Skip to main content

grit_lib/
pathspec.rs

1//! Git-compatible pathspec matching (magic tokens and global flags).
2//!
3//! ```rust
4//! use grit_lib::worktree::pathspec::parse_pathspecs_from_source;
5//!
6//! let specs = parse_pathspecs_from_source(b"README.md\n", false).expect("parse");
7//! assert_eq!(specs, ["README.md"]);
8//! ```
9//!
10//! Global flags are read from the same environment variables as Git:
11//! `GIT_LITERAL_PATHSPECS`, `GIT_GLOB_PATHSPECS`, `GIT_NOGLOB_PATHSPECS`,
12//! `GIT_ICASE_PATHSPECS`. The `grit` binary sets these from CLI flags such as
13//! `--literal-pathspecs` before dispatching subcommands.
14
15use crate::git_path::CanonicalizeExt;
16use std::borrow::Cow;
17use std::path::{Path, PathBuf};
18
19use hashbrown::HashSet;
20
21use crate::attributes::MacroTable;
22use crate::crlf::path_gitattribute_values;
23use crate::crlf::AttrRule;
24pub use crate::error::{AttrPathspecError, PathspecFromFileParseError, PathspecResolveError};
25use crate::precompose_config::argv_precompose_enabled;
26use crate::repo::Repository;
27use crate::unicode_normalization::precompose_utf8_path;
28use crate::wildmatch::{wildmatch, WM_CASEFOLD, WM_PATHNAME};
29
30/// Returns the length of the leading literal segment before the first glob metacharacter,
31/// matching Git's `simple_length()` (`*` `?` `[` `\`) on bytes.
32#[must_use]
33pub fn simple_length(match_str: &str) -> usize {
34    let b = match_str.as_bytes();
35    let mut len = 0usize;
36    for &c in b {
37        if matches!(c, b'*' | b'?' | b'[' | b'\\') {
38            break;
39        }
40        len += 1;
41    }
42    len
43}
44
45/// Whether the pattern uses wildcards after Git pathspec escaping rules.
46#[must_use]
47pub fn has_glob_chars(s: &str) -> bool {
48    simple_length(s) < s.len()
49}
50
51/// Read pathspec entries from raw file bytes (stdin or file), matching Git's
52/// `--pathspec-from-file` / `--pathspec-file-nul` rules.
53///
54/// * **NUL mode:** entries are separated by `NUL`; each segment is taken literally
55///   (no trimming or C-style unquoting).
56/// * **Line mode:** entries are separated by `LF`; optional `CR` before `LF`
57///   is stripped; optional trailing line without a final newline is included;
58///   double-quoted lines are C-unquoted (including octal escapes).
59pub fn parse_pathspecs_from_source(
60    data: &[u8],
61    nul_terminated: bool,
62) -> Result<Vec<String>, PathspecFromFileParseError> {
63    if nul_terminated {
64        let mut chunks: Vec<&[u8]> = data.split(|b| *b == 0).collect();
65        if chunks.last().copied() == Some(&[] as &[u8]) {
66            chunks.pop();
67        }
68        let mut out = Vec::with_capacity(chunks.len());
69        for chunk in chunks {
70            if chunk.is_empty() {
71                out.push(String::new());
72                continue;
73            }
74            let entry = std::str::from_utf8(chunk)
75                .map_err(|_| PathspecFromFileParseError::InvalidUtf8InNulRecord)?;
76            out.push(entry.to_string());
77        }
78        return Ok(out);
79    }
80
81    let text = String::from_utf8_lossy(data);
82    let mut out = Vec::new();
83    for raw in text.split_inclusive('\n') {
84        let line = raw.trim_end_matches('\n').trim_end_matches('\r');
85        if line.is_empty() {
86            out.push(String::new());
87            continue;
88        }
89        if line.starts_with('"') {
90            if line.ends_with('"') && line.len() >= 2 {
91                out.push(unquote_c_style_pathspec_line(line)?);
92            } else {
93                return Err(PathspecFromFileParseError::InvalidCStyleQuoting {
94                    line: line.to_string(),
95                });
96            }
97        } else {
98            out.push(line.to_string());
99        }
100    }
101    Ok(out)
102}
103
104/// Unquote a single `--pathspec-from-file` line that is wrapped in double quotes.
105fn unquote_c_style_pathspec_line(s: &str) -> Result<String, PathspecFromFileParseError> {
106    let bytes = s.as_bytes();
107    if bytes.first() != Some(&b'"') || bytes.last() != Some(&b'"') || bytes.len() < 2 {
108        return Err(PathspecFromFileParseError::InvalidCStyleQuoting {
109            line: s.to_string(),
110        });
111    }
112
113    let inner = &bytes[1..bytes.len() - 1];
114    let mut out = Vec::with_capacity(inner.len());
115    let mut i = 0;
116    while i < inner.len() {
117        if inner[i] != b'\\' {
118            out.push(inner[i]);
119            i += 1;
120            continue;
121        }
122        i += 1;
123        if i >= inner.len() {
124            return Err(PathspecFromFileParseError::InvalidEscapeAtEnd);
125        }
126        match inner[i] {
127            b'\\' => out.push(b'\\'),
128            b'"' => out.push(b'"'),
129            b'a' => out.push(7),
130            b'b' => out.push(8),
131            b'f' => out.push(12),
132            b'n' => out.push(b'\n'),
133            b'r' => out.push(b'\r'),
134            b't' => out.push(b'\t'),
135            b'v' => out.push(11),
136            c if c.is_ascii_digit() => {
137                if i + 2 >= inner.len() {
138                    return Err(PathspecFromFileParseError::TruncatedOctalEscape);
139                }
140                let oct = std::str::from_utf8(&inner[i..i + 3])
141                    .map_err(|_| PathspecFromFileParseError::InvalidOctalBytes)?;
142                out.push(
143                    u8::from_str_radix(oct, 8)
144                        .map_err(|_| PathspecFromFileParseError::InvalidOctalValue)?,
145                );
146                i += 2;
147            }
148            other => {
149                return Err(PathspecFromFileParseError::InvalidEscapeSequence {
150                    escape: char::from(other),
151                });
152            }
153        }
154        i += 1;
155    }
156    String::from_utf8(out).map_err(|_| PathspecFromFileParseError::InvalidUtf8InQuotedLine)
157}
158
159/// Raw bytes for [`ResolvedPathspecs::resolve`] from `--pathspec-from-file` (read by the caller).
160#[derive(Debug, Clone, PartialEq, Eq)]
161pub struct PathspecInput {
162    /// File or stdin contents.
163    pub data: Vec<u8>,
164    /// When true, entries are NUL-separated (`--pathspec-file-nul`).
165    pub nul_terminated: bool,
166}
167
168impl PathspecInput {
169    /// Build input from bytes already read from a file or stdin.
170    #[must_use]
171    pub fn new(data: Vec<u8>, nul_terminated: bool) -> Self {
172        Self {
173            data,
174            nul_terminated,
175        }
176    }
177}
178
179/// Pathspec argv fragments resolved against a repository work tree.
180#[derive(Debug, Clone, Default, PartialEq, Eq)]
181pub struct ResolvedPathspecs {
182    /// Work-tree-relative pathspec strings for matching.
183    pub specs: Vec<String>,
184    /// User-facing spellings aligned with `specs` (argv or file lines).
185    pub sources: Vec<String>,
186    /// True when the caller supplied argv paths and/or a from-file (not "whole tree").
187    pub scoped: bool,
188}
189
190impl ResolvedPathspecs {
191    /// Resolve `raw` argv pathspecs and optional `--pathspec-from-file` bytes against `repo`.
192    ///
193    /// `cwd` is the process working directory used for relative pathspec prefixes (never read
194    /// from the environment inside the library).
195    ///
196    /// # Examples
197    ///
198    /// ```
199    /// use grit_lib::worktree::pathspec::{PathspecInput, ResolvedPathspecs};
200    /// use grit_lib::repo::{InitOptions, Repository};
201    /// use std::fs;
202    ///
203    /// # fn main() -> Result<(), Box<dyn std::error::Error>> {
204    /// let dir = tempfile::tempdir()?;
205    /// let repo = Repository::init(dir.path(), &InitOptions::default())?;
206    /// fs::write(dir.path().join("a.txt"), b"x")?;
207    /// let resolved = ResolvedPathspecs::resolve(
208    ///     &repo,
209    ///     dir.path(),
210    ///     &["a.txt".to_string()],
211    ///     None,
212    /// )?;
213    /// assert_eq!(resolved.specs, ["a.txt"]);
214    /// let from_file = PathspecInput::new(b"a.txt\n".to_vec(), false);
215    /// let from_file = ResolvedPathspecs::resolve(&repo, dir.path(), &[], Some(from_file))?;
216    /// assert_eq!(from_file.specs, ["a.txt"]);
217    /// # Ok(())
218    /// # }
219    /// ```
220    pub fn resolve(
221        repo: &Repository,
222        cwd: &Path,
223        raw: &[String],
224        from_file: Option<PathspecInput>,
225    ) -> Result<Self, PathspecResolveError> {
226        let work_tree = repo
227            .work_tree()
228            .ok_or(PathspecResolveError::NeedsWorkTree)?;
229        let prefix = pathdiff(cwd, work_tree);
230        let scoped = from_file.is_some() || !raw.is_empty();
231
232        let mut sources = Vec::new();
233        if let Some(input) = from_file {
234            let parsed = parse_pathspecs_from_source(&input.data, input.nul_terminated)?;
235            sources.extend(parsed);
236        }
237        for sel in raw {
238            sources.push(sel.clone());
239        }
240
241        if scoped && sources.is_empty() {
242            return Err(PathspecResolveError::EmptyPathspec);
243        }
244        for sel in &sources {
245            if sel.is_empty() {
246                return Err(PathspecResolveError::EmptyPathspec);
247            }
248        }
249
250        let argv_precompose = repo.pathspec_precompose_enabled();
251        let specs = sources
252            .iter()
253            .map(|sel| {
254                resolve_pathspec_in_worktree_with_argv_precompose(
255                    sel,
256                    sel,
257                    work_tree,
258                    prefix.as_deref(),
259                    argv_precompose,
260                )
261            })
262            .collect::<Result<Vec<_>, _>>()?;
263
264        Ok(Self {
265            specs,
266            sources,
267            scoped,
268        })
269    }
270
271    /// Whether the invocation targets the entire work tree (no path arguments or from-file).
272    #[must_use]
273    pub fn is_entire_tree(&self) -> bool {
274        !self.scoped
275    }
276
277    /// Consume `self` and return resolved specs and user-facing sources.
278    ///
279    /// When [`Self::is_entire_tree`] is true, both vectors are empty (match legacy
280    /// "stage/remove/clean the whole tree" call sites).
281    #[must_use]
282    pub fn into_pathspec_lists(self) -> (Vec<String>, Vec<String>) {
283        if self.scoped {
284            (self.specs, self.sources)
285        } else {
286            (Vec::new(), Vec::new())
287        }
288    }
289
290    /// Resolved pathspec strings only; empty when the invocation is unscoped.
291    #[must_use]
292    pub fn into_specs(self) -> Vec<String> {
293        if self.scoped {
294            self.specs
295        } else {
296            Vec::new()
297        }
298    }
299}
300
301#[derive(Debug, Clone, Default)]
302struct PathspecMagic {
303    literal: bool,
304    glob: bool,
305    icase: bool,
306    exclude: bool,
307    /// `:(top)` / short `:/` — paths are relative to repo root.
308    top: bool,
309    prefix: Option<String>,
310    /// `:(attr:...)` requirements.
311    attr_requirements: Vec<AttrRequirement>,
312}
313
314#[derive(Debug, Clone, PartialEq, Eq)]
315enum AttrRequirement {
316    Set(String),
317    Unset(String),
318    Unspecified(String),
319    Value(String, String),
320}
321
322impl AttrRequirement {
323    fn name(&self) -> &str {
324        match self {
325            AttrRequirement::Set(name)
326            | AttrRequirement::Unset(name)
327            | AttrRequirement::Unspecified(name)
328            | AttrRequirement::Value(name, _) => name,
329        }
330    }
331}
332
333/// The library never reads the process environment, so Git's `GIT_*_PATHSPECS` globals are
334/// always unset here and every global flag takes its default. This runs on every pathspec
335/// comparison, so it must stay allocation-free.
336fn git_env_bool(_key: &str, default: bool) -> bool {
337    default
338}
339
340fn literal_global() -> bool {
341    git_env_bool("GIT_LITERAL_PATHSPECS", false)
342}
343
344/// Whether `GIT_LITERAL_PATHSPECS` is enabled (shell `*` and `?` are literal, not globs).
345#[must_use]
346pub fn literal_pathspecs_enabled() -> bool {
347    literal_global()
348}
349
350fn glob_global() -> bool {
351    git_env_bool("GIT_GLOB_PATHSPECS", false)
352}
353
354fn noglob_global() -> bool {
355    git_env_bool("GIT_NOGLOB_PATHSPECS", false)
356}
357
358fn icase_global() -> bool {
359    git_env_bool("GIT_ICASE_PATHSPECS", false)
360}
361
362/// Validates global pathspec environment flags the same way Git does.
363///
364/// Returns an error message suitable for `bail!` when flags are incompatible.
365pub fn validate_global_pathspec_flags() -> Result<(), String> {
366    let lit = literal_global();
367    let glob = glob_global();
368    let noglob = noglob_global();
369    let icase = icase_global();
370
371    if glob && noglob {
372        return Err("global 'glob' and 'noglob' pathspec settings are incompatible".to_string());
373    }
374    if lit && (glob || noglob || icase) {
375        return Err(
376            "global 'literal' pathspec setting is incompatible with all other global pathspec settings"
377                .to_string(),
378        );
379    }
380    Ok(())
381}
382
383fn is_valid_attr_name(name: &str) -> bool {
384    !name.is_empty()
385        && name
386            .bytes()
387            .all(|b| b.is_ascii_alphanumeric() || matches!(b, b'_' | b'-' | b'.'))
