grit_lib/

pathspec.rs

//! Git-compatible pathspec matching (magic tokens and global flags).
//!
//! ```rust
//! use grit_lib::worktree::pathspec::parse_pathspecs_from_source;
//!
//! let specs = parse_pathspecs_from_source(b"README.md\n", false).expect("parse");
//! assert_eq!(specs, ["README.md"]);
//! ```
//!
//! Global flags are read from the same environment variables as Git:
//! `GIT_LITERAL_PATHSPECS`, `GIT_GLOB_PATHSPECS`, `GIT_NOGLOB_PATHSPECS`,
//! `GIT_ICASE_PATHSPECS`. The `grit` binary sets these from CLI flags such as
//! `--literal-pathspecs` before dispatching subcommands.

use crate::git_path::CanonicalizeExt;
use std::borrow::Cow;
use std::path::{Path, PathBuf};

use hashbrown::HashSet;

use crate::attributes::MacroTable;
use crate::crlf::path_gitattribute_values;
use crate::crlf::AttrRule;
pub use crate::error::{AttrPathspecError, PathspecFromFileParseError, PathspecResolveError};
use crate::precompose_config::argv_precompose_enabled;
use crate::repo::Repository;
use crate::unicode_normalization::precompose_utf8_path;
use crate::wildmatch::{wildmatch, WM_CASEFOLD, WM_PATHNAME};

/// Returns the length of the leading literal segment before the first glob metacharacter,
/// matching Git's `simple_length()` (`*` `?` `[` `\`) on bytes.
#[must_use]
pub fn simple_length(match_str: &str) -> usize {
    let b = match_str.as_bytes();
    let mut len = 0usize;
    for &c in b {
        if matches!(c, b'*' | b'?' | b'[' | b'\\') {
            break;
        }
        len += 1;
    }
    len
}

/// Whether the pattern uses wildcards after Git pathspec escaping rules.
#[must_use]
pub fn has_glob_chars(s: &str) -> bool {
    simple_length(s) < s.len()
}

/// Read pathspec entries from raw file bytes (stdin or file), matching Git's
/// `--pathspec-from-file` / `--pathspec-file-nul` rules.
///
/// * **NUL mode:** entries are separated by `NUL`; each segment is taken literally
///   (no trimming or C-style unquoting).
/// * **Line mode:** entries are separated by `LF`; optional `CR` before `LF`
///   is stripped; optional trailing line without a final newline is included;
///   double-quoted lines are C-unquoted (including octal escapes).
pub fn parse_pathspecs_from_source(
    data: &[u8],
    nul_terminated: bool,
) -> Result<Vec<String>, PathspecFromFileParseError> {
    if nul_terminated {
        let mut chunks: Vec<&[u8]> = data.split(|b| *b == 0).collect();
        if chunks.last().copied() == Some(&[] as &[u8]) {
            chunks.pop();
        }
        let mut out = Vec::with_capacity(chunks.len());
        for chunk in chunks {
            if chunk.is_empty() {
                out.push(String::new());
                continue;
            }
            let entry = std::str::from_utf8(chunk)
                .map_err(|_| PathspecFromFileParseError::InvalidUtf8InNulRecord)?;
            out.push(entry.to_string());
        }
        return Ok(out);
    }

    let text = String::from_utf8_lossy(data);
    let mut out = Vec::new();
    for raw in text.split_inclusive('\n') {
        let line = raw.trim_end_matches('\n').trim_end_matches('\r');
        if line.is_empty() {
            out.push(String::new());
            continue;
        }
        if line.starts_with('"') {
            if line.ends_with('"') && line.len() >= 2 {
                out.push(unquote_c_style_pathspec_line(line)?);
            } else {
                return Err(PathspecFromFileParseError::InvalidCStyleQuoting {
                    line: line.to_string(),
                });
            }
        } else {
            out.push(line.to_string());
        }
    }
    Ok(out)
}

/// Unquote a single `--pathspec-from-file` line that is wrapped in double quotes.
fn unquote_c_style_pathspec_line(s: &str) -> Result<String, PathspecFromFileParseError> {
    let bytes = s.as_bytes();
    if bytes.first() != Some(&b'"') || bytes.last() != Some(&b'"') || bytes.len() < 2 {
        return Err(PathspecFromFileParseError::InvalidCStyleQuoting {
            line: s.to_string(),
        });
    }

    let inner = &bytes[1..bytes.len() - 1];
    let mut out = Vec::with_capacity(inner.len());
    let mut i = 0;
    while i < inner.len() {
        if inner[i] != b'\\' {
            out.push(inner[i]);
            i += 1;
            continue;
        }
        i += 1;
        if i >= inner.len() {
            return Err(PathspecFromFileParseError::InvalidEscapeAtEnd);
        }
        match inner[i] {
            b'\\' => out.push(b'\\'),
            b'"' => out.push(b'"'),
            b'a' => out.push(7),
            b'b' => out.push(8),
            b'f' => out.push(12),
            b'n' => out.push(b'\n'),
            b'r' => out.push(b'\r'),
            b't' => out.push(b'\t'),
            b'v' => out.push(11),
            c if c.is_ascii_digit() => {
                if i + 2 >= inner.len() {
                    return Err(PathspecFromFileParseError::TruncatedOctalEscape);
                }
                let oct = std::str::from_utf8(&inner[i..i + 3])
                    .map_err(|_| PathspecFromFileParseError::InvalidOctalBytes)?;
                out.push(
                    u8::from_str_radix(oct, 8)
                        .map_err(|_| PathspecFromFileParseError::InvalidOctalValue)?,
                );
                i += 2;
            }
            other => {
                return Err(PathspecFromFileParseError::InvalidEscapeSequence {
                    escape: char::from(other),
                });
            }
        }
        i += 1;
    }
    String::from_utf8(out).map_err(|_| PathspecFromFileParseError::InvalidUtf8InQuotedLine)
}

