `[case]` inside a case arm, where zsh prints `[]`
6160 // in every one. Most visible under xtrace, whose default PS4 ends in
6161 // `%_`, so every traced line inside a called function carried a stale
6162 // field. `( f )` was already correct only because the subshell forks.
6163 // Bug #1059.
6164 let saved_cmdstack: Vec<u8> =
6165 crate::ported::prompt::CMDSTACK.with(|s| std::mem::take(&mut *s.borrow_mut()));
6166 // c:Src/exec.c:4364 — a `return` out of a redirected compound
6167 // command still runs `fixfds(save)`. See
6168 // `unwind_redirect_scopes_to`.
6169 let redir_depth = self.redirect_scope_stack.len();
6170 // c:Src/exec.c:5632-5638 — `if ((osfc = sfcontext) == SFC_NONE)
6171 // sfcontext = SFC_DIRECT; … doshfunc(shf, args, 0); sfcontext = osfc;`.
6172 // Read by the job-text gates (c:2060, c:3537): a command run inside a
6173 // function is not given the source text `time` reports as %J.
6174 let osfc = crate::ported::exec::sfcontext.load(std::sync::atomic::Ordering::Relaxed);
6175 if osfc == crate::ported::zsh_h::SFC_NONE {
6176 crate::ported::exec::sfcontext.store(
6177 crate::ported::zsh_h::SFC_DIRECT,
6178 std::sync::atomic::Ordering::Relaxed,
6179 ); // c:5633
6180 }
6181 let _ctx = ExecutorContext::enter(self);
6182 let status = crate::ported::exec::doshfunc(&mut synth_shf, doshargs, false, body_runner);
6183 drop(_ctx);
6184 crate::ported::exec::sfcontext.store(osfc, std::sync::atomic::Ordering::Relaxed); // c:5638
6185 self.unwind_redirect_scopes_to(redir_depth);
6186 crate::ported::prompt::CMDSTACK.with(|s| *s.borrow_mut() = saved_cmdstack);
6187
6188 self.prompt_funcstack.pop();
6189 self.local_scope_depth -= 1;
6190
6191 // Honor explicit `return N` from inside the function body.
6192 if let Some(ret) = self.returning.take() {
6193 self.set_last_status(ret);
6194 Some(ret)
6195 } else {
6196 self.set_last_status(status);
6197 Some(status)
6198 }
6199 }
6200
6201 pub(crate) fn execute_external(
6202 &mut self,
6203 cmd: &str,
6204 args: &[String],
6205 redirects: &[Redirect],
6206 ) -> Result<i32, String> {
6207 // FORK_EVENTS is bumped at the real spawn site inside
6208 // execute_external_bg — this entry is only ONE of several
6209 // callers of that spawn (the common static-head command path
6210 // calls execute_external_bg directly), so counting here would
6211 // miss `time sleep 0` while double-counting this path.
6212 self.execute_external_bg(cmd, args, redirects, false, 0)
6213 }
6214
6215 /// `command -p cmd …` — `execute(args, cflags, defpath=1)`.
6216 ///
6217 /// c:Src/exec.c:3212-3214 — the `command` precommand modifier's `-p`
6218 /// sets `use_defpath = 1`, and c:4369 `execute(args, cflags,
6219 /// use_defpath)` carries it into the exec. C never touches `$path` for
6220 /// this: the default-path search happens inside `execute()` alone
6221 /// (c:810-828), so `$PATH`, the child's environment and `cmdnamtab` all
6222 /// stay exactly as they were.
6223 pub(crate) fn execute_external_defpath(
6224 &mut self,
6225 cmd: &str,
6226 args: &[String],
6227 redirects: &[Redirect],
6228 ) -> Result<i32, String> {
6229 self.execute_external_bg(cmd, args, redirects, false, 1)
6230 }
6231
6232 /// Port of `execute()` from `Src/exec.c:729` — C decl
6233 /// `execute(LinkList args, int flags, int defpath)`. `defpath` is the
6234 /// third C parameter and is non-zero only for `command -p`.
6235 fn execute_external_bg(
6236 &mut self,
6237 cmd: &str,
6238 args: &[String],
6239 _redirects: &[Redirect],
6240 background: bool,
6241 defpath: i32, // c:729
6242 ) -> Result<i32, String> {
6243 tracing::trace!(cmd, bg = background, "exec external");
6244 // c:Src/exec.c:3545-3547 — `setunderscore((args && nonempty(args)) ?
6245 // ((char *) getdata(lastnode(args))) : "")`. execcmd_exec sets `$_`
6246 // to the last word of the command it is about to run, in the PARENT,
6247 // before any builtin/plugin resolution or fork — so `cat /dev/null;
6248 // print $_` reports `/dev/null` and a pipeline's stages each leave
6249 // their own last word behind. This is the single funnel every
6250 // external spawn reaches (the static-head command path calls it
6251 // directly and bypasses ZshrsHost::exec / host_exec_external), so
6252 // the write belongs here. C's `args` list carries argv[0], hence the
6253 // fallback to `cmd` for a bare command.
6254 {
6255 let last = args.last().cloned().unwrap_or_else(|| cmd.to_string());
6256 crate::ported::params::set_zunderscore(std::slice::from_ref(&last));
6257 // c:3546
6258 }
6259 // !!! WARNING: RUST-ONLY — NO C COUNTERPART !!!
6260 // Native (Rust) plugin builtins registered via `zmodload -R`
6261 // (src/extensions/plugin_host.rs). fusevm compiles unknown
6262 // names into external execution, so a plugin command arrives
6263 // here as an "external". Resolve it BEFORE the PATH-unset guard
6264 // and the process spawn — plugin builtins are in-process and
6265 // need no PATH. This is the analog of C's `resolvebuiltin`
6266 // slot (Src/exec.c:2700), which likewise runs before the fork.
6267 // Bare names only: a `/`-qualified token is always a filesystem
6268 // path, never a plugin command name. Runs synchronously even
6269 // when backgrounded — zshrs is non-forking and an in-process
6270 // builtin has nothing to background.
6271 if !cmd.contains('/') {
6272 if let Some(status) = crate::plugin_host::dispatch(cmd, args) {
6273 return Ok(status);
6274 }
6275 }
6276 // !!! WARNING: RUST-ONLY — NO C COUNTERPART !!!
6277 // Host-registered native commands (`extensions/native_cmds.rs`) — the
6278 // sibling runtimes a fat binary links into the shell's address space:
6279 // `git` (zvcs), `arb` (arblang) and `stryke` (strykelang) in the
6280 // zshrs-native build. Same slot and same reason as the plugin-builtin
6281 // dispatch directly above: the compiler has never heard of these names,
6282 // so it lowered them to external execution and they arrive here — this
6283 // is where they must be caught, BEFORE the PATH guard and before the
6284 // spawn, because an in-process builtin needs no PATH and no process.
6285 //
6286 // Two escape hatches to the binary on disk stay open, and both are
6287 // checked here. A `/`-qualified token (`/usr/bin/git`) is a filesystem
6288 // path the user named, never a registry key. And `command git`
6289 // explicitly asks past the in-process one — the `command` handler
6290 // raises `native_cmds::force_external` around this call, exactly as
6291 // `command cat` already escapes the coreutils shadow.
6292 //
6293 // The registry's contract is full argv (argv[0] = the name as
6294 // invoked), which zvcs reads for its `git-<verb>` dashed form.
6295 //
6296 // Empty in the thin shell: one map lookup that always misses.
6297 if !cmd.contains('/')
6298 && !crate::native_cmds::is_forced_external()
6299 && crate::native_cmds::is_enabled(cmd)
6300 {
6301 let full: Vec<String> = std::iter::once(cmd.to_string())
6302 .chain(args.iter().cloned())
6303 .collect();
6304 if let Some(status) = crate::native_cmds::dispatch(cmd, &full) {
6305 return Ok(status);
6306 }
6307 }
6308 // c:Src/exec.c:824-876 — when arg0 has no `/`, C zsh requires
6309 // a PATH search. With PATH unset, the search yields no hit
6310 // and C emits `command not found: <cmd>`. Rust's
6311 // `Command::new(name)` delegates to libc `execvp`, which on
6312 // many platforms falls back to a built-in default PATH when
6313 // the env entry is missing — so `unset PATH; ls` still finds
6314 // `/bin/ls` and runs it, breaking the security boundary the
6315 // unset is supposed to establish (#416). Gate explicitly:
6316 // when cmd is a bare name (no `/`) and zshrs's own PATH
6317 // param is unset OR empty, emit the canonical
6318 // "command not found" diagnostic and return 127 BEFORE
6319 // touching libc.
6320 //
6321 // c:810-815 — `if (defpath) { … search_defpath(arg0, …) … }` runs
6322 // INSTEAD OF the `$path` walk, so `command -p` must not be judged by
6323 // `$PATH` at all: the oracle runs `PATH=; command -p awk …` fine.
6324 if !cmd.contains('/') && defpath == 0 {
6325 let path_set_and_nonempty = crate::ported::params::getsparam("PATH")
6326 .map(|p| !p.is_empty())
6327 .unwrap_or(false);
6328 if !path_set_and_nonempty {
6329 let sn =
6330 crate::ported::utils::scriptname_get().unwrap_or_else(|| "zshrs".to_string());
6331 // c:Src/exec.c:811 `zerr("command not found: %s", arg0)`
6332 // — the diagnostic carries the CURRENT line, not a
6333 // hardcoded 1. `lineno()` is the same counter zwarning
6334 // (utils.rs:179) uses and is live during VM execution
6335 // (verified: read-only / div-by-zero errors already
6336 // report the right line). Emitted directly (not via
6337 // zerr) to avoid setting errflag — command-not-found is
6338 // non-fatal and the script must continue.
6339 // Inline Rust FFI export: needs no PATH, so run it here rather
6340 // than reporting not-found when PATH is unset/empty.
6341 if let Some(rc) = self.try_registered_ffi_command(cmd, args) {
6342 return Ok(rc);
6343 }
6344 eprintln!("{}: command not found: {}", zerr_prefix(&sn), crate::ported::utils::nicedupstring(&cmd));
6345 return Ok(127);
6346 }
6347 }
6348 // c:Src/exec.c:2700-2724 resolvebuiltin — names registered via
6349 // `zmodload -ab MOD NAME` resolve through builtintab BEFORE
6350 // PATH search in C (execcmd's builtin lookup precedes the
6351 // external fork). Names the compiler didn't know as builtins
6352 // land here; consult the autoload ledger, load the module,
6353 // and re-dispatch through the builtin chokepoint. Without
6354 // this, `zmodload -ab zsh/bogus mybltn; mybltn` skipped the
6355 // C autoload-fire entirely (PATH miss → 127 instead of the
6356 // load_module diagnostic → 1).
6357 if !cmd.contains('/') {
6358 if let Some(rc) = crate::ported::exec::resolvebuiltin(cmd) {
6359 if rc != 0 {
6360 return Ok(1);
6361 }
6362 return Ok(crate::fusevm_bridge::dispatch_builtin_raw(
6363 cmd,
6364 args.to_vec(),
6365 ));
6366 }
6367 }
6368 // c:Src/exec.c:3655-3675 — execcmd_exec's external-command
6369 // resolution, the step that populates `cmdnamtab`:
6370 //
6371 // hn = cmdnamtab->getnode(cmdnamtab, cmdarg); c:3661
6372 // if (!hn && dohashcmd && strcmp(cmdarg, "..")) { c:3671
6373 // for (s = cmdarg; *s && *s != '/'; s++); c:3672
6374 // if (!*s) hn = (HashNode) hashcmd(cmdarg, ...); c:3674
6375 // }
6376 //
6377 // and c:831-869 in `execute()`, which turns the resulting node into
6378 // the pathname it hands `zexecve`: a HASHED entry execs `cn->u.cmd`
6379 // verbatim, an unhashed one execs `"<*cn->u.name>/<arg0>"`.
6380 //
6381 // Neither step ran here. This is the funnel every statically-spelled
6382 // external command reaches (`awk …` compiles to a fusevm exec op →
6383 // host_exec_external → here), and it handed the bare word straight to
6384 // `Command::new`, i.e. to libc `execvp`, which does its own PATH walk
6385 // and knows nothing about the shell's table. Two consequences:
6386 // `cmdnamtab` stayed empty for the whole life of the shell, so `hash`
6387 // listed nothing after running a command and everything reading the
6388 // table (`hash`, `unhash`, `whence`, `$commands`, completion) saw an
6389 // empty one; and an explicit `hash awk=/bin/echo` was ignored at exec
6390 // time because libc re-derived the path from `$PATH`. Only the
6391 // run-time-resolved head (`c=awk; $c …`) went through execcmd_exec
6392 // and hashed correctly, which is why the gap was invisible from the
6393 // ported side.
6394 //
6395 // `hashed_prog` also records that the pathname came from the table,
6396 // so the ENOENT arms below can fall back to the bare name the way
6397 // c:877 `execute_skip_exec:` falls through to a full `$path` walk
6398 // when the cmdnamtab candidate fails to exec.
6399 let mut hashed_prog: Option<String> = None;
6400 // c:3672-3673 — bare names only; a `/` in the word means the user
6401 // named a file and C never consults the table for it. c:3671 —
6402 // `..` is excluded by name.
6403 if !cmd.contains('/') && cmd != ".." {
6404 // c:3661 — `cmdnamtab->getnode(cmdnamtab, cmdarg)`.
6405 let mut have = crate::ported::hashtable::cmdnamtab_lock()
6406 .read()
6407 .ok()
6408 .and_then(|t| t.get(cmd).cloned());
6409 // c:3659 + c:3671 — `dohashcmd = isset(HASHCMDS)`.
6410 if have.is_none() && crate::ported::zsh_h::isset(crate::ported::zsh_h::HASHCMDS) {
6411 // c:3674 — `hashcmd(cmdarg, checkpath)`. Passing the whole
6412 // `$path` matches the sibling call in execcmd_exec
6413 // (exec.rs:11106) and hashcmd's own indexing of `pathchecked`.
6414 let dirs: Vec<String> = crate::ported::params::getsparam("PATH")
6415 .unwrap_or_default()
6416 .split(':')
6417 .map(String::from)
6418 .collect();
6419 have = crate::ported::exec::hashcmd(cmd, &dirs);
6420 }
6421 if have.is_some() {
6422 crate::extensions::emulation_output::bash_hash_hit(cmd);
6423 }
6424 if have.is_some() {
6425 // c:834 / c:861 — `cn->u.cmd` for a HASHED node, else
6426 // `"<*cn->u.name>/<arg0>"`. `get_full_path` is that arm.
6427 hashed_prog = crate::ported::hashtable::cmdnamtab_lock()
6428 .read()
6429 .ok()
6430 .and_then(|t| t.get_full_path(cmd))
6431 .map(|p| p.display().to_string());
6432 }
6433 }
6434 // c:Src/exec.c:3650-3652 + c:3678-3685 — AUTOCD. With no command
6435 // found, a one-word command with no redirections, in a shell
6436 // reading commands from its input (SHINSTDIN), that names a
6437 // directory — or, under CDABLEVARS, a parameter or user whose value
6438 // or home is one — runs as `cd -- DIR`:
6439 // trycd = isset(AUTOCD) && isset(SHINSTDIN) && (!redir ||
6440 // empty(redir)) && args && !empty(args) &&
6441 // !nextnode(firstnode(args)) && *peekfirst(args);
6442 // if (!hn && trycd && (s = cancd(peekfirst(args)))) { … "cd" … }
6443 // The ported execcmd_exec carries this; the compiled path every
6444 // literal command takes reaches only here, so `tmp` from `/`,
6445 // `/tmp` or a CDABLEVARS name reported "command not found".
6446 if hashed_prog.is_none()
6447 && defpath == 0
6448 && args.is_empty()
6449 && !cmd.is_empty()
6450 && crate::ported::zsh_h::isset(crate::ported::zsh_h::AUTOCD)
6451 && crate::ported::zsh_h::isset(crate::ported::zsh_h::SHINSTDIN)
6452 && !crate::fusevm_bridge::own_redirect_scope(self)
6453 {
6454 if let Some(dir) = crate::ported::exec::cancd(cmd) {
6455 return Ok(crate::fusevm_bridge::dispatch_builtin_raw(
6456 "cd",
6457 vec!["--".to_string(), dir],
6458 ));
6459 }
6460 }
6461 // c:Src/exec.c:810-828 — `/* for command -p, search the default path */
6462 // if (defpath) {
6463 // char pbuf[MAXCMDLEN];
6464 // if (!search_defpath(arg0, pbuf, MAXCMDLEN)) {
6465 // if (commandnotfound(arg0, args) == 0) _realexit();
6466 // zerr("command not found: %s", arg0);
6467 // _exit(127);
6468 // }
6469 // ee = zexecve(pbuf, argv, newenvp);
6470 // } else { … cmdnamtab / $path walk … }`
6471 //
6472 // The two arms are mutually exclusive in C, so under `-p` the exec'd
6473 // pathname is the DEFAULT_PATH hit and nothing else — no `$path` walk,
6474 // no bare-name retry, and a miss is final even when the name sits on
6475 // `$PATH`. The `cmdnamtab` population above still runs because it is
6476 // NOT part of `execute()`: it belongs to `execcmd_exec` in the parent
6477 // (c:3671-3674), which hashes `cmdarg` against `$path` regardless of
6478 // `use_defpath`. That asymmetry is visible in the oracle — after
6479 // `PATH=<dir-with-a-fake-awk>:…; command -p awk`, the REAL
6480 // `/usr/bin/awk` runs while the table records the fake one.
6481 //
6482 // c:798-808 runs before this and execs a `/`-bearing arg0 directly, so
6483 // `defpath` never applies to a name the user spelled as a path.
6484 let mut defpath_prog: Option<String> = None;
6485 if defpath != 0 && !cmd.contains('/') {
6486 // c:815 — `if (!search_defpath(arg0, pbuf, MAXCMDLEN))`
6487 match crate::ported::exec::search_defpath(cmd, libc::PATH_MAX as usize) {
6488 // c:822 — `ee = zexecve(pbuf, argv, newenvp)`
6489 Some(pbuf) => defpath_prog = Some(pbuf),
6490 None => {
6491 // c:816-817 — `if (commandnotfound(arg0, args) == 0)
6492 // _realexit();`. The `command_not_found_handler` hook
6493 // runs for a default-path miss exactly as it does for a
6494 // `$path` miss, and its status is the command's status.
6495 // The `$path` miss reaches the same hook from the ENOENT
6496 // arm further down, after the spawn has failed; there is
6497 // no spawn to fail here, so the call is explicit.
6498 let mut hook_args = Vec::with_capacity(args.len() + 1);
6499 hook_args.push(cmd.to_string());
6500 hook_args.extend_from_slice(args);
6501 if let Some(rc) =
6502 self.dispatch_function_call("command_not_found_handler", &hook_args)
6503 {
6504 return Ok(rc);
6505 }
6506 let sn = crate::ported::utils::scriptname_get()
6507 .unwrap_or_else(|| "zshrs".to_string());
6508 // c:818 — `zerr("command not found: %s", arg0)`. Emitted
6509 // directly rather than through `zerr` for the same reason
6510 // as the PATH-unset arm above: command-not-found is
6511 // non-fatal and must not raise errflag.
6512 eprintln!("{}: command not found: {}", zerr_prefix(&sn), crate::ported::utils::nicedupstring(&cmd));
6513 return Ok(127); // c:819 `_exit(127)`
6514 }
6515 }
6516 }
6517 // c:Src/exec.c:758-776 — "If ARGV0 is in the commands environment, we
6518 // use that as argv[0] for this external command" and unsetenv it; else
6519 // "if the pre-command `-' was given, we add `-' to the front of
6520 // argv[0] for this command."
6521 let carrier = crate::fusevm_bridge::take_exec_carrier();
6522 let job_text = crate::fusevm_bridge::take_job_text();
6523 let exec_dash = carrier & crate::fusevm_bridge::EXEC_CARRIER_DASH != 0;
6524 // c:Src/exec.c:4369 — this process was forked for the command
6525 // (BUILTIN_EXEC_FORKED_SIMPLE): exec it here rather than spawn.
6526 //
6527 // !!! WARNING: RUST-ONLY !!! C's multio helpers are forked tee/cat
6528 // processes (closemn, c:Src/exec.c:2288), so they outlive the execve
6529 // that replaces the shell. zshrs runs them as threads of the shell
6530 // (multios_scope_stack), and an execve kills every thread: the
6531 // concatenator of `cat <o1 <<<x` died before writing (cat read EOF)
6532 // and the splitter of `/bin/echo hi >a >b` left the command writing
6533 // into a pipe nobody reads (SIGPIPE, both files empty). While any
6534 // such thread is live the command is spawned and waited for instead,
6535 // which lets the threads finish before the shell exits. The command
6536 // still takes the shell's place for SHLVL, as it does in C.
6537 let exec_forked = !background && carrier & crate::fusevm_bridge::EXEC_CARRIER_FORKED != 0;
6538 let multio_threads_live = self.multios_scope_stack.iter().any(|scope| !scope.is_empty());
6539 let in_place = exec_forked && !multio_threads_live;
6540 if exec_forked && carrier & crate::fusevm_bridge::EXEC_CARRIER_SHLVL != 0 {
6541 crate::fusevm_bridge::exec_shlvl_decrement(); // c:Src/exec.c:4334-4336
6542 }
6543 let argv0_env = std::env::var("ARGV0").ok(); // c:760 zgetenv("ARGV0")
6544 let mut argv: Vec<String> = Vec::with_capacity(args.len() + 1);
6545 argv.push(match &argv0_env {
6546 Some(z) => z.clone(), // c:761
6547 None if exec_dash => format!("-{}", cmd), // c:775-776
6548 None => cmd.to_string(),
6549 });
6550 argv.extend_from_slice(args);
6551
6552 // c:Src/jobs.c — `time` reports only on JOBS (forked work). This
6553 // is the single chokepoint where an external process is actually
6554 // spawned (both fg and bg, all callers), AFTER the
6555 // command-not-found and resolvebuiltin early-returns above — so
6556 // counting here makes BUILTIN_TIME_SUBLIST report `time sleep 0`
6557 // / `time /usr/bin/true` (external → fork) while staying silent
6558 // for `time true` (builtin, never reaches this point). The
6559 // subshell entry counts separately (fusevm_bridge.rs:9573).
6560 FORK_EVENTS.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
6561 // c:Src/exec.c:4352 `closem(FDT_INTERNAL, 0)` before execve.
6562 crate::lowfd::cloexec_internal_fds();
6563
6564 // Redirect handling lives in fusevm's WithRedirectsBegin/End
6565 // ops at compile time; `_redirects` arrives empty here.
6566 let spawned = if background {
6567 execute_spawn(cmd, &argv, defpath_prog.as_deref(), hashed_prog.as_deref(), argv0_env.is_some(), false)
6568 .map_err(|eno| (eno, None))
6569 } else {
6570 // Queue signals across the wait so zshrs's SIGCHLD reaper
6571 // (waitpid(-1) in wait_for_processes, delivered on any thread)
6572 // can't reap this child before the wait does, and claim the
6573 // status back when it wins anyway — see foreground_status.
6574 let _wait_guard = crate::fusevm_bridge::ForegroundWaitGuard::enter();
6575 match execute_spawn(cmd, &argv, defpath_prog.as_deref(), hashed_prog.as_deref(), argv0_env.is_some(), in_place) {
6576 Ok(pid) => match crate::fusevm_bridge::wait_pid_status(pid) {
6577 Ok(status) => {
6578 use std::os::unix::process::ExitStatusExt as _;
6579 drop(_wait_guard);
6580 // c:Src/jobs.c:645-650 — update_job reports the finished job.
6581 crate::exec_jobs::foreground_job_report(
6582 status.into_raw(),
6583 job_text,
6584 !self.subshell_snapshots.is_empty(),
6585 );
6586 // c:Src/jobs.c:654-679 — update_job's foreground tail.
6587 crate::exec_jobs::foreground_job_signalled(status.into_raw());
6588 return Ok(crate::exec_jobs::wait_status_val(status));
6589 }
6590 Err(e) => Err((0, Some(e))),
6591 },
6592 Err(eno) => Err((eno, None)),
6593 }
6594 };
6595 let sn = crate::ported::utils::scriptname_get().unwrap_or_else(|| "zshrs".to_string());
6596 let eno = match spawned {
6597 Ok(pid) => {
6598 let cmd_str = format!("{} {}", cmd, args.join(" "));
6599 let job_id = self.jobs.add_pid_job(pid, cmd_str, JobState::Running);
6600 println!("[{}] {}", job_id, pid);
6601 return Ok(0);
6602 }
6603 Err((_, Some(e))) => return Err(format!("{}: {}: {}", sn, cmd, e)),
6604 Err((eno, None)) => eno,
6605 };
6606 // c:Src/exec.c:871-881 — `if (eno) zerr("%e: %s", eno, arg0); else
6607 // if (commandnotfound(arg0, args) == 0) _realexit(); else
6608 // zerr("command not found: %s", arg0); _exit((eno == EACCES || eno
6609 // == ENOEXEC) ? 126 : 127);`. Emitted directly rather than through
6610 // `zerr`: C prints from the doomed child, so the parent's errflag is
6611 // never raised and the script continues.
6612 if eno != 0 {
6613 eprintln!(
6614 "{}: {}: {}",
6615 zerr_prefix(&sn),
6616 crate::ported::utils::zsh_errno_msg(eno),
6617 cmd
6618 );
6619 return Ok(if eno == libc::EACCES || eno == libc::ENOEXEC { 126 } else { 127 });
6620 }
6621 // c:873 — the `command_not_found_handler` user hook; its status is
6622 // the command's status. Documented in zshmisc(1) under "Special
6623 // Functions". Bug #426.
6624 if !background {
6625 let mut hook_args = Vec::with_capacity(args.len() + 1);
6626 hook_args.push(cmd.to_string());
6627 hook_args.extend_from_slice(args);
6628 if let Some(rc) = self.dispatch_function_call("command_not_found_handler", &hook_args) {
6629 return Ok(rc);
6630 }
6631 }
6632 // Inline Rust FFI export: consulted after builtins, functions, PATH
6633 // search, and command_not_found_handler have all missed — real
6634 // commands keep priority. In the background an in-process FFI call
6635 // has nothing to background, so it runs synchronously.
6636 if let Some(rc) = self.try_registered_ffi_command(cmd, args) {
6637 return Ok(rc);
6638 }
6639 eprintln!("{}: command not found: {}", zerr_prefix(&sn), crate::ported::utils::nicedupstring(&cmd));
6640 Ok(127)
6641 }
6642 /// !!! WARNING: RUST-ONLY — NO C COUNTERPART !!!