388}
389
390fn split_attr_expr(expr: &str) -> Result<Vec<String>, AttrPathspecError> {
391    let mut parts = Vec::new();
392    let mut cur = String::new();
393    let mut in_value = false;
394    let mut escaped = false;
395
396    for ch in expr.chars() {
397        if escaped {
398            if ch.is_ascii_whitespace() {
399                return Err(AttrPathspecError::DanglingEscape);
400            }
401            if ch != ',' {
402                return Err(AttrPathspecError::UnsupportedEscape);
403            }
404            cur.push(ch);
405            escaped = false;
406            continue;
407        }
408        if in_value && ch == '\\' {
409            escaped = true;
410            continue;
411        }
412        if ch == '=' {
413            in_value = true;
414            cur.push(ch);
415            continue;
416        }
417        if ch.is_ascii_whitespace() {
418            if !cur.is_empty() {
419                parts.push(cur);
420                cur = String::new();
421            }
422            in_value = false;
423            continue;
424        }
425        cur.push(ch);
426    }
427
428    if escaped {
429        return Err(AttrPathspecError::DanglingEscape);
430    }
431    if !cur.is_empty() {
432        parts.push(cur);
433    }
434    Ok(parts)
435}
436
437fn parse_attr_requirements(expr: &str) -> Result<Vec<AttrRequirement>, AttrPathspecError> {
438    let name_only = |name: &str| {
439        if name.contains('=') || !is_valid_attr_name(name) {
440            Err(AttrPathspecError::InvalidName(name.to_owned()))
441        } else {
442            Ok(name.to_owned())
443        }
444    };
445    let mut out = Vec::new();
446    for token in split_attr_expr(expr)? {
447        let req = if let Some(name) = token.strip_prefix('-') {
448            AttrRequirement::Unset(name_only(name)?)
449        } else if let Some(name) = token.strip_prefix('!') {
450            AttrRequirement::Unspecified(name_only(name)?)
451        } else if let Some((name, value)) = token.split_once('=') {
452            AttrRequirement::Value(name_only(name)?, value.to_owned())
453        } else {
454            AttrRequirement::Set(name_only(&token)?)
455        };
456        out.push(req);
457    }
458    if out.is_empty() {
459        return Err(AttrPathspecError::Empty);
460    }
461    Ok(out)
462}
463
464/// The `attr:` clauses inside a long-form `:(...)` magic prefix (none for other specs).
465fn attr_clauses(spec: &str) -> Vec<String> {
466    if literal_global() {
467        return Vec::new();
468    }
469    let Some(rest) = spec.strip_prefix(":(") else {
470        return Vec::new();
471    };
472    let Some(close) = rest.find(')') else {
473        return Vec::new();
474    };
475    split_long_magic_tokens_raw(&rest[..close])
476        .into_iter()
477        .filter_map(|token| token.trim_start().strip_prefix("attr:").map(str::to_owned))
478        .collect()
479}
480
481/// Validate the `:(attr:...)` magic of every spec in `specs`.
482///
483/// Specs without attr magic are always accepted. A spec is rejected when it has more than one
484/// `attr:` clause or a clause that does not parse (empty clause, invalid attribute name, bad
485/// escape). A clause of only whitespace has no requirements and is accepted.
486///
487/// # Errors
488///
489/// [`crate::error::Error::InvalidAttrPathspec`] naming the first rejected spec and its
490/// [`AttrPathspecError`].
491pub fn validate_attr_pathspecs(specs: &[String]) -> crate::error::Result<()> {
492    for spec in specs {
493        let invalid = |reason| crate::error::Error::InvalidAttrPathspec {
494            spec: spec.clone(),
495            reason,
496        };
497        let clauses = attr_clauses(spec);
498        if clauses.len() > 1 {
499            return Err(invalid(AttrPathspecError::MultipleAttrClauses));
500        }
501        match clauses.first() {
502            Some(expr) if expr.is_empty() => return Err(invalid(AttrPathspecError::Empty)),
503            // Whitespace alone is an attr clause with no requirements, which Git accepts.
504            Some(expr) if expr.trim().is_empty() => {}
505            Some(expr) => {
506                parse_attr_requirements(expr).map_err(invalid)?;
507            }
508            None => {}
509        }
510    }
511    Ok(())
512}
513
514fn split_long_magic_tokens(magic_part: &str) -> Vec<String> {
515    split_long_magic_tokens_raw(magic_part)
516        .into_iter()
517        .map(|token| token.trim().to_string())
518        .collect()
519}
520
521/// Split long magic on unescaped `,`, keeping surrounding whitespace.
522fn split_long_magic_tokens_raw(magic_part: &str) -> Vec<String> {
523    let mut tokens = Vec::new();
524    let mut cur = String::new();
525    let mut escaped = false;
526    for ch in magic_part.chars() {
527        if escaped {
528            cur.push('\\');
529            cur.push(ch);
530            escaped = false;
531            continue;
532        }
533        if ch == '\\' {
534            escaped = true;
535            continue;
536        }
537        if ch == ',' {
538            tokens.push(std::mem::take(&mut cur));
539            continue;
540        }
541        cur.push(ch);
542    }
543    if escaped {
544        cur.push('\\');
545    }
546    tokens.push(cur);
547    tokens
548}
549
550fn parse_long_magic(rest_after_paren: &str) -> Option<(PathspecMagic, &str)> {
551    let close = rest_after_paren.find(')')?;
552    let magic_part = &rest_after_paren[..close];
553    let tail = &rest_after_paren[close + 1..];
554    let mut magic = PathspecMagic::default();
555    for raw in split_long_magic_tokens(magic_part) {
556        let token = raw.trim();
557        if token.is_empty() {
558            continue;
559        }
560        if let Some(p) = token.strip_prefix("prefix:") {
561            magic.prefix = Some(p.to_string());
562            continue;
563        }
564        if let Some(expr) = token.strip_prefix("attr:") {
565            if let Ok(reqs) = parse_attr_requirements(expr) {
566                magic.attr_requirements = reqs;
567            }
568            continue;
569        }
570        if token.eq_ignore_ascii_case("literal") {
571            magic.literal = true;
572        } else if token.eq_ignore_ascii_case("glob") {
573            magic.glob = true;
574        } else if token.eq_ignore_ascii_case("icase") {
575            magic.icase = true;
576        } else if token.eq_ignore_ascii_case("exclude") {
577            magic.exclude = true;
578        } else if token.eq_ignore_ascii_case("top") {
579            magic.top = true;
580        }
581    }
582    Some((magic, tail))
583}
584
585/// `elem` is the full pathspec beginning with `:` (short magic form, not `:(...)`).
586fn parse_short_magic(elem: &str) -> (PathspecMagic, &str) {
587    let bytes = elem.as_bytes();
588    let mut i = 1usize;
589    let mut magic = PathspecMagic::default();
590    while i < bytes.len() && bytes[i] != b':' {
591        let ch = bytes[i];
592        if ch == b'^' {
593            magic.exclude = true;
594            i += 1;
595            continue;
596        }
597        let is_magic = match ch {
598            b'!' => {
599                magic.exclude = true;
600                true
601            }
602            b'/' => {
603                magic.top = true;
604                true
605            } // short `:/` = top
606            _ => false,
607        };
608        if is_magic {
609            i += 1;
610            continue;
611        }
612        break;
613    }
614    if i < bytes.len() && bytes[i] == b':' {
615        i += 1;
616    }
617    (magic, &elem[i..])
618}
619
620/// Strip `:(magic)` / `:magic` prefix when not in literal-global mode.
621fn parse_element_magic(elem: &str) -> (PathspecMagic, &str) {
622    if !elem.starts_with(':') || literal_global() {
623        return (PathspecMagic::default(), elem);
624    }
625    if let Some(rest) = elem.strip_prefix(":(") {
626        return parse_long_magic(rest).unwrap_or((PathspecMagic::default(), elem));
627    }
628    parse_short_magic(elem)
629}
630
631fn combine_magic(element: PathspecMagic) -> PathspecMagic {
632    let mut m = element;
633    if literal_global() {
634        m.literal = true;
635    }
636    if glob_global() && !m.literal {
637        m.glob = true;
638    }
639    if icase_global() {
640        m.icase = true;
641    }
642    if noglob_global() && !m.glob {
643        m.literal = true;
644    }
645    m
646}
647
648fn strip_top_magic(mut pattern: &str) -> &str {
649    if let Some(r) = pattern.strip_prefix(":/") {
650        pattern = r;
651    }
652    pattern
653}
654
655/// Path prefix used for Bloom-filter lookups (`revision.c` `convert_pathspec_to_bloom_keyvec`).
656///
657/// `cwd_from_repo_root` is the path from the repository work tree to the process cwd, using `/`
658/// separators and no leading slash (empty string at repo root). Used for `:(top)` / `:/`.
659#[must_use]
660pub fn bloom_lookup_prefix_with_cwd(
661    spec: &str,
662    cwd_from_repo_root: Option<&str>,
663) -> Option<String> {
664    let (elem_magic, raw_pattern) = parse_element_magic(spec);
665    let magic = combine_magic(elem_magic);
666    if magic.exclude || magic.icase {
667        return None;
668    }
669    let pattern = strip_top_magic(raw_pattern);
670    if pattern.is_empty() {
671        return None;
672    }
673    let combined = if magic.top {
674        let cwd = cwd_from_repo_root.unwrap_or("").trim_end_matches('/');
675        if cwd.is_empty() {
676            pattern.to_string()
677        } else {
678            format!("{cwd}/{pattern}")
679        }
680    } else if let Some(prefix) = magic.prefix.as_deref() {
681        let base = prefix.trim_end_matches('/');
682        if base.is_empty() {
683            pattern.to_string()
684        } else {
685            normalize_relative_path_str(&format!("{base}/{pattern}"))
686        }
687    } else {
688        pattern.to_string()
689    };
690    let pattern = combined.as_str();
691    let mut len = simple_length(pattern);
692    if len != pattern.len() {
693        while len > 0 && pattern.as_bytes()[len - 1] != b'/' {
694            len -= 1;
695        }
696    }
697    while len > 0 && pattern.as_bytes()[len - 1] == b'/' {
698        len -= 1;
699    }
700    if len == 0 {
701        return None;
702    }
703    Some(combined[..len].to_string())
704}
705
706/// Bloom filter lookup prefix for a pathspec at the repository root.
707#[must_use]
708pub fn bloom_lookup_prefix(spec: &str) -> Option<String> {
709    bloom_lookup_prefix_with_cwd(spec, None)
710}
711
712/// Whether every pathspec can participate in Bloom precomputation (Git `forbid_bloom_filters`).
713#[must_use]
714pub fn pathspecs_allow_bloom(specs: &[String]) -> bool {
715    specs.iter().all(|s| {
716        !s.is_empty() && !pathspec_is_exclude(s) && bloom_lookup_prefix_with_cwd(s, None).is_some()
717    })
718}
719
720/// Whether `path` is included when Git applies a pathspec list with optional `:(exclude)` entries.
721///
722/// A path is rejected if any exclude pathspec matches it. When at least one non-exclude pathspec is
723/// present, the path must also match one of those positives (`OR` semantics).
724#[must_use]
725pub fn path_allowed_by_pathspec_list(specs: &[String], path: &str) -> bool {
726    let mut has_positive = false;
727    let mut positive_match = false;
728    for s in specs {
729        let (elem, raw_pattern) = parse_element_magic(s);
730        let magic = combine_magic(elem);
731        if magic.exclude {
732            if path_matches_pathspec_tail(raw_pattern, path, magic) {
733                return false;
734            }
735            continue;
736        }
737        has_positive = true;
738        if pathspec_matches(s, path) {
739            positive_match = true;
740        }
741    }
742    !has_positive || positive_match
743}
744
745/// True when `spec` matches `path` for pathspec bookkeeping (positive match or exclude hit).
746#[must_use]
747pub fn pathspec_contributes_match(spec: &str, path: &str) -> bool {
748    pathspec_matches(spec, path) || pathspec_exclude_matches(spec, path)
749}
750
751/// Path under `prefix:` magic (e.g. `sub/`) for repository-relative `path`, including the
752/// directory entry `sub` when the prefix is `sub/`.
753fn path_for_match_under_magic_prefix<'a>(path: &'a str, prefix: &str) -> Option<&'a str> {
754    let base = prefix.trim_end_matches('/');
755    if base.is_empty() {
756        return Some(path);
757    }
758    if path == base {
759        return Some("");
760    }
761    if let Some(rest) = path.strip_prefix(prefix) {
762        return Some(rest);
763    }
764    let anchored = if prefix.ends_with('/') {
765        prefix.to_string()
766    } else {
767        format!("{prefix}/")
768    };
769    path.starts_with(&anchored).then(|| &path[anchored.len()..])
770}
771
772fn path_matches_pathspec_tail(raw_pattern: &str, path: &str, magic: PathspecMagic) -> bool {
773    if magic.literal && magic.glob {
774        return false;
775    }
776    let pattern = strip_top_magic(raw_pattern);
777    let path_for_match = if let Some(prefix) = magic.prefix.as_deref() {
778        match path_for_match_under_magic_prefix(path, prefix) {
779            Some(rest) => rest,
780            None => return false,
781        }
782    } else {
783        path
784    };
785    pathspec_matches_tail(pattern, path_for_match, magic)
786}
787
788/// True if `path` is matched by `spec` (Git pathspec syntax, including magic and globals).
789///
790/// Same as [`matches_pathspec`] (default file context; exclude specs never match positively here).
791/// See [`matches_pathspec_list`].
792#[must_use]
793pub fn pathspec_matches(spec: &str, path: &str) -> bool {
794    matches_pathspec(spec, path)
795}
796
797/// Returns whether `spec` uses Git's exclude magic (`:(exclude)`, `:!`, `:^`, etc.).
798#[must_use]
799pub fn pathspec_is_exclude(spec: &str) -> bool {
800    let (elem_magic, _) = parse_element_magic(spec);
801    combine_magic(elem_magic).exclude
802}
803
804/// Whether a path-limited revision tree walk should recurse into directory `path`.
805///
806/// Uses [`pathspec_wants_descent_into_tree`] for non-exclude specs. Exclude-only lists never
807/// trigger that helper alone, but Git's implicit match-all still requires walking non-excluded
808/// directories so nested paths can match [`matches_pathspec_list`].
809#[must_use]
810pub fn pathspec_wants_descent_for_rev_tree(path: &str, specs: &[String]) -> bool {
811    if specs.is_empty() {
812        return true;
813    }
814    for spec in specs {
815        if !pathspec_is_exclude(spec) && pathspec_wants_descent_into_tree(spec, path) {
816            return true;
817        }
818    }
819    if !specs.iter().any(|s| pathspec_is_exclude(s)) {
820        return false;
821    }
822    let ctx = PathspecMatchContext {
823        is_directory: true,
824        is_git_submodule: false,
825        precompose_paths: false,
826    };
827    for spec in specs {
828        if pathspec_is_exclude(spec)
829            && pathspec_exclude_element_matches_with_context(spec, path, ctx)
830        {
831            return false;
832        }
833    }
834    !specs.iter().any(|s| !pathspec_is_exclude(s))
835}
836
837/// Whether a path-limited tree walk must look at entry `path`: it matches `specs`, or it is a
838/// directory that [`pathspec_wants_descent_for_rev_tree`] says may hold a match.
839pub(crate) fn pathspec_entry_is_relevant(path: &str, is_tree: bool, specs: &[String]) -> bool {
840    matches_pathspec_list(path, specs)
841        || (is_tree && pathspec_wants_descent_for_rev_tree(path, specs))
842}
843
844/// Whether a path-limited tree walk should recurse into directory `full_name` for pathspec `spec`.
845///
846/// Used by revision history path limiting ([`crate::revwalk::rev_list`]) and diff tree pruning.
847/// Matches Git `tree-walk.c` `do_match` recursive
848/// directory interest: a directory is entered when a positive spec's leading literal segment
849/// matches it, or when the pattern begins with a glob metacharacter (`simple_length` zero).
850///
851/// Exclude-only patterns never trigger descent alone.
852#[must_use]
853pub fn pathspec_wants_descent_into_tree(spec: &str, full_name: &str) -> bool {
854    if pathspec_is_exclude(spec) {
855        return false;
856    }
857    let (elem_magic, raw_pattern) = parse_element_magic(spec);
858    let magic = combine_magic(elem_magic);
859    if magic.exclude {
860        return false;
861    }
862    let pattern = strip_top_magic(raw_pattern);
863    let pattern = pattern.strip_prefix("./").unwrap_or(pattern);
864    if pattern.is_empty() || pattern == "." {
865        return true;
866    }
867    if let Some(prefix) = magic.prefix.as_deref() {
868        let base = prefix.trim_end_matches('/');
869        if !base.is_empty() && (full_name == base || base.starts_with(&format!("{full_name}/"))) {
870            return true;
871        }
872    }
873    let dir_prefix = format!("{full_name}/");
874    if pattern.starts_with(&dir_prefix) {
875        return true;
876    }
877    if ps_str_eq(pattern, full_name, magic.icase) {
878        return true;
879    }
880    pathspec_recursive_descent_wanted(pattern, magic, full_name)
881}
882
883fn pathspec_literal_descent_wanted(pattern: &str, dir: &str) -> bool {
884    if pattern.is_empty() {
885        return false;
886    }
887    if let Some(prefix) = pattern.strip_suffix('/') {
888        return dir == prefix || path_is_under_dir(dir, prefix, false);
889    }
890    if dir == pattern {
891        return false;
892    }
893    dir.len() < pattern.len()
894        && path_starts_with(pattern, dir, false)
895        && pattern.as_bytes()[dir.len()] == b'/'
896}
897
898fn pathspec_recursive_descent_wanted(pattern: &str, magic: PathspecMagic, dir: &str) -> bool {
899    if magic.literal {
900        return pathspec_literal_descent_wanted(pattern, dir);
901    }
902    if simple_length(pattern) == 0 {
903        return true;
904    }
905
906    let icase = magic.icase;
907    let glob = magic.glob;
908    let components: Vec<&str> = dir.split('/').filter(|c| !c.is_empty()).collect();
909    pathspec_descent_match_components(pattern, icase, glob, &components)
910}
911
912fn pathspec_first_component(pat: &str) -> (&str, &str) {
913    match pat.find('/') {
914        Some(i) => (&pat[..i], &pat[i + 1..]),
915        None => (pat, ""),
916    }
917}
918
919fn pathspec_dir_name_matches_component(
920    pat_comp: &str,
921    dir_name: &str,
922    icase: bool,
923    glob: bool,
924) -> bool {
925    if pat_comp.is_empty() {
926        return true;
927    }
928    let nwl = simple_length(pat_comp);
929    if nwl == 0 {
930        return true;
931    }
932    if nwl >= pat_comp.len() {
933        return ps_str_eq(pat_comp, dir_name, icase);
934    }
935    let lit = &pat_comp[..nwl];
936    let lit_bytes = lit.as_bytes();
937    let dir_bytes = dir_name.as_bytes();
938    if icase {
939        if dir_bytes.len() < lit_bytes.len()
940            || !dir_bytes[..lit_bytes.len()].eq_ignore_ascii_case(lit_bytes)
941        {
942            return false;
943        }
944    } else if dir_bytes.len() < lit_bytes.len() || &dir_bytes[..lit_bytes.len()] != lit_bytes {
945        return false;
946    }
947    let flags = if icase { WM_CASEFOLD } else { 0 };
948    let wm_flags = if glob { flags | WM_PATHNAME } else { flags };
949    wildmatch(pat_comp.as_bytes(), dir_name.as_bytes(), wm_flags)
950}
951
952fn pathspec_descent_match_components(
953    pat: &str,
954    icase: bool,
955    glob: bool,
956    dir_components: &[&str],
957) -> bool {
958    if dir_components.is_empty() {
959        return simple_length(pat) == 0 || !pat.is_empty();
960    }
961    if simple_length(pat) == 0 {
962        return true;
963    }
964    let (pat_comp, pat_rest) = pathspec_first_component(pat);
965    if !pathspec_dir_name_matches_component(pat_comp, dir_components[0], icase, glob) {
966        return false;
967    }
968    pathspec_descent_match_components(pat_rest, icase, glob, &dir_components[1..])
969}
970
971#[cfg(test)]
972mod pathspec_descent_tests {
973    use super::*;
974
975    #[test]
976    fn descent_wildcard_non_final_component() {
977        assert!(pathspec_wants_descent_into_tree("*/g.c", "a"));
978        assert!(pathspec_wants_descent_into_tree("a/*/f.c", "a"));
979        assert!(pathspec_wants_descent_into_tree("a/*/f.c", "a/b"));
980        assert!(pathspec_wants_descent_into_tree(":(glob)**/f.c", "a"));
981        assert!(pathspec_wants_descent_into_tree(":(glob)*/g.c", "a"));
982        assert!(pathspec_wants_descent_into_tree(":(icase)A/g.c", "a"));
983        assert!(!pathspec_wants_descent_into_tree(":(literal)*g.c", "a"));
984        assert!(!pathspec_wants_descent_into_tree("a/*.c", "b"));
985        assert!(pathspec_wants_descent_into_tree("b*/log.c", "builtin"));
986        assert!(!pathspec_wants_descent_into_tree("b*/log.c", "other"));
987        assert!(pathspec_wants_descent_into_tree("bu?ltin/log.c", "builtin"));
988        assert!(pathspec_wants_descent_into_tree(
989            ":(icase)B*/LOG.C",
990            "builtin"
991        ));
992        assert!(!pathspec_wants_descent_into_tree(":(icase)b*/log.c", "é"));
993    }
994
995    #[test]
996    fn glob_doublestar_path_matches_nested_file() {
997        let specs = vec![":(glob)**/f.c".to_owned()];
998        assert!(matches_pathspec_list("a/b/f.c", &specs));
999    }
1000}
1001
1002/// Like [`matches_pathspec_set_for_object`], but uses [`matches_ls_tree_pathspec`] for each
1003/// element so `ls-files` / index filtering agrees with `ls-tree` on patterns such as `a[a]`.
1004#[must_use]
1005pub fn matches_pathspec_set_for_object_ls_tree(
1006    specs: &[String],
1007    path: &str,
1008    mode: u32,
1009    attr_rules: &[AttrRule],
1010) -> bool {
1011    if specs.is_empty() {
1012        return true;
1013    }
1014    let mut positives: Vec<&str> = Vec::new();
1015    let mut excludes: Vec<&str> = Vec::new();
1016    for s in specs {
1017        if pathspec_is_exclude(s) {
1018            excludes.push(s.as_str());
1019        } else {
1020            positives.push(s.as_str());
1021        }
1022    }
1023    let positive_ok = if positives.is_empty() {
1024        true
1025    } else {
1026        positives
1027            .iter()
1028            .any(|s| matches_ls_tree_pathspec(s, path, mode, attr_rules))
1029    };
1030    if !positive_ok {
1031        return false;
1032    }
1033    for ex in excludes {
1034        if matches_ls_tree_pathspec(ex, path, mode, attr_rules) {
1035            return false;
1036        }
1037    }
1038    true
1039}
1040
1041/// True if `path` matches the combined pathspec list: any positive spec (or all paths when there
1042/// are only excludes, matching Git `parse_pathspec`), and not matched by any exclude spec.
1043#[must_use]
1044pub fn matches_pathspec_set_for_object(
1045    specs: &[String],
1046    path: &str,
1047    mode: u32,
1048    attr_rules: &[AttrRule],
1049) -> bool {
1050    if specs.is_empty() {
1051        return true;
1052    }
1053    let mut positives: Vec<&str> = Vec::new();
1054    let mut excludes: Vec<&str> = Vec::new();
1055    for s in specs {
1056        if pathspec_is_exclude(s) {
1057            excludes.push(s.as_str());
1058        } else {
1059            positives.push(s.as_str());
1060        }
1061    }
1062    let positive_ok = if positives.is_empty() {
1063        true
1064    } else {
1065        positives
1066            .iter()
1067            .any(|s| matches_pathspec_for_object(s, path, mode, attr_rules))
1068    };
1069    if !positive_ok {
1070        return false;
1071    }
1072    for ex in excludes {
1073        if matches_pathspec_for_object(ex, path, mode, attr_rules) {
1074            return false;
1075        }
1076    }
1077    true
1078}
1079
1080/// True if `spec` uses `:(top)` or short `:/` (repo-root-relative) magic.
1081#[must_use]
1082pub fn pathspec_has_top(spec: &str) -> bool {
1083    let (elem_magic, _) = parse_element_magic(spec);
1084    combine_magic(elem_magic).top
1085}
1086
1087/// True if `spec` carries an `attr:` clause in long-form `:(...)` magic, so matching it needs
1088/// `.gitattributes` rules. Malformed clauses count too; [`validate_attr_pathspecs`] rejects them.
1089///
1090/// Such specs only match through the attribute-aware entry points:
1091/// [`crate::repo::Repository::pathspec_matcher`], which loads the repository's attributes and
1092/// validates the clauses, or [`matches_pathspec_list_for_object`] with caller-supplied rules.
1093#[must_use]
1094pub fn pathspec_has_attr_magic(spec: &str) -> bool {
1095    // Attr magic only exists in long form; skip the allocating parse for every other spec.
1096    spec.starts_with(":(") && spec.contains("attr:") && !attr_clauses(spec).is_empty()
1097}
1098
1099/// A resolved pathspec list bound to one repository's matching rules, built by
1100/// [`crate::repo::Repository::pathspec_matcher`].
1101///
1102/// Holds the repository's attributes when a spec carries `:(attr:...)` magic, so testing many
1103/// paths reads them once. Attr requirements see `[attr]` macro expansion and match attribute
1104/// patterns case-insensitively when `core.ignorecase` is set, as Git does. Paths are treated as regular files: a spec ending in `/` does not match
1105/// a path equal to its directory, and `builtin_objectmode` reads as `100644`. Use
1106/// [`matches_pathspec_list_for_object`] when the entry mode matters.
1107#[derive(Debug, Clone)]
1108pub struct PathspecMatcher<'s> {
1109    specs: Cow<'s, [String]>,
1110    precompose: bool,
1111    attrs: Option<std::sync::Arc<crate::attributes::ParsedGitAttributes>>,
1112    /// Match attribute patterns case-insensitively (`core.ignorecase`).
1113    attr_icase: bool,
1114}
1115
1116impl<'s> PathspecMatcher<'s> {
1117    pub(crate) fn new(
1118        specs: &'s [String],
1119        precompose: bool,
1120        attrs: Option<std::sync::Arc<crate::attributes::ParsedGitAttributes>>,
1121        attr_icase: bool,
1122    ) -> Self {
1123        // The attribute-aware matcher has no precompose context, so normalize the specs up front.
1124        let specs = if precompose && attrs.is_some() {
1125            Cow::Owned(
1126                specs
1127                    .iter()
1128                    .map(|spec| precompose_utf8_path(spec).into_owned())
1129                    .collect(),
1130            )
1131        } else {
1132            Cow::Borrowed(specs)
1133        };
1134        Self {
1135            specs,
1136            precompose,
1137            attrs,
1138            attr_icase,
1139        }
1140    }
1141
1142    /// Whether `path` is selected: it matches at least one positive spec (any path when there
1143    /// are none or the list is empty) and no `:(exclude)` spec.
1144    #[must_use]
1145    pub fn matches(&self, path: &str) -> bool {
1146        const REGULAR_FILE: u32 = 0o100644;
1147        let Some(attrs) = &self.attrs else {
1148            return matches_pathspec_list_with_context(
1149                path,
1150                &self.specs,
1151                pathspec_match_context_with_precompose(self.precompose),
1152            );
1153        };
1154        let path = if self.precompose {
1155            precompose_utf8_path(path)
1156        } else {
1157            Cow::Borrowed(path)
1158        };
1159        let lookup = AttrLookup::new(&attrs.rules, &attrs.macros, self.attr_icase);
1160        matches_pathspec_list_with_attrs(&path, REGULAR_FILE, lookup, &self.specs)
1161    }
1162
1163    /// Whether a changed file is selected, checking its current `path` and, for a rename or
1164    /// copy, its `old_path` too (Git limits diffs on either side of a rename).
1165    #[must_use]
1166    pub fn matches_change(&self, path: &str, old_path: Option<&str>) -> bool {
1167        self.specs.is_empty()
1168            || self.matches(path)
1169            || old_path.is_some_and(|old| old != path && self.matches(old))
1170    }
1171
1172    /// Whether the list is empty (every path is selected).
1173    #[must_use]
1174    pub fn is_empty(&self) -> bool {
1175        self.specs.is_empty()
1176    }
1177}
1178
1179fn pathspec_match_one_positive(path: &str, magic: PathspecMagic, raw_pattern: &str) -> bool {
1180    if magic.literal && magic.glob {
1181        return false;
1182    }
1183    let pattern = strip_top_magic(raw_pattern);
1184    let path_for_match = if let Some(prefix) = magic.prefix.as_deref() {
1185        match path_for_match_under_magic_prefix(path, prefix) {
1186            Some(rest) => rest,
1187            None => return false,
1188        }
1189    } else {
1190        path
1191    };
1192    pathspec_matches_tail(pattern, path_for_match, magic)
1193}
1194
1195/// Attribute state that `:(attr:...)` requirements are evaluated against.
1196#[derive(Clone, Copy)]
1197pub(crate) struct AttrLookup<'a> {
1198    rules: &'a [AttrRule],
1199    /// `[attr]` macro definitions; `None` means no macros are defined.
1200    macros: Option<&'a MacroTable>,
1201    /// Match attribute patterns case-insensitively (`core.ignorecase`).