6643 /// Inline Rust FFI fallback: when `cmd` names a function exported by a
6644 /// `rust { ... }` block (registered by the `__rust_compile` builtin) run it
6645 /// as a command. Consulted only when `cmd` resolved to nothing else — not a
6646 /// builtin, function, plugin, external on `$PATH`, or
6647 /// `command_not_found_handler` — so real commands keep priority. Positional
6648 /// args are marshalled as strings; fusevm coerces each to the export's
6649 /// signature (`i64` / `f64` / `*const c_char`). The return value is printed
6650 /// to stdout (the redirect-aware process fd 1) and the command exits 0.
6651 /// Bare names only — a `/`-qualified token is a filesystem path, never an
6652 /// FFI export. Returns `None` when `cmd` is not a registered export, so the
6653 /// caller emits its normal "command not found".
6654 fn try_registered_ffi_command(&self, cmd: &str, args: &[String]) -> Option<i32> {
6655 if cmd.contains('/') || !fusevm::ffi::is_registered(cmd) {
6656 return None;
6657 }
6658 let vals: Vec<fusevm::Value> = args.iter().map(|a| fusevm::Value::str(a.clone())).collect();
6659 match fusevm::ffi::try_call(cmd, &vals) {
6660 Some(Ok(v)) => {
6661 use std::io::Write as _;
6662 let mut out = io::stdout().lock();
6663 let _ = writeln!(out, "{}", v.to_str());
6664 let _ = out.flush();
6665 Some(0)
6666 }
6667 Some(Err(e)) => {
6668 eprintln!("zshrs: {e}");
6669 Some(1)
6670 }
6671 // Registered a moment ago but the entry vanished (registry race) —
6672 // treat as unresolved and let the caller report command-not-found.
6673 None => None,
6674 }
6675 }
6676
6677 /// Parse `cmd_str` via parse_init+parse and pull out the first Simple
6678 /// command's words, untokenized + variable-expanded, ready to spawn
6679 /// as argv. Used by process-substitution where we need raw argv to
6680 /// hand to `Command::new`. Returns empty vec if the cmd isn't a
6681 /// simple shape — pipelines / compound forms aren't process-sub
6682 /// friendly anyway.
6683 fn simple_cmd_words(&mut self, cmd_str: &str) -> Vec<String> {
6684 // Mirror Src/init.c-style errflag save/clear/check around the
6685 // parse. Process-sub argv extraction silently bails on syntax
6686 // errors (matches zsh's behavior when the inner command can't
6687 // be parsed).
6688 let saved_errflag = errflag.load(Ordering::Relaxed);
6689 errflag.fetch_and(!ERRFLAG_ERROR, Ordering::Relaxed);
6690 // Context-isolated nested parse (c:Src/exec.c:283 parse_string) —
6691 // same rationale as run_command_substitution: process-sub argv
6692 // extraction runs during execution and must not clobber the outer
6693 // single-event reader's lexer/input position.
6694 let prog = parse_isolated(cmd_str);
6695 let parse_failed = (errflag.load(Ordering::Relaxed) & ERRFLAG_ERROR) != 0;
6696 errflag.store(saved_errflag, Ordering::Relaxed);
6697 if parse_failed {
6698 return Vec::new();
6699 }
6700 let first = match prog.lists.first() {
6701 Some(l) => l,
6702 None => return Vec::new(),
6703 };
6704 let pipe = &first.sublist.pipe;
6705 if let crate::parse::ZshCommand::Simple(simple) = &pipe.cmd {
6706 simple
6707 .words
6708 .iter()
6709 .map(|w| {
6710 // Untokenize then variable-expand — text-based
6711 // word expansion for the spawned argv.
6712 let untoked = crate::lex::untokenize(w);
6713 singsub(&untoked)
6714 })
6715 .collect()
6716 } else {
6717 Vec::new()
6718 }
6719 }
6720 /// `run_command_substitution` — see implementation.
6721 /// The SQLite mirror, opened the first time anything asks for it.
6722 ///
6723 /// Returns `None` when no cache file exists yet (or it failed to open),
6724 /// which is the same answer the eager constructor produced.
6725 pub fn compsys_cache(&self) -> Option<&CompsysCache> {
6726 self.compsys_cache
6727 .get_or_init(|| {
6728 let cache_path = crate::compsys::cache::default_cache_path();
6729 if !cache_path.exists() {
6730 tracing::debug!("compsys: no cache at {}", cache_path.display());
6731 return None;
6732 }
6733 let db_size = fs::metadata(&cache_path).map(|m| m.len()).unwrap_or(0);
6734 match CompsysCache::open(&cache_path) {
6735 Ok(c) => {
6736 tracing::info!(
6737 db_bytes = db_size,
6738 path = %cache_path.display(),
6739 "compsys: sqlite mirror opened (dbview/SQL inspection only; rkyv shards are the authoritative cache)"
6740 );
6741 Some(c)
6742 }
6743 Err(e) => {
6744 tracing::warn!(error = %e, "compsys: failed to open cache");
6745 None
6746 }
6747 }
6748 })
6749 .as_ref()
6750 }
6751
6752 pub fn run_command_substitution(&mut self, cmd_str: &str) -> String {
6753 // c:Src/subst.c / Src/lex.c — the text inside `$(…)` is a FRESH
6754 // command line. The double quotes that may surround the substitution
6755 // apply to its RESULT, not to the words inside it: in `"$(f $x)"` the
6756 // `$x` is unquoted. `in_dq_context` is the runtime signal the
6757 // `${(flags)…}` bridges read for paramsubst's `qt` (c:1625), and it
6758 // stayed set for the whole body, so every flag-expansion inside a
6759 // DQ command substitution ran as if quoted.
6760 //
6761 // What that broke: `qt` suppresses RC_EXPAND_PARAM's word removal, so
6762 // under `setopt rcexpandparam` an EMPTY array kept a word instead of
6763 // deleting it (c:4327's `while ((x = *aval++))` emits nothing for an
6764 // empty array; the `!plan9` single-empty-word path at c:4261 is the
6765 // one that must NOT run):
6766 // setopt rcexpandparam
6767 // f() { declare -a x; print "n=$(set -- H ${(q)x}; print $#)" }
6768 // f # zsh: n=1, zshrs was n=2
6769 // Only the `"$(…)"` spelling was affected — unquoted `$(…)`,
6770 // backticks, and `v=$(…)` were all already correct, which is what
6771 // made it look like a quoting bug rather than an option bug.
6772 //
6773 // Bit through compsys: completion runs with rcexpandparam ON, and
6774 // `_git`'s __git_recent_commits passes `${(q)commit_opts}` to
6775 // `_call_program` inside `"$(…)"`. The stray empty word became a
6776 // bogus `''` argument to `git rev-list`, the command failed, and
6777 // `git checkout <TAB>` lost its whole recent-commits group.
6778 //
6779 // `SUBEXP_SCALAR_CTX` carries the same thing one level down — it is
6780 // what a NESTED expansion reads as `subexp_dq` (subst.rs:18873) to
6781 // learn that its OUTER `${…}` was quoted. A `$(…)` inside a quoted
6782 // outer expansion is still a fresh command line, so it has to be
6783 // cleared too:
6784 // setopt rcexpandparam
6785 // f() { declare -a co; local -a c
6786 // c=("${(f)"$(cmd HEAD ${(q)co})"}") }
6787 // is `_git`'s exact shape, and the leaked context flipped c:4354's
6788 // `mark_empty`, keeping the empty element that plan9 must delete.
6789 let saved_dq = std::mem::replace(&mut self.in_dq_context, 0);
6790 let saved_subexp = crate::ported::subst::SUBEXP_SCALAR_CTX.with(|c| c.replace(0));
6791 let out = self.run_command_substitution_inner(cmd_str, false);
6792 crate::ported::subst::SUBEXP_SCALAR_CTX.with(|c| c.set(saved_subexp));
6793 self.in_dq_context = saved_dq;
6794 out
6795 }
6796
6797 /// ksh93 funsub `${ list; }` / mksh valsub `${| list; }` — capture the
6798 /// output of `cmd_str` WITHOUT the subshell isolation `$( … )` applies.
6799 ///
6800 /// ksh(1), Command Substitution: "${ command;} … the command is
6801 /// executed in the current shell environment", so an assignment or a
6802 /// `cd` inside survives:
6803 /// `ksh -c 'x=0; y=${ x=5; print -n out; }; print "x=$x y=$y"'`
6804 /// → `x=5 y=out`, where the same body in `$( … )` leaves `x` at 0.
6805 /// mksh behaves identically for both of its forms.
6806 ///
6807 /// Same capture machinery as `$( … )` — only the parent-state
6808 /// snapshot/restore is skipped, which is exactly the difference the
6809 /// two references document.
6810 ///
6811 /// !!! RUST-ONLY ENTRY POINT — zsh has no funsub/valsub !!!
6812 pub fn run_shared_state_substitution(&mut self, cmd_str: &str) -> String {
6813 let saved_dq = std::mem::replace(&mut self.in_dq_context, 0);
6814 let saved_subexp = crate::ported::subst::SUBEXP_SCALAR_CTX.with(|c| c.replace(0));
6815 let out = self.run_command_substitution_inner(cmd_str, true);
6816 crate::ported::subst::SUBEXP_SCALAR_CTX.with(|c| c.set(saved_subexp));
6817 self.in_dq_context = saved_dq;
6818 out
6819 }
6820
6821 /// `shared_state`: skip the parent-state snapshot/restore that makes
6822 /// `$( … )` a subshell. Only the ksh/mksh funsub-valsub entry point
6823 /// passes true.
6824 fn run_command_substitution_inner(&mut self, cmd_str: &str, shared_state: bool) -> String {
6825 // `$(< FILE)` — zsh shorthand for "read FILE contents". Faster
6826 // than spawning `cat`. The leading `<` (after stripping
6827 // whitespace) means "read this file". Trailing newline is
6828 // stripped (same as command-substitution).
6829 let trimmed = cmd_str.trim_start();
6830 // Only treat as `$(<file)` shorthand when the SINGLE leading `<`
6831 // is followed by a filename, not another `<`. `$(<<<"hi" cat)`
6832 // starts with `<<<` (here-string) and must go through the full
6833 // parse path, not the read-file shortcut.
6834 if let Some(rest) = trimmed.strip_prefix('<').filter(|s| !s.starts_with('<')) {
6835 let filename = rest.trim();
6836 // c:Src/lex.c — the `$(<file)` shortcut ONLY applies when
6837 // the body is exactly `<` + ONE word. Anything else (extra
6838 // args, redirects, semicolons, pipes) is a regular command
6839 // list and must go through the full parse path so `2>/dev/null`
6840 // / `>file` / `|cmd` / `; next` etc. work. Without this
6841 // gate, `$(< file 2>/dev/null)` treated `file 2>/dev/null`
6842 // as the literal filename and errored on the missing file.
6843 // Bug #615.
6844 let is_single_word = !filename.is_empty()
6845 && !filename.chars().any(|c| {
6846 matches!(
6847 c,
6848 ' ' | '\t'
6849 | '\n'
6850 | ';'
6851 | '&'
6852 | '|'
6853 | '<'
6854 | '>'
6855 | '('
6856 | ')'
6857 | '`'
6858 | '"'
6859 | '\''
6860 )
6861 });
6862 if is_single_word {
6863 // Expand any leading $ / tilde in the filename so
6864 // `$(< $f)` and `$(< ~/x)` work.
6865 let resolved = if filename.contains('$') || filename.starts_with('~') {
6866 singsub(filename)
6867 } else {
6868 filename.to_string()
6869 };
6870 let resolved = resolved.to_string();
6871 // c:Src/exec.c `readoutput` — `$(<file)` reads the file a
6872 // BYTE at a time and metafies (`if (imeta(c)) { *ptr++ =
6873 // Meta; ... }`); it does not require UTF-8.
6874 match crate::script_bytes::read_script_file(&resolved) {
6875 Ok(contents) => {
6876 return contents.trim_end_matches('\n').to_string();
6877 }
6878 Err(e) => {
6879 // c:Src/exec.c:4797-4800 — `zwarn("%e: %s", errno, s);
6880 // lastval = cmdoutval = 1; return newlinklist();`.
6881 // zwarn carries the real `<script>:<line>:` prefix.
6882 crate::ported::utils::zwarn(&format!(
6883 "{}: {}",
6884 crate::ported::utils::zsh_errno_msg(
6885 e.raw_os_error().unwrap_or(libc::ENOENT)
6886 ),
6887 resolved
6888 ));
6889 self.set_last_status(1); // c:4799
6890 crate::ported::exec::cmdoutval.store(1, Ordering::Relaxed);
6891 return String::new();
6892 }
6893 }
6894 }
6895 // Multi-word / has-redirects → fall through to full parse.
6896 }
6897
6898 // Port of getoutput(char *cmd, int qt) from Src/exec.c. Parse and compile via
6899 // the lex+parse free ported + ZshCompiler pipeline, run on a
6900 // sub-VM with the host wired up. Stdout is captured through
6901 // an in-process pipe via dup2 — no fork. The sub-VM emits
6902 // Op::Exec for unknown command names, which forks/execs
6903 // through the host.
6904
6905 // Set up the stdout-capture pipe. We dup the original stdout
6906 // so post-run we can restore it; the write end is dup2'd onto
6907 // STDOUT_FILENO so all output the sub-VM emits (including from
6908 // forked children, which inherit fd 1) lands in the pipe.
6909 //
6910 // c:Src/exec.c:4753 — `if (mpipe(pipes) < 0)`. mpipe (c:5160)
6911 // moves BOTH pipe ends to fd >= 10 via movefd and marks them
6912 // FDT_INTERNAL. This is load-bearing: zsh's invariant is that
6913 // shell-internal fds never live below 10, so user redirections
6914 // like `exec 9>&-` (which close fd<10 unconditionally, no
6915 // FDT_INTERNAL guard — c:Src/exec.c:3856-3868) can never hit
6916 // them. A raw pipe() here landed the read end on fd 9 when
6917 // fresh-HOME init held fds 3-7, and A04redirect's %prep
6918 // `exec 9>&-` closed our own capture pipe → SIGPIPE killed
6919 // the whole shell.
6920 let (read_fd, write_fd) = {
6921 let mut fds = [0i32; 2];
6922 if crate::ported::exec::mpipe(&mut fds) < 0 {
6923 return String::new();
6924 }
6925 // The read end stays open while the body runs in process; C's
6926 // forked body closes it (c:Src/exec.c:4839), so it must not sit
6927 // where the body's `{var}` allocations land.
6928 (crate::lowfd::movefd_past_script(fds[0]), fds[1])
6929 };
6930 // c:Src/utils.c:1996 — `movefd(dup(fd))`: saved copies of the
6931 // user-visible fds are shell-internal (FDT_INTERNAL), and since
6932 // they exist only for the in-process body they sit past the
6933 // script's `{var}` range too (movefd_past_script).
6934 // A CLOSED stdout saves as -1 and is closed again on the way out
6935 // (c:Src/utils.c:2047-2048 redup `if (x < 0) zclose(y)`). C's forked
6936 // getoutput child needs no save at all (c:4840 `redup(pipes[1], 1)`),
6937 // so the body runs either way: bailing out here lost everything the
6938 // body did, `{ x=$(print b >&2) } >&-` never printed `b`.
6939 let saved_stdout =
6940 crate::lowfd::movefd_past_script(unsafe { libc::dup(libc::STDOUT_FILENO) });
6941 // Flush Rust's stdout BufWriter against the ORIGINAL fd before
6942 // dup2 swaps fd 1 to the capture pipe. Without this, bytes left
6943 // buffered by a prior `print -n` get drained to fd 1 AFTER the
6944 // dup2, which routes them into the cmd-subst's pipe — they end
6945 // up in the captured result and disappear from terminal output.
6946 //
6947 // Bug #10 in docs/BUGS.md — `print -n "A"; v=$(true); print -n
6948 // "B"; v=$(true); print -n "C"; echo` printed only `C` because
6949 // `A` and `B` were redirected into the empty cmd-subst's pipe
6950 // and discarded as its "output". C zsh's getoutput() forks, so
6951 // the child inherits the buffer COPY and the parent's buffer
6952 // stays untouched; zshrs runs cmd-subst in-process so the
6953 // parent buffer is the only one — must flush before the swap.
6954 let _ = io::stdout().flush();
6955 // c:Bug #56 — publish the saved outer stdout so a trap firing
6956 // during the nested run routes body output to the parent's
6957 // real stdout instead of the cmdsub's pipe-bound fd 1.
6958 // Same as saved_stdout above.
6959 let saved_stderr_for_trap =
6960 crate::lowfd::movefd_past_script(unsafe { libc::dup(libc::STDERR_FILENO) });
6961 // A job the body forks must not keep them: they belong to the
6962 // parent C never forked the body from (see mark_saved_fd).
6963 crate::fusevm_bridge::mark_saved_fd(saved_stdout);
6964 crate::fusevm_bridge::mark_saved_fd(saved_stderr_for_trap);
6965 crate::fusevm_bridge::CMDSUBST_OUTER_FDS
6966 .with(|s| s.borrow_mut().push((saved_stdout, saved_stderr_for_trap)));
6967 unsafe {
6968 libc::dup2(write_fd, libc::STDOUT_FILENO);
6969 }
6970 // zclose (not raw close) so the FDT_INTERNAL mark set by mpipe
6971 // is cleared from fdtable — c:Src/utils.c:2137.
6972 crate::ported::utils::zclose(write_fd);
6973
6974 // Drain the capture pipe CONCURRENTLY on a background reader
6975 // thread. The sub-VM (and any children it forks, which inherit
6976 // fd 1) writes to the pipe; reading it only AFTER vm.run()
6977 // returns deadlocks the moment the output exceeds the OS pipe
6978 // buffer (~64KB): the writer blocks on a full pipe that nothing
6979 // is draining, so vm.run() never returns. `$(alias)` over
6980 // zpwr's 2000+ aliases (~177KB) hung the whole shell a few
6981 // prompts in (thefuck's `fuck()` init runs `TF_SHELL_ALIASES=
6982 // $(alias)`). C's getoutput (Src/exec.c) forks the writer child
6983 // so the parent reads concurrently; this reader thread is the
6984 // in-process analog. It does only raw fd reads (no shell state /
6985 // thread-locals). EOF arrives once every write end closes — fd 1
6986 // restored below plus any forked child exiting.
6987 let reader_handle = crate::signal_thread::spawn(move || {
6988 let mut buf: Vec<u8> = Vec::new();
6989 let mut chunk = [0u8; 65536];
6990 loop {
6991 let n = unsafe {
6992 libc::read(
6993 read_fd,
6994 chunk.as_mut_ptr() as *mut libc::c_void,
6995 chunk.len(),
6996 )
6997 };
6998 if n < 0 {
6999 // Retry on EINTR (a signal interrupted the read);
7000 // any other error ends the drain.
7001 let e = std::io::Error::last_os_error().raw_os_error().unwrap_or(0);
7002 if e == libc::EINTR {
7003 continue;
7004 }
7005 break;
7006 }
7007 if n == 0 {
7008 break; // EOF — all write ends closed.
7009 }
7010 buf.extend_from_slice(&chunk[..n as usize]);
7011 }
7012 buf
7013 });
7014
7015 // c:Src/exec.c:1161 — forked cmdsub child runs entersubsh()
7016 // which does `zsh_subshell++`; in-process equivalent (RAII,
7017 // restored on every return path below).
7018 // A funsub/valsub is NOT a subshell — ksh(1) says the command runs
7019 // "in the current shell environment" — so it must not bump the
7020 // nesting counter `$ZSH_SUBSHELL` / `$BASH_SUBSHELL` reads.
7021 let _subshell_bump = if shared_state {
7022 None
7023 } else {
7024 Some(crate::fusevm_bridge::CmdSubstSubshellBump::enter())
7025 };
7026 // A funsub/valsub shares the shell, so only a real substitution drops -e.
7027 let _bash_errexit = (!shared_state).then(crate::dash_mode::BashCmdsubstErrexit::enter);
7028
7029 // c:Src/exec.c:1208-1209 — the same forked child clears
7030 // `opts[USEZLE]` and `zleactive`. Without it a substitution run
7031 // from inside a widget still looks "in ZLE", so `fc` refuses with
7032 // "no interactive history within ZLE" (c:Src/builtin.c:1523-1527)
7033 // and history-based completers come back empty. Placed here rather
7034 // than in exec::getoutput so the bridge's own cmdsubst paths
7035 // (BUILTIN_CMD_SUBST_TEXT, backtick) are covered too.
7036 let _subsh_state = crate::ported::exec::SubshStateGuard::enter();
7037
7038 // Parse + compile + run.
7039 // Push CS_CMDSUBST for `%_` xtrace prefix — direct port of
7040 // Src/exec.c:4783 `cmdpush(CS_CMDSUBST);` around execode().
7041 // Trace lines emitted by the inner program inherit this token
7042 // so their PS4 prefix shows "cmdsubst" matching zsh -x.
7043 cmdpush(crate::ported::zsh_h::CS_CMDSUBST as u8); // c:zsh.h:2799
7044 // Save LINENO so the inner cmdsubst's line counter doesn't
7045 // leak into the outer trace — direct port of Src/exec.c:1407
7046 // `oldlineno = lineno;` followed by `lineno = oldlineno;`
7047 // restore at line 1640. Inner program parses fresh as line 1
7048 // and increments from there; once it returns, the outer
7049 // line at the `$(…)` site must read the original outer
7050 // lineno (so xtrace renders `+:5:> echo …` not `+:1:> …`).
7051 let saved_lineno = getsparam("LINENO");
7052 // Anchor the inner program's lineno to the outer's current
7053 // $LINENO so xtrace inside the cmdsubst renders the outer
7054 // line. zsh's execlist preserves lineno across the inner
7055 // exec — for our sub-VM (fresh compile) nested_lineno_base maps
7056 // inner line N → outer_lineno + (N - 1), outer 0 included.
7057 let outer_lineno: u64 = self
7058 .scalar("LINENO")
7059 .and_then(|s| s.parse::<u64>().ok())
7060 .unwrap_or(0);
7061 // Mirror Src/init.c errflag save/clear/check pattern around
7062 // the nested parse so an inner syntax error doesn't bleed into
7063 // the outer execution.
7064 let saved_errflag = errflag.load(Ordering::Relaxed);
7065 errflag.fetch_and(!ERRFLAG_ERROR, Ordering::Relaxed);
7066 // Context-isolated nested parse (c:Src/exec.c:283 parse_string).
7067 // The outer loop()/parse_event reader may be mid-stream when this
7068 // cmd-subst executes (single-event mode), so a destructive
7069 // parse_init/lex_init would clobber its next read. parse_isolated
7070 // brackets the parse with zcontext_save/restore + inpush/inpop.
7071 let parsed = parse_isolated_at(cmd_str, Some(outer_lineno)); // c:4778 parse_string(cmd, 0)
7072 let parse_failed = (errflag.load(Ordering::Relaxed) & ERRFLAG_ERROR) != 0;
7073 errflag.store(saved_errflag, Ordering::Relaxed);
7074 let prog = if parse_failed { None } else { Some(parsed) };
7075 let mut cmd_status: Option<i32> = None;
7076 if let Some(prog) = prog {
7077 let mut compiler = crate::compile_zsh::ZshCompiler::new();
7078 compiler.nested_lineno_base = Some(outer_lineno); // c:Src/exec.c:4778
7079 let chunk = compiler.compile(&prog);
7080 if !chunk.ops.is_empty() {
7081 crate::fusevm_disasm::maybe_print_stdout("run_command_substitution", &chunk);
7082 // c:Src/exec.c:4783 — `$(...)` runs in a subshell, so
7083 // assignments / setopt / cd / trap changes inside
7084 // mustn't leak to the parent. zsh forks; we run
7085 // in-process and snapshot/restore manually. Same
7086 // snapshot shape used by host_subshell_begin/end for
7087 // the `(...)` subshell form.
7088 let paramtab_snap = crate::ported::params::paramtab()
7089 .read()
7090 .ok()
7091 .map(|t| t.clone())
7092 .unwrap_or_default();
7093 let paramtab_hashed_snap = crate::ported::params::paramtab_hashed_storage()
7094 .lock()
7095 .ok()
7096 .map(|m| m.clone())
7097 .unwrap_or_default();
7098 let pparams_snap = self.pparams();
7099 let opts_snap = crate::ported::options::opt_state_snapshot();
7100 // c:Src/exec.c:1161 — a command substitution runs in a
7101 // subshell, so IFS changes inside it must NOT leak to the
7102 // parent. IFS lives in the external `ifs_lock` global (not
7103 // paramtab), so the paramtab snapshot above doesn't cover
7104 // it: `echo $(IFS=:; set -- a b c; echo "$*")` set IFS=":"
7105 // which both produced "a:b:c" AND then word-split the
7106 // UNQUOTED result on the leaked ":" → "a b c". Snapshot the
7107 // global IFS here and restore it (with inittyptab) below.
7108 let ifs_snap = crate::ported::params::ifs_lock()
7109 .lock()
7110 .map(|g| g.clone())
7111 .unwrap_or_default();
7112 let traps_snap = crate::ported::builtin::traps_table()
7113 .lock()
7114 .map(|t| t.clone())
7115 .unwrap_or_default();
7116 // c:Src/exec.c:4783 — function definitions / unfunction
7117 // inside `$(...)` must also be isolated from the parent.
7118 // C zsh's getoutput() forks, so the child's shfunctab
7119 // mutations die with the child. zshrs's in-process
7120 // cmd-subst needs to snapshot/restore the function
7121 // tables manually alongside the param/opts/trap snaps
7122 // already in this block. Bug #455.
7123 let shfunctab_snap = crate::ported::hashtable::shfunctab_lock()
7124 .read()
7125 .ok()
7126 .map(|t| t.snapshot())
7127 .unwrap_or_default();
7128 let functions_compiled_snap = self.functions_compiled.clone();
7129 let function_source_snap = self.function_source.clone();
7130 // c:Src/exec.c:4816 — what the forked child owns a copy of:
7131 // hash tables, `environ`, cwd, umask, rlimits, …; see
7132 // SubshForkCopy.
7133 // A funsub/valsub runs in the current shell and keeps it all.
7134 let tables_snap = (!shared_state).then(crate::ported::exec::SubshForkCopy::save);
7135 let locale_snap = (!shared_state).then(subsh_locale_save).flatten();
7136 let fd_frame = (!shared_state).then(crate::ported::exec::SubshFdFrame::enter);
7137 // c:Src/exec.c:1127-1131 — the child's `entersubsh` unsets
7138 // every trap that is not function-form (`$(trap)` lists
7139 // nothing), in ITS copy of `sigtrapped[]`. Keep the
7140 // parent's for the restore, and route signals that arrive
7141 // meanwhile to it — same as `( … )`, see
7142 // fusevm_bridge::subshell_defer_signal.