1202    icase: bool,
1203}
1204
1205impl<'a> AttrLookup<'a> {
1206    pub(crate) fn new(rules: &'a [AttrRule], macros: &'a MacroTable, icase: bool) -> Self {
1207        Self {
1208            rules,
1209            macros: Some(macros),
1210            icase,
1211        }
1212    }
1213
1214    fn rules_only(rules: &'a [AttrRule]) -> Self {
1215        Self {
1216            rules,
1217            macros: None,
1218            icase: false,
1219        }
1220    }
1221}
1222
1223fn attr_requirements_match(
1224    requirements: &[AttrRequirement],
1225    attrs: AttrLookup<'_>,
1226    path: &str,
1227    is_dir: bool,
1228    mode: u32,
1229) -> bool {
1230    let no_macros = MacroTable::default();
1231    let assigned = path_gitattribute_values(
1232        attrs.rules,
1233        attrs.macros.unwrap_or(&no_macros),
1234        attrs.icase,
1235        path,
1236        is_dir,
1237    );
1238    requirements.iter().all(|req| {
1239        let value: Option<Cow<'_, str>> = if req.name() == "builtin_objectmode" {
1240            (mode != 0).then(|| Cow::Owned(format!("{mode:06o}")))
1241        } else {
1242            assigned
1243                .get(req.name())
1244                .map(|v| Cow::Borrowed(v.display()))
1245                .filter(|v| v != "unspecified")
1246        };
1247        match req {
1248            AttrRequirement::Set(_) => value.as_deref() == Some("set"),
1249            AttrRequirement::Unset(_) => value.as_deref() == Some("unset"),
1250            AttrRequirement::Unspecified(_) => value.is_none(),
1251            AttrRequirement::Value(_, expected) => value.as_deref() == Some(expected.as_str()),
1252        }
1253    })
1254}
1255
1256fn matches_pathspec_element_with_context(
1257    spec: &str,
1258    path: &str,
1259    ctx: PathspecMatchContext,
1260) -> bool {
1261    let (elem_magic, raw_pattern) = parse_element_magic(spec);
1262    let magic = combine_magic(elem_magic);
1263    if magic.exclude {
1264        return false;
1265    }
1266    if magic.literal && magic.glob {
1267        return false;
1268    }
1269    if !magic.attr_requirements.is_empty() {
1270        return false;
1271    }
1272    if magic.literal || magic.glob || magic.icase {
1273        return pathspec_matches(spec, path);
1274    }
1275    let pattern = strip_top_magic(raw_pattern);
1276    let path_for_match = if let Some(prefix) = magic.prefix.as_deref() {
1277        match path_for_match_under_magic_prefix(path, prefix) {
1278            Some(rest) => rest,
1279            None => return false,
1280        }
1281    } else {
1282        path
1283    };
1284    matches_pathspec_with_context(pattern, path_for_match, ctx)
1285}
1286
1287fn pathspec_exclude_element_matches_with_context(
1288    spec: &str,
1289    path: &str,
1290    ctx: PathspecMatchContext,
1291) -> bool {
1292    let (elem_magic, raw_pattern) = parse_element_magic(spec);
1293    let mut magic = combine_magic(elem_magic);
1294    if !magic.exclude {
1295        return false;
1296    }
1297    magic.exclude = false;
1298    if magic.literal && magic.glob {
1299        return false;
1300    }
1301    if !magic.attr_requirements.is_empty() {
1302        // Attribute pathspecs need `.gitattributes` context; use
1303        // [`matches_pathspec_list_for_object`] for those.
1304        return false;
1305    }
1306    if magic.literal || magic.glob || magic.icase {
1307        return pathspec_match_one_positive(path, magic, raw_pattern);
1308    }
1309    let pattern = strip_top_magic(raw_pattern);
1310    let path_for_match = if let Some(prefix) = magic.prefix.as_deref() {
1311        match path_for_match_under_magic_prefix(path, prefix) {
1312            Some(rest) => rest,
1313            None => return false,
1314        }
1315    } else {
1316        path
1317    };
1318    matches_pathspec_with_context(pattern, path_for_match, ctx)
1319}
1320
1321/// True if `path` is matched by an exclude pathspec's pattern. Returns `false` if `spec` is not
1322/// an exclude pathspec.
1323#[must_use]
1324pub fn pathspec_exclude_matches(spec: &str, path: &str) -> bool {
1325    pathspec_exclude_element_matches_with_context(spec, path, PathspecMatchContext::default())
1326}
1327
1328/// When every pathspec is an exclude and none use `:(top)` / `:/`, Git prepends an implicit
1329/// positive that matches only under the process cwd (relative to the work tree), not the whole
1330/// repository (`PATHSPEC_PREFER_CWD` in `pathspec.c`). `cwd_from_repo_root` is that prefix
1331/// without a trailing slash, or empty at the work tree root.
1332#[must_use]
1333pub fn extend_pathspec_list_implicit_cwd(
1334    specs: &[String],
1335    cwd_from_repo_root: Option<&str>,
1336) -> Vec<String> {
1337    if specs.is_empty() {
1338        return specs.to_vec();
1339    }
1340    if !specs.iter().all(|s| pathspec_is_exclude(s)) {
1341        return specs.to_vec();
1342    }
1343    let any_top = specs.iter().any(|s| pathspec_has_top(s));
1344    if any_top {
1345        return specs.to_vec();
1346    }
1347    let Some(cwd) = cwd_from_repo_root.map(str::trim).filter(|s| !s.is_empty()) else {
1348        return specs.to_vec();
1349    };
1350    let cwd = cwd.trim_end_matches('/');
1351    if cwd.is_empty() {
1352        return specs.to_vec();
1353    }
1354    let mut out = Vec::with_capacity(specs.len() + 1);
1355    // Git `parse_pathspec` with `PATHSPEC_PREFER_CWD` adds `.` (see pathspec.c), not `{cwd}/`.
1356    out.push(".".to_owned());
1357    out.extend_from_slice(specs);
1358    out
1359}
1360
1361/// Git `match_pathspec` semantics over a pathspec list: OR of positive specs minus OR of exclude
1362/// specs. If every element is exclude-only, Git implicitly prepends `.` (match all); this
1363/// function does the same.
1364///
1365/// No `.gitattributes` rules are available here, so a positive `:(attr:...)` spec never matches
1366/// and an exclude `:(exclude,attr:...)` spec never excludes. Use
1367/// [`crate::repo::Repository::paths_match_pathspecs`] or [`matches_pathspec_list_for_object`]
1368/// when specs may carry attr magic (see [`pathspec_has_attr_magic`]).
1369#[must_use]
1370pub fn matches_pathspec_list(path: &str, specs: &[String]) -> bool {
1371    matches_pathspec_list_with_context(path, specs, PathspecMatchContext::default())
1372}
1373
1374/// Like [`matches_pathspec_list`], but uses `ctx` for non-magic pathspec elements (trailing `/`).
1375///
1376/// Like [`matches_pathspec_list`], this does not evaluate `:(attr:...)` magic: positive attr specs
1377/// never match and attr excludes never exclude.
1378#[must_use]
1379pub fn matches_pathspec_list_with_context(
1380    path: &str,
1381    specs: &[String],
1382    ctx: PathspecMatchContext,
1383) -> bool {
1384    if specs.is_empty() {
1385        return true;
1386    }
1387    let mut has_exclude = false;
1388    let mut has_positive = false;
1389    let mut positive = false;
1390    for spec in specs {
1391        if pathspec_is_exclude(spec) {
1392            has_exclude = true;
1393        } else {
1394            has_positive = true;
1395            positive = positive || matches_pathspec_element_with_context(spec, path, ctx);
1396        }
1397    }
1398    if has_positive && !positive {
1399        return false;
1400    }
1401    if !has_exclude {
1402        return true;
1403    }
1404    let excluded = specs.iter().any(|s| {
1405        pathspec_is_exclude(s) && pathspec_exclude_element_matches_with_context(s, path, ctx)
1406    });
1407    !excluded
1408}
1409
1410/// Whether any of `sorted_paths` matches the single pathspec `spec`.
1411///
1412/// Same result as testing every path with [`matches_pathspec_list_with_context`] and `[spec]`,
1413/// but a plain literal pathspec (no magic, no wildcards, no `./`) is answered with a binary
1414/// search instead of a scan, so checking many pathspecs against a large index stays cheap.
1415///
1416/// `sorted_paths` must be sorted in byte order (the [`str`] ordering) for the result to be
1417/// correct; duplicates are allowed.
1418pub(crate) fn sorted_paths_match_pathspec<S: AsRef<str>>(
1419    spec: &str,
1420    sorted_paths: &[S],
1421    ctx: PathspecMatchContext,
1422) -> bool {
1423    let list = [spec.to_owned()];
1424    let matches_linearly = |p: &S| matches_pathspec_list_with_context(p.as_ref(), &list, ctx);
1425    if !is_plain_literal_pathspec(spec, ctx) {
1426        return sorted_paths.iter().any(matches_linearly);
1427    }
1428    let spec_nfc = literal_spec_for_lookup(spec, ctx);
1429    if sorted_literal_lookup(&spec_nfc, sorted_paths) {
1430        return true;
1431    }
1432    // A hit on the NFC spelling is always a match; a path stored in another normalization form
1433    // can only match through a non-ASCII segment, so a miss rechecks just those paths.
1434    ctx.precompose_paths
1435        && sorted_paths
1436            .iter()
1437            .filter(|p| !p.as_ref().is_ascii())
1438            .any(matches_linearly)
1439}
1440
1441/// Binary-search `sorted_paths` for the plain literal `spec` (exact path or directory prefix).
1442fn sorted_literal_lookup<S: AsRef<str>>(spec: &str, sorted_paths: &[S]) -> bool {
1443    let has_path_with_prefix = |prefix: &str| {
1444        let at = sorted_paths.partition_point(|p| p.as_ref() < prefix);
1445        sorted_paths
1446            .get(at)
1447            .is_some_and(|p| p.as_ref().starts_with(prefix))
1448    };
1449    if spec.ends_with('/') {
1450        return has_path_with_prefix(spec);
1451    }
1452    let at = sorted_paths.partition_point(|p| p.as_ref() < spec);
1453    if sorted_paths.get(at).is_some_and(|p| p.as_ref() == spec) {
1454        return true;
1455    }
1456    has_path_with_prefix(&format!("{spec}/"))
1457}
1458
1459/// The spelling of a plain literal `spec` that [`matches_pathspec_with_context`] compares paths
1460/// against: NFC when the context precomposes paths, else `spec` itself.
1461fn literal_spec_for_lookup(spec: &str, ctx: PathspecMatchContext) -> Cow<'_, str> {
1462    if ctx.precompose_paths {
1463        precompose_utf8_path(spec)
1464    } else {
1465        Cow::Borrowed(spec)
1466    }
1467}
1468
1469/// Repository-relative paths when every spec is a plain literal file path (no trailing `/`).
1470///
1471/// Used to narrow index/worktree diffs to a small path set without scanning the full index.
1472/// Returns `None` when `ctx` precomposes paths: another normalization form of a spec could also
1473/// match, and only a scan of the candidate paths can find it.
1474pub(crate) fn exact_literal_pathspec_set(
1475    specs: &[String],
1476    ctx: PathspecMatchContext,
1477) -> Option<std::collections::BTreeSet<String>> {
1478    if specs.is_empty() || ctx.precompose_paths {
1479        return None;
1480    }
1481    if !specs
1482        .iter()
1483        .all(|s| is_plain_literal_pathspec(s, ctx) && !s.ends_with('/'))
1484    {
1485        return None;
1486    }
1487    Some(specs.iter().cloned().collect())
1488}
1489
1490/// Whether `spec` matches exactly the path `spec` and everything under directory `spec`
1491/// (no magic, no wildcards, no `./`, and no context that changes trailing-`/` handling).
1492///
1493/// With [`PathspecMatchContext::precompose_paths`], "the path `spec`" means equality after NFC
1494/// normalization of both sides; see [`literal_spec_for_lookup`].
1495fn is_plain_literal_pathspec(spec: &str, ctx: PathspecMatchContext) -> bool {
1496    !ctx.is_directory
1497        && !ctx.is_git_submodule
1498        && !spec.is_empty()
1499        && spec != "."
1500        && !spec.starts_with("./")
1501        && !spec.starts_with(':')
1502        && simple_length(spec) == spec.len()
1503        && !(literal_global() || glob_global() || noglob_global() || icase_global())
1504}
1505
1506/// A pathspec list prepared once for matching many paths with the same context.
1507///
1508/// Gives the same answers as [`matches_pathspec_list_with_context`] with the list and context
1509/// it was built from. When every element is a plain literal (see
1510/// [`sorted_paths_match_pathspec`]), a path is answered with hash lookups of the path and its
1511/// leading directories instead of comparing it against every element, which keeps matching a
1512/// large index against many pathspecs linear in the index size.
1513pub(crate) struct PathspecListMatcher<'a> {
1514    specs: &'a [String],
1515    ctx: PathspecMatchContext,
1516    literals: Option<LiteralPathspecSet>,
1517}
1518
1519/// Plain literal pathspecs split by how they match, spelled as [`literal_spec_for_lookup`] does.
1520struct LiteralPathspecSet {
1521    /// Specs without a trailing `/`: they match the path itself.
1522    exact: HashSet<String>,
1523    /// Directories whose contents match: every spec, with one trailing `/` removed.
1524    dirs: HashSet<String>,
1525}
1526
1527impl<'a> PathspecListMatcher<'a> {
1528    /// Prepare `specs` for repeated matching under `ctx`.
1529    pub(crate) fn new(specs: &'a [String], ctx: PathspecMatchContext) -> Self {
1530        let all_plain =
1531            !specs.is_empty() && specs.iter().all(|s| is_plain_literal_pathspec(s, ctx));
1532        let literals = all_plain.then(|| {
1533            let mut set = LiteralPathspecSet {
1534                exact: HashSet::with_capacity(specs.len()),
1535                dirs: HashSet::with_capacity(specs.len()),
1536            };
1537            for spec in specs {
1538                let spec = literal_spec_for_lookup(spec, ctx);
1539                match spec.strip_suffix('/') {
1540                    Some(dir) => {
1541                        set.dirs.insert(dir.to_owned());
1542                    }
1543                    None => {
1544                        set.exact.insert(spec.to_string());
1545                        set.dirs.insert(spec.into_owned());
1546                    }
1547                }
1548            }
1549            set
1550        });
1551        Self {
1552            specs,
1553            ctx,
1554            literals,
1555        }
1556    }
1557
1558    /// Whether `path` matches the pathspec list.
1559    pub(crate) fn matches(&self, path: &str) -> bool {
1560        let Some(set) = &self.literals else {
1561            return matches_pathspec_list_with_context(path, self.specs, self.ctx);
1562        };
1563        let path = literal_spec_for_lookup(path, self.ctx);
1564        let path = path.as_ref();
1565        if set.exact.contains(path) {
1566            return true;
1567        }
1568        path.bytes()
1569            .enumerate()
1570            .any(|(i, b)| b == b'/' && set.dirs.contains(&path[..i]))
1571    }
1572}
1573
1574/// `matches_pathspec_list` for tree/index objects with mode and `.gitattributes` rules.
1575///
1576/// `attr_rules` are applied without `[attr]` macro definitions and with case-sensitive
1577/// patterns; [`crate::repo::Repository::pathspec_matcher`] also honors macros and
1578/// `core.ignorecase`.
1579#[must_use]
1580pub fn matches_pathspec_list_for_object(
1581    path: &str,
1582    mode: u32,
1583    attr_rules: &[AttrRule],
1584    specs: &[String],
1585) -> bool {
1586    matches_pathspec_list_with_attrs(path, mode, AttrLookup::rules_only(attr_rules), specs)
1587}
1588
1589pub(crate) fn matches_pathspec_list_with_attrs(
1590    path: &str,
1591    mode: u32,
1592    attrs: AttrLookup<'_>,
1593    specs: &[String],
1594) -> bool {
1595    if specs.is_empty() {
1596        return true;
1597    }
1598    let mut has_exclude = false;
1599    let mut has_positive = false;
1600    let mut positive = false;
1601    for spec in specs {
1602        if pathspec_is_exclude(spec) {
1603            has_exclude = true;
1604        } else {
1605            has_positive = true;
1606            positive = positive || matches_pathspec_for_object_with_attrs(spec, path, mode, attrs);
1607        }
1608    }
1609    if has_positive && !positive {
1610        return false;
1611    }
1612    if !has_exclude {
1613        return true;
1614    }
1615    !specs.iter().any(|s| {
1616        pathspec_is_exclude(s) && matches_pathspec_exclude_for_object(s, path, mode, attrs)
1617    })
1618}
1619
1620fn matches_pathspec_exclude_for_object(
1621    spec: &str,
1622    path: &str,
1623    mode: u32,
1624    attrs: AttrLookup<'_>,
1625) -> bool {
1626    let (elem_magic, raw_pattern) = parse_element_magic(spec);
1627    let mut magic = combine_magic(elem_magic);
1628    if !magic.exclude {
1629        return false;
1630    }
1631    magic.exclude = false;
1632    if magic.literal && magic.glob {
1633        return false;
1634    }
1635    let ctx = context_from_mode_bits(mode);
1636    let is_dir_for_attr = path.ends_with('/') || ctx.is_directory || ctx.is_git_submodule;
1637    if !magic.attr_requirements.is_empty()
1638        && !attr_requirements_match(&magic.attr_requirements, attrs, path, is_dir_for_attr, mode)
1639    {
1640        return false;
1641    }
1642    let pattern = strip_top_magic(raw_pattern);
1643    let path_for_match = if let Some(prefix) = magic.prefix.as_deref() {
1644        match path_for_match_under_magic_prefix(path, prefix) {
1645            Some(rest) => rest,
1646            None => return false,
1647        }
1648    } else {
1649        path
1650    };
1651    if magic.literal || magic.glob || magic.icase {
1652        pathspec_matches_tail(pattern, path_for_match, magic)
1653    } else {
1654        matches_pathspec_with_context(pattern, path_for_match, ctx)
1655    }
1656}
1657
1658fn pathspec_matches_tail(pattern: &str, path: &str, magic: PathspecMagic) -> bool {
1659    if pattern.is_empty() {
1660        return true;
1661    }
1662
1663    let flags = if magic.icase { WM_CASEFOLD } else { 0 };
1664
1665    if magic.literal {
1666        return literal_prefix_match(pattern, path);
1667    }
1668
1669    let wm_flags = if magic.glob {
1670        flags | WM_PATHNAME
1671    } else {
1672        flags
1673    };
1674
1675    let pattern_bytes = pattern.as_bytes();
1676    let path_bytes = path.as_bytes();
1677    let simple = simple_length(pattern);
1678
1679    // Git `match_pathspec_item`: exact / directory prefix before `git_fnmatch`.
1680    // Only when the pattern has no glob metacharacters (`simple_length` spans the whole pattern);
1681    // otherwise a pattern like `a[a]` must not match children via `a[a]/` prefix (t6130 vs ls-tree).
1682    if ps_str_eq(pattern, path, magic.icase) {
1683        return true;
1684    }
1685    if simple == pattern.len() {
1686        if let Some(prefix) = pattern.strip_suffix('/') {
1687            if ps_str_eq(prefix, path, magic.icase) {
1688                return true;
1689            }
1690            if path_is_under_dir(path, prefix, magic.icase) {
1691                return true;
1692            }
1693        } else if path_is_under_dir(path, pattern, magic.icase) {
1694            return true;
1695        }
1696    }
1697
1698    // `:(glob)**/*.txt` at repo root: Git matches `untracked.txt` (leading `**/` is optional).
1699    if magic.glob && !path.contains('/') && pattern.starts_with("**/") {
1700        if wildmatch(pattern_bytes, path_bytes, wm_flags) {
1701            return true;
1702        }
1703        if let Some(suffix) = pattern.strip_prefix("**/") {
1704            if wildmatch(suffix.as_bytes(), path_bytes, wm_flags) {
1705                return true;
1706            }
1707        }
1708    }
1709
1710    // Wildcard: require literal bytes up to `simple_length`, then wildmatch the tail only.
1711    if simple < pattern.len() {
1712        if path_bytes.len() < simple {
1713            return false;
1714        }
1715        let path_lit = &path_bytes[..simple];
1716        let pat_lit = &pattern_bytes[..simple];
1717        let same = if magic.icase {
1718            path_lit.eq_ignore_ascii_case(pat_lit)
1719        } else {
1720            path_lit == pat_lit
1721        };
1722        if !same {
1723            return false;
1724        }
1725        let pat_rest = &pattern[simple..];
1726        let path_rest = &path[simple..];
1727        return wildmatch(pat_rest.as_bytes(), path_rest.as_bytes(), wm_flags);
1728    }
1729
1730    ps_str_eq(pattern, path, magic.icase) || path_is_under_dir(path, pattern, magic.icase)
1731}
1732
1733fn ps_str_eq(a: &str, b: &str, icase: bool) -> bool {
1734    if icase {
1735        a.eq_ignore_ascii_case(b)
1736    } else {
1737        a == b
1738    }
1739}
1740
1741fn path_starts_with(path: &str, prefix: &str, icase: bool) -> bool {
1742    if icase {
1743        path.get(..prefix.len())
1744            .is_some_and(|head| head.eq_ignore_ascii_case(prefix))
1745    } else {
1746        path.starts_with(prefix)
1747    }
1748}
1749
1750/// Whether `path` starts with `dir` followed by `/` (the path lies inside directory `dir`).
1751fn path_is_under_dir(path: &str, dir: &str, icase: bool) -> bool {
1752    path.as_bytes().get(dir.len()) == Some(&b'/') && path_starts_with(path, dir, icase)
1753}
1754
1755fn literal_prefix_match(pattern: &str, path: &str) -> bool {
1756    if let Some(prefix) = pattern.strip_suffix('/') {
1757        return path == prefix || path_is_under_dir(path, prefix, false);
1758    }
1759    path == pattern || path_is_under_dir(path, pattern, false)
1760}
1761
1762/// Literal pathspec match for `ls-tree` when the pattern has no `*`/`?` (brackets stay literal).
1763fn ls_tree_literal_match(pattern: &str, path: &str, ctx: PathspecMatchContext) -> bool {
1764    if let Some(prefix) = pattern.strip_suffix('/') {
1765        if path_is_under_dir(path, prefix, false) {
1766            return true;
1767        }
1768        if path == prefix {
1769            return ctx.is_directory || ctx.is_git_submodule;
1770        }
1771        return false;
1772    }
1773    path == pattern || path_is_under_dir(path, pattern, false)
1774}
1775
1776/// Optional path metadata for literal pathspecs with a trailing `/` (tree-walk / diff-tree).
1777///
1778/// Git treats `dir/` as “directory or git submodule only”: a regular file `dir`
1779/// does not match, but a tree entry `dir` or gitlink `dir` does.
1780#[derive(Debug, Clone, Copy, Default, PartialEq, Eq)]
1781pub struct PathspecMatchContext {
1782    /// The index/tree entry is a directory (mode `040000`).
1783    pub is_directory: bool,
1784    /// The entry is a git submodule / gitlink (`160000`).
1785    pub is_git_submodule: bool,
1786    /// NFC-normalize pathspec and path when matching (per-repository `core.precomposeunicode`).
1787    pub precompose_paths: bool,
1788}
1789
1790/// Returns whether `path` matches the pathspec `spec` with default (file) context.
1791///
1792/// For pathspecs ending in `/`, a path equal to the prefix matches only when
1793/// [`PathspecMatchContext`] indicates a directory or submodule; see
1794/// [`matches_pathspec_with_context`].
1795///
1796/// `:(attr:...)` requirements are not evaluated here (only the pattern part is matched); use
1797/// [`matches_pathspec_for_object`] or [`crate::repo::Repository::paths_match_pathspecs`] for
1798/// specs where [`pathspec_has_attr_magic`] is true.
1799#[must_use]
1800pub fn matches_pathspec(spec: &str, path: &str) -> bool {
1801    matches_pathspec_with_context(spec, path, PathspecMatchContext::default())
1802}
1803
1804/// Like [`matches_pathspec`], but uses `ctx` for trailing-`/` literal pathspecs and for
1805/// wildcard pathspecs where the pattern continues after a directory boundary (Git
1806/// `matches_pathspec` + directory semantics).
1807#[must_use]
1808pub fn matches_pathspec_with_context(spec: &str, path: &str, ctx: PathspecMatchContext) -> bool {
1809    let spec_nfc: Cow<'_, str> = if ctx.precompose_paths {
1810        precompose_utf8_path(spec)
1811    } else {
1812        Cow::Borrowed(spec)
1813    };
1814    let path_nfc: Cow<'_, str> = if ctx.precompose_paths {
1815        precompose_utf8_path(path)
1816    } else {
1817        Cow::Borrowed(path)
1818    };
1819    let spec = spec_nfc.as_ref();
1820    let path = path_nfc.as_ref();
1821
1822    let trimmed = spec.strip_prefix("./").unwrap_or(spec);
1823    if trimmed == "." || trimmed.is_empty() {
1824        return true;
1825    }
1826
1827    let (elem_magic, raw_pattern) = parse_element_magic(trimmed);
1828    let magic = combine_magic(elem_magic);
1829
1830    if magic.literal && magic.glob {
1831        return false;
1832    }
1833    if magic.exclude {
1834        return false;
1835    }
1836
1837    let pattern = strip_top_magic(raw_pattern);
1838    let path_for_match = if let Some(prefix) = magic.prefix.as_deref() {
1839        match path_for_match_under_magic_prefix(path, prefix) {
1840            Some(rest) => rest,
1841            None => return false,
1842        }
1843    } else {
1844        path
1845    };
1846
1847    if magic.literal {
1848        if let Some(prefix) = pattern.strip_suffix('/') {
1849            if path_is_under_dir(path_for_match, prefix, false) {
1850                return true;
1851            }
1852            if path_for_match == prefix {
1853                return ctx.is_directory || ctx.is_git_submodule;
1854            }
1855            return false;
1856        }
1857        return path_for_match == pattern || path_is_under_dir(path_for_match, pattern, false);
1858    }
1859
1860    // No wildcards and trailing `/`: directory-only semantics (Git `matches_pathspec`).
1861    if let Some(prefix) = pattern.strip_suffix('/') {
1862        if simple_length(pattern) == pattern.len() {
1863            if path_is_under_dir(path_for_match, prefix, false) {
1864                return true;
1865            }
1866            if path_for_match == prefix {
1867                return ctx.is_directory || ctx.is_git_submodule;
1868            }
1869            return false;
1870        }
1871    }
1872
1873    if pathspec_matches_tail(pattern, path_for_match, magic) {
1874        return true;
1875    }
1876
1877    if (ctx.is_directory || ctx.is_git_submodule)
1878        && !path_for_match.is_empty()
1879        && pattern.len() > path_for_match.len()
1880        && pattern.as_bytes().get(path_for_match.len()) == Some(&b'/')
1881        && pattern.starts_with(path_for_match)
1882        && simple_length(pattern) < pattern.len()
1883    {
1884        return true;
1885    }
1886
1887    false
1888}
1889
1890/// Parse a Git mode string (e.g. `100644`, `040000`) into a [`PathspecMatchContext`].
1891#[must_use]
1892pub fn context_from_mode_octal(mode: &str) -> PathspecMatchContext {
1893    let Ok(bits) = u32::from_str_radix(mode, 8) else {
1894        return PathspecMatchContext::default();
1895    };
1896    context_from_mode_bits(bits)
1897}
1898
1899/// Classify a raw Git mode (e.g. from an index or tree entry) for pathspec matching.
1900#[must_use]
1901pub fn context_from_mode_bits(mode: u32) -> PathspecMatchContext {
1902    let ty = mode & 0o170000;
1903    PathspecMatchContext {
1904        is_directory: ty == 0o040000,
1905        is_git_submodule: ty == 0o160000,
1906        precompose_paths: false,
1907    }
1908}
1909
1910/// Pathspec matching for `ls-tree` after Git forces `pathspec.has_wildcard = 0` (`ls-tree.c`).
1911///
1912/// Metacharacters `*` / `?` still participate in [`wildmatch`]; `[` and `\\` are **not** glob
1913/// starters unless a `*` or `?` appears — so `a[a]` matches the literal directory `a[a]` (t3102),
1914/// while `a*` matches `a/one`, `aa/two`, `a[a]/three`, …
1915#[must_use]
1916pub fn matches_ls_tree_pathspec(
1917    spec: &str,
1918    path: &str,
1919    mode: u32,
1920    attr_rules: &[AttrRule],
1921) -> bool {
1922    let (elem_magic, raw_pattern) = parse_element_magic(spec);
1923    let mut magic = combine_magic(elem_magic);
1924    magic.exclude = false;
1925
1926    if magic.literal && magic.glob {
1927        return false;
1928    }
1929
1930    let ctx = context_from_mode_bits(mode);
1931    let is_dir_for_attr = path.ends_with('/') || ctx.is_directory || ctx.is_git_submodule;
1932
1933    if !magic.attr_requirements.is_empty()
1934        && !attr_requirements_match(
1935            &magic.attr_requirements,
1936            AttrLookup::rules_only(attr_rules),
1937            path,
1938            is_dir_for_attr,
1939            mode,
1940        )
1941    {
1942        return false;
1943    }
1944
1945    let pattern = strip_top_magic(raw_pattern);
1946    let path_for_match = if let Some(prefix) = magic.prefix.as_deref() {
1947        match path_for_match_under_magic_prefix(path, prefix) {
1948            Some(rest) => rest,
1949            None => return false,
1950        }
1951    } else {
1952        path
1953    };
1954
1955    if magic.literal || magic.glob || magic.icase {
1956        return pathspec_matches_tail(pattern, path_for_match, magic);
1957    }
1958
1959    let spec_nfc: Cow<'_, str> = Cow::Borrowed(pattern);
1960    let path_nfc: Cow<'_, str> = Cow::Borrowed(path_for_match);
1961    let pattern = spec_nfc.as_ref();
1962    let path = path_nfc.as_ref();
1963
1964    let trimmed = pattern.strip_prefix("./").unwrap_or(pattern);
1965    if trimmed == "." || trimmed.is_empty() {
1966        return true;
1967    }
1968
1969    let uses_star_or_question = trimmed.contains('*') || trimmed.contains('?');
1970    if !uses_star_or_question {
1971        return ls_tree_literal_match(trimmed, path, ctx);
1972    }
1973
1974    let nwl = simple_length(trimmed);
1975    let flags = 0u32;
1976    if nwl == trimmed.len() {
1977        return wildmatch(trimmed.as_bytes(), path.as_bytes(), flags);
1978    }
1979    let lit = trimmed.as_bytes().get(..nwl).unwrap_or_default();
1980    let path_b = path.as_bytes();
1981    if path_b.len() < nwl {
1982        return false;
1983    }
1984    if &path_b[..nwl] != lit {
1985        return false;
1986    }
1987    let pat_rest = &trimmed[nwl..];
1988    let path_rest = &path[nwl..];
1989    wildmatch(pat_rest.as_bytes(), path_rest.as_bytes(), flags)
1990}
1991
1992/// Match a pathspec against a tree path, using `.gitattributes` for `:(attr:...)`.
1993///
1994/// `mode` supplies directory/gitlink context for literal pathspecs ending in `/`.