7143 let sigtrapped_snap = (!shared_state).then(|| {
7144 let parent = crate::ported::signals::sigtrapped
7145 .lock()
7146 .map(|g| g.clone())
7147 .unwrap_or_default();
7148 crate::fusevm_bridge::subshell_signal_enter();
7149 crate::fusevm_bridge::entersubsh_reset_traps();
7150 parent
7151 });
7152 // c:Src/exec.c:4782 — getoutput's child runs
7153 // `entersubsh(ESUB_PGRP|ESUB_NOMONITOR)`, and c:1219
7154 // `if (flags & ESUB_PGRP) clearjobtab(monitor)` hands
7155 // that child an EMPTY job table. The oldjobtab snapshot
7156 // (c:Src/jobs.c:1800) is monitor-only, so a
7157 // non-interactive shell keeps nothing at all — which is
7158 // why zsh prints nothing for `sleep 5 & print $(jobs)`.
7159 // The `(...)` and pipeline-stage paths already call
7160 // clearjobtab in their forked children; cmd-subst runs
7161 // in-process, so snapshot the globals clearjobtab
7162 // mutates and restore them below. freejob (c:1457) is
7163 // struct-local — no waitpid/kill — so the restore is
7164 // exact.
7165 // c:Src/exec.c:4782 — same fork, one more thing it copies:
7166 // the completion-match arena (c:Src/Zle/compcore.c:124-259).
7167 // A `compadd` run inside `$(…)` lands in the CHILD's
7168 // `matches`/`amatches`/`mgroup`, which die with it, so the
7169 // completing parent never sees those matches. zshrs's
7170 // in-process cmd-subst shares the arena, so `_tmux`'s
7171 // `desc="$(_tmux-backup)"` description probe leaked five
7172 // whole completion groups into `tmux <TAB>` (551 matches vs
7173 // zsh's 450). Snapshot/restore it by hand, exactly as the
7174 // param/opts/trap/job snaps above do.
7175 let comp_arena_snap = crate::comp_match_handles::comp_arena_save();
7176 // !!! WARNING: RUST-ONLY LOCK DISCIPLINE !!! zhandler's SIGCHLD
7177 // branch locks JOBTAB (c:Src/signals.c:429-431 → update_bg_job)
7178 // on whichever thread the signal interrupts. C's jobtab is a
7179 // plain array; here a SIGCHLD landing while this thread holds
7180 // the JOBTAB mutex (the snapshot, clearjobtab, and the restore
7181 // below, which also drops the child's table under the lock)
7182 // blocks the handler forever: `echo $(echo a &) w` hung.
7183 // Queue signals across both windows (c:Src/signals.c:410-424).
7184 crate::ported::signals_h::queue_signals();
7185 let jobtab_snap = crate::ported::jobs::JOBTAB
7186 .get()
7187 .and_then(|t| t.lock().ok().map(|g| g.clone()));
7188 let maxjob_snap = crate::ported::jobs::MAXJOB
7189 .get()
7190 .and_then(|m| m.lock().ok().map(|g| *g));
7191 let thisjob_snap = crate::ported::jobs::THISJOB
7192 .get()
7193 .and_then(|t| t.lock().ok().map(|g| *g));
7194 // curjob/prevjob (c:Src/jobs.c) are plain globals in C,
7195 // so the forked child's setcurjob calls never reach the
7196 // parent — restore them alongside the table, else the
7197 // `+`/`-` markers are lost after a `$(jobs)`.
7198 let curjob_snap = crate::ported::jobs::CURJOB
7199 .get()
7200 .and_then(|t| t.lock().ok().map(|g| *g));
7201 let prevjob_snap = crate::ported::jobs::PREVJOB
7202 .get()
7203 .and_then(|t| t.lock().ok().map(|g| *g));
7204 // Pipeline job slots the body takes come from the cleared
7205 // table below; their records go when the parent's comes back.
7206 let slots_snap = crate::exec_jobs::execpline_slots_mark();
7207 {
7208 let monitor = crate::ported::zsh_h::isset(crate::ported::zsh_h::MONITOR) as i32;
7209 crate::ported::jobs::clearjobtab(&mut self.jobs, monitor);
7210 }
7211 crate::ported::signals_h::unqueue_signals();
7212 let mut vm = fusevm::VM::new(chunk);
7213 register_builtins(&mut vm);
7214 vm.set_shell_host(Box::new(ZshrsHost));
7215 // Seed inner $? with the outer's last_status so the
7216 // sub-shell inherits the parent's exit code. Direct
7217 // port of Src/exec.c:4783 around execcmd_exec — the
7218 // child inherits `lastval` at fork time, so `false;
7219 // echo $(echo $?)` reads 1, not the freshly-zeroed
7220 // sub-VM default. Without this, every cmd-subst
7221 // started with $?==0 regardless of the parent's
7222 // last command.
7223 vm.last_status = self.last_status();
7224 // `exit N` inside a cmd-subst should terminate ONLY
7225 // the sub-shell (C zsh: cmd-subst forks, the child
7226 // `_exit(N)`s; status reaches the parent as
7227 // cmd-subst exit). zshrs runs in-process, so we
7228 // route through the SUBSHELL_DEPTH-gated deferred
7229 // path inside zexit (builtin.rs:7713): bump
7230 // SUBSHELL_DEPTH so `exit` sets EXIT_PENDING/
7231 // EXIT_VAL instead of calling realexit (which would
7232 // process::exit and kill the parent shell). After
7233 // the sub-VM returns, harvest EXIT_PENDING/EXIT_VAL
7234 // as the cmd-subst's status, then restore the
7235 // parent's flags so the outer VM continues normally.
7236 use crate::ported::builtin::{
7237 BREAKS, EXIT_PENDING, EXIT_VAL, RETFLAG, SHELL_EXITING, SUBSHELL_DEPTH,
7238 };
7239 use std::sync::atomic::Ordering::Relaxed;
7240 let saved_exit_pending = EXIT_PENDING.swap(0, Relaxed);
7241 let saved_exit_val = EXIT_VAL.swap(0, Relaxed);
7242 let saved_shell_exiting = SHELL_EXITING.swap(0, Relaxed);
7243 let saved_retflag = RETFLAG.swap(0, Relaxed);
7244 let saved_breaks = BREAKS.swap(0, Relaxed);
7245 // c:Src/exec.c:4784 — `execode(prog, 0, 1, "cmdsubst");`.
7246 // execode (c:1245-1266) APPENDS its `context` argument to
7247 // `zsh_eval_context` for the duration of the body, so code
7248 // inside `$(…)` / backticks sees `cmdarg:cmdsubst` where the
7249 // top level sees just `cmdarg`. zshrs pushed "shfunc" at the
7250 // function-call site but never pushed "cmdsubst", so
7251 // `$(print $ZSH_EVAL_CONTEXT)` reported `cmdarg` and
7252 // `$(f)` reported `cmdarg:shfunc` instead of
7253 // `cmdarg:cmdsubst:shfunc`. Popped on every return path by the
7254 // guard below, mirroring execode's stack discipline.
7255 // Bug #1065.
7256 let sync_eval_ctx = |stack: &[String]| {
7257 crate::ported::exec::EvalContextFrame::sync(stack);
7258 };
7259 if let Ok(mut ctx) = crate::ported::exec::zsh_eval_context.lock() {
7260 ctx.push("cmdsubst".to_string());
7261 sync_eval_ctx(&ctx);
7262 }
7263 struct CmdsubstEvalCtxGuard<F: Fn(&[String])>(F);
7264 impl<F: Fn(&[String])> Drop for CmdsubstEvalCtxGuard<F> {
7265 fn drop(&mut self) {
7266 if let Ok(mut ctx) = crate::ported::exec::zsh_eval_context.lock() {
7267 ctx.pop();
7268 (self.0)(&ctx);
7269 }
7270 }
7271 }
7272 let _cs_eval_ctx_guard = CmdsubstEvalCtxGuard(sync_eval_ctx);
7273 SUBSHELL_DEPTH.fetch_add(1, Relaxed);
7274 let _ctx = ExecutorContext::enter(self);
7275 let _ = vm.run();
7276 let inner_exit_pending = EXIT_PENDING.load(Relaxed);
7277 let inner_exit_val = EXIT_VAL.load(Relaxed);
7278 let inner_status = if inner_exit_pending != 0 {
7279 inner_exit_val & 0xFF
7280 } else {
7281 vm.last_status
7282 };
7283 cmd_status = Some(inner_status);
7284 SUBSHELL_DEPTH.fetch_sub(1, Relaxed);
7285 // c:Src/exec.c — `$(…)` is a FORK in C: an errflag
7286 // abort inside the child ends the child (its lastval
7287 // becomes the cmd-subst status) and the flag dies
7288 // with the child process — the parent's lists keep
7289 // running. zsh 5.9: `v=$(typeset -A q; q=(odd));
7290 // echo "after $?"` prints `after 1`. Mirror the fork
7291 // isolation by clearing ERRFLAG_ERROR at the
7292 // cmd-subst boundary.
7293 //
7294 // ERRFLAG_HARD dies here too: `${u:?msg}` inside the
7295 // child sets it (c:Src/subst.c:3344) then `_exit(1)`s
7296 // (c:3353) — C's parent never sees the bit. A leaked
7297 // HARD bit makes every later zerr() silent
7298 // (c:Src/utils.c:175-177) and silently fails every
7299 // later parse. Same fix as subshell_end.
7300 // A nofork `${ … }` is no fork: c:Src/subst.c:2056 leaves the
7301 // body's errflag set, so the command around it aborts.
7302 if !shared_state {
7303 errflag.fetch_and(
7304 !(ERRFLAG_ERROR | crate::ported::zsh_h::ERRFLAG_HARD),
7305 Relaxed,
7306 );
7307 }
7308 // c:Src/exec.c:4783 execcmdoutsubst — `$(...)` is a
7309 // subshell, and zsh fires the EXIT trap when the
7310 // subshell ends BUT only if the trap was installed
7311 // INSIDE the subshell. An EXIT trap inherited from
7312 // the parent fires when the parent shell exits, not
7313 // again at cmdsub end. Detect "installed inside" by
7314 // comparing the current traps_table["EXIT"] entry
7315 // against the pre-cmdsub snapshot — fire only when
7316 // the body differs (newly set, removed, or replaced).
7317 // Pop the body before execute_script to avoid the
7318 // re-fire inside execute_script_zsh_pipeline's own
7319 // EXIT-handler tail at vm_helper.rs:1490. Bug #354.
7320 let snap_exit = traps_snap.get("EXIT").cloned();
7321 let live_exit = crate::ported::builtin::traps_table()
7322 .lock()
7323 .ok()
7324 .and_then(|t| t.get("EXIT").cloned());
7325 if !shared_state && live_exit != snap_exit {
7326 if let Some(body) = live_exit {
7327 if let Ok(mut t) = crate::ported::builtin::traps_table().lock() {
7328 t.remove("EXIT");
7329 }
7330 let _ = crate::ported::exec::execute_script(&body);
7331 }
7332 }
7333 // c:Src/signals.c::dotrap(SIGEXIT) — also fire the
7334 // TRAPEXIT() function-named form (ZSIG_FUNC) — but
7335 // only if it was defined INSIDE the subshell (the
7336 // parent's TRAPEXIT fires at parent exit, not here).
7337 // ZSIG_FUNC bit on sigtrapped[SIGEXIT] tells us
7338 // whether a TRAPEXIT function is registered; it is read
7339 // BEFORE the snapshot restore, and an inherited one (flag
7340 // already set in the parent's snapshot) is left alone.
7341 if let Some(parent) = &sigtrapped_snap {
7342 let func = crate::ported::zsh_h::ZSIG_FUNC as i32;
7343 let exit_idx = crate::signals_h::SIGEXIT as usize;
7344 let live = crate::ported::signals::sigtrapped
7345 .lock()
7346 .ok()
7347 .and_then(|g| g.get(exit_idx).copied())
7348 .unwrap_or(0);
7349 let inherited = parent.get(exit_idx).copied().unwrap_or(0);
7350 if live & func != 0 && inherited & func == 0 {
7351 crate::ported::signals::intrap.fetch_add(1, Ordering::SeqCst); // c:1123
7352 let _ = crate::ported::exec::execute_script("TRAPEXIT");
7353 crate::ported::signals::intrap.fetch_sub(1, Ordering::SeqCst); // c:1236
7354 }
7355 }
7356 // Restore parent's exit / loop / function-return
7357 // state so the outer VM continues normally.
7358 //
7359 // A nofork `${ … }` runs in the CURRENT shell, so only `return`
7360 // stops at it (c:Src/subst.c:2049-2053 `if (retflag) { retflag
7361 // = 0; breaks = obreaks; }`): an `exit` stays pending for the
7362 // caller's c:2094 zexit and a `break` keeps unwinding the loop
7363 // around the command.
7364 if shared_state {
7365 if RETFLAG.load(Relaxed) != 0 {
7366 RETFLAG.store(saved_retflag, Relaxed); // c:2051
7367 BREAKS.store(saved_breaks, Relaxed); // c:2052
7368 }
7369 } else {
7370 EXIT_PENDING.store(saved_exit_pending, Relaxed);
7371 EXIT_VAL.store(saved_exit_val, Relaxed);
7372 SHELL_EXITING.store(saved_shell_exiting, Relaxed);
7373 RETFLAG.store(saved_retflag, Relaxed);
7374 BREAKS.store(saved_breaks, Relaxed);
7375 }
7376 // Restore parent state. The inner cmd-subst's stdout
7377 // (the captured pipe contents) is the only thing
7378 // that leaks out.
7379 //
7380 // A funsub/valsub skips ALL of it: that is the entire
7381 // difference between `${ list; }` and `$(list)`.
7382 if !shared_state {
7383 // First: rolling a module back looks its parameters up
7384 // in the LIVE paramtab (see SubshForkCopy::restore).
7385 if let Some(snap) = tables_snap {
7386 snap.restore();
7387 }
7388 // Before fd 1 and 2 go back to the parent below.
7389 drop(fd_frame);
7390 if let Ok(mut t) = crate::ported::params::paramtab().write() {
7391 *t = paramtab_snap;
7392 }
7393 if let Ok(mut m) = crate::ported::params::paramtab_hashed_storage().lock() {
7394 *m = paramtab_hashed_snap;
7395 }
7396 self.set_pparams(pparams_snap);
7397 crate::ported::options::opt_state_restore(opts_snap);
7398 // Restore the parent's IFS (subshell isolation): the body's
7399 // `IFS=` must not leak out and word-split the parent's use
7400 // of the cmdsub result. Runtime word-splitting reads the
7401 // IFS *string* (this `ifs_lock` global), so restoring it is
7402 // sufficient. Deliberately do NOT call inittyptab() here —
7403 // that rewrites the process-global typtab the LEXER reads
7404 // on every character, and firing it per-cmdsub races
7405 // concurrent lexing in zshrs's worker threads, producing
7406 // spurious "parse error" flakes (HEAD ran clean 3/3; the
7407 // per-cmdsub inittyptab flaked ~50%). The typtab only
7408 // affects re-lexing — the parent is already compiled — and
7409 // leaving it at the body's value is strictly less divergent
7410 // than the prior behavior, which leaked the whole IFS.
7411 // Except when the body CHANGED IFS: its `ifssetfn` (or the
7412 // unset's c:Src/params.c:4748-4749) already rebuilt the typtab
7413 // that word splitting reads, so the parent's table has to come
7414 // back with its value (`IFS=:; print $(IFS=,; :); s="p,q";
7415 // print -rl -- ${=s}` is one word in zsh). Only that case
7416 // rebuilds, so the per-cmdsub race above does not return.
7417 let ifs_changed = crate::ported::params::ifs_lock()
7418 .lock()
7419 .map(|mut g| {
7420 let changed = *g != ifs_snap;
7421 *g = ifs_snap;
7422 changed
7423 })
7424 .unwrap_or(false);
7425 if ifs_changed {
7426 crate::ported::utils::inittyptab();
7427 }
7428 // The body's `LC_ALL=` / `LANG=` changed only the
7429 // forked child's locale in C.
7430 subsh_locale_restore(locale_snap.as_deref());
7431 if let Ok(mut t) = crate::ported::builtin::traps_table().lock() {
7432 *t = traps_snap;
7433 }
7434 // The trap flags, and the `sigaction`s the body installed
7435 // for its own traps, go back to the parent's.
7436 if let Some(parent) = &sigtrapped_snap {
7437 let child = match crate::ported::signals::sigtrapped.lock() {
7438 Ok(mut st) => std::mem::replace(&mut *st, parent.clone()),
7439 Err(_) => Vec::new(),
7440 };
7441 crate::fusevm_bridge::subshell_restore_signal_dispositions(parent, &child);
7442 }
7443 // Restore function tables (parallel to the trap/param
7444 // restore above). Bug #455.
7445 if let Ok(mut t) = crate::ported::hashtable::shfunctab_lock().write() {
7446 t.restore(shfunctab_snap);
7447 }
7448 self.functions_compiled = functions_compiled_snap;
7449 self.function_source = function_source_snap;
7450 // Discard anything the substitution added to the completion
7451 // arena — the in-process stand-in for the forked child's
7452 // address space going away (see comp_arena_save above).
7453 crate::comp_match_handles::comp_arena_restore(comp_arena_snap);
7454 // Undo the clearjobtab above — in C the cleared table
7455 // belongs to the forked child and dies with it, so the
7456 // parent's table must come back untouched. Signals stay
7457 // queued while the job mutexes are held (see the snapshot).
7458 crate::ported::signals_h::queue_signals();
7459 if let (Some(js), Some(t)) = (jobtab_snap, crate::ported::jobs::JOBTAB.get()) {
7460 if let Ok(mut g) = t.lock() {
7461 *g = js;
7462 }
7463 }
7464 if let (Some(mj), Some(m)) = (maxjob_snap, crate::ported::jobs::MAXJOB.get()) {
7465 if let Ok(mut g) = m.lock() {
7466 *g = mj;
7467 }
7468 }
7469 if let (Some(tj), Some(t)) = (thisjob_snap, crate::ported::jobs::THISJOB.get())
7470 {
7471 if let Ok(mut g) = t.lock() {
7472 *g = tj;
7473 }
7474 }
7475 if let (Some(cj), Some(t)) = (curjob_snap, crate::ported::jobs::CURJOB.get()) {
7476 if let Ok(mut g) = t.lock() {
7477 *g = cj;
7478 }
7479 }
7480 if let (Some(pj), Some(t)) = (prevjob_snap, crate::ported::jobs::PREVJOB.get())
7481 {
7482 if let Ok(mut g) = t.lock() {
7483 *g = pj;
7484 }
7485 }
7486 crate::exec_jobs::execpline_slots_forget_to(slots_snap);
7487 crate::ported::signals_h::unqueue_signals();
7488 // Now that the parent's traps are back, run the ones
7489 // whose signal arrived during the body.
7490 if sigtrapped_snap.is_some() {
7491 crate::fusevm_bridge::subshell_signal_leave();
7492 }
7493 } // if !shared_state
7494 }
7495 }
7496 // Restore LINENO so outer xtrace sees the outer line. LINENO
7497 // carries PM_READONLY (matching zsh's `integer-readonly-special`
7498 // GSU), so the restore must bypass the generic readonly guard
7499 // exactly like BUILTIN_SET_LINENO (fusevm_bridge.rs:5156) — write
7500 // the param's `u_val` directly and mirror the file-static /
7501 // lexer line counters. The previous `set_scalar` went through the
7502 // readonly-checked path: harmless on the `-c` route (LINENO not
7503 // yet flagged readonly there) but fatal on the faithful
7504 // loop()/zsh_main route, where every `$(...)` in piped/redirected
7505 // input died with `read-only variable: LINENO`.
7506 if let Some(ln) = saved_lineno {
7507 let n: crate::ported::zsh_h::zlong = ln.parse().unwrap_or(0);
7508 if let Ok(mut tab) = crate::ported::params::paramtab().write() {
7509 if let Some(pm) = tab.get_mut("LINENO") {
7510 // c:Src/utils.c:121 `zlong lineno` — the value lives in the C
7511 // GLOBAL, reached through LINENO's GSU. A `typeset -h +g LINENO`
7512 // local shadow has no PM_SPECIAL and no GSU, so C's `lineno = N`
7513 // never touches it; skip the paramtab mirror for the same reason.
7514 if (pm.node.flags & crate::ported::zsh_h::PM_SPECIAL as i32) != 0 {
7515 pm.u_val = n;
7516 pm.node.flags &= !(crate::ported::zsh_h::PM_UNSET as i32);
7517 }
7518 }
7519 }
7520 crate::ported::utils::set_lineno(n as i32);
7521 crate::ported::lex::set_lineno(n as u64);
7522 }
7523 cmdpop();
7524 // Propagate the inner cmd's status to the parent shell. zsh:
7525 // `a=$(false); echo $?` → 1 because cmd-subst status leaks to
7526 // $?. Set last_status on the executor so $? reads the right
7527 // value for callers that don't have a SetStatus(0) overwrite
7528 // (echo, test, etc.). Bare assignment paths still get the
7529 // SetStatus(0) from compile_simple — that's a separate gap.
7530 // Empty cmd-subst (`\`\``, `$()`) resets status to 0 per
7531 // Src/exec.c — the inner ran no command so the "last
7532 // command's exit" is the implicit success of "did nothing".
7533 // Without this branch, a prior command's non-zero status
7534 // leaked through the empty cmd-subst.
7535 let final_status = cmd_status.unwrap_or(0);
7536 self.set_last_status(final_status);
7537 // c:Src/exec.c:4775 — `getoutput` (the C cmd-subst path used by
7538 // both `$(…)` and `` `…` ``) propagates the inner exit through
7539 // `cmdoutval`, then the caller does `LASTVAL = cmdoutval`. Mirror
7540 // by writing the cmd-subst's exit into the ported `cmdoutval`
7541 // global so `getoutput()`'s post-call `LASTVAL = cmdoutval` (at
7542 // exec.rs:559-562) and the C-equivalent `cmdoutval = lastval`
7543 // bookkeeping in execcmd_exec's assignment paths both see the
7544 // real exit. Without this, backtick assignments (`a=\`false\`;
7545 // echo $?`) reported 0 because getoutput's caller path read a
7546 // cmdoutval that was never updated by the in-process hook.
7547 crate::ported::exec::cmdoutval.store(final_status, std::sync::atomic::Ordering::Relaxed);
7548
7549 // Flush any buffered Rust-side stdout so it reaches the pipe
7550 // before we restore.
7551 let _ = io::stdout().flush();
7552
7553 // Pop the trap-routing stack BEFORE restoring stdout so any
7554 // trap that fires during the restore goes to the cmdsub's
7555 // pipe (matching what zsh's forked cmdsub would do — the
7556 // child's fd 1 is the pipe right up until the child exits).
7557 crate::fusevm_bridge::CMDSUBST_OUTER_FDS.with(|s| {
7558 s.borrow_mut().pop();
7559 });
7560 // c:Bug #353 — restore fd 2 from the saved outer stderr. A
7561 // body that ran `exec 2>&1` (no command, just redirects)
7562 // would have committed fd 2 → the cmdsub's pipe write end.
7563 // In zsh's forked cmdsub the committed redirect dies with
7564 // the child; zshrs's in-process cmdsub would leak the dup
7565 // back to the parent and keep the pipe write-end alive,
7566 // blocking the parent's read on the read_end forever.
7567 // Always restoring fd 2 here rolls back any commit so the
7568 // pipe write-end count drops to zero when we drop the
7569 // local write_fd reference (which already happened above).
7570 if saved_stderr_for_trap >= 0 {
7571 unsafe {
7572 libc::dup2(saved_stderr_for_trap, libc::STDERR_FILENO);
7573 }
7574 crate::ported::utils::zclose(saved_stderr_for_trap);
7575 }
7576 // Restore stdout and read what was captured.
7577 if saved_stdout >= 0 {
7578 unsafe {
7579 libc::dup2(saved_stdout, libc::STDOUT_FILENO);
7580 }
7581 crate::ported::utils::zclose(saved_stdout);
7582 } else {
7583 unsafe { libc::close(libc::STDOUT_FILENO) }; // c:2048 zclose(y)
7584 }
7585 // Collect the concurrently-drained output. With fd 1 restored
7586 // above, the last shell-side write end is closed, so the reader
7587 // hits EOF and join() returns the full buffer regardless of
7588 // size — no pipe-full deadlock. The reader only read (never
7589 // closed) read_fd, so zclose still clears its FDT_INTERNAL mark
7590 // (c:Src/utils.c:2137).
7591 let bytes = reader_handle.join().unwrap_or_default();
7592 crate::ported::utils::zclose(read_fd);
7593 // This is the `$( cmd )` counterpart of `readoutput`'s c:4835-4849
7594 // conversion, and it has to be just as lossless: with
7595 // `String::from_utf8_lossy` a helper that emitted a latin-1 byte —
7596 // which is routine for the legacy completion corpus, where
7597 // `_call_program` shells out to tools whose descriptions predate
7598 // UTF-8 — had that byte replaced by U+FFFD before any completer saw
7599 // it. `decode_script_bytes` Meta-encodes only what is not valid
7600 // UTF-8, and `utils::unmetafy_str` restores the byte on output.
7601 let mut output = crate::script_bytes::decode_script_bytes(&bytes);
7602
7603 // POSIX: trailing newlines stripped from cmd-sub result.
7604 // c:Src/subst.c:1908 + c:2064-2069 — a nofork `${ … }` does its own
7605 // trim (`trim = !qt` under zsh emulation: ONE newline unquoted, NONE
7606 // quoted), so its caller gets the bytes untouched. Stripping them all
7607 // here left `"${ print INNER } $?"` as `INNER 7`, zsh `INNER\n 7`.
7608 while !shared_state && output.ends_with('\n') {
7609 output.pop();
7610 }
7611 // !!! RUST-ONLY: provenance tap. `$(…)` is a lineage ORIGIN —
7612 // these bytes did not exist in the shell before the inner list
7613 // ran. This is the in-process cmd-subst funnel (the host
7614 // `ShellHost::cmd_subst` path taps the same event for chunks
7615 // that reach the VM as sub-chunks instead of source text).
7616 if crate::provenance::active() {
7617 crate::provenance::on_cmd_subst(cmd_str, &output);
7618 }
7619 if parse_failed {
7620 // c:Src/exec.c:4778-4780 — `getoutput` returns NULL when
7621 // `parse_string` fails, and its caller raises the error:
7622 // c:Src/subst.c:400-402 `zerr("parse error in command
7623 // substitution"); return NULL;`, abandoning the command. Only a
7624 // body parsed at run time gets here (`` `…` ``); the lexer has
7625 // already parsed a `$(…)` body.
7626 crate::ported::utils::zerr("parse error in command substitution");
7627 }
7628 output
7629 }
7630}
7631
7632#[cfg(test)]
7633mod tests {
7634 use super::*;
7635
7636 #[test]
7637 fn test_simple_echo() {
7638 let _g = crate::test_util::global_state_lock();
7639 let mut exec = ShellExecutor::new();
7640 let status = exec.execute_script("true").unwrap();
7641 assert_eq!(status, 0);
7642 }
7643
7644 /// A `zsh/parameter` param is an autoload stub until it is READ; the read
7645 /// materializes it. `$parameters` enumerations type a stub as "undefined"
7646 /// (Src/Modules/parameter.c:49-50) and never resolve it, which is what
7647 /// puts those names in the right `_parameters -g` bucket.
7648 ///
7649 /// Reference behavior (`zsh -f`):
7650 /// `m=( ${(kv)parameters} ); print $m[aliases]` → undefined
7651 /// `${parameters[aliases]}` first, then the same scan → association-…
7652 #[test]
7653 fn module_params_are_autoload_stubs_until_read() {
7654 let _g = crate::test_util::global_state_lock();
7655 // `jobstates` is not touched by shell STARTUP, but
7656 // `MATERIALIZED_MODULE_PARAMS` is a process-global side-set and 10k
7657 // tests share one process — `compsys::_jobs` and friends read the job
7658 // parameters, so by the time this runs the stub may already have been
7659 // materialized. The lock serialises but restores nothing, so put the
7660 // name back to its untouched state here rather than assume nothing
7661 // else in the binary ever reads it. The assertions below are
7662 // unchanged: an untouched stub reads as a stub, and a read
7663 // materializes it.
7664 unmark_module_param_used("jobstates");
7665 assert!(
7666 module_param_is_autoload_stub("jobstates"),
7667 "untouched module param must read as a stub"
7668 );
7669 mark_module_param_used("jobstates");
7670 assert!(
7671 !module_param_is_autoload_stub("jobstates"),
7672 "a read must materialize it"
7673 );
7674 // A core special (not module-provided) is never a stub.
7675 assert!(!module_param_is_autoload_stub("path"));
7676 assert!(!module_param_is_autoload_stub("PATH"));
7677 }
7678
7679 /// Phase 3 diagnostic: a worker must be able to run a USER-DEFINED function
7680 /// (defined on the main executor) — the function source lives in the shared
7681 /// shfunctab, and the worker lazy-compiles it from there. Its `typeset -g`
7682 /// must reach the global param table. This is what `async_precmd` needs.
7683 #[test]
7684 fn phase3_worker_runs_user_defined_function() {
7685 let _g = crate::test_util::global_state_lock();
7686 let mut main = ShellExecutor::new();
7687 main.execute_script("phase3fn() { typeset -g PHASE3_FN_RESULT=fn_ran }")
7688 .unwrap();
7689 // sanity: it ran on main? (define only — not called yet)
7690 assert_eq!(getsparam("PHASE3_FN_RESULT"), None);
7691
7692 let pool = std::sync::Arc::new(crate::worker::WorkerPool::new(2));
7693 let pool2 = std::sync::Arc::clone(&pool);
7694 let (tx, rx) = std::sync::mpsc::channel::<()>();
7695 pool.submit(move || {
7696 let mut wex = ShellExecutor::new_worker(pool2);
7697 let _ = wex.execute_script_zsh_pipeline("phase3fn");
7698 let _ = tx.send(());
7699 });
7700 rx.recv().expect("worker completed");
7701 assert_eq!(
7702 getsparam("PHASE3_FN_RESULT"),
7703 Some("fn_ran".to_string()),
7704 "worker could not run the user-defined function from shared shfunctab"
7705 );
7706 }
7707
7708 /// Phase 1 of the in-process thread-execution model: prove a shell body
7709 /// runs on a POOL WORKER THREAD via `new_worker` and that its `typeset -g`
7710 /// lands in the GLOBAL (RwLock-synchronized) param table. Each worker writes
7711 /// a DISTINCT key, so a green run shows: (a) `ExecutorContext::enter` +
7712 /// `execute_script_zsh_pipeline` work off the main thread, and (b) N
7713 /// concurrent writers don't corrupt the shared table. This is the linchpin
7714 /// for converting the subprocess-forking parallel builtins to threads.
7715 #[test]
7716 fn phase1_worker_shell_writes_reach_global_paramtab() {
7717 let _g = crate::test_util::global_state_lock();
7718 // Seed the globals (options, default params) exactly as a live session
7719 // would — workers SHARE these; new_worker() never re-seeds them.
7720 let _main = ShellExecutor::new();
7721
7722 let pool = std::sync::Arc::new(crate::worker::WorkerPool::new(4));
7723 const N: usize = 16;
7724 let (tx, rx) = std::sync::mpsc::channel::<usize>();
7725 for i in 0..N {
7726 let tx = tx.clone();
7727 let pool_for_worker = std::sync::Arc::clone(&pool);
7728 pool.submit(move || {
7729 // Lightweight per-worker executor; shares the global tables.
7730 let mut wex = ShellExecutor::new_worker(pool_for_worker);
7731 let _ =
7732 wex.execute_script_zsh_pipeline(&format!("typeset -g PHASE1_WK_{i}=val_{i}"));
7733 let _ = tx.send(i);
7734 });
7735 }
7736 drop(tx);
7737 // Barrier: wait for all N workers.
7738 let mut done = 0usize;
7739 while rx.recv().is_ok() {
7740 done += 1;
7741 }
7742 assert_eq!(done, N, "all {N} workers completed");
7743
7744 // Every worker's write must be visible in the global param table.
7745 for i in 0..N {
7746 assert_eq!(
7747 getsparam(&format!("PHASE1_WK_{i}")),
7748 Some(format!("val_{i}")),
7749 "worker {i} typeset -g did not reach the global paramtab"
7750 );
7751 }
7752 }
7753
7754 #[test]
7755 fn test_if_true() {
7756 let _g = crate::test_util::global_state_lock();
7757 let mut exec = ShellExecutor::new();
7758 let status = exec.execute_script("if true; then true; fi").unwrap();
7759 assert_eq!(status, 0);
7760 }
7761
7762 #[test]
7763 fn test_if_false() {
7764 let _g = crate::test_util::global_state_lock();
7765 let mut exec = ShellExecutor::new();
7766 let status = exec
7767 .execute_script("if false; then true; else false; fi")
7768 .unwrap();
7769 assert_eq!(status, 1);
7770 }
7771
7772 #[test]
7773 fn test_for_loop() {
7774 let _g = crate::test_util::global_state_lock();
7775 let mut exec = ShellExecutor::new();
7776 exec.execute_script("for i in a b c; do true; done")
7777 .unwrap();
7778 assert_eq!(exec.last_status(), 0);
7779 }
7780
7781 #[test]
7782 fn test_and_list() {
7783 let _g = crate::test_util::global_state_lock();
7784 let mut exec = ShellExecutor::new();
7785 let status = exec.execute_script("true && true").unwrap();
7786 assert_eq!(status, 0);
7787
7788 let status = exec.execute_script("true && false").unwrap();
7789 assert_eq!(status, 1);
7790 }
7791
7792 #[test]
7793 fn test_or_list() {
7794 let _g = crate::test_util::global_state_lock();
7795 let mut exec = ShellExecutor::new();
7796 let status = exec.execute_script("false || true").unwrap();
7797 assert_eq!(status, 0);
7798 }
7799
7800 /// Pin: `forklevel` matches the C global declared at
7801 /// `Src/exec.c:1052` (`int forklevel;`). Like `int` in C, the
7802 /// Rust port is an AtomicI32 starting at 0 (no fork has occurred
7803 /// at process start). Per `Src/exec.c:1221` (`forklevel =
7804 /// locallevel;`), every subshell entry copies `locallevel` into
7805 /// the global; the SIGPIPE handler at `Src/signals.c:808` reads
7806 /// it back to distinguish the top-level shell from a subshell.
7807 #[test]
7808 fn test_forklevel_default_zero_and_roundtrip() {
7809 let _g = crate::test_util::global_state_lock();
7810 use std::sync::atomic::Ordering;
7811 let prev = crate::ported::exec::FORKLEVEL.load(Ordering::Relaxed);
7812 // Default state at process start: zero (matches C's BSS init
7813 // of `int forklevel;` to 0).
7814 crate::ported::exec::FORKLEVEL.store(0, Ordering::Relaxed);
7815 assert_eq!(crate::ported::exec::FORKLEVEL.load(Ordering::Relaxed), 0);
7816 // Simulate the c:1221 store: `forklevel = locallevel;`.
7817 crate::ported::exec::FORKLEVEL.store(3, Ordering::Relaxed);
7818 assert_eq!(crate::ported::exec::FORKLEVEL.load(Ordering::Relaxed), 3);
7819 crate::ported::exec::FORKLEVEL.store(prev, Ordering::Relaxed);
7820 }
7821
7822 /// `zexecve_recover` is the PARENT-side half of C's `#!`-recovery block
7823 /// (c:Src/exec.c:534-627). Every `execve` in that block is preceded by
7824 /// `winch_unblock()` (c:565/570/580/584/626) because C is inside the
7825 /// forked child and is about to be replaced. This port returns to the
7826 /// caller instead, so those unblocks left the shell's standing
7827 /// `winch_block()` (c:Src/init.c:1458) dropped with nothing to re-arm
7828 /// it — and a SIGWINCH delivered inside a widget makes `calclist`
7829 /// (c:Src/Zle/compresult.c:1495) count display strings against a width
7830 /// they were not built at.
7831 ///
7832 /// The mask is per-thread, so blocking here cannot leak into other
7833 /// tests.
7834 #[test]
7835 #[cfg(unix)]
7836 fn zexecve_recover_keeps_sigwinch_blocked_in_the_parent() {
7837 fn sigwinch_blocked() -> bool {
7838 unsafe {
7839 let mut cur: libc::sigset_t = std::mem::zeroed();
7840 libc::pthread_sigmask(libc::SIG_BLOCK, std::ptr::null(), &mut cur);
7841 libc::sigismember(&cur, libc::SIGWINCH) == 1
7842 }
7843 }
7844
7845 let dir = std::env::temp_dir().join(format!(
7846 "zshrs_winch_recover_{}_{:?}",
7847 std::process::id(),
7848 std::thread::current().id()
7849 ));
7850 std::fs::create_dir_all(&dir).unwrap();
7851
7852 // c:611-627 — shebang-less text file: the `/bin/sh` fallback.
7853 let noshebang = dir.join("noshebang");
7854 std::fs::write(&noshebang, "echo hi\n").unwrap();
7855 // c:576-585 — `#!` line naming an absolute interpreter.
7856 let shebang = dir.join("shebang");
7857 std::fs::write(&shebang, "#!/bin/cat -v\necho hi\n").unwrap();
7858
7859 let saved = sigwinch_blocked();
7860 crate::ported::signals_h::winch_block(); // c:Src/init.c:1458
7861 assert!(sigwinch_blocked(), "precondition: winch_block() holds");
7862
7863 let argv = vec!["script".to_string()];
7864 let sh = zexecve_recover(noshebang.to_str().unwrap(), &argv, libc::ENOEXEC);
7865 assert_eq!(
7866 sh.as_ref().map(|(p, _)| p.as_str()),
7867 Ok("/bin/sh"),
7868 "c:625-627 — a shebang-less script re-execs through /bin/sh"
7869 );
7870 assert!(
7871 sigwinch_blocked(),
7872 "c:626 — the /bin/sh recovery must not unblock SIGWINCH in the parent"
7873 );
7874
7875 let interp = zexecve_recover(shebang.to_str().unwrap(), &argv, libc::ENOEXEC);
7876 assert_eq!(
7877 interp.as_ref().map(|(p, _)| p.as_str()),
7878 Ok("/bin/cat"),
7879 "c:576-585 — the `#!` line names the interpreter"
7880 );
7881 assert!(
7882 sigwinch_blocked(),
7883 "c:580/c:584 — the `#!` recovery must not unblock SIGWINCH in the parent"
7884 );
7885
7886 if !saved {
7887 crate::ported::signals_h::winch_unblock();
7888 }
7889 let _ = std::fs::remove_dir_all(&dir);
7890 }
7891
7892 /// c:Src/exec.c:544 `*argv = pth;` — the shebang-less `/bin/sh`
7893 /// fallback (c:611-627) hands sh the PATH being executed and applies the
7894 /// c:616 `-`/`+` guard to it, not to the command word the user typed.
7895 /// `PATH=-dir tstcmd` (A05execution.ztst, workers/52515) resolves to
7896 /// `-dir/tstcmd`, which sh must see as an operand.
7897 #[test]
7898 #[cfg(unix)]
7899 fn zexecve_recover_sh_fallback_passes_the_path_not_the_word() {
7900 let dir = std::env::temp_dir().join(format!(
7901 "zshrs_recover_argv_{}_{:?}",
7902 std::process::id(),
7903 std::thread::current().id()
7904 ));
7905 std::fs::create_dir_all(&dir).unwrap();
7906 let script = dir.join("tstcmd");
7907 std::fs::write(&script, "echo no shebang\n").unwrap();
7908 let pth = script.to_str().unwrap().to_string();
7909 // A word spelled like an option must neither be handed to sh nor
7910 // trip the guard: only the executed path counts.
7911 let argv = vec!["-tstcmd".to_string(), "arg".to_string()];
7912 let got = zexecve_recover(&pth, &argv, libc::ENOEXEC);
7913 let _ = std::fs::remove_dir_all(&dir);
7914 let want: Vec<String> = vec!["sh".to_string(), pth.clone(), "arg".to_string()];
7915 assert_eq!(got, Ok(("/bin/sh".to_string(), want)));
7916 }
7917}
7918
7919// Plugin-Framework-Agnostic State-Modification Recorder hook helpers.
7920/// Recorder helper: emit one record for an array/scalar mutation
7921/// targeting a path-family parameter (path/fpath/manpath/module_path/
7922/// cdpath, lower- or upper-cased), or one `assign` record for any
7923/// other name. Centralises the path-family list so `BUILTIN_SET_ARRAY`,
7924/// `BUILTIN_APPEND_ARRAY`, and `BUILTIN_APPEND_SCALAR_OR_PUSH` share
7925/// the same routing.
7926///
7927/// `is_append` distinguishes `arr=(...)` from `arr+=(...)` so the
7928/// emitted event carries the APPEND attr bit and replay can choose
7929/// between fresh-set and extend semantics.
7930///
7931/// `attrs` carries any pre-existing type info from
7932/// `recorder_attrs_for(name)` (readonly/export/global) — array shape
7933/// and APPEND get OR'd in by emit_array_assign.
7934#[cfg(feature = "recorder")]
7935pub(crate) fn emit_path_or_assign(
7936 name: &str,
7937 values: &[String],
7938 attrs: crate::recorder::ParamAttrs,
7939 is_append: bool,
7940 ctx: &crate::recorder::RecordCtx,
7941) {
7942 let lower = name.to_ascii_lowercase();
7943 let kind_name: Option<&'static str> = match lower.as_str() {
7944 "path" => Some("path"),
7945 "fpath" => Some("fpath"),
7946 "manpath" => Some("manpath"),
7947 "module_path" => Some("module_path"),
7948 "cdpath" => Some("cdpath"),
7949 _ => None,
7950 };
7951 match kind_name {
7952 Some(k) => {
7953 for v in values {
7954 crate::recorder::emit_path_mod(v, k, ctx.clone());
7955 // Each fpath addition also surfaces every `_completion`
7956 // file inside the directory — matches zinit-report's
7957 // per-plugin "Completions:" listing. Only fpath dirs
7958 // get this treatment; PATH dirs hold executables, not
7959 // completion functions.
7960 if k == "fpath" {
7961 crate::recorder::discover_completions_in_fpath_dir(v, ctx);
7962 }
7963 }
7964 }
7965 None => {
7966 // Non-path arrays: emit ONE `assign` event with the
7967 // ordered element list preserved in value_array. Replay
7968 // reconstructs `name=(elem1 elem2 ...)` exactly without
7969 // having to re-split a joined string.
7970 crate::recorder::emit_array_assign(
7971 name,
7972 values.to_vec(),
7973 attrs,
7974 is_append,
7975 ctx.clone(),
7976 );
7977 }
7978 }
7979}
7980
7981use std::os::unix::fs::MetadataExt;
7982
7983bitflags::bitflags! {
7984 /// Flags for zfork()
7985 #[derive(Debug, Clone, Copy, Default)]
7986 pub struct ForkFlags: u32 {
7987 const NOJOB = 1 << 0; // Don't add to job table
7988 const NEWGRP = 1 << 1; // Create new process group
7989 const FGTTY = 1 << 2; // Take foreground terminal
7990 const KEEPSIGS = 1 << 3; // Keep signal handlers
7991 }
7992}
7993
7994bitflags::bitflags! {
7995 /// Flags for entersubsh()
7996 #[derive(Debug, Clone, Copy, Default)]
7997 pub struct SubshellFlags: u32 {
7998 const NOMONITOR = 1 << 0; // Disable job control
7999 const KEEPFDS = 1 << 1; // Keep file descriptors
8000 const KEEPTRAPS = 1 << 2; // Keep trap handlers
8001 }
8002}
8003
8004/// Result of fork operation
8005#[derive(Debug)]
8006/// `fork()` outcome (parent / child / error).
8007/// Mirrors the integer return of `zfork()` from Src/exec.c:349.
8008pub enum ForkResult {
8009 /// `Parent` variant.
8010 Parent(i32), // Contains child PID
8011 /// `Child` variant.
8012 Child,
8013}
8014
8015/// Redirection mode
8016#[derive(Debug, Clone, Copy)]
8017/// File-redirection mode (`>` / `>>` / `<` / etc.).
8018/// Mirrors the `REDIR_*` enum from Src/zsh.h.
8019pub enum RedirMode {
8020 /// `Dup` variant.
8021 Dup,
8022 /// `Close` variant.
8023 Close,
8024}
8025
8026/// Builtin command type
8027#[derive(Debug, Clone, Copy)]
8028/// Builtin classification.
8029/// Mirrors the `BINF_*` flag set Src/builtin.c uses to
8030/// classify special vs regular builtins.
8031pub enum BuiltinType {
8032 /// `Normal` variant.
8033 Normal,
8034 /// `Disabled` variant.
8035 Disabled,
8036}
8037
8038use crate::fusevm_bridge::with_executor;
8039use crate::ported::glob::*;
8040use crate::ported::hist::*;
8041use crate::ported::jobs::*;
8042use crate::ported::math::*;
8043use crate::ported::module::*;
8044use crate::ported::modules::cap::*;
8045use crate::ported::modules::terminfo::*;
8046use crate::ported::options::*;
8047use crate::ported::params::*;
8048use crate::ported::pattern::*;
8049use crate::ported::prompt::*;
8050use crate::ported::signals::*;
8051use crate::ported::subst::*;
8052use crate::ported::utils::{zerr, zerrnam, zwarn, zwarnnam};
8053use ::regex::{Error as RegexError, Regex, RegexBuilder};
8054
8055pub use crate::ported::modules::regex::posix_ere_bracket_escape;
8056
8057impl ShellExecutor {
8058 /// Every option name in `ZSH_OPTIONS_SET` (port of `optns[]` at
8059 /// `Src/options.c:79+`).
8060 pub(crate) fn all_zsh_options() -> Vec<&'static str> {
8061 ZSH_OPTIONS_SET.iter().copied().collect()
8062 }
8063
8064 /// `name → default-on` map via canonical `default_on_options`
8065 /// (port of `defset()` macro at `Src/options.c:73`).
8066 pub(crate) fn default_options() -> HashMap<String, bool> {
8067 let on = default_on_options();
8068 Self::all_zsh_options()
8069 .into_iter()
8070 .map(|n| (n.to_string(), on.contains(n)))
8071 .collect()
8072 }
8073}
8074impl ShellExecutor {
8075 /// PURE PASSTHRU to the canonical `params::getsparam` (C port of
8076 /// `Src/params.c::getsparam`). Every special-name case the old
8077 /// 316-line body handled lives in `params::lookup_special_var` +
8078 /// `getsparam`'s paramtab/env walk. Returns an empty string for
8079 /// unset names (matching the old fn's signature; callers that
8080 /// need the set/unset distinction call `scalar` / `has_scalar`
8081 /// directly).
8082 pub(crate) fn get_variable(&self, name: &str) -> String {
8083 getsparam(name).unwrap_or_default()
8084 }
8085}
8086
8087// Source-form registration step used by the autoload-load path
8088// (`dispatch_function_call` / `run_function_body_only`). Decides
8089// whether to feed the file body to the funcdef pipeline VERBATIM
8090// (zsh-style: the body's own `function NAME() {...}` definition
8091// registers the function on execution) or to WRAP it in
8092// `NAME() {...}` (ksh-style or multi-statement: the body is just
8093// commands or includes additional statements past the def).
8094//
8095// The classification mirrors c:Src/exec.c:5725 + 5750 — KSHAUTOLOAD-
8096// equivalent vs zsh-style autoload. The structural check is the
8097// canonical `stripkshdef` (Src/exec.c:6291, ported at exec.rs:10548):
8098// parse the body to Eprog, run stripkshdef, and check whether it
8099// returned a stripped (different-length wordcode) Eprog. When it
8100// did, the file is the single-funcdef shape `[function] NAME [()] {
8101// INNER }`; running the file source directly through the funcdef
8102// pipeline registers NAME via the WC_FUNCDEF opcode at
8103// fusevm_bridge.rs:6330, matching C's `shf->funcdef = stripkshdef(
8104// prog, name)` semantics (the inner body becomes the function's
8105// body). When it didn't strip — single statement that isn't a
8106// funcdef, or multiple list nodes (e.g. `function ztm() {...}` +
8107// trailing `ztm "$@"` self-call) — we fall back to wrap-and-run so
8108// the canonical funcdef opcode still fires and any extra
8109// statements run inside the registered body, matching C's
8110// behavior of using the whole prog as funcdef in that case.
8111/// Restore `noaliases` on scope exit.
8112///
8113/// C's `loadautofn` (Src/exec.c:5684-5704) saves `noaliases`, sets it from the
8114/// function's PM_UNALIASED bit for the duration of the body parse, and restores
8115/// it unconditionally. The zshrs autoload block has early `return` paths, so the
8116/// restore has to ride on Drop rather than a trailing statement — otherwise a
8117/// `-U` autoload that failed to load would leave alias expansion disabled for
8118/// the rest of the shell.
8119struct NoAliasesRestore(bool);
8120
8121impl Drop for NoAliasesRestore {
8122 fn drop(&mut self) {
8123 crate::ported::lex::set_noaliases(self.0); // c:5704
8124 }
8125}
8126
8127/// c:Src/exec.c:5735 `loadautofnsetfile(shf, fdir)` + c:5751 — put the load
8128/// directory (and the absolute-path marker) back on a function whose body
8129/// zshrs just re-registered through the funcdef pipeline.
8130///
8131/// C loads an autoloaded body IN PLACE on the existing `Shfunc`, so
8132/// `filename`, PM_LOADDIR and PM_ABSPATH_USED all survive the load:
8133/// `shf->node.flags &= ~PM_UNDEFINED` (c:5751) is the only flag C clears.
8134/// zshrs re-registers the body as SOURCE through the WC_FUNCDEF pipeline,
8135/// which builds a FRESH node whose `filename` is the enclosing script and
8136/// whose flag word starts at zero. Both have to be reinstated, or `whence -v`
8137/// names the calling script instead of the definition file, and the
8138/// PM_LOADDIR|PM_ABSPATH_USED pair that `add_autoload_function`
8139/// (Src/builtin.c:3310-3323) tests — to hand a sibling `autoload -Uz NAME`
8140/// the caller's directory — is gone.
8141///
8142/// One helper rather than four inline copies: all four
8143/// `autoload_register_source` call sites need the identical restore.
8144///
8145/// `ksh_style` must be sampled BEFORE the re-registration (the fresh node's
8146/// flag word no longer carries PM_KSHSTORED / PM_ZSHSTORED, so re-deriving it
8147/// here would read the wrong answer): c:Src/exec.c:5792-5806 is the one arm
8148/// where C does NOT reload the body in place. It runs the whole FILE at top
8149/// level (`execode(prog, 1, 0, "evalautofunc")`, c:5795) and then REFETCHES
8150/// the node (`shf = shfunctab->getnode(shfunctab, n)`, c:5797) because the
8151/// file's own `NAME() { … }` created a brand-new Shfunc. There is no
8152/// `loadautofnsetfile` call on that arm, so zsh itself loses `filename`,
8153/// PM_LOADDIR and PM_ABSPATH_USED there. Verified against the reference shell:
8154///
8155/// ```text
8156/// $ zsh -f -c 'autoload -k /D/kw; kw'
8157/// ksh-kw ran
8158/// kw: ksib: function definition file not found
8159/// $ zsh -f -c 'autoload -k /D/kw; kw >/dev/null; whence -v kw'
8160/// kw is a shell function from zsh
8161/// ```
8162///
8163/// Restoring on that arm would make a ksh-autoloaded function inherit its
8164/// directory to siblings where zsh does not.
8165fn restore_loaddir(name: &str, dir: &str, abspath_used: bool, ksh_style: bool) {
8166 if ksh_style {
8167 return; // c:5792-5806 — no loadautofnsetfile on the ksh arm
8168 }
8169 if let Ok(mut tab) = crate::ported::hashtable::shfunctab_lock().write() {
8170 if let Some(shf) = tab.get_mut(name) {
8171 crate::ported::exec::loadautofnsetfile(shf, Some(dir)); // c:5735
8172 if abspath_used {
8173 shf.node.flags |= crate::ported::zsh_h::PM_ABSPATH_USED as i32; // c:5751
8174 }
8175 }
8176 }
8177}
8178
8179/// !!! WARNING: RUST-ONLY HELPER — NO C COUNTERPART !!!
8180///
8181/// C never has to ask this question. `Src/Modules/parameter.c:295-315`
8182/// (`setfunction`, the `functions[NAME]=BODY` setter) parses the assigned text
8183/// with `parse_string(val, 1)` and stores the result AS THE BODY —
8184/// `shf->funcdef = dupeprog(prog, 0)` (c:313) — so the function is fully
8185/// defined the moment the assignment runs. `loadautofn`
8186/// (`Src/exec.c:5723-5806`), which reads a FILE and therefore has to decide
8187/// between "the file IS the body" and "the file DEFINES the function"
8188/// (c:5781 `ksh == 2 || (ksh == 1 && isset(KSHAUTOLOAD))`), is a different
8189/// entry point that a parameter assignment never reaches.
8190///
8191/// zshrs stores the assigned text on `shfunc.body` with no compiled chunk, so
8192/// the first CALL lands in the same autoload prelude a real `autoload` stub
8193/// uses — and that prelude was applying `loadautofn`'s file rules to it. The
8194/// two cases are told apart by the flag `loadautofn` stamps and a parameter
8195/// assignment does not: `loadautofnsetfile` sets PM_LOADDIR (c:5657) on the
8196/// node it loaded, while `setfunction` leaves the flag word at `dis` (c:314).
8197/// The `loaded_dir` capture at both call sites already reads the same bit for
8198/// the same reason.
8199fn body_came_from_param_assignment(flags: i32) -> bool {
8200 (flags as u32 & crate::ported::zsh_h::PM_LOADDIR) == 0
8201}
8202
8203/// !!! WARNING: RUST-ONLY HELPER — NO C COUNTERPART !!!