1995#[must_use]
1996pub fn matches_pathspec_for_object(
1997    spec: &str,
1998    path: &str,
1999    mode: u32,
2000    attr_rules: &[AttrRule],
2001) -> bool {
2002    matches_pathspec_for_object_with_attrs(spec, path, mode, AttrLookup::rules_only(attr_rules))
2003}
2004
2005fn matches_pathspec_for_object_with_attrs(
2006    spec: &str,
2007    path: &str,
2008    mode: u32,
2009    attrs: AttrLookup<'_>,
2010) -> bool {
2011    let (elem_magic, raw_pattern) = parse_element_magic(spec);
2012    let mut magic = combine_magic(elem_magic);
2013    magic.exclude = false;
2014
2015    if magic.literal && magic.glob {
2016        return false;
2017    }
2018
2019    let ctx = context_from_mode_bits(mode);
2020    let is_dir_for_attr = path.ends_with('/') || ctx.is_directory || ctx.is_git_submodule;
2021
2022    if !magic.attr_requirements.is_empty()
2023        && !attr_requirements_match(&magic.attr_requirements, attrs, path, is_dir_for_attr, mode)
2024    {
2025        return false;
2026    }
2027
2028    let pattern = strip_top_magic(raw_pattern);
2029    let path_for_match = if let Some(prefix) = magic.prefix.as_deref() {
2030        match path_for_match_under_magic_prefix(path, prefix) {
2031            Some(rest) => rest,
2032            None => return false,
2033        }
2034    } else {
2035        path
2036    };
2037    if magic.literal || magic.glob || magic.icase {
2038        pathspec_matches_tail(pattern, path_for_match, magic)
2039    } else {
2040        matches_pathspec_with_context(pattern, path_for_match, ctx)
2041    }
2042}
2043
2044/// Resolved path lies outside the repository work tree (Git `prefix_path_gently` failure).
2045#[derive(Debug, Clone)]
2046pub struct PathOutsideRepository {
2047    /// User-facing pathspec token (argv element).
2048    pub elt: String,
2049    /// Resolved absolute path outside the work tree.
2050    pub path: String,
2051    /// Canonical work tree root.
2052    pub work_tree: PathBuf,
2053}
2054
2055impl std::fmt::Display for PathOutsideRepository {
2056    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
2057        write!(
2058            f,
2059            "{}: '{}' is outside repository at '{}'",
2060            self.elt,
2061            self.path,
2062            self.work_tree.display()
2063        )
2064    }
2065}
2066
2067impl From<PathOutsideRepository> for PathspecResolveError {
2068    fn from(outside: PathOutsideRepository) -> Self {
2069        Self::OutsideRepository {
2070            elt: outside.elt,
2071            path: outside.path,
2072            work_tree: outside.work_tree.display().to_string(),
2073        }
2074    }
2075}
2076
2077/// Resolve a magic pathspec relative to a current-directory prefix.
2078///
2079/// This keeps the `cwd` prefix case-sensitive (via an internal `prefix:` magic
2080/// token) while still honoring magic options like `icase` for the tail.
2081/// Returns `None` when `spec` is not a parseable magic pathspec.
2082pub fn resolve_magic_pathspec(spec: &str, cwd_prefix: &str) -> Option<String> {
2083    if !spec.starts_with(":(") {
2084        return None;
2085    }
2086    let close_idx = spec.find(')')?;
2087    let magic_prefix = &spec[..=close_idx];
2088    let tail = &spec[close_idx + 1..];
2089    Some(resolve_magic_pathspec_parts(magic_prefix, tail, cwd_prefix))
2090}
2091
2092fn resolve_magic_pathspec_parts(magic_prefix: &str, tail: &str, cwd_prefix: &str) -> String {
2093    if has_magic_prefix_token(magic_prefix) {
2094        return format!("{magic_prefix}{tail}");
2095    }
2096
2097    if let Some(rooted_tail) = tail.strip_prefix('/') {
2098        return format!("{magic_prefix}{}", normalize_relative_path_str(rooted_tail));
2099    }
2100
2101    let combined = if cwd_prefix.is_empty() {
2102        normalize_relative_path_str(tail)
2103    } else {
2104        normalize_relative_path_str(&format!("{cwd_prefix}{tail}"))
2105    };
2106
2107    let cwd_base = normalize_relative_path_str(cwd_prefix.trim_end_matches('/'));
2108    if !cwd_base.is_empty()
2109        && (combined == cwd_base || combined.starts_with(&format!("{cwd_base}/")))
2110    {
2111        let after_base = combined
2112            .strip_prefix(&cwd_base)
2113            .unwrap_or(combined.as_str());
2114        let remainder = after_base.strip_prefix('/').unwrap_or(after_base);
2115        let magic_with_prefix = inject_magic_prefix_token(magic_prefix, &format!("{cwd_base}/"));
2116        return format!("{magic_with_prefix}{remainder}");
2117    }
2118
2119    format!("{magic_prefix}{combined}")
2120}
2121
2122fn has_magic_prefix_token(magic_prefix: &str) -> bool {
2123    let Some(inner) = magic_prefix
2124        .strip_prefix(":(")
2125        .and_then(|s| s.strip_suffix(')'))
2126    else {
2127        return false;
2128    };
2129    inner
2130        .split(',')
2131        .map(str::trim)
2132        .any(|token| token.starts_with("prefix:"))
2133}
2134
2135fn inject_magic_prefix_token(magic_prefix: &str, prefix: &str) -> String {
2136    let Some(inner) = magic_prefix
2137        .strip_prefix(":(")
2138        .and_then(|s| s.strip_suffix(')'))
2139    else {
2140        return magic_prefix.to_string();
2141    };
2142    if inner.trim().is_empty() {
2143        format!(":(prefix:{prefix})")
2144    } else {
2145        format!(":({inner},prefix:{prefix})")
2146    }
2147}
2148
2149/// Drop repeated leading `./` / `.//` from a repo-root pathspec (Git `prefix_path` behavior).
2150///
2151/// When stripping exposes a leading `:`, wrap with `:(literal)` so a filesystem path like
2152/// `:!foo` is not reinterpreted as short exclude magic.
2153fn normalize_repo_root_relative_pathspec(pathspec: &str) -> String {
2154    let mut spec = pathspec;
2155    let mut stripped = false;
2156    while let Some(rest) = spec.strip_prefix("./") {
2157        stripped = true;
2158        spec = rest.trim_start_matches('/');
2159    }
2160    if !stripped {
2161        return pathspec.to_owned();
2162    }
2163    if spec.is_empty() {
2164        return ".".to_owned();
2165    }
2166    if spec.starts_with(':') && !literal_pathspecs_enabled() {
2167        return format!(":(literal){spec}");
2168    }
2169    spec.to_owned()
2170}
2171
2172/// Normalize user pathspec separators on Windows (`\` → `/`) before resolution.
2173fn normalize_pathspec_separators(pathspec: &str) -> std::borrow::Cow<'_, str> {
2174    #[cfg(windows)]
2175    {
2176        if !pathspec.starts_with(':') && pathspec.contains('\\') {
2177            return std::borrow::Cow::Owned(pathspec.replace('\\', "/"));
2178        }
2179    }
2180    std::borrow::Cow::Borrowed(pathspec)
2181}
2182
2183fn normalize_relative_path_str(path: &str) -> String {
2184    let mut parts: Vec<String> = Vec::new();
2185    for component in std::path::Path::new(path).components() {
2186        match component {
2187            std::path::Component::CurDir => {}
2188            std::path::Component::ParentDir => {
2189                parts.pop();
2190            }
2191            std::path::Component::Normal(seg) => {
2192                parts.push(seg.to_string_lossy().to_string());
2193            }
2194            std::path::Component::RootDir | std::path::Component::Prefix(_) => {}
2195        }
2196    }
2197    parts.join("/")
2198}
2199
2200/// Current directory relative to `work_tree`, or `None` if cwd is the work tree root.
2201#[must_use]
2202pub fn pathdiff(cwd: &Path, work_tree: &Path) -> Option<String> {
2203    use crate::git_path::{path_for_disk_compare, path_to_git_slash};
2204
2205    let cwd_cmp = path_for_disk_compare(cwd);
2206    let wt_cmp = path_for_disk_compare(work_tree);
2207
2208    if cwd_cmp == wt_cmp {
2209        return None;
2210    }
2211
2212    let rel = cwd_cmp.strip_prefix(&wt_cmp).ok()?;
2213    let git = path_to_git_slash(rel);
2214    if git.is_empty() {
2215        None
2216    } else {
2217        Some(git)
2218    }
2219}
2220
2221/// For short exclude magic from a subdirectory, Git applies the worktree prefix to the pattern
2222/// tail (via `prefix_path`) while keeping argv spelling. Long-form `:(exclude,prefix:…)` matches
2223/// [`resolve_worktree_pathspec`] for `:(exclude)…` under a cwd prefix.
2224fn prepend_cwd_to_short_exclude_pathspec(spec: &str, cwd: &str) -> Option<String> {
2225    let cwd = cwd.trim_end_matches('/');
2226    if cwd.is_empty() {
2227        return None;
2228    }
2229    let (elem_magic, pattern) = parse_element_magic(spec);
2230    let magic = combine_magic(elem_magic);
2231    if !magic.exclude || magic.top {
2232        return None;
2233    }
2234    if pattern.is_empty() || pattern.starts_with('/') {
2235        return None;
2236    }
2237    if magic.glob || magic.literal || magic.icase || !magic.attr_requirements.is_empty() {
2238        return None;
2239    }
2240    Some(format!(":(exclude,prefix:{cwd}/){pattern}"))
2241}
2242
2243/// Build [`PathspecMatchContext`] with optional NFC path normalization for comparisons.
2244#[must_use]
2245pub fn pathspec_match_context_with_precompose(precompose_paths: bool) -> PathspecMatchContext {
2246    PathspecMatchContext {
2247        precompose_paths,
2248        ..PathspecMatchContext::default()
2249    }
2250}
2251
2252/// Like [`matches_pathspec_list`], with an explicit `precompose_paths` flag.
2253#[must_use]
2254pub fn matches_pathspec_list_precompose(
2255    path: &str,
2256    specs: &[String],
2257    precompose_paths: bool,
2258) -> bool {
2259    matches_pathspec_list_with_context(
2260        path,
2261        specs,
2262        pathspec_match_context_with_precompose(precompose_paths),
2263    )
2264}
2265
2266/// True when argv pathspec strings must not be NFC-normalized (magic syntax).
2267#[must_use]
2268pub fn argv_pathspec_preserves_spelling(pathspec: &str) -> bool {
2269    pathspec.starts_with(":(")
2270        || pathspec.starts_with(":/")
2271        || pathspec.starts_with(":!")
2272        || pathspec.starts_with(":^")
2273}
2274
2275/// NFC-normalize a literal command-line pathspec (Git `precompose_argv_prefix`).
2276#[must_use]
2277pub fn precompose_argv_pathspec(pathspec: &str) -> Cow<'_, str> {
2278    if argv_pathspec_preserves_spelling(pathspec) {
2279        Cow::Borrowed(pathspec)
2280    } else {
2281        precompose_utf8_path(pathspec)
2282    }
2283}
2284
2285/// Resolve a pathspec string to a path relative to the repository work tree.
2286///
2287/// `prefix` is the current directory relative to the work tree (no trailing slash),
2288/// or `None` when cwd is the work tree root.
2289#[must_use]
2290pub fn resolve_pathspec(pathspec: &str, work_tree: &Path, prefix: Option<&str>) -> String {
2291    resolve_pathspec_with_argv_precompose(pathspec, work_tree, prefix, false)
2292}
2293
2294/// Like [`resolve_pathspec`], but NFC-normalizes literal argv path segments when requested.
2295#[must_use]
2296pub fn resolve_pathspec_with_argv_precompose(
2297    pathspec: &str,
2298    work_tree: &Path,
2299    prefix: Option<&str>,
2300    argv_precompose: bool,
2301) -> String {
2302    let pathspec = if argv_precompose {
2303        precompose_argv_pathspec(pathspec)
2304    } else {
2305        Cow::Borrowed(pathspec)
2306    };
2307    resolve_pathspec_impl(pathspec.as_ref(), work_tree, prefix)
2308}
2309
2310fn resolve_pathspec_impl(pathspec: &str, work_tree: &Path, prefix: Option<&str>) -> String {
2311    let pathspec = normalize_pathspec_separators(pathspec);
2312    let pathspec = pathspec.as_ref();
2313    // Git: `.` at repo root means "match the whole tree" (not an empty pathspec).
2314    // An empty resolved pathspec would match nothing and breaks `grep -- . t` max-depth.
2315    if pathspec == "." {
2316        return match prefix {
2317            Some(p) if !p.is_empty() => p.to_owned(),
2318            _ => ".".to_owned(),
2319        };
2320    }
2321    if pathspec.contains("../") || pathspec.starts_with("../") {
2322        let cwd = match prefix.filter(|p| !p.is_empty()) {
2323            Some(p) => work_tree.join(p),
2324            None => work_tree.to_path_buf(),
2325        };
2326        let abs = cwd.join(pathspec);
2327        let mut parts: Vec<std::ffi::OsString> = Vec::new();
2328        for component in abs.components() {
2329            use std::path::Component;
2330            match component {
2331                Component::ParentDir => {
2332                    parts.pop();
2333                }
2334                Component::CurDir => {}
2335                other => parts.push(other.as_os_str().to_os_string()),
2336            }
2337        }
2338        let abs_norm: PathBuf = parts.iter().collect();
2339        if let Some(rel) = crate::git_path::strip_worktree_prefix(&abs_norm, work_tree) {
2340            return rel;
2341        }
2342    }
2343    if Path::new(pathspec).is_absolute() {
2344        let abs = Path::new(pathspec);
2345        if let Some(rel) = crate::git_path::strip_worktree_prefix(abs, work_tree) {
2346            return rel;
2347        }
2348        return pathspec.to_owned();
2349    }
2350
2351    if pathspec.starts_with(':') {
2352        if let Some(p) = prefix {
2353            if !p.is_empty() && !literal_pathspecs_enabled() {
2354                let cwd_ps = format!("{}/", p.trim_end_matches('/'));
2355                if pathspec.starts_with(":(") {
2356                    if let Some(resolved) = resolve_magic_pathspec(pathspec, &cwd_ps) {
2357                        return resolved;
2358                    }
2359                    return pathspec.to_owned();
2360                }
2361                if pathspec_is_exclude(pathspec) {
2362                    if let Some(fixed) = prepend_cwd_to_short_exclude_pathspec(pathspec, p) {
2363                        return fixed;
2364                    }
2365                }
2366            }
2367        }
2368        if let Some(rest) = pathspec.strip_prefix(":/") {
2369            // `:/!foo` / `:/^bar` — `:/` is `:(top)`; the tail is still short magic, not a literal path.
2370            if rest.starts_with('!') || rest.starts_with('^') {
2371                return pathspec.to_owned();
2372            }
2373            return rest.to_owned();
2374        }
2375        // Long magic `:(...)` must stay intact — `:(exclude)path` is not the same as `path`
2376        // (t6132-pathspec-exclude, grep --untracked with exclude pathspecs).