8204///
8205/// The text to run to install a `functions[NAME]=BODY` assignment: ALWAYS the
8206/// `NAME() { … }` wrap, never the verbatim run that
8207/// [`autoload_definition_source`] falls back to for a file whose contents are
8208/// already a definition of `NAME`. c:Src/Modules/parameter.c:313 makes the
8209/// parsed text the funcdef unconditionally, so text that happens to READ as
8210/// `NAME () { … }` is a BODY whose execution defines `NAME` — measured
8211/// against the reference shell:
8212///
8213/// ```text
8214/// $ zsh -f -c 'functions[g]="g () { print inner }"; g; print "[${functions[g]}]"'
8215/// [ print inner]
8216/// ```
8217///
8218/// Nothing is printed: the call ran the definition. Running that text verbatim
8219/// at install time instead printed `inner` and defined `g` a call too early.
8220fn param_definition_source(name: &str, body: &str) -> String {
8221 format!("{name}() {{\n{body}\n}}")
8222}
8223
8224/// c:Src/exec.c:5781 — `if (ksh == 2 || (ksh == 1 && isset(KSHAUTOLOAD)))`,
8225/// the ksh-style load branch. `ksh` derives from the stub's stored-style bits
8226/// per c:5762-5766 (`PM_KSHSTORED ? 2 : PM_ZSHSTORED ? 0 : 1`; a decisive
8227/// `.zwc` header flag was already folded into these bits by `loadautofn`).
8228///
8229/// Two zshrs steps need the same answer — `autoload_register_source` (wrap vs
8230/// verbatim) and `restore_loaddir` (whether the load kept the original node)
8231/// — so the decision lives in one place.
8232fn autoload_is_ksh_style(name: &str) -> bool {
8233 let flags = crate::ported::utils::getshfunc(name)
8234 .map(|f| f.node.flags as u32)
8235 .unwrap_or(0);
8236 let ksh = if flags & crate::ported::zsh_h::PM_KSHSTORED != 0 {
8237 2 // c:5765
8238 } else if flags & crate::ported::zsh_h::PM_ZSHSTORED != 0 {
8239 0 // c:5766
8240 } else {
8241 1 // c:5766
8242 };
8243 ksh == 2 || (ksh == 1 && crate::ported::zsh_h::isset(crate::ported::zsh_h::KSHAUTOLOAD))
8244 // c:5781
8245}
8246
8247/// The cache key for an autoloaded function: `(resolved fpath dir,
8248/// SHA-256 of the definition text)`, or `None` when the directory
8249/// cannot be pinned down.
8250///
8251/// `loadautofn` records the resolved fpath directory on the shfunc
8252/// (`filename` + `PM_LOADDIR`, c:Src/exec.c:5657); a function whose
8253/// `filename` is still the placeholder `"zsh"` was not resolved through
8254/// `$fpath` and is not cached.
8255///
8256/// The hash is over `registered` — the exact string about to be
8257/// compiled — and NOT over a `stat` of `<dir>/<name>`. Those are not the
8258/// same thing: `getfpfunc` prefers a `<dir>.zwc` digest over the plain
8259/// file whenever the digest is newer (c:Src/parse.c:3771-3777), so the
8260/// body being installed may have no relationship to the bytes of the
8261/// file that path names. Stamping the path let a chunk built from one
8262/// text be served for another.
8263///
8264/// `from_wordcode` salts the digest because it changes what the same bytes
8265/// COMPILE TO: a wordcode-derived body is compiled under [`ZwcRelexGuard`],
8266/// so it resolves quotes and aliases the way `zcompile` did, while a
8267/// plain-file body of those same bytes resolves them against the live
8268/// RCQUOTES / alias state. One cache line must never be served for both.
8269fn autoload_source_key(
8270 name: &str,
8271 registered: &str,
8272 from_wordcode: bool,
8273) -> Option<(String, [u8; 32])> {
8274 let dir = crate::ported::utils::getshfunc(name)
8275 .and_then(|f| f.filename)
8276 .filter(|d| d != "zsh")?;
8277 let sha = if from_wordcode {
8278 crate::autoload_cache::source_digest(&format!("\0zwc\0{registered}"))
8279 } else {
8280 crate::autoload_cache::source_digest(registered)
8281 };
8282 Some((dir, sha))
8283}
8284
8285/// !!! WARNING: RUST-ONLY HELPER — NO C COUNTERPART !!!
8286///
8287/// Raises `noerrs` (c:Src/utils.c:175 — the gate every `zerr`/`zwarn`
8288/// consults) for the life of the guard, so a pass that only exists because
8289/// zshrs parses text where C carries a parsed `Eprog` cannot report an error
8290/// C would report exactly once. `ERRFLAG_ERROR` is still set while muted, so
8291/// callers that read errflag still see the failure.
8292struct ZerrMute(i32);
8293
8294impl ZerrMute {
8295 fn enter() -> Self {
8296 let mut g = crate::ported::utils::noerrs_lock().lock().unwrap();
8297 let prev = *g;
8298 *g = 1;
8299 Self(prev)
8300 }
8301}
8302
8303impl Drop for ZerrMute {
8304 fn drop(&mut self) {
8305 if let Ok(mut g) = crate::ported::utils::noerrs_lock().lock() {
8306 *g = self.0;
8307 }
8308 }
8309}
8310
8311/// What an autoload of one name will hand the compiler.
8312struct AutoloadSource {
8313 /// The definition text: the file body verbatim, or the `name() { … }` wrap.
8314 text: String,
8315 /// The body is a `.zwc` deparse, so the compile needs [`ZwcRelexGuard`].
8316 from_wordcode: bool,
8317 /// c:Src/exec.c:6264 — `parse_string` rejected the file and has ALREADY
8318 /// reported the error against the loadee's name. C returns NULL from
8319 /// `getfpfunc`, so `loadautofn` abandons the load right there (c:5726) and
8320 /// prints nothing more. zshrs still runs `text` through the compiler
8321 /// (that is where the body becomes a chunk), so the caller silences THAT
8322 /// pass instead — otherwise the same parse error lands twice, the second
8323 /// time one line off because of the wrap's leading `name() {` line.
8324 parse_failed: bool,
8325}
8326
8327/// Returns the text to run to install `name`, plus whether that text came out
8328/// of a `.zwc` (see [`autoload_note_wordcode_body`]) and therefore has to be
8329/// lexed under [`ZwcRelexGuard`] wherever it is lexed, and whether the file's
8330/// one parse rejected it.
8331fn autoload_register_source(name: &str, body: &str) -> AutoloadSource {
8332 let ksh_style = autoload_is_ksh_style(name);
8333 // `loadautofn` took this body from autoloads.rkyv: the registration it
8334 // was stored with is the one the parse below would produce.
8335 if let Some((text, from_wordcode, parse_failed)) =
8336 crate::autoload_cache::take_hit(name, body, ksh_style)
8337 {
8338 return AutoloadSource {
8339 text,
8340 from_wordcode,
8341 parse_failed,
8342 };
8343 }
8344 // c:Src/exec.c:5725 `stripkshdef(prog, …)` — the ksh-vs-zsh wrap decision
8345 // below PARSES `body`, so it is itself a second lex of the deparse and
8346 // needs the same pin as the compile that follows it.
8347 let from_wordcode = autoload_body_from_wordcode(name, body);
8348 let _relex = from_wordcode.then(ZwcRelexGuard::enter);
8349 let (text, parse_failed) = autoload_definition_source(name, body, ksh_style);
8350 crate::autoload_cache::commit_search(name, body, &text, parse_failed, ksh_style);
8351 AutoloadSource {
8352 text,
8353 from_wordcode,
8354 parse_failed,
8355 }
8356}
8357
8358/// !!! WARNING: RUST-ONLY HELPER — NO C COUNTERPART !!!
8359///
8360/// Pins the lexer to the spelling a `.zwc` deparse was written in, for the
8361/// duration of one compile.
8362///
8363/// C never needs this: the compiled arm of `source()` is `execode(prog, 1,
8364/// 0, "filecode")` (c:Src/init.c:1621) and the ksh-autoload arm is
8365/// `execode(prog, 1, 0, "evalautofunc")` (c:Src/exec.c:5795) — the wordcode
8366/// runs as it stands and NOTHING is lexed. A `.zwc` is resolved once, at
8367/// `zcompile` time. zshrs has no execute-the-wordcode path: it deparses back
8368/// to source with `getpermtext` and lexes that, so every piece of LEXER-TIME
8369/// state that C bypassed has to be neutralised by hand, or the second lex
8370/// resolves the text differently from the first.
8371///
8372/// Two such pieces are known to change the result:
8373///
8374/// * **RCQUOTES.** `untokenize` (c:Src/exec.c:2134) renders every quote
8375/// null through `ztokens[Snull - Pound]`, and that entry is a bare
8376/// single quote (c:Src/lex.c:38), so a closing null followed by an
8377/// opening one deparses to two adjacent quotes. Under RCQUOTES the lexer
8378/// reads that pair inside a quoted word as one LITERAL quote
8379/// (c:Src/lex.c:1328) rather than as two delimiters. `zsh-expand`'s
8380/// plugin entry does `setopt rcquotes` at its line 39, so every later
8381/// `.zwc`-sourced alias with adjacent quoted segments — the whole
8382/// `zsh-openshift-aliases` set — gained literal quotes zsh never had.
8383///
8384/// * **Aliases.** `checkalias` (c:Src/lex.c:1909) fires from the lexer, so
8385/// an alias installed before the `source` rewrote words INSIDE the
8386/// compiled program: with `alias mycmd=…` live, a `.zwc` whose second
8387/// line is `mycmd` ran the alias where zsh runs the function. Global
8388/// aliases are worse — they rewrite any word, so `print -r -- GA: x`
8389/// became `print -r -- GA: LEAKED` under `alias -g x=LEAKED`. This is
8390/// C's `noaliases` (c:Src/lex.c:135), the same switch `par_case` uses to
8391/// keep `in` from being alias-expanded.
8392///
8393/// Both are restored on drop, so the restore survives a panic out of the
8394/// compiler — and, more importantly, the program's own RUNTIME lexing is
8395/// unaffected: a `setopt rcquotes` the `.zwc` performs still takes effect
8396/// and still outlives the source, and a function or `eval` body it runs is
8397/// lexed later, against the live alias table.
8398///
8399/// One residual case is out of reach here and needs a real
8400/// execute-the-wordcode path: a `.zwc` COMPILED while RCQUOTES was set holds
8401/// an unescaped literal quote in its tokenized word (c:Src/lex.c:1329 adds
8402/// the character with no `Bnull` prefix), and `untokenize` cannot tell that
8403/// apart from a quote null. No option state at re-lex time recovers it.
8404///
8405/// `noaliases` is a thread-local (`LEX_NOALIASES`), but the option store is
8406/// process-wide, so the RCQUOTES window is visible to a worker thread that
8407/// lexes concurrently. The window is one synchronous compile.
8408pub(crate) struct ZwcRelexGuard {
8409 rcquotes: bool,
8410 noaliases: bool,
8411}
8412
8413impl ZwcRelexGuard {
8414 pub(crate) fn enter() -> Self {
8415 let rcquotes = crate::ported::zsh_h::isset(crate::ported::zsh_h::RCQUOTES);
8416 if rcquotes {
8417 crate::ported::options::opt_state_set("rcquotes", false);
8418 }
8419 let noaliases = crate::ported::lex::noaliases();
8420 crate::ported::lex::set_noaliases(true); // c:Src/lex.c:1909
8421 Self {
8422 rcquotes,
8423 noaliases,
8424 }
8425 }
8426}
8427
8428impl Drop for ZwcRelexGuard {
8429 fn drop(&mut self) {
8430 if self.rcquotes {
8431 crate::ported::options::opt_state_set("rcquotes", true);
8432 }
8433 crate::ported::lex::set_noaliases(self.noaliases);
8434 }
8435}
8436
8437/// !!! WARNING: RUST-ONLY HELPER — NO C COUNTERPART !!!
8438///
8439/// The RCQUOTES state each shell-defined function's body was resolved
8440/// under at DEFINITION time, keyed by name and digested body text.
8441///
8442/// C has no such fact to carry. `execfuncdef` compiles the body into
8443/// wordcode when the definition runs (c:Src/exec.c:5389-5391 —
8444/// `shf->funcdef = dupeprog(...)`), and `printshfuncnode` renders THAT
8445/// program back with `getpermtext(f->funcdef, NULL, 1)`
8446/// (c:Src/hashtable.c:954). Every lexer-time decision is already settled
8447/// in the wordcode, so printing cannot revisit it — `functions f` gives
8448/// the same text no matter which options are set when it runs.
8449///
8450/// zshrs has no `Eprog` for a shell-defined function: `shfunc::body` keeps
8451/// the raw source and `printshfuncnode`'s stand-in RE-LEXES it at print
8452/// time. Under RCQUOTES an adjacent quote pair inside a quoted word lexes
8453/// as one LITERAL quote (c:Src/lex.c:1328) instead of as two delimiters,
8454/// so one unchanged function deparsed two different ways across a
8455/// `setopt rcquotes` — docs/BUGS.md #1105. Pinning the print-time lex to
8456/// defaults is wrong in the other direction: a function defined WHILE
8457/// RCQUOTES was set has to keep printing the RCQUOTES resolution, which is
8458/// why the state has to be recorded rather than assumed.
8459///
8460/// The state is sampled where the definition INSTALLS (the
8461/// `BUILTIN_REGISTER_COMPILED_FN` handler), not where its body was lexed.
8462/// In C those are the same instant, because zsh lexes one command list at
8463/// a time as it runs it; zshrs compiles a whole script ahead of running
8464/// it, so a runtime `setopt rcquotes` on an earlier line of the same file
8465/// is already in force at install time but was NOT in force at the ahead
8466/// -of-time lex. Install time is the one that matches zsh's printed
8467/// output (`setopt rcquotes` on line 1, `f() { … }` on line 2 →
8468/// C prints the RCQUOTES resolution), and it collapses onto lex time the
8469/// day the compiler stops running ahead of execution.
8470///
8471/// The mark is a DIGEST of the body rather than a bare flag, exactly like
8472/// [`AUTOLOAD_WORDCODE_BODY`]: `functions[name]=…`, an autoload, and a
8473/// plain redefinition all install different text, and a stale entry
8474/// simply stops matching instead of pinning the lexer for a body it never
8475/// described.
8476static FUNCDEF_LEX_RCQUOTES: Mutex<Option<HashMap<String, ([u8; 32], bool)>>> = Mutex::new(None);
8477
8478/// !!! WARNING: RUST-ONLY HELPER — NO C COUNTERPART !!!
8479///
8480/// Record the RCQUOTES state `name`'s definition was installed under.
8481/// See [`FUNCDEF_LEX_RCQUOTES`]; called from the funcdef install path
8482/// (c:Src/exec.c:5389 `shf->funcdef = …`, where C bakes the same fact
8483/// into the wordcode instead).
8484pub(crate) fn funcdef_note_rcquotes(name: &str, body: &str, rcquotes: bool) {
8485 let digest = crate::autoload_cache::source_digest(body);
8486 // The state the body was LEXED under wins when the parser recorded it
8487 // (see FUNCDEF_PARSE_RCQUOTES); install time is only the fallback.
8488 let rcquotes = FUNCDEF_PARSE_RCQUOTES
8489 .lock()
8490 .as_ref()
8491 .and_then(|m| m.get(&digest).copied())
8492 .unwrap_or(rcquotes);
8493 FUNCDEF_LEX_RCQUOTES
8494 .lock()
8495 .get_or_insert_with(HashMap::new)
8496 .insert(
8497 name.to_string(),
8498 (digest, rcquotes),
8499 );
8500}
8501
8502/// !!! WARNING: RUST-ONLY HELPER STATE — NO C COUNTERPART !!!
8503///
8504/// RCQUOTES as it stood when the parser lexed each captured function body,
8505/// keyed by the body digest. C bakes the `''` decision into the wordcode at
8506/// lex time (c:Src/lex.c:1326-1328), and zsh lexes one list at a time, so
8507/// `setopt rcquotes; f() { print 'a''b' }` on ONE line lexes `f` before the
8508/// option is on and prints `'a''b'`, while a definition on a later line (or
8509/// through `eval`) prints `'a'b'`. Install time cannot tell those apart.
8510static FUNCDEF_PARSE_RCQUOTES: Mutex<Option<HashMap<[u8; 32], bool>>> = Mutex::new(None);
8511
8512/// !!! WARNING: RUST-ONLY HELPER STATE — NO C COUNTERPART !!!
8513///
8514/// Digests of function bodies the parser captured, which are ALIAS-RESOLVED.
8515///
8516/// C's par_funcdef compiles the body to wordcode after alias expansion, so
8517/// `functions` (getpermtext, c:Src/hashtable.c:954) shows exactly what the
8518/// parse saw, whatever the alias table holds later: an alias removed or
8519/// redefined after the definition changes nothing, and one defined after the
8520/// parse (a `-c` string is parsed whole, c:Src/init.c:1568 execstring) never
8521/// appears. `funcdef_capture::body_text` hands the parser that same
8522/// post-expansion text, so re-lexing it with aliases ON expands a second time
8523/// — `alias ls='ls -G'` listed `ls -G -G x` — and a body recorded here is
8524/// always re-lexed with aliases off. Keyed by digest like
8525/// [`FUNCDEF_LEX_RCQUOTES`]; text that did not come from a capture
8526/// (`functions[f]=…`, an autoload file) has no entry and keeps the live table.
8527static FUNCDEF_ALIAS_RESOLVED: Mutex<Option<std::collections::HashSet<[u8; 32]>>> =
8528 Mutex::new(None);
8529
8530/// !!! WARNING: RUST-ONLY HELPER — NO C COUNTERPART !!!
8531///
8532/// Called by the parser right after it captures a function body. See
8533/// [`FUNCDEF_ALIAS_RESOLVED`].
8534pub(crate) fn funcdef_note_resolved_body(body: Option<&str>) {
8535 let Some(body) = body else { return };
8536 let digest = crate::autoload_cache::source_digest(body);
8537 FUNCDEF_ALIAS_RESOLVED
8538 .lock()
8539 .get_or_insert_with(std::collections::HashSet::new)
8540 .insert(digest);
8541 // Same parse-time capture point: record the RCQUOTES the body was lexed
8542 // under (FUNCDEF_PARSE_RCQUOTES).
8543 FUNCDEF_PARSE_RCQUOTES
8544 .lock()
8545 .get_or_insert_with(HashMap::new)
8546 .insert(digest, crate::ported::zsh_h::isset(crate::ported::zsh_h::RCQUOTES));
8547}
8548
8549/// !!! WARNING: RUST-ONLY HELPER — NO C COUNTERPART !!!
8550///
8551/// Pin RCQUOTES to the value `name`'s body was defined under, for the
8552/// duration of one deparse of that body.
8553///
8554/// This is [`ZwcRelexGuard`]'s sibling: same problem (zshrs lexes text
8555/// where C runs wordcode), same RAII shape, but the target value is the
8556/// RECORDED one rather than "off" — the definition may itself have been
8557/// made under RCQUOTES, and then the RCQUOTES resolution is the correct
8558/// one to print. With no record for this exact body (a function installed
8559/// through a path that does not record — `functions[f]=…`, the ported
8560/// walker, an autoload stub) nothing is pinned and the live option stands,
8561/// which is the behaviour that predates the record.
8562///
8563/// The option store is process-wide, so the pin is visible to a worker
8564/// thread that lexes concurrently; the window is one synchronous deparse,
8565/// as it is for [`ZwcRelexGuard`].
8566pub(crate) fn funcdef_lex_pin(name: &str, body: &str) -> FuncdefLexPin {
8567 // c:Src/hashtable.c:954 renders the STORED wordcode, where every alias the
8568 // body's parse expanded is already baked in and nothing else can expand.
8569 // The re-lex runs with `noaliases` (c:Src/lex.c:135) whenever the text is
8570 // already resolved: the parser captured it (FUNCDEF_ALIAS_RESOLVED), or —
8571 // for an autoload body not yet parsed, as after `autoload +X` — it was
8572 // loaded under `autoload -U` (PM_UNALIASED, c:Src/exec.c:5746) or rendered
8573 // from a `.zwc`. Any other body keeps the live alias table.
8574 let noaliases = crate::ported::lex::noaliases();
8575 // c:Src/exec.c:5389 execfuncdef — C deparses the wordcode parsed when the
8576 // function was DEFINED, and a definition inside `emulate sh -c` is parsed
8577 // under that emulation (stamped as `shf->sticky`, c:5527-5532). Re-lex
8578 // under the same emulation (installemulation + its on/off options, as
8579 // doshfunc applies it at c:5977-6010), so an sh-mode `( one | two )` case
8580 // pattern keeps its separate alternatives and prints `(one | two)`. Only a
8581 // function carrying a sticky emulation is touched: `eval` / `-z` / `+X`
8582 // bodies keep the live state.
8583 let sticky_restore = crate::ported::utils::getshfunc(name)
8584 .and_then(|f| f.sticky.clone())
8585 .filter(|s| crate::ported::exec::sticky_emulation_differs(Some(s)) != 0)
8586 .map(|s| install_sticky_emulation(&s));
8587 let captured = FUNCDEF_ALIAS_RESOLVED
8588 .lock()
8589 .as_ref()
8590 .is_some_and(|set| set.contains(&crate::autoload_cache::source_digest(body)));
8591 let deparse_noaliases = captured
8592 || crate::ported::utils::getshfunc(name).is_some_and(|f| {
8593 (f.node.flags as u32 & crate::ported::zsh_h::PM_UNALIASED) != 0
8594 })
8595 || autoload_body_from_wordcode(name, body);
8596 if deparse_noaliases {
8597 crate::ported::lex::set_noaliases(true); // c:Src/lex.c:1909
8598 }
8599 let want = {
8600 let slot = FUNCDEF_LEX_RCQUOTES.lock();
8601 slot.as_ref()
8602 .and_then(|map| map.get(name))
8603 .filter(|(sha, _)| *sha == crate::autoload_cache::source_digest(body))
8604 .map(|(_, rcquotes)| *rcquotes)
8605 };
8606 let live = crate::ported::zsh_h::isset(crate::ported::zsh_h::RCQUOTES);
8607 match want {
8608 Some(w) if w != live => {
8609 crate::ported::options::opt_state_set("rcquotes", w);
8610 FuncdefLexPin {
8611 restore: Some(live),
8612 noaliases,
8613 sticky_restore,
8614 }
8615 }
8616 _ => FuncdefLexPin {
8617 restore: None,
8618 noaliases,
8619 sticky_restore,
8620 },
8621 }
8622}
8623
8624/// !!! WARNING: RUST-ONLY HELPER — NO C COUNTERPART !!!
8625///
8626/// RAII half of [`funcdef_lex_pin`]. `restore` is `None` when nothing was
8627/// changed, so the common path costs one hash lookup and no option write.
8628pub(crate) struct FuncdefLexPin {
8629 restore: Option<bool>,
8630 noaliases: bool,
8631 /// Options and emulation in force before a sticky emulation was applied
8632 /// for the re-lex; put back on drop.
8633 sticky_restore: Option<(Vec<bool>, i32, i32, bool)>,
8634}
8635
8636impl Drop for FuncdefLexPin {
8637 fn drop(&mut self) {
8638 if let Some(live) = self.restore {
8639 crate::ported::options::opt_state_set("rcquotes", live);
8640 }
8641 crate::ported::lex::set_noaliases(self.noaliases);
8642 if let Some(saved) = self.sticky_restore.take() {
8643 restore_sticky_emulation(saved);
8644 }
8645 }
8646}
8647
8648/// Options and emulation cells saved by [`install_sticky_emulation`].
8649type SavedEmulation = (Vec<bool>, i32, i32, bool);
8650
8651/// !!! WARNING: RUST-ONLY HELPER — NO C COUNTERPART !!!
8652///
8653/// The option half of doshfunc's sticky-emulation entry (c:Src/exec.c:
8654/// 5991-6008): `installemulation(emulation, opts)` then the on/off option
8655/// lists. Returns what it overwrote for [`restore_sticky_emulation`].
8656fn install_sticky_emulation(s: &crate::ported::zsh_h::emulation_options) -> SavedEmulation {
8657 use std::sync::atomic::Ordering;
8658 let size = crate::ported::zsh_h::OPT_SIZE as usize;
8659 let saved: Vec<bool> = (0..size)
8660 .map(|optno| {
8661 optno > 0
8662 && crate::ported::options::opt_state_get(crate::ported::zsh_h::opt_name(optno as i32))
8663 .unwrap_or(false)
8664 })
8665 .collect();
8666 let saved_emu = (
8667 crate::ported::options::emulation.load(Ordering::Relaxed),
8668 crate::ported::options::EMULATION.load(Ordering::Relaxed),
8669 crate::ported::options::FULLY_EMULATING.load(Ordering::Relaxed),
8670 );
8671 let mut new_opts = [-1i8; crate::ported::zsh_h::OPT_SIZE as usize];
8672 crate::ported::options::installemulation(s.emulation, &mut new_opts); // c:5993
8673 for (optno, &v) in new_opts.iter().enumerate().skip(1) {
8674 if v >= 0 {
8675 crate::ported::options::opt_state_set(crate::ported::zsh_h::opt_name(optno as i32), v == 1);
8676 }
8677 }
8678 for on in &s.on_opts {
8679 crate::ported::options::opt_state_set(crate::ported::zsh_h::opt_name(*on as i32), true); // c:5995-6001
8680 }
8681 for off in &s.off_opts {
8682 crate::ported::options::opt_state_set(crate::ported::zsh_h::opt_name(*off as i32), false); // c:6002-6008
8683 }
8684 let emu_base = s.emulation & !crate::ported::zsh_h::EMULATE_FULLY;
8685 crate::ported::options::emulation.store(emu_base, Ordering::Relaxed);
8686 crate::ported::options::EMULATION.store(emu_base, Ordering::Relaxed);
8687 crate::ported::options::FULLY_EMULATING
8688 .store((s.emulation & crate::ported::zsh_h::EMULATE_FULLY) != 0, Ordering::Relaxed);
8689 (saved, saved_emu.0, saved_emu.1, saved_emu.2)
8690}
8691
8692/// !!! WARNING: RUST-ONLY HELPER — NO C COUNTERPART !!!
8693///
8694/// Undo [`install_sticky_emulation`].
8695fn restore_sticky_emulation((opts, emu, emu_cell, fully): SavedEmulation) {
8696 use std::sync::atomic::Ordering;
8697 for (optno, on) in opts.into_iter().enumerate().skip(1) {
8698 crate::ported::options::opt_state_set(crate::ported::zsh_h::opt_name(optno as i32), on);
8699 }
8700 crate::ported::options::emulation.store(emu, Ordering::Relaxed);
8701 crate::ported::options::EMULATION.store(emu_cell, Ordering::Relaxed);
8702 crate::ported::options::FULLY_EMULATING.store(fully, Ordering::Relaxed);
8703}
8704
8705/// !!! WARNING: RUST-ONLY HELPER — NO C COUNTERPART !!!
8706///
8707/// C loads an autoloaded body from INSIDE doshfunc: the stub's funcdef is
8708/// the `mkautofn` trampoline, so `execautofn` → `loadautofn` → `getfpfunc`
8709/// (c:Src/exec.c:5640, 5697) parses the file after doshfunc has entered the
8710/// stub's sticky emulation (c:5978-6010), and the body is installed IN
8711/// PLACE on the same Shfunc, keeping `shf->sticky` (stamped by
8712/// `shfunc_set_sticky` in bin_autoload, c:Src/builtin.c:3766). zshrs loads
8713/// and registers the body before doshfunc runs, so this guard opens the
8714/// same window for the load: the stub's emulation for the parse, and the
8715/// stub's snapshot as the pending `sticky` so the re-registration stamps it
8716/// onto the loaded function.
8717struct AutoloadStickyGuard {
8718 saved_opts: SavedEmulation,
8719 saved_sticky: Option<crate::ported::zsh_h::Emulation_options>,
8720}
8721
8722impl AutoloadStickyGuard {
8723 fn enter(stub: &crate::ported::zsh_h::shfunc) -> Option<Self> {
8724 let s = stub.sticky.as_deref()?;
8725 if crate::ported::exec::sticky_emulation_differs(Some(s)) == 0 {
8726 return None;
8727 }
8728 let dup = crate::ported::exec::sticky_emulation_dup(s, 1);
8729 let saved_opts = install_sticky_emulation(&dup);
8730 let saved_sticky = std::mem::replace(
8731 &mut *crate::ported::options::sticky
8732 .lock()
8733 .unwrap_or_else(|e| e.into_inner()),
8734 Some(dup),
8735 );
8736 Some(Self {
8737 saved_opts,
8738 saved_sticky,
8739 })
8740 }
8741}
8742
8743impl Drop for AutoloadStickyGuard {
8744 fn drop(&mut self) {
8745 *crate::ported::options::sticky
8746 .lock()
8747 .unwrap_or_else(|e| e.into_inner()) = self.saved_sticky.take();
8748 restore_sticky_emulation(std::mem::take(&mut self.saved_opts));
8749 }
8750}
8751
8752/// !!! WARNING: RUST-ONLY HELPER — NO C COUNTERPART !!!
8753///
8754/// Names whose installed autoload body was rendered back from WORDCODE
8755/// (`getpermtext` over a `.zwc` program) rather than read from a source file,
8756/// each mapped to a digest of that body.
8757///
8758/// C has no such fact to carry. `loadautofn` hands the dump's program
8759/// straight to `shf->funcdef = stripkshdef(prog, …)` (c:Src/exec.c:5753-5755),
8760/// and the ksh-style arm runs it with `execode(prog, 1, 0, "evalautofunc")`
8761/// (c:Src/exec.c:5795): a dump-loaded funcdef is never lexed again, at load
8762/// time or ever. zshrs executes function bodies as TEXT, so `loadautofn`
8763/// deparses the program and the REAL compile happens later — at the call that
8764/// installs the function, through [`ShellExecutor::run_autoload_definition`].
8765/// The two are separated by arbitrary user code, so "this text came from
8766/// wordcode" has to survive the gap or the deparse is re-lexed against
8767/// whatever RCQUOTES / alias state is live at call time. See
8768/// [`ZwcRelexGuard`] for exactly what that costs; the `source` leg of the same
8769/// bug is fixed by [`ShellExecutor::execute_zwc_program`].
8770///
8771/// The mark is a DIGEST of the body, not a bare flag, so it self-invalidates.
8772/// `functions[name]=…` (c:Src/Modules/parameter.c setpmfunction) and a later
8773/// load of the same name from a plain file both install different text, and
8774/// the digest simply stops matching — no writer anywhere else has to know this
8775/// table exists, and a stale entry cannot pin the lexer for a body that was
8776/// never wordcode.
8777static AUTOLOAD_WORDCODE_BODY: Mutex<Option<HashMap<String, [u8; 32]>>> = Mutex::new(None);
8778
8779/// !!! WARNING: RUST-ONLY HELPER — NO C COUNTERPART !!!
8780///
8781/// Record (`Some`) or drop (`None`) the "body came from wordcode" mark for
8782/// `name`. Called from `loadautofn` (c:Src/exec.c:5682) on every load, so a
8783/// plain-file load clears a mark an earlier dump load left.
8784pub(crate) fn autoload_note_wordcode_body(name: &str, body: Option<&str>) {
8785 let mut slot = AUTOLOAD_WORDCODE_BODY.lock();
8786 match body {
8787 Some(text) => {
8788 slot.get_or_insert_with(HashMap::new)
8789 .insert(name.to_string(), crate::autoload_cache::source_digest(text));
8790 }
8791 None => {
8792 if let Some(map) = slot.as_mut() {
8793 map.remove(name);
8794 }
8795 }
8796 }
8797}
8798
8799/// !!! WARNING: RUST-ONLY HELPER — NO C COUNTERPART !!!
8800///
8801/// True when `body` is the exact text [`autoload_note_wordcode_body`] recorded
8802/// for `name` — i.e. this is a `.zwc` deparse about to be lexed a second time.
8803pub(crate) fn autoload_body_from_wordcode(name: &str, body: &str) -> bool {
8804 let slot = AUTOLOAD_WORDCODE_BODY.lock();
8805 slot.as_ref()
8806 .and_then(|map| map.get(name))
8807 .is_some_and(|sha| *sha == crate::autoload_cache::source_digest(body))
8808}
8809
8810/// The exact source text an autoload of `name` installs — either the
8811/// file body verbatim (ksh style, or a file that already defines the
8812/// function) or `name() { <body> }`.
8813///
8814/// Split out of [`autoload_register_source`] with the ksh decision
8815/// passed IN so the prewarm (`autoload_prewarm`, which has no shfunc
8816/// flags to consult because nothing is registered yet) compiles
8817/// byte-identical text to what the loader will run. The two drifting
8818/// apart is precisely what made the pre-v2 shard unusable: it cached a
8819/// different program than the one the loader installs.
8820/// Returns the definition text plus whether that parse REJECTED the body
8821/// (c:Src/exec.c:6264 `r = parse_string(d, 1)` returning NULL, which makes
8822/// `loadautofn` abandon the load at c:5726 with no further diagnostic).
8823pub(crate) fn autoload_definition_source(
8824 name: &str,
8825 body: &str,
8826 ksh_style: bool,
8827) -> (String, bool) {
8828 // c:Src/exec.c:5781 — a ksh-style load executes the file contents at top
8829 // level (c:5795 `execode(prog, 1, 0, "evalautofunc")`) and expects the
8830 // file itself to define the function — so the body goes through the
8831 // pipeline VERBATIM, never wrapped.
8832 if ksh_style {
8833 return (body.to_string(), false); // c:5795 execode(prog, ..., "evalautofunc")
8834 }
8835 // c:6257-6265 — `getfpfunc` parses the file under the LOADEE's name:
8836 // char *oldscriptname = scriptname;
8837 // scriptname = dupstring(s);
8838 // r = parse_string(d, 1);
8839 // scriptname = oldscriptname;
8840 // so a syntax error in an autoloaded file is reported as
8841 // `_files:40: parse error: …`, not against the frame that triggered the
8842 // load. zshrs runs the only parse of the raw body right here, so the
8843 // same window belongs around it — without it the diagnostic came out as
8844 // `g:40:` for `g(){ emulate ksh -c _files }`.
8845 //
8846 // `scriptname` is one process-wide cell where C's is a single-thread
8847 // global, and a worker-thread compile prints nothing anyway (`zwarning`,
8848 // utils.rs:150, routes its diagnostics to the log) — so only the shell
8849 // thread takes the window, and no worker can blank out a prefix the shell
8850 // thread is in the middle of writing.
8851 let swap_scriptname = !crate::worker::in_worker_thread();
8852 let old_scriptname = crate::ported::utils::scriptname_get(); // c:6257
8853 if swap_scriptname {
8854 crate::ported::utils::set_scriptname(Some(name.to_string())); // c:6263
8855 }
8856 let saved_errflag = crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed);
8857 crate::ported::utils::errflag.fetch_and(
8858 !crate::ported::utils::ERRFLAG_ERROR,
8859 std::sync::atomic::Ordering::Relaxed,
8860 );
8861 let prog = crate::ported::exec::parse_string(body, 1); // c:6329 `r = parse_string(d, 1);` — reset_lineno: errors count lines within the file
8862 let parse_failed = (crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed)
8863 & crate::ported::utils::ERRFLAG_ERROR)
8864 != 0
8865 || prog.is_none();
8866 crate::ported::utils::errflag.store(saved_errflag, std::sync::atomic::Ordering::Relaxed);
8867 if swap_scriptname {
8868 crate::ported::utils::set_scriptname(old_scriptname); // c:6265
8869 }
8870 let stripped = prog
8871 .map(|prog| {
8872 let original_len = prog.prog.len();
8873 // stripkshdef returns the input untouched when the prog
8874 // doesn't match the single-`function NAME` shape, and a
8875 // shorter (body-only) prog when it does. Compare the
8876 // wordcode length to detect the strip without owning the
8877 // post-strip Eprog (we only need the yes/no answer here).
8878 let prog_box = Box::new(prog);
8879 crate::ported::exec::stripkshdef(Some(prog_box), name)
8880 .map(|p| p.prog.len() != original_len)
8881 .unwrap_or(false)
8882 })
8883 .unwrap_or(false);
8884 let text = if stripped {
8885 body.to_string()
8886 } else {
8887 format!("{name}() {{\n{body}\n}}")
8888 };
8889 (text, parse_failed)
8890}
8891
8892// zsh_eval_context push/pop/sync relocated 2026-06-12 INTO doshfunc
8893// (src/ported/exec.rs) — its sole caller, and `zsh_eval_context` is
8894// that module's own static. The shell-visible mirror writes inline
8895// at the push site + the guard's Drop. No bridge indirection.
8896
8897impl ShellExecutor {
8898 /// Execute the trap body for a signal name from the REPL signal
8899 /// loop (bins/zshrs.rs CtrlC/CtrlD dispatch). Thin passthru to
8900 /// `traps_table` lookup + `execute_script` — kept as a method
8901 /// because the REPL loop owns `&mut ShellExecutor` and needs a
8902 /// single call point. The async signal-handler dispatch path
8903 /// goes through `crate::ported::signals::dotrap` instead.
8904 pub fn run_trap(&mut self, signal: &str) {
8905 let action = crate::ported::builtin::traps_table()
8906 .lock()
8907 .ok()
8908 .and_then(|t| t.get(signal).cloned());
8909 if let Some(body) = action {
8910 if !body.is_empty() {
8911 let _ = self.execute_string_without_exit_hooks(&body);
8912 }
8913 }
8914 }
8915}
8916
8917impl ShellExecutor {
8918 /// Expand glob pattern via canonical `glob_path` (port of
8919 /// `Src/glob.c::zglob`). Adds executor-side `current_command_glob_failed`
8920 /// cell so the dispatch layer skips the current command on NOMATCH +
8921 /// looks_like_glob instead of exiting the shell.
8922 pub fn expand_glob(&self, pattern: &str) -> Vec<String> {
8923 // c:Src/glob.c:1230-1233 — `if (unset(GLOBOPT) || !haswilds(ostr) ||
8924 // unset(EXECOPT)) { untokenize(ostr); return; }`: under NO_EXEC zglob
8925 // never globs, so no word (the command word included) can report
8926 // NOMATCH.
8927 if !crate::ported::zsh_h::isset(crate::ported::zsh_h::EXECOPT) {
8928 return vec![crate::ported::lex::untokenize(pattern)];
8929 }
8930 let errflag_before =
8931 crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed);
8932 let expanded = glob_path(pattern);
8933 if !expanded.is_empty() {
8934 // c:Src/glob.c:1871-1872 — `if (matchct) badcshglob |= 2;`
8935 // (at least one expansion on this command line worked).
8936 // Only real glob patterns count — C's zglob early-returns
8937 // before the matchct accounting for non-wild words, so
8938 // gate on haswilds like the failure path below. Consumed
8939 // per command by fusevm_bridge::consume_badcshglob.
8940 if crate::ported::zsh_h::isset(crate::ported::zsh_h::CSHNULLGLOB) {
8941 let mut pattern_tok = pattern.to_string();
8942 crate::ported::glob::tokenize(&mut pattern_tok);
8943 if crate::ported::pattern::haswilds(&pattern_tok) {
8944 crate::ported::glob::BADCSHGLOB
8945 .fetch_or(2, std::sync::atomic::Ordering::Relaxed);
8946 }
8947 }
8948 return expanded;
8949 }
8950 // c:Src/glob.c:1786-1788 — `if (errflag) { restore_globstate(saved);
8951 // return; }`. A qualifier-parse error returns from `zglob` outright,
8952 // so C never reaches the c:1873-1886 nullglob/nomatch dispatch below.
8953 // The port has to re-derive `gf_nullglob` from the pattern because
8954 // `glob_path` hands back only a `Vec` — and that SECOND qualifier
8955 // parse re-runs every diagnostic the first one already emitted. It is
8956 // normally invisible because `zerr` suppresses itself while
8957 // ERRFLAG_ERROR is set (c:Src/utils.c:175), but a subscript qualifier
8958 // runs the lexer (`getindex` → `parse_subscript` → `strinbeg` →
8959 // `hbegin`, c:Src/hist.c:1115 `errflag &= ~ERRFLAG_ERROR`), which
8960 // clears exactly that bit — so `*(N[1,])` printed `bad math
8961 // expression: empty string` twice. Bail out where C's `return` lands.
8962 if crate::ported::utils::errflag.load(std::sync::atomic::Ordering::Relaxed) != 0 {
8963 // A diagnostic THIS glob raised (`bad pattern: a[B`, a bad
8964 // qualifier) belongs to the same per-command carrier as the
8965 // c:1877 NOMATCH below. C runs an external command's prefork
8966 // expansion in the child forked at Src/exec.c:3719, so the
8967 // `zerr` there only costs that child (status 1) and the list
8968 // goes on; a builtin expands in the shell and the errflag aborts
8969 // the list. The dispatch layer makes exactly that split for
8970 // `current_command_glob_failed`; without it `/bin/echo a[B;
8971 // print after` aborted the list and the `find` shadow ran with
8972 // the raw word after printing the diagnostic.
8973 if errflag_before == 0 {
8974 self.current_command_glob_failed.set(true);
8975 }
8976 return Vec::new();
8977 }
8978 // c:1871-1872 — `if (matchct) badcshglob |= 2;`: files DID match and
8979 // only the `[first,last]` subscript left none (`*([100])`), so this
8980 // is not a failed expansion — the word simply expands to nothing.
8981 if crate::ported::glob::CURGLOBDATA
8982 .lock()
8983 .unwrap_or_else(|e| e.into_inner())
8984 .matchct
8985 > 0
8986 {
8987 if crate::ported::zsh_h::isset(crate::ported::zsh_h::CSHNULLGLOB) {
8988 crate::ported::glob::BADCSHGLOB.fetch_or(2, std::sync::atomic::Ordering::Relaxed);
8989 }
8990 return Vec::new();
8991 }
8992 // No matches. Mirror zsh's `setopt nullglob` / `nomatch`
8993 // dispatch (Src/glob.c:1873-1886) here because glob_path
8994 // returns an empty Vec without knowing executor state.
8995 // c:Src/glob.c:1567-1569 `gf_nullglob` per-glob — the `(N)`
8996 // qualifier acts like `setopt nullglob` for this expression
8997 // alone. parse_qualifiers detects the suffix `(...)` block;
8998 // the resulting `qualifiers.nullglob` mirrors C's gf_nullglob
8999 // carrier.
9000 let per_glob_nullglob = crate::ported::glob::parse_qualifiers(pattern)
9001 .1
9002 .map(|q| q.nullglob)
9003 .unwrap_or(false);
9004 let nullglob = opt_state_get("nullglob").unwrap_or(false) || per_glob_nullglob;
9005 if nullglob {
9006 // c:Src/glob.c:1888-1894 —
9007 // `else if (in_expandredir) {`
9008 // `/* if completing for redirection, we can't remove the`
9009 // ` pattern even if NULL_GLOB is in effect */`
9010 // `zerr("redirection failed (no match): %s", ostr);`
9011 // `zfree(matchbuf, 0);`
9012 // `restore_globstate(saved);`
9013 // `return;`
9014 // `}`
9015 // Reached ONLY when gf_nullglob is set (the `else if` chain at
9016 // c:1873 owns every other no-match case), which is exactly the
9017 // `> file(N)` shape: dropping the word would leave the
9018 // redirection with no target at all, so `echo > nope(N)` failed
9019 // with an empty filename in `no such file or directory:`.
9020 if crate::ported::glob::IN_EXPANDREDIR.load(std::sync::atomic::Ordering::SeqCst) != 0 {
9021 zerr(&format!(
9022 "redirection failed (no match): {}",
9023 crate::ported::lex::untokenize(pattern)
9024 )); // c:1891
9025 self.current_command_glob_failed.set(true);
9026 return Vec::new(); // c:1894
9027 }
9028 return Vec::new();
9029 }
9030 let nomatch = opt_state_get("nomatch").unwrap_or(true);
9031 // Use canonical `haswilds` (port of Src/pattern.c:4306-4376)
9032 // instead of the Rust-only `looks_like_glob`. C zsh's
9033 // `Src/glob.c:1876` NOMATCH branch fires whenever the input
9034 // tripped haswilds during the `zglob` entry check —
9035 // including patterns whose internal `(` / `)` form a group
9036 // or alternation but don't end with `)` (e.g. `abc(a)def`,
9037 // `(abc`). The previous `looks_like_glob` only caught
9038 // trailing-`(...)` qualifiers, leaving mid-word groups and
9039 // unclosed parens to fall through to the literal-passthrough
9040 // branch. #170 in docs/BUGS.md.
9041 //
9042 // haswilds scans TOKENIZED strings (C's zglob gets the
9043 // lexer-tokenized word at Src/glob.c:1230); this entry point
9044 // receives untokenized fast-path patterns, so tokenize a
9045 // local copy first — the same preparation C applies to
9046 // runtime-built strings (compcore.c:2231 tokenizes fignore
9047 // entries before its haswilds call). tokenize Bnull's
9048 // backslash-escaped metachars, so `\*` stays literal here
9049 // exactly as in C. Bug #627: plain multibyte text (`↔`)
9050 // passes through tokenize unchanged and matches no token.
9051 let mut pattern_tok = pattern.to_string();
9052 crate::ported::glob::tokenize(&mut pattern_tok); // c:Src/glob.c:3548
9053 let is_glob = crate::ported::pattern::haswilds(&pattern_tok);
9054 // c:Src/glob.c:1874-1875 — `if (isset(CSHNULLGLOB)) {
9055 // badcshglob |= 1; }` — the else-if chain means neither the
9056 // NOMATCH error nor the literal passthrough runs: the failed
9057 // word is silently DROPPED here, and the per-command boundary
9058 // (fusevm_bridge::consume_badcshglob, Src/subst.c:505-507)
9059 // emits the csh-style `no match` iff NO glob on the line
9060 // matched.
9061 if is_glob && crate::ported::zsh_h::isset(crate::ported::zsh_h::CSHNULLGLOB) {
9062 crate::ported::glob::BADCSHGLOB.fetch_or(1, std::sync::atomic::Ordering::Relaxed);
9063 return Vec::new();
9064 }
9065 if nomatch && is_glob {
9066 // c:Src/glob.c:1876-1880 — `else if (isset(NOMATCH)) {`
9067 // `zerr("no matches found: %s", ostr);`
9068 // `zfree(matchbuf, 0);`
9069 // `restore_globstate(saved);`
9070 // `return;`
9071 // `}`
9072 // C aborts via ERRFLAG_ERROR set by zerr() at c:Src/utils.c
9073 // and the matchbuf/state cleanup. The Rust port mirrors
9074 // both: zerr() in utils.rs sets ERRFLAG_ERROR via
9075 // `errflag.fetch_or(ERRFLAG_ERROR, ...)` already; we then
9076 // re-set explicitly (defensive — historically this line
9077 // had `fetch_and(!ERRFLAG_ERROR)` which CLEARED the flag
9078 // immediately after zerr, making `echo /never/*` print
9079 // the literal and exit 0 instead of erroring like zsh —
9080 // parity bug #13).
9081 // c:1877 `zerr("no matches found: %s", ostr);` — `ostr` is
9082 // the TOKENIZED word, and zerrmsg's `%s` arm renders it
9083 // through `nicezputs` → `sb_niceformat`, which calls
9084 // `untokenize(ums)` (Src/utils.c). Without that step the
9085 // token bytes are dropped by the terminal and the message
9086 // reads `no matches found: /tmp/nope_.txt`.
9087 zerr(&format!(
9088 "no matches found: {}",
9089 crate::ported::lex::untokenize(pattern)
9090 )); // c:1877
9091 self.current_command_glob_failed.set(true);
9092 // c:Src/glob.c:1876-1880 — zerr sets ERRFLAG_ERROR and
9093 // glob_failed cell carries the signal. The ERRFLAG_ERROR
9094 // clear (so subsequent sublists run) now lives at the
9095 // dispatcher's post-command-boundary at
9096 // fusevm_bridge.rs:299 where current_command_glob_failed
9097 // is consumed — matches C's execlist behavior of clearing
9098 // command-error errflag between sublists.
9099 return Vec::new(); // c:1880 return
9100 }
9101 // Pattern has no glob meta — pass through literally.
9102 // c:Src/glob.c:1882-1886 — `/* treat as an ordinary string */
9103 // untokenize(matchptr->name = dupstring(ostr));`. The word
9104 // arrives here in LEXER-TOKENIZED form (c:1221 `ostr =
9105 // getdata(np)`), so the literal fallback MUST untokenize or the
9106 // raw token bytes reach stdout: `unsetopt nomatch; echo
9107 // /tmp/nope_*.txt` printed `/tmp/nope_\u{e187}.txt`.
9108 vec![crate::ported::lex::untokenize(pattern)]
9109 }
9110 /// True iff the literal `pattern` actually contains a glob metachar
9111 /// in a position that would have triggered globbing. Used to avoid
9112 /// spurious "no matches" errors when expand_glob is called on a
9113 /// plain path that happened to route through this code (e.g. some
9114 /// fast paths bridge unconditionally).
9115 pub(crate) fn looks_like_glob(pattern: &str) -> bool {
9116 // A trailing `(qualifier)` is itself a glob trigger — e.g.
9117 // `path(L+10)` should be treated as a glob even when the
9118 // body has no `*`/`?`/`[...]`.
9119 let has_qual_suffix = if let Some(open) = pattern.rfind('(') {
9120 pattern.ends_with(')') && open + 1 < pattern.len() - 1
9121 } else {
9122 false
9123 };
9124 // Strip trailing `(...)` qualifier so we test the pattern body.
9125 let body = if let Some(open) = pattern.rfind('(') {
9126 if pattern.ends_with(')') {
9127 &pattern[..open]
9128 } else {
9129 pattern
9130 }
9131 } else {
9132 pattern
9133 };
9134 // Walk character-by-character so escaped metachars (`\*`, `\?`,
9135 // `\[`) are NOT counted as glob triggers. zsh: `echo \*` prints
9136 // a literal `*`; without the unescaped check, looks_like_glob
9137 // returned true on the bare `*` and the runtime glob expansion
9138 // aborted with NOMATCH.
9139 let chars: Vec<char> = body.chars().collect();
9140 let mut i = 0;
9141 let mut has_unescaped_star = false;
9142 let mut has_unescaped_question = false;
9143 let mut has_unescaped_bracket_open: Option<usize> = None;
9144 while i < chars.len() {
9145 let c = chars[i];
9146 if c == '\\' && i + 1 < chars.len() {
9147 // Escaped char — skip both.
9148 i += 2;
9149 continue;
9150 }
9151 match c {
9152 '*' => has_unescaped_star = true,
9153 '?' => has_unescaped_question = true,
9154 '[' if has_unescaped_bracket_open.is_none() => {
9155 has_unescaped_bracket_open = Some(i);
9156 }
9157 _ => {}
9158 }
9159 i += 1;
9160 }
9161 // `[` only counts when there's a matching `]` after it.
9162 let has_bracket_class = has_unescaped_bracket_open
9163 .map(|i| body[i + 1..].contains(']'))
9164 .unwrap_or(false);
9165 // `<N-M>` numeric range glob is also a trigger — match shape
9166 // `<` + optional digits + `-` + optional digits + `>` outside
9167 // any bracket expression.
9168 let has_numeric_range =
9169 body.contains('<') && body.contains('>') && !extract_numeric_ranges(body).is_empty();
9170 has_unescaped_star
9171 || has_unescaped_question
9172 || has_bracket_class
9173 || has_qual_suffix
9174 || has_numeric_range
9175 }
9176}
9177
9178impl ShellExecutor {
9179 pub(crate) fn copy_dir_recursive(src: &Path, dest: &Path) -> io::Result<()> {
9180 if !dest.exists() {
9181 fs::create_dir_all(dest)?;
9182 }
9183 for entry in fs::read_dir(src)? {
9184 let entry = entry?;
9185 let file_type = entry.file_type()?;
9186 let src_path = entry.path();
9187 let dest_path = dest.join(entry.file_name());
9188
9189 if file_type.is_dir() {
9190 Self::copy_dir_recursive(&src_path, &dest_path)?;
9191 } else {
9192 fs::copy(&src_path, &dest_path)?;
9193 }
9194 }
9195 Ok(())
9196 }
9197}
9198
9199// Magic-assoc scan-by-name aggregator. C's per-table getfn/scanfn
9200// pointers in paramdef[] (Src/Modules/parameter.c:825+) handle this
9201// indirectly via paramtab dispatch; this Rust-only helper exposes a
9202// single `partab_get` / `partab_scan_keys` entry that the bridge
9203// uses for name → keys lookup.
9204use std::cell::RefCell;
9205thread_local! {
9206 static SCAN_KEYS: RefCell<Vec<String>> = const { RefCell::new(Vec::new()) };
9207}
9208
9209/// Lookup helper for `${name[key]}` magic-assoc reads — dispatches
9210/// through canonical `PARTAB` (Src/Modules/parameter.c:2235 ports).
9211/// Returns `None` if name isn't a known magic-assoc.
9212/// Module parameters that have actually been touched this session.
9213///
9214/// zshrs-original bookkeeping for a C behavior that falls out of the module
9215/// system there. zsh registers `zsh/parameter`'s params (`aliases`,
9216/// `commands`, `functions`, …) as PM_AUTOLOAD stubs in `realparamtab`;
9217/// touching ONE of them materializes only that name — its siblings stay
9218/// stubs even though the module is now loaded. `paramtypestr`
9219/// (Src/Modules/parameter.c:49-50) reports a PM_AUTOLOAD node as
9220/// "undefined", which is what an enumeration of `$parameters` shows.