2377        if pathspec.starts_with(":(") {
2378            return pathspec.to_owned();
2379        }
2380        return pathspec.to_owned();
2381    }
2382
2383    match prefix {
2384        Some(p) if !p.is_empty() => {
2385            normalize_relative_path_str(&PathBuf::from(p).join(pathspec).to_string_lossy())
2386        }
2387        _ => normalize_repo_root_relative_pathspec(pathspec),
2388    }
2389}
2390
2391/// Resolve a pathspec and ensure it lies inside `work_tree` (used by `git add`, etc.).
2392///
2393/// Returns [`PathOutsideRepository`] when resolution stays absolute, matching Git's
2394/// `'%s' is outside repository at '%s'` fatal (t7010).
2395pub fn resolve_pathspec_in_worktree(
2396    elt: &str,
2397    pathspec: &str,
2398    work_tree: &Path,
2399    prefix: Option<&str>,
2400) -> Result<String, PathOutsideRepository> {
2401    resolve_pathspec_in_worktree_with_argv_precompose(elt, pathspec, work_tree, prefix, false)
2402}
2403
2404/// Like [`resolve_pathspec_in_worktree`], NFC-normalizing `pathspec` when `argv_precompose` is set.
2405///
2406/// Callers holding a [`Repository`] should use [`Repository::resolve_worktree_pathspec`], which
2407/// computes `argv_precompose` once per handle instead of reloading config per pathspec.
2408pub(crate) fn resolve_pathspec_in_worktree_with_argv_precompose(
2409    elt: &str,
2410    pathspec: &str,
2411    work_tree: &Path,
2412    prefix: Option<&str>,
2413    argv_precompose: bool,
2414) -> Result<String, PathOutsideRepository> {
2415    let resolved =
2416        resolve_pathspec_with_argv_precompose(pathspec, work_tree, prefix, argv_precompose);
2417    if Path::new(&resolved).is_absolute() {
2418        let wt = work_tree
2419            .canonicalize_simplified()
2420            .unwrap_or_else(|_| work_tree.to_path_buf());
2421        return Err(PathOutsideRepository {
2422            elt: elt.to_string(),
2423            path: resolved,
2424            work_tree: wt,
2425        });
2426    }
2427    Ok(resolved)
2428}
2429
2430/// Normalize a worktree file path for porcelain commands (`blame`, `log`, …).
2431///
2432/// Accepts repo-relative or absolute paths under `work_tree`.
2433#[must_use]
2434pub fn normalize_worktree_file_path(
2435    file_path: &str,
2436    work_tree: &Path,
2437    prefix: Option<&str>,
2438) -> String {
2439    normalize_worktree_file_path_with_git_dir(file_path, work_tree, prefix, None)
2440}
2441
2442/// Like [`normalize_worktree_file_path`], honoring `core.precomposeunicode` argv normalization for `git_dir`.
2443#[must_use]
2444pub fn normalize_worktree_file_path_with_git_dir(
2445    file_path: &str,
2446    work_tree: &Path,
2447    prefix: Option<&str>,
2448    git_dir: Option<&Path>,
2449) -> String {
2450    let argv_precompose = argv_precompose_enabled(git_dir);
2451    let resolved =
2452        resolve_pathspec_with_argv_precompose(file_path, work_tree, prefix, argv_precompose);
2453    if Path::new(&resolved).is_absolute() {
2454        file_path.to_string()
2455    } else {
2456        resolved
2457    }
2458}
2459
2460#[cfg(test)]
2461mod tree_entry_pathspec_tests {
2462    use super::*;
2463
2464    #[test]
2465    fn t6130_bracket_filename_matches_pathspec() {
2466        assert!(matches_pathspec("f[o][o]", "f[o][o]"));
2467        assert!(matches_pathspec(":(glob)f[o][o]", "f[o][o]"));
2468    }
2469
2470    #[test]
2471    fn literal_prefix_and_exact() {
2472        assert!(matches_pathspec("path1", "path1/file1"));
2473        assert!(matches_pathspec_with_context(
2474            "path1/",
2475            "path1/file1",
2476            PathspecMatchContext::default()
2477        ));
2478        assert!(matches_pathspec("file0", "file0"));
2479        assert!(!matches_pathspec("path", "path1/file1"));
2480    }
2481
2482    #[test]
2483    fn dir_prefix_needs_slash_boundary() {
2484        assert!(matches_pathspec("a/b", "a/b/c"));
2485        assert!(!matches_pathspec("a/b", "a/bc"));
2486        assert!(!matches_pathspec("a/b", "a"));
2487        assert!(matches_pathspec(":(literal)dir", "dir/x"));
2488        assert!(!matches_pathspec(":(literal)dir", "dirx"));
2489        assert!(matches_pathspec(":(icase)DIR", "dir/x"));
2490        assert!(!matches_pathspec(":(icase)DIR", "dirx"));
2491        assert!(matches_pathspec("dir/", "dir/x"));
2492        assert!(!matches_pathspec("dir/", "dir"));
2493    }
2494
2495    #[test]
2496    fn sorted_lookup_agrees_with_linear_scan() {
2497        let mut paths: Vec<&str> = vec![
2498            "a",
2499            "a-b",
2500            "a.txt",
2501            "a/b",
2502            "a/b/c.txt",
2503            "a/bc",
2504            "a0",
2505            "b/x[1].txt",
2506            "dir/sub/file",
2507            "dir-sub",
2508            "f[o][o]",
2509            "z",
2510            "caf\u{e9}/menu",
2511            "cafe\u{301}/old",
2512            "\u{212a}.txt",
2513        ];
2514        paths.sort_unstable();
2515        let specs = [
2516            "caf\u{e9}",
2517            "cafe\u{301}/",
2518            "cafe\u{301}/old",
2519            "K.txt",
2520            "\u{212a}.txt",
2521            "a",
2522            "a/",
2523            "a/b",
2524            "a/b/",
2525            "a/bc",
2526            "a-",
2527            "a-b",
2528            "a.txt",
2529            "b",
2530            "b/",
2531            "dir",
2532            "dir/sub",
2533            "dir/sub/",
2534            "dir/su",
2535            "dir-sub",
2536            "f[o][o]",
2537            "*.txt",
2538            "a/*",
2539            ":(glob)**/*.txt",
2540            ":(literal)f[o][o]",
2541            ":(icase)A/B",
2542            ":!a",
2543            ".",
2544            "./a/b",
2545            "",
2546            "z",
2547            "zz",
2548            "/",
2549            "a//",
2550        ];
2551        for ctx in [
2552            PathspecMatchContext::default(),
2553            PathspecMatchContext {
2554                precompose_paths: true,
2555                ..PathspecMatchContext::default()
2556            },
2557            PathspecMatchContext {
2558                is_directory: true,
2559                ..PathspecMatchContext::default()
2560            },
2561        ] {
2562            for spec in specs {
2563                let list = [spec.to_owned()];
2564                let linear = paths
2565                    .iter()
2566                    .any(|p| matches_pathspec_list_with_context(p, &list, ctx));
2567                assert_eq!(
2568                    sorted_paths_match_pathspec(spec, &paths, ctx),
2569                    linear,
2570                    "spec {spec:?} ctx {ctx:?}"
2571                );
2572            }
2573        }
2574        let empty: [&str; 0] = [];
2575        assert!(!sorted_paths_match_pathspec(
2576            "a",
2577            &empty,
2578            PathspecMatchContext::default()
2579        ));
2580    }
2581
2582    #[test]
2583    fn list_matcher_agrees_with_linear_matching() {
2584        let paths = [
2585            "a",
2586            "a-b",
2587            "a.txt",
2588            "a/",
2589            "a/b",
2590            "a/b/c.txt",
2591            "a/bc",
2592            "a//x",
2593            "a0",
2594            "b/x[1].txt",
2595            "dir/sub/file",
2596            "dir-sub",
2597            "f[o][o]",
2598            "z",
2599            "zz/a",
2600            "é/x",
2601            "e\u{301}/y",
2602            "\u{212a}.txt",
2603            "K.txt",
2604            "",
2605        ];
2606        let lists: &[&[&str]] = &[
2607            &[],
2608            &["a"],
2609            &["a/"],
2610            &["a//"],
2611            &["a/b", "dir/sub/", "z"],
2612            &["a/b/", "a/bc", "dir", "zz"],
2613            &["dir/su", "dir-sub", "f[o][o]"],
2614            &["é", "a-"],
2615            &["e\u{301}", "z"],
2616            &["e\u{301}/y", "K.txt"],
2617            &["\u{212a}.txt"],
2618            &["a", "*.txt"],
2619            &["dir", ":!dir/sub"],
2620            &[":!a"],
2621            &["./a", "z"],
2622            &[".", "a"],
2623            &["", "a"],
2624            &[":(icase)A/B", "z"],
2625        ];
2626        for ctx in [
2627            PathspecMatchContext::default(),
2628            PathspecMatchContext {
2629                precompose_paths: true,
2630                ..PathspecMatchContext::default()
2631            },
2632            PathspecMatchContext {
2633                is_directory: true,
2634                ..PathspecMatchContext::default()
2635            },
2636        ] {
2637            for list in lists {
2638                let specs: Vec<String> = list.iter().map(|s| (*s).to_owned()).collect();
2639                let matcher = PathspecListMatcher::new(&specs, ctx);
2640                for path in paths {
2641                    assert_eq!(
2642                        matcher.matches(path),
2643                        matches_pathspec_list_with_context(path, &specs, ctx),
2644                        "path {path:?} specs {specs:?} ctx {ctx:?}"
2645                    );
2646                }
2647            }
2648        }
2649    }
2650
2651    #[test]
2652    fn list_matcher_uses_literal_lookup_only_for_plain_lists() {
2653        let ctx = PathspecMatchContext::default();
2654        let plain = vec!["a/b".to_owned(), "dir/".to_owned()];
2655        assert!(PathspecListMatcher::new(&plain, ctx).literals.is_some());
2656        let glob = vec!["a/b".to_owned(), "*.txt".to_owned()];
2657        assert!(PathspecListMatcher::new(&glob, ctx).literals.is_none());
2658        let exclude = vec!["a/b".to_owned(), ":!a/b/c".to_owned()];
2659        assert!(PathspecListMatcher::new(&exclude, ctx).literals.is_none());
2660        assert!(PathspecListMatcher::new(&[], ctx).literals.is_none());
2661    }
2662
2663    #[test]
2664    fn precomposed_literal_specs_keep_fast_lookups() {
2665        let ctx = PathspecMatchContext {
2666            precompose_paths: true,
2667            ..PathspecMatchContext::default()
2668        };
2669        let specs: Vec<String> = (0..100).map(|i| format!("d{i}/file.txt")).collect();
2670        let matcher = PathspecListMatcher::new(&specs, ctx);
2671        assert!(
2672            matcher.literals.is_some(),
2673            "core.precomposeunicode must not force a per-spec scan of every path"
2674        );
2675        assert!(matcher.matches("d42/file.txt"));
2676        assert!(!matcher.matches("d42/other.txt"));
2677        let nfd = vec!["cafe\u{301}".to_owned()];
2678        let matcher = PathspecListMatcher::new(&nfd, ctx);
2679        assert!(matcher.literals.is_some());
2680        assert!(matcher.matches("caf\u{e9}/menu"));
2681        assert!(matcher.matches("cafe\u{301}/menu"));
2682        assert!(exact_literal_pathspec_set(&specs, ctx).is_none());
2683    }
2684
2685    #[test]
2686    fn ls_tree_bracket_in_name_is_literal_prefix() {
2687        assert!(matches_ls_tree_pathspec(
2688            "a[a]",
2689            "a[a]/three",
2690            0o100644,
2691            &[]
2692        ));
2693        assert!(!matches_pathspec_with_context(
2694            "a[a]",
2695            "a[a]/three",
2696            PathspecMatchContext::default()
2697        ));
2698    }
2699
2700    #[test]
2701    fn wildcards_cross_slash_by_default() {
2702        assert!(matches_pathspec("f*", "file0"));
2703        assert!(matches_pathspec("*file1", "path1/file1"));
2704        assert!(matches_pathspec_with_context(
2705            "path1/f*",
2706            "path1",
2707            PathspecMatchContext {
2708                is_directory: true,
2709                ..Default::default()
2710            }
2711        ));
2712        assert!(matches_pathspec("path1/*file1", "path1/file1"));
2713    }
2714
2715    #[test]
2716    fn glob_double_star_txt_at_repo_root() {
2717        assert!(pathspec_matches(":(glob)**/*.txt", "untracked.txt"));
2718        assert!(pathspec_matches(":(glob)**/*.txt", "d/untracked.txt"));
2719    }
2720
2721    #[test]
2722    fn trailing_slash_directory_only() {
2723        assert!(!matches_pathspec_with_context(
2724            "file0/",
2725            "file0",
2726            PathspecMatchContext::default()
2727        ));
2728        assert!(matches_pathspec_with_context(
2729            "file0/",
2730            "file0",
2731            PathspecMatchContext {
2732                is_directory: true,
2733                ..Default::default()
2734            }
2735        ));
2736        assert!(matches_pathspec_with_context(
2737            "submod/",
2738            "submod",
2739            PathspecMatchContext {
2740                is_git_submodule: true,
2741                ..Default::default()
2742            }
2743        ));
2744    }
2745
2746    #[test]
2747    fn exclude_top_short_magic_subtracts_from_positive() {
2748        let specs = vec!["*".to_string(), ":/!sub2".to_string()];
2749        assert!(matches_pathspec_list("sub/file", &specs));
2750        assert!(!matches_pathspec_list("sub2/file", &specs));
2751        assert!(pathspec_exclude_matches(":/!sub2", "sub2/file"));
2752    }
2753}
2754
2755#[cfg(test)]
2756mod exact_literal_pathspec_set_tests {
2757    use super::*;
2758
2759    #[test]
2760    fn exact_literal_pathspec_set_accepts_file_paths() {
2761        let ctx = PathspecMatchContext::default();
2762        let set = exact_literal_pathspec_set(&["large.txt".to_string()], ctx).expect("set");
2763        assert_eq!(set.len(), 1);
2764        assert!(set.contains("large.txt"));
2765    }
2766
2767    #[test]
2768    fn exact_literal_pathspec_set_rejects_directory_trailing_slash() {
2769        let ctx = PathspecMatchContext::default();
2770        assert!(exact_literal_pathspec_set(&["src/".to_string()], ctx).is_none());
2771    }
2772
2773    #[test]
2774    fn exact_literal_pathspec_set_rejects_glob_magic() {
2775        let ctx = PathspecMatchContext::default();
2776        assert!(exact_literal_pathspec_set(&[":(glob)*.rs".to_string()], ctx).is_none());
2777    }
2778}
2779
2780#[cfg(test)]