9221/// zshrs seeds all of them eagerly (init_partab_params), so without this
9222/// set every one reported its real type and `${(@k)parameters[(R)a*]}`
9223/// matched 56 names against zsh's 18 — putting them in the wrong
9224/// `_parameters -g` bucket (`unset <TAB>`: 418 entries vs zsh's 496).
9225static MATERIALIZED_MODULE_PARAMS: std::sync::OnceLock<Mutex<HashSet<String>>> =
9226 std::sync::OnceLock::new();
9227
9228/// Record that `name` was read/written, so `$parameters` stops reporting it
9229/// as an unmaterialized autoload stub. See [`MATERIALIZED_MODULE_PARAMS`].
9230pub fn mark_module_param_used(name: &str) {
9231 let set = MATERIALIZED_MODULE_PARAMS.get_or_init(|| Mutex::new(HashSet::new()));
9232 let first_touch = {
9233 let mut g = set.lock();
9234 // Drop the guard before the module load below — `boot_` runs shell
9235 // code (setsparam/setiparam) that can re-enter this function.
9236 g.insert(name.to_string())
9237 };
9238 if first_touch {
9239 materialize_module_param(name);
9240 }
9241}
9242
9243/// The side effect C's `loadparamnode` (`Src/params.c:563-585`) has beyond
9244/// clearing PM_AUTOLOAD: `(void)ensurefeature(mn, "p:", nam)`
9245/// (`Src/module.c:3419-3432`) actually LOADS the owning module, running its
9246/// `setup_`/`boot_`. `zsh/watch`'s `boot_` (`Src/Modules/watch.c:750-753`)
9247/// seeds `WATCHFMT`/`LOGCHECK` when absent, so in zsh
9248/// `${parameters[watch]}` leaves `${parameters[LOGCHECK]}` == "integer".
9249///
9250/// !!! WARNING: RUST-ONLY HELPER !!!
9251/// C has no separate function: `loadparamnode` calls `ensurefeature`
9252/// inline. zshrs models PM_AUTOLOAD as a side-set rather than a node flag
9253/// (see [`MATERIALIZED_MODULE_PARAMS`]), so the load side effect needs its
9254/// own hook off the marking point. The `try_lock` and the re-entrancy guard
9255/// are also Rust-only: C serialises on `queue_signals`, whereas zshrs's
9256/// `MODULESTAB` is a real mutex that several callers of
9257/// `mark_module_param_used` already hold.
9258fn materialize_module_param(name: &str) {
9259 use std::cell::Cell;
9260 thread_local! {
9261 static LOADING: Cell<bool> = const { Cell::new(false) };
9262 }
9263 // c:Src/params.c:566 — only PM_AUTOLOAD stubs carry `pm->u.str` (the
9264 // owning module name); anything else falls straight through.
9265 let Some((_, modname)) = AUTOLOAD_PARAMS.iter().find(|(p, _)| *p == name) else {
9266 return;
9267 };
9268 if LOADING.with(|f| f.get()) {
9269 return;
9270 }
9271 // The whole load chain wants `&mut modulestab`. Every other caller takes
9272 // the same lock for a moment; if one of them is mid-flight (or is our own
9273 // caller), skip rather than deadlock — the mark itself already landed.
9274 let Ok(mut tab) = crate::ported::module::MODULESTAB.try_lock() else {
9275 return;
9276 };
9277 // c:Src/module.c:2352 — require_module short-circuits on an already
9278 // booted module, but check first so the common case never pays for the
9279 // find_module/alias walk.
9280 if tab
9281 .modules
9282 .get(*modname)
9283 .is_some_and(|m| (m.node.flags & crate::ported::zsh_h::MOD_INIT_B) != 0)
9284 {
9285 return;
9286 }
9287 LOADING.with(|f| f.set(true));
9288 // c:3419-3432 ensurefeature(mn, "p:", nam) — `silent` is 0 in C, but a
9289 // failure here is not user-visible in the autoload path (c:571-580 only
9290 // errors when the parameter is still undefined afterwards), and zshrs's
9291 // require_module warns on any module it cannot static-link.
9292 let _ = crate::ported::module::ensurefeature(&mut tab, modname, "p:", Some(name)); // c:3419
9293 LOADING.with(|f| f.set(false));
9294}
9295
9296/// Undo [`mark_module_param_used`] — the `pd->pm = NULL` half of C's
9297/// `deleteparamdef` (`Src/module.c:1128-1179`), reached from
9298/// `setparamdefs` (`c:1176-1180`) whenever a module feature is turned OFF
9299/// (`zmodload -F zsh/parameter -p:functions`, or a single-feature load
9300/// leaving the siblings disabled).
9301///
9302/// !!! WARNING: RUST-ONLY HELPER !!!
9303/// C has no separate function: `deleteparamdef` removes the node from
9304/// `paramtab` outright. zshrs seeds every magic parameter eagerly and models
9305/// PM_AUTOLOAD as the [`MATERIALIZED_MODULE_PARAMS`] side-set, so "removed"
9306/// is "back to being an unmaterialized stub".
9307pub fn unmark_module_param_used(name: &str) {
9308 let set = MATERIALIZED_MODULE_PARAMS.get_or_init(|| Mutex::new(HashSet::new()));
9309 set.lock().remove(name);
9310}
9311
9312/// True when `name` is still an untouched module-parameter stub — i.e. zsh
9313/// would report it as PM_AUTOLOAD ("undefined") when enumerating
9314/// `$parameters`. See [`MATERIALIZED_MODULE_PARAMS`].
9315pub fn module_param_is_autoload_stub(name: &str) -> bool {
9316 if !AUTOLOAD_PARAMS.iter().any(|(p, _)| *p == name) {
9317 return false;
9318 }
9319 MATERIALIZED_MODULE_PARAMS
9320 .get_or_init(|| Mutex::new(HashSet::new()))
9321 .lock()
9322 .contains(name)
9323 .eq(&false)
9324}
9325
9326/// True when the magic special parameter `name` (a `partab[]` row from
9327/// `Src/Modules/parameter.c:2235-2298` — `options`, `functions`,
9328/// `commands`, `parameters`, `dirstack`, …) is currently SHADOWED by a
9329/// plain user parameter of the same name.
9330///
9331/// Rust-only helper with no C counterpart BY CONSTRUCTION: C zsh keeps
9332/// specials and user parameters in the ONE `paramtab` hash, so ordinary
9333/// hash lookup already implements this. `createparam`
9334/// (c:Src/params.c:1090-1115) finds the existing special node, stashes
9335/// it in `pm->old`, and inserts a fresh plain node under the same key;
9336/// every later `getvalue` / `fetchvalue` / `gethashparam` therefore hits
9337/// the plain node and the special's `gsu` callbacks are unreachable
9338/// until `endparamscope` restores `pm->old`. zshrs instead keeps the
9339/// magic rows in SEPARATE static tables (`PARTAB`, `PARTAB_ARRAY`) and
9340/// matches them BY NAME, so a name-only match resurrected the special
9341/// even while a local shadowed it. This predicate re-imposes C's
9342/// shadowing on the split-table layout; it is architecture glue, not a
9343/// port.
9344///
9345/// `init_partab_params` (below) seeds every magic row into `paramtab`
9346/// with `PM_SPECIAL` (C's `SPECIALPMDEF` macro), and `local`/`typeset`
9347/// replaces that node with one carrying no `PM_SPECIAL`, so the live
9348/// node's `PM_SPECIAL` bit is exactly C's "is the special still the
9349/// visible binding" test.
9350///
9351/// Returns false when a MODULE-GATED row (`sysparams`, `errnos`,
9352/// `mapfile`, `langinfo`) has no `paramtab` node and its module is not
9353/// loaded: those are seeded on
9354/// demand by `seed_partab_param`, and the PARTAB walk must still answer
9355/// for them (their own `module` gate decides).
9356///
9357/// For every other magic row an ABSENT node means the binding is gone —
9358/// `unset` removed it. C reaches the same answer through one gate:
9359/// `Src/params.c:2264-2266` `if (!pm || ((pm->node.flags & PM_UNSET) &&
9360/// !(pm->node.flags & PM_DECLARED))) return NULL;` in `fetchvalue`, the
9361/// single choke point every `${X}` / `${#X}` / `${(k)X}` / `${(t)X}` /
9362/// `${X[k]}` read passes through. Both of its arms show up here:
9363/// * `!pm` — `unset functions` at a point where the name is still the
9364/// `PM_AUTOLOAD` stub (`Src/module.c:1218-1223`) finds a PLAIN
9365/// `PM_SCALAR` node with neither `PM_SPECIAL` nor `PM_READONLY`, so
9366/// `unsetparam_pm`'s c:3851-3852 keep-the-node test
9367/// (`(flags & (PM_SPECIAL|PM_REMOVABLE)) == PM_SPECIAL`) is false and
9368/// c:3874 `paramtab->removenode` drops it outright.
9369/// * `PM_UNSET && !PM_DECLARED` — once the special HAS been
9370/// materialized, c:3851-3852 keeps the node and `stdunsetfn`
9371/// (c:3939) marking `PM_UNSET` is the entire record of the unset;
9372/// `setpmfunctions(pm, NULL)` returns immediately on `if (!ht)
9373/// return` (`Src/Modules/parameter.c:361-362`), so `shfunctab` — the
9374/// real table behind the row — survives untouched and `ff` still runs.
9375///
9376/// Both arms mean the same thing for a split-table port: the magic row
9377/// is no longer the visible binding for this name, exactly as when a
9378/// `local` shadows it. Answering that one question here is what lets
9379/// every PARTAB dispatch site keep a single guard.
9380///
9381/// Symptom this fixes: git's `git-completion.bash`
9382/// `__git_resolve_builtins` does `local options; eval
9383/// "options=\${$var-}"` (git 2.55.0
9384/// share/zsh/site-functions/git-completion.bash:500-501). `$options`
9385/// read back the zsh/parameter option table (`off on off …`) instead of
9386/// the local, so `git checkout --<TAB>` completed nothing.
9387pub fn magic_special_shadowed(name: &str) -> bool {
9388 // Only `partab[]` names can be shadowed in this sense — an ordinary
9389 // user assoc / array has no special behind it, and its own paramtab
9390 // node legitimately carries no PM_SPECIAL.
9391 if !PARTAB.iter().any(|e| e.name == name) && !PARTAB_ARRAY.iter().any(|e| e.name == name) {
9392 return false;
9393 }
9394 // A module-gated row (`errnos`, `sysparams`, `mapfile`, `langinfo`)
9395 // gets its paramtab node from the owning module's setparamdefs
9396 // (Src/module.c:1169-1196), so while that module is loaded an absent
9397 // node means its `p:` feature is off: deleteparamdef (c:1190) removed
9398 // it, or the module was loaded with other features only. Read the
9399 // module state before taking the paramtab lock.
9400 let gated_loaded = module_gated_partab_module(name)
9401 .map(|m| crate::ported::module::MODULESTAB.lock().unwrap().is_loaded(m));
9402 crate::ported::params::paramtab()
9403 .read()
9404 .map_or(false, |tab| {
9405 let Some(pm) = tab.get(name) else {
9406 // c:Src/params.c:2264 `if (!pm ...) return NULL` —
9407 // `unset` removed the node (see the doc comment). Only
9408 // a valid reading once the rows have been seeded. A
9409 // gated row whose module is not loaded is left to the
9410 // module gate in partab_get / partab_array_get.
9411 return match gated_loaded {
9412 Some(loaded) => loaded,
9413 None => PARTAB_SEEDED.load(std::sync::atomic::Ordering::Acquire),
9414 };
9415 };
9416 {
9417 // c:Src/params.c:2264-2266 — `(pm->node.flags & PM_UNSET)
9418 // && !(pm->node.flags & PM_DECLARED)`: the materialized
9419 // special was unset and the node kept (c:3851-3852).
9420 let f = pm.node.flags as u32;
9421 if (f & crate::ported::zsh_h::PM_UNSET) != 0
9422 && (f & crate::ported::zsh_h::PM_DECLARED) == 0
9423 {
9424 return true;
9425 }
9426 }
9427 {
9428 // c:Src/module.c:1029-1052 checkaddparam — `if (pm->level ||
9429 // !(pm->node.flags & PM_AUTOLOAD))` is C's OWN test for "is
9430 // this node a blocker or the module's own placeholder": a
9431 // GLOBAL PM_AUTOLOAD node is the autoload STUB
9432 // `add_autoparam` planted (c:1222-1223 `setsparam(pnam,
9433 // module); pm->node.flags |= PM_AUTOLOAD` — its VALUE is the
9434 // owning module's name), and C replaces it with the real
9435 // special via `unsetparam_pm` (c:1051) + `createspecialhash`
9436 // (c:1068) the moment the module loads. Reading the name is
9437 // what triggers that: c:Src/params.c:563-585 loadparamnode
9438 // runs `ensurefeature(mn, "p:", nam)` and re-fetches the node.
9439 // So a stub NEVER makes the special unreachable — treating it
9440 // as a shadow left `${options}` reading the stub's own scalar
9441 // value ("zsh/parameter") and killed every magic-assoc read
9442 // for that name for the rest of the session. A LOCAL stub
9443 // (pm->level != 0) still blocks, exactly as c:1032 says.
9444 let f = pm.node.flags as u32;
9445 if pm.level == 0 && (f & crate::ported::zsh_h::PM_AUTOLOAD) != 0 {
9446 return false;
9447 }
9448 (f & crate::ported::zsh_h::PM_SPECIAL) == 0
9449 }
9450 })
9451}
9452
9453/// [`magic_special_shadowed`] AND the shadow is not itself a hash.
9454///
9455/// "No longer the magic row" is not the same as "no longer a hash". C
9456/// decides the subscript dispatch — and every whole-map read — from the
9457/// TYPE of the node paramtab actually returned:
9458/// c:Src/params.c:2270 `if (PM_TYPE(pm->node.flags) & (PM_ARRAY|PM_HASHED))`
9459/// c:Src/params.c:1597-1606 getarg —
9460/// if (ishash) {
9461/// HashTable ht = v->pm->gsu.h->getfn(v->pm);
9462/// ...
9463/// if (!(v->pm = (Param) ht->getnode(ht, s))) ...
9464/// So a shadow that IS itself PM_HASHED (`local -A commands`, or
9465/// `local -A +h commands` as in Completion/Zsh/Type/_command_names:70)
9466/// owns a table of its own and must be SERVED FROM IT, while a scalar or
9467/// array shadow (`local options`, `local -a options`) must fall through
9468/// to the non-hash arms.
9469///
9470/// Bailing on every shadowed magic name made the hashed shadow invisible
9471/// to each reader in turn: the keyed read (`$commands[fake]`), then
9472/// `${#commands}`, `${(k)commands}`, `${(kv)commands}` and
9473/// `${commands[(I)pat]}`. The values were in `paramtab_hashed_storage`
9474/// the whole time — `${commands[@]}` found them — only these guards
9475/// refused to look. Every dispatch site that means "the magic row is not
9476/// the visible binding, and there is no hash behind it either" asks
9477/// through here so the qualification cannot drift between them.
9478pub fn magic_special_shadowed_by_nonhash(name: &str) -> bool {
9479 magic_special_shadowed(name)
9480 && crate::ported::params::paramtab()
9481 .read()
9482 .ok()
9483 .and_then(|t| t.get(name).map(|pm| pm.node.flags as u32))
9484 // c:2270 — PM_TYPE decides; absent node ⇒ nothing to serve; c:2264-2266 — nor
9485 // does an unset, undeclared node (`zmodload zsh/parameter; unset functions`).
9486 .map_or(true, |f| {
9487 use crate::ported::zsh_h::{PM_DECLARED, PM_TYPE, PM_UNSET};
9488 PM_TYPE(f) != PM_HASHED || (f & (PM_UNSET | PM_DECLARED)) == PM_UNSET
9489 })
9490}
9491
9492pub fn partab_get(name: &str, key: &str) -> Option<String> {
9493 // C's paramtab lookup would already have found the shadowing local;
9494 // the split-table port needs the explicit check.
9495 if magic_special_shadowed(name) {
9496 return None;
9497 }
9498 mark_module_param_used(name);
9499 // c:Src/Modules/system.c:902,904 — `sysparams` and `errnos` are
9500 // bound by zsh/system's boot_/setup_ chain. Same for `mapfile`
9501 // from zsh/mapfile. Without explicit `zmodload`, these names
9502 // are unset in zsh; gate the PARTAB dispatch here so they
9503 // resolve via the empty-fallback path (matching ${sysparams[k]:-x}
9504 // taking the default). Bug #69 in docs/BUGS.md.
9505 if let Some(modname) = module_gated_partab_module(name) {
9506 if !crate::ported::module::MODULESTAB
9507 .lock()
9508 .unwrap()
9509 .is_loaded(modname)
9510 {
9511 return None;
9512 }
9513 }
9514 for entry in PARTAB.iter() {
9515 if entry.name == name {
9516 // c:Src/Modules/parameter.c — a row getfn (`getpmfunction`,
9517 // `getpmcommand`, `getpmoption`, …) answers a missing key with a
9518 // node carrying `PM_UNSET` (e.g. getfunction, parameter.c:449 `pm->node.flags |=
9519 // (PM_UNSET|PM_SPECIAL)`), which `IS_UNSET_VALUE` (params.c:479)
9520 // and `issetvar` (params.c:764-765) read as unset. Keep that
9521 // distinction: dropping the flag turned a miss into a set empty
9522 // value, so `[[ -v functions[nosuch] ]]` was true.
9523 return (entry.getfn)(std::ptr::null_mut(), key)
9524 .filter(|p| p.node.flags as u32 & crate::ported::zsh_h::PM_UNSET == 0)
9525 .and_then(|p| p.u_str);
9526 }
9527 }
9528 None
9529}
9530
9531/// Returns the owning module name for partab entries that are
9532/// bound by an explicit zmodload — `sysparams`/`errnos` from
9533/// zsh/system, `mapfile` from zsh/mapfile. Other partab entries
9534/// (aliases/commands/functions/...) are part of zsh/main and
9535/// always available.
9536fn module_gated_partab_module(name: &str) -> Option<&'static str> {
9537 match name {
9538 "sysparams" | "errnos" => Some("zsh/system"),
9539 "mapfile" => Some("zsh/mapfile"),
9540 "langinfo" => Some("zsh/langinfo"),
9541 ".zle.esc" | ".zle.sgr" => Some("zsh/hlgroup"),
9542 ".sh.match" => Some("zsh/ksh93"),
9543 _ => None,
9544 }
9545}
9546
9547/// Publish a value into a read-only special from shell-INTERNAL code.
9548///
9549/// C binds these params to C variables through a gsu vtable —
9550/// `compvarscalar_gsu` for `$QIPREFIX`/`$QISUFFIX`
9551/// (Src/Zle/complete.c:1308-1324), `keymap_gsu` for `$KEYMAP`
9552/// (Src/Zle/zle_params.c:151) — and the shell's own writes go straight
9553/// to that variable. PM_READONLY is only consulted on the ASSIGNMENT
9554/// path (`assignsparam`, Src/params.c), so the bit stops a user's
9555/// `QIPREFIX=x` without ever standing in the way of the completion
9556/// machinery's own publish.
9557///
9558/// zshrs keeps the value in the param itself, so the internal publish
9559/// has to step around the same gate explicitly: drop PM_READONLY,
9560/// assign through the canonical path, put the bit back.
9561pub fn set_readonly_special(name: &str, value: &str) {
9562 use crate::ported::zsh_h::PM_READONLY;
9563 let was_readonly = crate::ported::params::paramtab()
9564 .write()
9565 .ok()
9566 .and_then(|mut tab| {
9567 tab.get_mut(name).map(|pm| {
9568 let ro = (pm.node.flags & PM_READONLY as i32) != 0;
9569 pm.node.flags &= !(PM_READONLY as i32);
9570 ro
9571 })
9572 })
9573 .unwrap_or(false);
9574 let _ = crate::ported::params::setsparam(name, value);
9575 if was_readonly {
9576 if let Ok(mut tab) = crate::ported::params::paramtab().write() {
9577 if let Some(pm) = tab.get_mut(name) {
9578 pm.node.flags |= PM_READONLY as i32;
9579 }
9580 }
9581 }
9582}
9583
9584/// PM_ARRAY lookup for `${name}` / `${name[N]}` — walks
9585/// PARTAB_ARRAY and dispatches the whole-array getfn (Src/Modules/
9586/// parameter.c:2239-2291 ports). Returns `None` if name isn't a
9587/// known PM_ARRAY magic-assoc.
9588pub fn partab_array_get(name: &str) -> Option<Vec<String>> {
9589 // c:Src/params.c:1090-1115 createparam — see `partab_get`.
9590 if magic_special_shadowed(name) {
9591 return None;
9592 }
9593 mark_module_param_used(name);
9594 // Bug #69 — gate module-bound PARTAB names on the owning
9595 // module's MOD_LINKED && !MOD_UNLOAD state.
9596 if let Some(modname) = module_gated_partab_module(name) {
9597 if !crate::ported::module::MODULESTAB
9598 .lock()
9599 .unwrap()
9600 .is_loaded(modname)
9601 {
9602 return None;
9603 }
9604 }
9605 for entry in PARTAB_ARRAY.iter() {
9606 if entry.name == name {
9607 return Some((entry.getfn)(std::ptr::null_mut()));
9608 }
9609 }
9610 None
9611}
9612
9613/// Scan helper for `${(k)name}` — enumerates keys via canonical
9614/// scanfn, collected into Vec via SCAN_KEYS thread-local.
9615pub fn partab_scan_keys(name: &str) -> Option<Vec<String>> {
9616 // c:Src/params.c:1090-1115 createparam — see `partab_get`.
9617 if magic_special_shadowed(name) {
9618 return None;
9619 }
9620 mark_module_param_used(name);
9621 // Bug #69 — gate module-bound PARTAB names on the owning
9622 // module's MOD_LINKED && !MOD_UNLOAD state.
9623 if let Some(modname) = module_gated_partab_module(name) {
9624 if !crate::ported::module::MODULESTAB
9625 .lock()
9626 .unwrap()
9627 .is_loaded(modname)
9628 {
9629 return None;
9630 }
9631 }
9632 for entry in PARTAB.iter() {
9633 if entry.name == name {
9634 SCAN_KEYS.with(|k| k.borrow_mut().clear());
9635 // c:Src/Modules/parameter.c — a param-table ScanFunc receives
9636 // `&pm.node` of a fully populated `struct param`; the Rust side
9637 // models that as `ParamScanFunc = fn(¶m, i32)`.
9638 fn cb(pm: &crate::ported::zsh_h::param, _flags: i32) {
9639 SCAN_KEYS.with(|k| k.borrow_mut().push(pm.node.nam.clone()));
9640 }
9641 // c:Src/params.c:3138 — `paramvalarr(…, SCANPM_WANTKEYS)`: keys
9642 // only, so a scanfn need not materialize the value side.
9643 (entry.scanfn)(
9644 std::ptr::null_mut(),
9645 Some(cb),
9646 crate::ported::zsh_h::SCANPM_WANTKEYS as i32,
9647 );
9648 return Some(SCAN_KEYS.with(|k| k.borrow().clone()));
9649 }
9650 }
9651 None
9652}
9653/// Populate paramtab with PM_SPECIAL placeholder Params for every
9654/// PARTAB / PARTAB_ARRAY entry — Rust-only init helper, no direct
9655/// C counterpart (closest is `handlefeatures` walking `partab[]`
9656/// in `Src/Modules/parameter.c:2341` boot/enables chain).
9657///
9658/// Each magic-assoc name gets a Param with `entry.flags | PM_SPECIAL`.
9659/// Value reads still route through `partab_get` / `partab_array_get`;
9660/// having the Param in paramtab makes `paramtab.get(name)` return
9661/// Some(Param) so `${+name}` / `${(t)name}` / `typeset -p name` see
9662/// the entry. Without this, those reads returned empty for every
9663/// magic-assoc (aliases, commands, functions, etc.).
9664///
9665/// Called from ShellExecutor::new() since zshrs's bin entry skips
9666/// the canonical module-bootstrap chain.
9667pub fn init_partab_params() {
9668 use crate::ported::modules::parameter::{PARTAB, PARTAB_ARRAY};
9669 use crate::ported::zsh_h::{
9670 hashnode, param, Param, PM_HIDE, PM_HIDEVAL, PM_READONLY, PM_SPECIAL,
9671 };
9672 let mut tab = match paramtab().write() {
9673 Ok(t) => t,
9674 Err(_) => return,
9675 };
9676 // c:Src/zsh.h SPECIALPMDEF macro: `flags | PM_SPECIAL | PM_HIDE |
9677 // PM_HIDEVAL`. All magic-assoc/array params get HIDE+HIDEVAL added
9678 // by the macro itself.
9679 //
9680 // PM_READONLY is preserved on the stub for params that legitimately
9681 // need user-write protection (reswords, dis_reswords, patchars,
9682 // dis_patchars — all compute via getfn and have no legitimate
9683 // internal-write path). Other specials that DO have internal-write
9684 // paths (e.g. funcstack from function-call tracking) get the bit
9685 // stripped so the runtime can mutate their u_arr. Bug #374.
9686 // `parameters` is computed entirely by getpmparameter and has no
9687 // internal-write path either (zsh: PM_READONLY_SPECIAL, c:2287), so it
9688 // keeps the bit — `${parameters[parameters]}` reads
9689 // `association-readonly-hide-hideval-special` in zsh.
9690 // The remaining names below are the rest of C's PM_READONLY_SPECIAL
9691 // rows whose `partab[]` entry has a NULL gsu (c:2237/2243/2255/2265/
9692 // :2272/2276-2280/2284/2296-2298 + Src/Zle/zleparameter.c:133): they
9693 // are computed purely by their getfn/scanfn, so there is nothing for
9694 // the runtime to write and the bit is safe to keep. Without them
9695 // `${(t)builtins}` and friends reported
9696 // `association-hide-hideval-special` where zsh reports
9697 // `association-readonly-hide-hideval-special`.
9698 let user_protected: &[&str] = &[
9699 "parameters",
9700 "reswords",
9701 "dis_reswords",
9702 "patchars",
9703 "dis_patchars",
9704 "historywords",
9705 "errnos",
9706 "keymaps",
9707 "builtins", // c:2237
9708 "dis_builtins", // c:2243
9709 "functions_source", // c:2265
9710 "dis_functions_source", // c:2247
9711 "history", // c:2272
9712 "jobdirs", // c:2276
9713 "jobstates", // c:2278
9714 "jobtexts", // c:2280
9715 "modules", // c:2284
9716 "userdirs", // c:2296
9717 "usergroups", // c:2297
9718 "widgets", // c:Src/Zle/zleparameter.c:133
9719 // c:2279-2280 — `SPECIALPMDEF("funcstack", PM_ARRAY|
9720 // PM_READONLY_SPECIAL, &funcstack_gsu, NULL, NULL)`. The bit was
9721 // being stripped here on the theory that the runtime writes
9722 // `funcstack`'s `u_arr`; it does not — `funcstackgetfn`
9723 // (PARTAB_ARRAY, parameter.rs:4726-4732) computes the value from
9724 // the `FUNCSTACK` global on every read and the row's `setfn` is
9725 // `None`, so there is nothing to protect against. Without the bit
9726 // `${(t)funcstack}` read `array-hide-hideval-special` where zsh
9727 // reads `array-readonly-hide-hideval-special`, which put it in the
9728 // wrong `_parameters -g '^*(readonly|association)*'` bucket and
9729 // added one candidate zsh does not offer.
9730 "funcstack", // c:2279
9731 // Same argument as `funcstack` above, for the three sibling trace
9732 // arrays: `SPECIALPMDEF(..., PM_ARRAY|PM_READONLY_SPECIAL, ...)` in C
9733 // and `setfn: None` in PARTAB_ARRAY (parameter.rs:4717/4725/4741), so
9734 // they are getfn-computed with no internal-write path to protect.
9735 "funcfiletrace", // c:2275
9736 "funcsourcetrace", // c:2277
9737 "functrace", // c:2285
9738 // c:Src/Modules/termcap.c:312 / Src/Modules/terminfo.c:305 —
9739 // `SPECIALPMDEF("termcap", PM_READONLY, NULL, gettermcap, scantermcap)`
9740 // and the terminfo twin: NULL gsu, value produced entirely by the
9741 // getnode/scan fns, so the readonly bit has nothing to fight.
9742 "termcap", // c:Src/Modules/termcap.c:312
9743 "terminfo", // c:Src/Modules/terminfo.c:305
9744 // c:Src/Builtins/sched.c:382 — `SPECIALPMDEF(
9745 // "zsh_scheduled_events", PM_ARRAY|PM_READONLY, &sched_gsu,
9746 // NULL, NULL)`. Same shape as `funcstack` above: `schedgetfn`
9747 // (sched.rs:582) walks the schedcmds list on every read and the
9748 // PARTAB_ARRAY row's `setfn` is `None`, so no internal write
9749 // needs the bit cleared. `sched`/`sched -N` mutate the list,
9750 // never the param.
9751 "zsh_scheduled_events", // c:Src/Builtins/sched.c:382
9752 ];
9753 let mk_pm = |name: &str, flags: i32| -> Param {
9754 let keep_readonly = user_protected.contains(&name);
9755 let pre_readonly_mask = if keep_readonly {
9756 !0i32
9757 } else {
9758 !(PM_READONLY as i32)
9759 };
9760 Box::new(param {
9761 node: hashnode {
9762 next: None,
9763 nam: name.to_string(),
9764 flags: (flags & pre_readonly_mask)
9765 | PM_SPECIAL as i32
9766 | PM_HIDE as i32
9767 | PM_HIDEVAL as i32,
9768 },
9769 u_data: 0,
9770 u_tied: None,
9771 u_arr: None,
9772 u_str: None,
9773 u_val: 0,
9774 u_dval: 0.0,
9775 u_hash: None,
9776 gsu_s: None,
9777 gsu_i: None,
9778 gsu_f: None,
9779 gsu_a: None,
9780 gsu_h: None,
9781 base: 0,
9782 width: 0,
9783 env: None,
9784 ename: None,
9785 old: None,
9786 level: 0,
9787 })
9788 };
9789 // c:Src/Modules/system.c:902,904 + Src/Modules/mapfile.c — these
9790 // params are provided by modules that real zsh requires explicit
9791 // `zmodload` for. Seeding them unconditionally makes
9792 // `${+sysparams}` return 1 by default (bug #69 in docs/BUGS.md),
9793 // diverging from zsh which returns 0 until the user runs
9794 // `zmodload zsh/system`. Skip here; `seed_partab_param` below adds
9795 // them on demand from the module's load path.
9796 let module_gated: &[&str] = &[
9797 "sysparams", // zsh/system
9798 "errnos", // zsh/system
9799 "mapfile", // zsh/mapfile
9800 "langinfo", // zsh/langinfo
9801 ".zle.esc", // zsh/hlgroup
9802 ".zle.sgr", // zsh/hlgroup
9803 ".sh.match", // zsh/ksh93
9804 ];
9805 // c:Src/module.c:1065 `addparamdef` — `checkaddparam` (c:1026) finds
9806 // the PM_AUTOLOAD stub `init_bltinmods` planted, calls
9807 // `unsetparam_pm(pm, 0, 1)` which UNLINKS the node (c:1052), and only
9808 // then does `createparam` re-add it — so the real param takes a FRESH
9809 // chain slot, it does not inherit the stub's. `hashtable_nodes::
9810 // insert` is `addhashnode2` (Src/hashtable.c:168), which replaces an
9811 // existing key IN PLACE (c:187-203) and would pin the special to the
9812 // stub's slot; remove first to reproduce C. Visible in
9813 // `${(k)parameters}`: without the remove, `dis_reswords` and
9814 // `usergroups` came out one position off from `zsh -f`.
9815 for entry in PARTAB.iter() {
9816 if module_gated.contains(&entry.name) {
9817 continue;
9818 }
9819 tab.remove(entry.name); // c:1052 unsetparam_pm
9820 tab.insert(entry.name.to_string(), mk_pm(entry.name, entry.flags));
9821 }
9822 for entry in PARTAB_ARRAY.iter() {
9823 if module_gated.contains(&entry.name) {
9824 continue;
9825 }
9826 tab.remove(entry.name); // c:1052 unsetparam_pm
9827 tab.insert(entry.name.to_string(), mk_pm(entry.name, entry.flags));
9828 }
9829 // See `PARTAB_SEEDED` — every magic row now has its paramtab node,
9830 // so from here on a MISSING node is a real "no binding" answer.
9831 PARTAB_SEEDED.store(true, std::sync::atomic::Ordering::Release);
9832}
9833
9834/// !!! WARNING: RUST-ONLY HELPER !!!
9835///
9836/// True once [`init_partab_params`] has finished planting a paramtab
9837/// node for every `PARTAB` / `PARTAB_ARRAY` row.
9838///
9839/// No C counterpart BY CONSTRUCTION. In C the magic rows only ENTER
9840/// `paramtab` when `zsh/parameter` boots (`handlefeatures` →
9841/// `addparamdef` → `createspecialhash`, `Src/module.c:1065`), and before
9842/// that the name still resolves — to the `PM_AUTOLOAD` stub
9843/// `init_bltinmods` planted (`Src/module.c:1218-1223`). Either way C's
9844/// `paramtab->getnode(name)` answers, so C never has to distinguish
9845/// "not seeded yet" from "unset". zshrs seeds the rows from
9846/// `ShellExecutor::new` (vm_helper.rs:2566) and matches `PARTAB` BY NAME
9847/// out of a separate static table, so a magic read that runs BEFORE that
9848/// seeding would see an absent node. `magic_special_shadowed` reads
9849/// absence as "unset" (C: `getnode` → NULL → `fetchvalue` NULL,
9850/// `Src/params.c:2264-2266`), which is only a valid inference once the
9851/// seeding has run; this flag is that precondition.
9852static PARTAB_SEEDED: std::sync::atomic::AtomicBool = std::sync::atomic::AtomicBool::new(false);
9853
9854/// Insert a single PARTAB / PARTAB_ARRAY entry into paramtab. Called
9855/// from `zmodload <module>` once the module's boot completes, so that
9856/// `${+sysparams}` (etc.) flip from 0 → 1 only after explicit load.
9857/// No direct C counterpart — the C path runs through the module's
9858/// `setup_/boot_` chain which adds the SPECIALPMDEF entry via the
9859/// general hashtable machinery. Bug #69 in docs/BUGS.md.
9860pub fn seed_partab_param(name: &str) {
9861 use crate::ported::modules::parameter::{PARTAB, PARTAB_ARRAY};
9862 use crate::ported::zsh_h::{hashnode, param, PM_HIDE, PM_HIDEVAL, PM_READONLY, PM_SPECIAL};
9863 let mut tab = match crate::ported::params::paramtab().write() {
9864 Ok(t) => t,
9865 Err(_) => return,
9866 };
9867 // c:Src/module.c:1026-1052 `checkaddparam` — the node already present may
9868 // be the PM_AUTOLOAD STUB that `add_autoparam` planted (c:1218-1223,
9869 // ported at src/ported/module.rs), not a seeded special: forcing
9870 // MODULESTAB registers `p:builtins` and friends as autoload scalars. C
9871 // does NOT treat that as "already added" — `unsetparam_pm(pm, 0, 1)`
9872 // UNLINKS the stub (c:1052) and only then does `createparam` install the
9873 // real special (c:1065). The bulk path here already models that
9874 // (`tab.remove(entry.name); // c:1052 unsetparam_pm`); the single-name
9875 // sibling omitted it, so `builtins` stayed PM_SCALAR and `gethkparam`'s
9876 // PM_TYPE == PM_HASHED test fell through to None.
9877 //
9878 // A genuine user parameter shadowing the name carries no PM_AUTOLOAD, so
9879 // it still short-circuits exactly as before.
9880 if tab
9881 .get(name)
9882 .is_some_and(|pm| (pm.node.flags as u32 & crate::ported::zsh_h::PM_AUTOLOAD) == 0)
9883 {
9884 return; // a real special is already seeded
9885 }
9886 let flags = PARTAB
9887 .iter()
9888 .find(|e| e.name == name)
9889 .map(|e| e.flags)
9890 .or_else(|| {
9891 PARTAB_ARRAY
9892 .iter()
9893 .find(|e| e.name == name)
9894 .map(|e| e.flags)
9895 });
9896 let Some(flags) = flags else {
9897 return;
9898 };
9899 // c:1052 `unsetparam_pm(pm, 0, 1)` — unlink the autoload stub, but only
9900 // once we know a real special is going in to replace it. Doing it before
9901 // the PARTAB lookup above would destroy an existing node for any name
9902 // that is NOT a known special and then return, which is strictly worse
9903 // than the early-return it replaced.
9904 tab.remove(name);
9905 let pm = Box::new(param {
9906 node: hashnode {
9907 next: None,
9908 nam: name.to_string(),
9909 // Keep C's PM_READONLY. `init_partab_params` strips it from
9910 // rows the RUNTIME writes internally (funcstack pushes and
9911 // friends) and keeps it on the getfn/scanfn-computed rows —
9912 // see its `user_protected` list. Every name reaching THIS
9913 // seeder is a zmodload-gated row
9914 // (`module_gated_params_for`), and all of them are the
9915 // computed kind: `SPECIALPMDEF("sysparams", PM_READONLY,
9916 // NULL, getpmsysparams, scanpmsysparams)`
9917 // (Src/Modules/system.c:906), `errnos` (c:904), `langinfo`
9918 // (Src/Modules/langinfo.c:455); `mapfile` carries flags 0 in
9919 // C so it is unaffected either way. Stripping the bit made
9920 // `${(t)sysparams}` read `association-hide-hideval-special`
9921 // against zsh's `association-readonly-hide-hideval-special`,
9922 // and let `unset sysparams` succeed where zsh rejects with
9923 // `read-only variable: sysparams`.
9924 flags: flags | PM_SPECIAL as i32 | PM_HIDE as i32 | PM_HIDEVAL as i32,
9925 },
9926 u_data: 0,
9927 u_tied: None,
9928 u_arr: None,
9929 u_str: None,
9930 u_val: 0,
9931 u_dval: 0.0,
9932 u_hash: None,
9933 gsu_s: None,
9934 gsu_i: None,
9935 gsu_f: None,
9936 gsu_a: None,
9937 gsu_h: None,
9938 base: 0,
9939 width: 0,
9940 env: None,
9941 ename: None,
9942 old: None,
9943 level: 0,
9944 });
9945 tab.insert(name.to_string(), pm);
9946}
9947
9948/// Default autoloadable parameters: name → owning module. Port of the
9949/// `autofeatures` `p:` rows in Src/Modules/parameter.mdd, watch.mdd,
9950/// termcap.mdd, terminfo.mdd, Src/Zle/zleparameter.mdd and
9951/// Src/Builtins/sched.mdd, registered at startup through
9952/// `setautofeatures` → `add_autoparam` (Src/module.c:1198-1229): each
9953/// name becomes a scalar paramtab stub whose VALUE is the module name,
9954/// flagged PM_AUTOLOAD (module.c:1218-1219). Matches `zmodload -ap`
9955/// output of the reference zsh build.
9956pub const AUTOLOAD_PARAMS: &[(&str, &str)] = &[
9957 // Src/Modules/watch.mdd:5 autofeatures
9958 ("WATCH", "zsh/watch"),
9959 ("watch", "zsh/watch"),
9960 // Src/Modules/parameter.mdd:5 autofeatures
9961 ("aliases", "zsh/parameter"),
9962 ("builtins", "zsh/parameter"),
9963 ("commands", "zsh/parameter"),
9964 ("dirstack", "zsh/parameter"),
9965 ("dis_aliases", "zsh/parameter"),
9966 ("dis_builtins", "zsh/parameter"),
9967 ("dis_functions", "zsh/parameter"),
9968 ("dis_functions_source", "zsh/parameter"),
9969 ("dis_galiases", "zsh/parameter"),
9970 ("dis_patchars", "zsh/parameter"),
9971 ("dis_reswords", "zsh/parameter"),
9972 ("dis_saliases", "zsh/parameter"),
9973 ("funcfiletrace", "zsh/parameter"),
9974 ("funcsourcetrace", "zsh/parameter"),
9975 ("funcstack", "zsh/parameter"),
9976 ("functions", "zsh/parameter"),
9977 ("functions_source", "zsh/parameter"),
9978 ("functrace", "zsh/parameter"),
9979 ("galiases", "zsh/parameter"),
9980 ("history", "zsh/parameter"),
9981 ("historywords", "zsh/parameter"),
9982 ("jobdirs", "zsh/parameter"),
9983 ("jobstates", "zsh/parameter"),
9984 ("jobtexts", "zsh/parameter"),
9985 ("modules", "zsh/parameter"),
9986 ("nameddirs", "zsh/parameter"),
9987 ("options", "zsh/parameter"),
9988 ("parameters", "zsh/parameter"),
9989 ("patchars", "zsh/parameter"),
9990 ("reswords", "zsh/parameter"),
9991 ("saliases", "zsh/parameter"),
9992 ("userdirs", "zsh/parameter"),
9993 ("usergroups", "zsh/parameter"),
9994 // Src/Zle/zleparameter.mdd:5 autofeatures
9995 ("keymaps", "zsh/zleparameter"),
9996 ("widgets", "zsh/zleparameter"),
9997 // Src/Modules/termcap.mdd:5 / terminfo.mdd:5 autofeatures
9998 ("termcap", "zsh/termcap"),
9999 ("terminfo", "zsh/terminfo"),
10000 // Src/Builtins/sched.mdd:5 autofeatures
10001 ("zsh_scheduled_events", "zsh/sched"),
10002];
10003
10004/// Autoload stubs whose owning module is NOT loaded — the rows zsh's
10005/// `typeset` listings print as `undefined NAME` (printparamnode's
10006/// PM_AUTOLOAD pmtypes row, Src/params.c:6011 + the PM_AUTOLOAD
10007/// NAMEONLY arm at Src/params.c:6146-6155). Once a module loads, its
10008/// stubs drop out and the real params list instead.
10009pub fn autoload_param_stubs() -> Vec<(&'static str, &'static str)> {
10010 use crate::ported::zsh_h::{MOD_INIT_B, MOD_UNLOAD};
10011 let tab = crate::ported::module::MODULESTAB.lock().unwrap();
10012 AUTOLOAD_PARAMS
10013 .iter()
10014 .copied()
10015 .filter(|(_, m)| {
10016 // "Boot ran" is MOD_INIT_B && !MOD_UNLOAD (the criterion
10017 // printmodulenode uses, src/ported/module.rs:246 — C's
10018 // `m->u.handle` union check at Src/module.c:218-241).
10019 // modulestab::is_loaded checks MOD_LINKED which
10020 // register_builtin_modules pre-seeds for EVERY compiled-in
10021 // module, so it would report zsh/parameter "loaded" in a
10022 // fresh `zsh -f` where real zsh still shows the stubs.
10023 !tab.modules.get(*m).is_some_and(|md| {
10024 (md.node.flags & MOD_INIT_B) != 0 && (md.node.flags & MOD_UNLOAD) == 0
10025 })
10026 })
10027 .collect()
10028}
10029
10030/// Names provided by `zsh/system` / `zsh/mapfile` etc. that are
10031/// gated on explicit `zmodload`. Used by the bin_zmodload path to
10032/// re-seed paramtab after the module's boot completes.
10033pub fn module_gated_params_for(module: &str) -> &'static [&'static str] {
10034 match module {
10035 "zsh/system" => &["sysparams", "errnos"],
10036 "zsh/mapfile" => &["mapfile"],
10037 "zsh/langinfo" => &["langinfo"],
10038 "zsh/hlgroup" => &[".zle.esc", ".zle.sgr"],
10039 "zsh/ksh93" => &[".sh.match"],
10040 _ => &[],
10041 }
10042}
10043impl ShellExecutor {
10044 /// `enter_posix_mode` — see implementation.
10045 pub fn enter_posix_mode(&mut self) {
10046 self.posix_mode = true;
10047 self.plugin_cache = None;
10048 self.compsys_cache = std::cell::OnceCell::new();
10049 self.compinit_pending = None;
10050 self.worker_pool = std::sync::Arc::new(crate::worker::WorkerPool::new(1));
10051 // Direct call to the canonical `emulate()` port
10052 // (Src/options.c:533) — `-R` semantics = fully=true.
10053 // bin_emulate goes through dispatch_builtin which needs an
10054 // ExecutorContext that isn't set up yet at apply_cli_flags
10055 // time; the underlying emulate() doesn't need one.
10056 crate::ported::options::emulate("sh", true);
10057 }
10058 /// `enter_ksh_mode` — see implementation.
10059 pub fn enter_ksh_mode(&mut self) {
10060 self.plugin_cache = None;
10061 self.compsys_cache = std::cell::OnceCell::new();
10062 self.compinit_pending = None;
10063 self.worker_pool = std::sync::Arc::new(crate::worker::WorkerPool::new(1));
10064 crate::ported::options::emulate("ksh", true);
10065 }
10066 /// `enter_dash_mode` — strict-dash (Debian Almquist Shell) runtime.
10067 /// Same executor setup as [`enter_posix_mode`] (dash IS `sh` for every
10068 /// option), but calls `emulate("dash")` so the Rust-only DASH_STRICT
10069 /// flag is raised (and NOT cleared, as `emulate("sh")` would). See
10070 /// `src/extensions/dash_mode.rs`.
10071 pub fn enter_dash_mode(&mut self) {
10072 self.posix_mode = true;
10073 self.plugin_cache = None;
10074 self.compsys_cache = std::cell::OnceCell::new();
10075 self.compinit_pending = None;
10076 self.worker_pool = std::sync::Arc::new(crate::worker::WorkerPool::new(1));
10077 crate::ported::options::emulate("dash", true);
10078 }
10079}
10080
10081/// Thin (text, pattern) → bool wrapper over the canonical
10082/// `patcompile()` + `pattry()` pair from `Src/pattern.c`. Argument
10083/// order is flipped so callers read naturally. Lives in vm_helper.rs
10084/// (non-port file) as the public convenience entry for extensions
10085/// and the VM bridge; `src/ported/*` files inline the compile+match
10086/// idiom directly to preserve PORT.md Rule 1 faithfulness.
10087pub fn glob_match_static(s: &str, pattern: &str) -> bool {
10088 let Some(prog) = patcompile(
10089 &{
10090 let mut __pat_tok = (pattern).to_string();
10091 crate::ported::glob::tokenize(&mut __pat_tok);
10092 __pat_tok
10093 },
10094 PAT_HEAPDUP as i32,
10095 None,
10096 ) else {
10097 return false;
10098 };
10099 // c:Src/pattern.c:2570-2621 — `else if (prog->patnpar && !(patflags &
10100 // PAT_FILE))`: when the caller passes NO nump/begp/endp, `pattryrefs`
10101 // ITSELF publishes $match / $mbegin / $mend. That arm is ported in
10102 // pattern.rs, so this layer must not re-derive them from begp/endp.
10103 //
10104 // Re-deriving required compensating for the EXCLUSIVE end index
10105 // `pattryrefs` used to return, and the compensation was wrong twice over:
10106 // * it would double-correct the c:2562-2564 `- 1` now applied there, and
10107 // collide with the `endp[i] < 0` unset-group sentinel — an empty
10108 // capture at offset 0 reports (0, -1) and was misread as an UNMATCHED
10109 // alternation branch;
10110 // * `saturating_sub(1)` clamped at 0, so under KSHARRAYS an empty capture
10111 // at offset 0 gave `mend=0` where C computes `0 + 0 + 0 - 1` = -1.
10112 // Measured: `setopt ksharrays; [[ abc = (#b)(x#)abc ]]` → zsh mend=-1,
10113 // zshrs mend=0, while the substitution path (which already uses the
10114 // c:2570-2621 arm) printed -1 correctly.
10115 let matched = pattry(&prog, s);
10116 // $MATCH / $MBEGIN / $MEND are set by the MATCHER (pattern.rs, c:2526),
10117 // which is where C decides it — gated on the GLOBAL patglobflags so a later
10118 // `(#M)` can turn GF_MATCHREF back off (c:1099-1100).
10119 //
10120 // This layer used to re-do it with `pattern.contains("(#m)")`, which can
10121 // only ever answer "on": it re-set $MATCH after the matcher had correctly
10122 // declined to, so `(#m)(#M)a*` reported a match string where zsh leaves it
10123 // unset. Wrong mechanism (a substring test cannot see which flag came last)
10124 // and wrong layer (the matcher already has the compiled flags).
10125 matched
10126}
10127
10128pub use crate::ported::lex::untokenize_ztokens;
10129
10130pub use crate::ported::utils::unmetafy_str;
10131
10132pub use crate::ported::utils::zsh_errno_msg;
10133
10134// ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
10135// PM_NAMEREF bridge helpers (typeset -n / named references).
10136//
10137// Rust-only adapters around the canonical C nameref machinery in
10138// Src/params.c (resolve_nameref_rec c:6332, setscope c:6382,
10139// upscope c:6455) and Src/builtin.c (bin_typeset nameref arm
10140// c:3117-3150). zshrs's paramtab is a name-keyed HashMap handing
10141// out clones, so the chain walk operates by NAME against the live
10142// table instead of by Param pointer — same hop rule, same loop
10143// detection, same upscope old-chain walk. The ported fns in
10144// params.rs/builtin.rs call into these at the exact C deref points
10145// (getparamnode c:570-575, getvalue/fetchvalue c:2247-2270,
10146// assignsparam c:3252/3258, assignaparam c:3392-3398, bin_unset
10147// c:3939-3951, typeset_single c:2032-2050).
10148// ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
10149
10150pub use crate::ported::params::*;
10151
10152/// True for an `fpath` entry that is a host zsh **distribution's own**
10153/// function tree — the copy of `Completion/**` + `Functions/**` that
10154/// `make install` flattens into the zsh package.
10155///
10156/// c:Src/init.c:1132-1143 — zsh's `setupvals` seeds `fpath` with
10157/// `FIXED_FPATH_DIR`, `SITEFPATH_DIR`, then
10158/// `<prefix>/share/zsh/<version>/functions`. zshrs keeps the first two and
10159/// drops the last, from the compiled-in default and from an inherited
10160/// `FPATH` alike. This is startup construction only: like zsh, a later
10161/// shell-level `fpath=( … )` stores exactly what was assigned.
10162///
10163/// The distribution tree goes because zshrs ships its own: `vendor/zsh` is
10164/// packed into the binary by `build.rs` and materialised into
10165/// `~/.zshrs/functions`, which takes the host tree's index on `fpath`. A host tree is not a
10166/// fallback for that — it is a second, differently-versioned copy of the
10167/// same `compinit`, `_git` and `_describe`, and letting it linger means a
10168/// Homebrew zsh upgrade can change how zshrs completes.
10169///
10170/// `share/zsh/site-functions` STAYS. Nothing in it comes from zsh: it is
10171/// where third-party formulae install their completions (`_brew`,
10172/// `_docker`, `_kubectl`, `_gh` …), the bundle has no copy of them, and
10173/// dropping the directory would silently disable completion for every
10174/// tool installed through a package manager.
10175///
10176/// The two shapes to reject differ in depth — Homebrew's Cellar layout has
10177/// no version component under `share/zsh` — so the test is a `share`
10178/// component followed by `zsh`, plus a final component of exactly
10179/// `functions`:
10180/// /usr/share/zsh/5.9/functions drop
10181/// /opt/homebrew/Cellar/zsh/5.9.2/share/zsh/functions drop
10182/// /opt/homebrew/share/zsh/site-functions keep
10183/// ~/.zinit/plugins/…/src keep
10184pub(crate) fn is_host_zsh_function_tree(p: &Path) -> bool {
10185 if p.file_name() != Some(OsStr::new("functions")) {
10186 return false;
10187 }
10188 let mut comps = p.components().map(|c| c.as_os_str());
10189 while let Some(c) = comps.next() {
10190 if c == "share" && comps.clone().next() == Some(OsStr::new("zsh")) {
10191 return true;
10192 }
10193 }
10194 false
10195}
10196
10197/// Default `fpath` when `FPATH` is absent from the environment.
10198///
10199/// c:Src/init.c:1143 — `SITEFPATH_DIR`, listed by zsh even when it does
10200/// not exist on disk. zshrs adds the same directory under whichever
10201/// prefixes are actually present, because that is where package managers
10202/// put third-party completions and the bundled tree has none of them.
10203///
10204/// The versioned `<prefix>/share/zsh/<version>/functions` that C also
10205/// seeds here is deliberately NOT included — see
10206/// [`is_host_zsh_function_tree`]. `~/.zshrs/functions` supplies it, and is
10207/// appended by both constructors.
10208///
10209/// The prefix list is the set of places a package manager puts
10210/// `share/zsh/site-functions` on the platforms zshrs targets. `/usr/local`
10211/// and `/usr` cover Linux distributions and a from-source install;
10212/// `/opt/homebrew` is Homebrew on Apple silicon, `/usr/local` doubles as
10213/// Homebrew on Intel macOS, and `/home/linuxbrew/.linuxbrew` is
10214/// Homebrew's Linux prefix — which is NOT `/opt/homebrew`, so a Linux box
10215/// with brew-installed completions found none of them before it was
10216/// listed. Only directories that exist are added, so naming a prefix
10217/// costs nothing on a machine that lacks it.
10218fn default_fpath() -> Vec<PathBuf> {
10219 let mut out: Vec<PathBuf> = vec![PathBuf::from("/usr/local/share/zsh/site-functions")];
10220 for pfx in [
10221 "/usr/local",
10222 "/opt/homebrew",
10223 "/home/linuxbrew/.linuxbrew",
10224 "/usr",
10225 ] {
10226 let site = PathBuf::from(format!("{pfx}/share/zsh/site-functions"));
10227 if site.is_dir() && !out.contains(&site) {
10228 out.push(site);
10229 }
10230 }
10231 out
10232}