2781mod pathspec_list_tests {
2782    use super::*;
2783    use crate::crlf::parse_gitattributes_content;
2784
2785    #[test]
2786    fn exclude_removes_paths_matching_icase_positive() {
2787        let specs = vec![
2788            ":(icase)*.txt".to_string(),
2789            ":(exclude)submodule/subsub/*".to_string(),
2790        ];
2791        assert!(path_allowed_by_pathspec_list(&specs, "submodule/g.txt"));
2792        assert!(!path_allowed_by_pathspec_list(
2793            &specs,
2794            "submodule/subsub/e.txt"
2795        ));
2796    }
2797
2798    #[test]
2799    fn prefixed_attr_exclude_removes_matching_child_path() {
2800        let specs = vec![
2801            "sub".to_string(),
2802            ":(exclude,attr:labelB,prefix:sub/)".to_string(),
2803        ];
2804        let exclude_only = vec![":(exclude,attr:labelB,prefix:sub/)".to_string()];
2805        let attrs = parse_gitattributes_content("fileB labelB\n");
2806        assert!(!matches_pathspec_list_for_object(
2807            "sub/fileB",
2808            0o100644,
2809            &attrs,
2810            &specs,
2811        ));
2812        assert!(!matches_pathspec_list_for_object(
2813            "sub/fileB",
2814            0o100644,
2815            &attrs,
2816            &exclude_only,
2817        ));
2818    }
2819}
2820
2821#[cfg(test)]
2822mod resolve_pathspec_tests {
2823    use super::*;
2824
2825    fn wt() -> PathBuf {
2826        PathBuf::from("/tmp/grit-pathspec-worktree")
2827    }
2828
2829    #[test]
2830    fn strips_leading_dot_slash_at_repo_root() {
2831        let wt = wt();
2832        assert_eq!(resolve_pathspec("./-", &wt, None), "-");
2833        assert_eq!(resolve_pathspec("./a/b", &wt, None), "a/b");
2834        assert_eq!(resolve_pathspec("././a", &wt, None), "a");
2835        assert_eq!(resolve_pathspec(".//a", &wt, None), "a");
2836        assert_eq!(resolve_pathspec("./", &wt, None), ".");
2837        assert_eq!(resolve_pathspec(".", &wt, None), ".");
2838    }
2839
2840    #[test]
2841    fn prefixed_cwd_resolution_unchanged() {
2842        let wt = wt();
2843        assert_eq!(
2844            resolve_pathspec("file", &wt, Some("sub/dir")),
2845            "sub/dir/file"
2846        );
2847        assert_eq!(resolve_pathspec(":(glob)*.rs", &wt, None), ":(glob)*.rs");
2848    }
2849
2850    #[test]
2851    fn dot_slash_colon_filename_stays_literal_not_exclude_magic() {
2852        let wt = wt();
2853        let resolved = resolve_pathspec("./:!foo", &wt, None);
2854        assert_eq!(resolved, ":(literal):!foo");
2855        assert!(!pathspec_is_exclude(&resolved));
2856        assert!(pathspec_matches(&resolved, ":!foo"));
2857        assert!(!pathspec_matches(&resolved, "other"));
2858    }
2859
2860    #[test]
2861    fn resolve_absolute_path_under_worktree() {
2862        let tmp = tempfile::tempdir().expect("tempdir");
2863        let wt = tmp.path().join("repo");
2864        std::fs::create_dir_all(wt.join("d")).expect("mkdir");
2865        let file = wt.join("d/b.txt");
2866        std::fs::write(&file, b"x").expect("write");
2867        let wt = wt.canonicalize_simplified().expect("canonicalize");
2868        let abs = file.canonicalize_simplified().expect("canonicalize file");
2869        let spec = abs.to_string_lossy();
2870        assert_eq!(
2871            resolve_pathspec(&spec, &wt, None),
2872            "d/b.txt",
2873            "absolute forward-slash pathspec"
2874        );
2875    }
2876
2877    #[cfg(windows)]
2878    mod windows {
2879        use super::*;
2880
2881        #[test]
2882        fn resolve_backslash_relative_and_absolute_pathspecs() {
2883            let tmp = tempfile::tempdir().expect("tempdir");
2884            let wt = tmp.path().join("repo");
2885            std::fs::create_dir_all(wt.join("d")).expect("mkdir");
2886            let file_a = wt.join("d/a.txt");
2887            let file_b = wt.join("d/b.txt");
2888            std::fs::write(&file_a, b"a").expect("write");
2889            std::fs::write(&file_b, b"b").expect("write");
2890            let wt = wt.canonicalize_simplified().expect("canonicalize worktree");
2891
2892            assert_eq!(
2893                resolve_pathspec(r"d\a.txt", &wt, None),
2894                "d/a.txt",
2895                "backslash relative pathspec (issue #915 A)"
2896            );
2897
2898            let abs_back = file_b.canonicalize_simplified().expect("canonicalize");
2899            let abs_back_spec = abs_back.display().to_string();
2900            assert_eq!(
2901                resolve_pathspec(&abs_back_spec, &wt, None),
2902                "d/b.txt",
2903                "absolute backslash pathspec (issue #915 B)"
2904            );
2905
2906            let abs_forward = abs_back_spec.replace('\\', "/");
2907            assert_eq!(
2908                resolve_pathspec(&abs_forward, &wt, None),
2909                "d/b.txt",
2910                "absolute forward-slash pathspec (issue #915 C)"
2911            );
2912        }
2913    }
2914
2915    #[test]
2916    fn nfd_pathspec_matches_nfc_index_path_when_precompose_enabled() {
2917        let nfc_path = "caf\u{00e9}.txt";
2918        let nfd_spec = format!("cafe\u{0301}.txt");
2919        assert!(!matches_pathspec_list(&nfc_path, &[nfd_spec.clone()]));
2920        assert!(matches_pathspec_list_precompose(
2921            nfc_path,
2922            &[nfd_spec.clone()],
2923            true
2924        ));
2925    }
2926
2927    #[test]
2928    fn argv_precompose_normalizes_nfd_literal_pathspec() {
2929        let nfd = format!("cafe\u{0301}.txt");
2930        let nfc = precompose_argv_pathspec(&nfd);
2931        assert_eq!(nfc.as_ref(), "caf\u{00e9}.txt");
2932        assert!(argv_pathspec_preserves_spelling(":(glob)*.rs"));
2933        assert_eq!(
2934            precompose_argv_pathspec(":(glob)*.rs").as_ref(),
2935            ":(glob)*.rs"
2936        );
2937    }
2938}
2939
2940#[cfg(test)]
2941mod resolved_pathspecs_tests {
2942    use super::*;
2943    use crate::repo::{InitOptions, Repository};
2944
2945    fn repo_with_tree() -> (tempfile::TempDir, Repository) {
2946        let dir = tempfile::tempdir().expect("tempdir");
2947        let repo = Repository::init(dir.path(), &InitOptions::default()).expect("init");
2948        std::fs::write(dir.path().join("top.txt"), b"a").expect("write");
2949        std::fs::create_dir_all(dir.path().join("sub")).expect("mkdir");
2950        std::fs::write(dir.path().join("sub/inner.txt"), b"b").expect("write");
2951        (dir, repo)
2952    }
2953
2954    #[test]
2955    fn resolved_pathspecs_subdir_prefix() {
2956        let (dir, repo) = repo_with_tree();
2957        let sub = dir.path().join("sub");
2958        let resolved =
2959            ResolvedPathspecs::resolve(&repo, &sub, &["inner.txt".to_string()], None).expect("ok");
2960        assert_eq!(resolved.specs, ["sub/inner.txt"]);
2961        assert!(resolved.scoped);
2962    }
2963
2964    #[test]
2965    fn resolved_pathspecs_top_level_magic() {
2966        let (dir, repo) = repo_with_tree();
2967        let resolved =
2968            ResolvedPathspecs::resolve(&repo, dir.path(), &[":/top.txt".to_string()], None)
2969                .expect("ok");
2970        assert_eq!(resolved.specs, ["top.txt"]);
2971    }
2972
2973    #[test]
2974    fn resolved_pathspecs_parent_outside_repo() {
2975        let (dir, repo) = repo_with_tree();
2976        let outside = dir.path().parent().expect("parent").join("outside.txt");
2977        std::fs::write(&outside, b"x").expect("write outside");
2978        let spec = outside.to_string_lossy().into_owned();
2979        let err =
2980            ResolvedPathspecs::resolve(&repo, dir.path(), &[spec], None).expect_err("outside");
2981        assert!(matches!(
2982            err,
2983            PathspecResolveError::OutsideRepository { .. }
2984        ));
2985    }
2986
2987    #[test]
2988    fn resolved_pathspecs_empty_argv() {
2989        let (dir, repo) = repo_with_tree();
2990        let err = ResolvedPathspecs::resolve(&repo, dir.path(), &[String::new()], None)
2991            .expect_err("empty");
2992        assert_eq!(err, PathspecResolveError::EmptyPathspec);
2993    }
2994
2995    #[test]
2996    fn resolved_pathspecs_from_file_lf_and_quoted() {
2997        let (dir, repo) = repo_with_tree();
2998        let input = PathspecInput::new(b"top.txt\n\"sub/inner.txt\"\n".to_vec(), false);
2999        let resolved = ResolvedPathspecs::resolve(&repo, dir.path(), &[], Some(input)).expect("ok");
3000        assert_eq!(resolved.specs, ["top.txt", "sub/inner.txt"]);
3001    }
3002
3003    #[test]
3004    fn resolved_pathspecs_from_file_nul() {
3005        let (dir, repo) = repo_with_tree();
3006        let input = PathspecInput::new(b"top.txt\0sub/\0".to_vec(), true);
3007        let resolved = ResolvedPathspecs::resolve(&repo, dir.path(), &[], Some(input)).expect("ok");
3008        assert_eq!(resolved.specs, ["top.txt", "sub/"]);
3009    }
3010
3011    #[test]
3012    fn resolved_pathspecs_from_file_nul_preserves_literal_bytes() {
3013        let (dir, repo) = repo_with_tree();
3014        std::fs::write(dir.path().join("  spaced  "), b"x").expect("write");
3015        let quoted_dir = dir.path().join("\"double\"");
3016        std::fs::create_dir_all(&quoted_dir).expect("mkdir");
3017        std::fs::write(quoted_dir.join("f"), b"y").expect("write");
3018        let input = PathspecInput::new(b"  spaced  \0\"double\"\0".to_vec(), true);
3019        let resolved = ResolvedPathspecs::resolve(&repo, dir.path(), &[], Some(input)).expect("ok");
3020        assert_eq!(resolved.sources, ["  spaced  ", "\"double\""]);
3021    }
3022
3023    #[test]
3024    fn resolved_pathspecs_from_file_blank_line_errors() {
3025        let (dir, repo) = repo_with_tree();
3026        let input = PathspecInput::new(b"\n".to_vec(), false);
3027        let err = ResolvedPathspecs::resolve(&repo, dir.path(), &[], Some(input)).expect_err("err");
3028        assert_eq!(err, PathspecResolveError::EmptyPathspec);
3029    }
3030
3031    #[test]
3032    fn resolved_pathspecs_from_file_nul_empty_record_errors() {
3033        let (dir, repo) = repo_with_tree();
3034        let input = PathspecInput::new(b"a\0\0b".to_vec(), true);
3035        let err = ResolvedPathspecs::resolve(&repo, dir.path(), &[], Some(input)).expect_err("err");
3036        assert_eq!(err, PathspecResolveError::EmptyPathspec);
3037    }
3038
3039    #[test]
3040    fn resolved_pathspecs_from_file_invalid_quote() {
3041        let (dir, repo) = repo_with_tree();
3042        let input = PathspecInput::new(b"\"bad\\z\"\n".to_vec(), false);
3043        let err = ResolvedPathspecs::resolve(&repo, dir.path(), &[], Some(input)).expect_err("err");
3044        assert!(matches!(
3045            err,
3046            PathspecResolveError::FromFile(
3047                PathspecFromFileParseError::InvalidEscapeSequence { .. }
3048            )
3049        ));
3050        let input = PathspecInput::new(b"\"unterminated\n".to_vec(), false);
3051        let err = ResolvedPathspecs::resolve(&repo, dir.path(), &[], Some(input)).expect_err("err");
3052        assert!(matches!(
3053            err,
3054            PathspecResolveError::FromFile(PathspecFromFileParseError::InvalidCStyleQuoting { .. })
3055        ));
3056    }
3057
3058    #[test]
3059    fn resolved_pathspecs_entire_tree_when_unscoped() {
3060        let (dir, repo) = repo_with_tree();
3061        let resolved = ResolvedPathspecs::resolve(&repo, dir.path(), &[], None).expect("ok");
3062        assert!(resolved.is_entire_tree());
3063        assert!(resolved.specs.is_empty());
3064    }
3065}
3066
3067#[cfg(test)]
3068mod rev_walk_pathspec_resolution_tests {
3069    use super::*;
3070    use crate::config::environment::{Environment, RepositoryOptions};
3071    use crate::repo::{InitOptions, Repository};
3072
3073    fn repo_from_sub(worktree: &std::path::Path, sub: &std::path::Path) -> Repository {
3074        let mut env = Environment::empty();
3075        env.cwd = sub.to_path_buf();
3076        let mut opts = RepositoryOptions::empty();
3077        opts.environment = env;
3078        Repository::discover(Some(worktree), &opts).expect("discover")
3079    }
3080
3081    #[test]
3082    fn glob_magic_from_subdirectory_gets_prefix_and_matches() {
3083        let dir = tempfile::tempdir().expect("tempdir");
3084        let wt = dir.path();
3085        std::fs::create_dir_all(wt.join("sub")).expect("mkdir");
3086        std::fs::write(wt.join("sub/f.txt"), b"x").expect("write");
3087        Repository::init(wt, &InitOptions::default()).expect("init");
3088        let repo = repo_from_sub(wt, &wt.join("sub"));
3089        let specs = repo
3090            .resolve_rev_walk_pathspecs(&[":(glob)**/f.txt".to_owned()])
3091            .expect("resolve");
3092        assert_eq!(specs, [":(glob,prefix:sub/)**/f.txt"]);
3093        assert!(pathspec_matches(&specs[0], "sub/f.txt"));
3094        assert!(!pathspec_matches(&specs[0], "other/f.txt"));
3095    }
3096
3097    #[test]
3098    fn short_exclude_from_subdirectory_uses_prefix_magic() {
3099        let dir = tempfile::tempdir().expect("tempdir");
3100        let wt = dir.path();
3101        std::fs::create_dir_all(wt.join("sub")).expect("mkdir");
3102        Repository::init(wt, &InitOptions::default()).expect("init");
3103        let repo = repo_from_sub(wt, &wt.join("sub"));
3104        let specs = repo
3105            .resolve_rev_walk_pathspecs(&[":!f.txt".to_owned()])
3106            .expect("resolve");
3107        assert_eq!(specs, [".", ":(exclude,prefix:sub/)f.txt"]);
3108        assert!(matches_pathspec_list("other/g.txt", &specs));
3109        assert!(!matches_pathspec_list("sub/f.txt", &specs));
3110    }
3111
3112    #[test]
3113    fn glob_with_prefix_descends_into_cwd_directory() {
3114        let spec = ":(glob,prefix:sub/)**/f.txt";
3115        assert!(pathspec_wants_descent_into_tree(spec, "sub"));
3116        assert!(pathspec_entry_is_relevant(
3117            "sub/f.txt",
3118            false,
3119            &[spec.to_owned()]
3120        ));
3121    }
3122}