/// Raw bytes for [`ResolvedPathspecs::resolve`] from `--pathspec-from-file` (read by the caller).
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct PathspecInput {
    /// File or stdin contents.
    pub data: Vec<u8>,
    /// When true, entries are NUL-separated (`--pathspec-file-nul`).
    pub nul_terminated: bool,
}

impl PathspecInput {
    /// Build input from bytes already read from a file or stdin.
    #[must_use]
    pub fn new(data: Vec<u8>, nul_terminated: bool) -> Self {
        Self {
            data,
            nul_terminated,
        }
    }
}

/// Pathspec argv fragments resolved against a repository work tree.
#[derive(Debug, Clone, Default, PartialEq, Eq)]
pub struct ResolvedPathspecs {
    /// Work-tree-relative pathspec strings for matching.
    pub specs: Vec<String>,
    /// User-facing spellings aligned with `specs` (argv or file lines).
    pub sources: Vec<String>,
    /// True when the caller supplied argv paths and/or a from-file (not "whole tree").
    pub scoped: bool,
}

impl ResolvedPathspecs {
    /// Resolve `raw` argv pathspecs and optional `--pathspec-from-file` bytes against `repo`.
    ///
    /// `cwd` is the process working directory used for relative pathspec prefixes (never read
    /// from the environment inside the library).
    ///
    /// # Examples
    ///
    /// ```
    /// use grit_lib::worktree::pathspec::{PathspecInput, ResolvedPathspecs};
    /// use grit_lib::repo::{InitOptions, Repository};
    /// use std::fs;
    ///
    /// # fn main() -> Result<(), Box<dyn std::error::Error>> {
    /// let dir = tempfile::tempdir()?;
    /// let repo = Repository::init(dir.path(), &InitOptions::default())?;
    /// fs::write(dir.path().join("a.txt"), b"x")?;
    /// let resolved = ResolvedPathspecs::resolve(
    ///     &repo,
    ///     dir.path(),
    ///     &["a.txt".to_string()],
    ///     None,
    /// )?;
    /// assert_eq!(resolved.specs, ["a.txt"]);
    /// let from_file = PathspecInput::new(b"a.txt\n".to_vec(), false);
    /// let from_file = ResolvedPathspecs::resolve(&repo, dir.path(), &[], Some(from_file))?;
    /// assert_eq!(from_file.specs, ["a.txt"]);
    /// # Ok(())
    /// # }
    /// ```
    pub fn resolve(
        repo: &Repository,
        cwd: &Path,
        raw: &[String],
        from_file: Option<PathspecInput>,
    ) -> Result<Self, PathspecResolveError> {
        let work_tree = repo
            .work_tree()
            .ok_or(PathspecResolveError::NeedsWorkTree)?;
        let prefix = pathdiff(cwd, work_tree);
        let scoped = from_file.is_some() || !raw.is_empty();

        let mut sources = Vec::new();
        if let Some(input) = from_file {
            let parsed = parse_pathspecs_from_source(&input.data, input.nul_terminated)?;
            sources.extend(parsed);
        }
        for sel in raw {
            sources.push(sel.clone());
        }

        if scoped && sources.is_empty() {
            return Err(PathspecResolveError::EmptyPathspec);
        }
        for sel in &sources {
            if sel.is_empty() {
                return Err(PathspecResolveError::EmptyPathspec);
            }
        }

        let argv_precompose = repo.pathspec_precompose_enabled();
        let specs = sources
            .iter()
            .map(|sel| {
                resolve_pathspec_in_worktree_with_argv_precompose(
                    sel,
                    sel,
                    work_tree,
                    prefix.as_deref(),
                    argv_precompose,
                )
            })
            .collect::<Result<Vec<_>, _>>()?;

        Ok(Self {
            specs,
            sources,
            scoped,
        })
    }

    /// Whether the invocation targets the entire work tree (no path arguments or from-file).
    #[must_use]
    pub fn is_entire_tree(&self) -> bool {
        !self.scoped
    }

    /// Consume `self` and return resolved specs and user-facing sources.
    ///
    /// When [`Self::is_entire_tree`] is true, both vectors are empty (match legacy
    /// "stage/remove/clean the whole tree" call sites).
    #[must_use]
    pub fn into_pathspec_lists(self) -> (Vec<String>, Vec<String>) {
        if self.scoped {
            (self.specs, self.sources)
        } else {
            (Vec::new(), Vec::new())
        }
    }

    /// Resolved pathspec strings only; empty when the invocation is unscoped.
    #[must_use]
    pub fn into_specs(self) -> Vec<String> {
        if self.scoped {
            self.specs
        } else {
            Vec::new()
        }
    }
}

#[derive(Debug, Clone, Default)]
struct PathspecMagic {
    literal: bool,
    glob: bool,
    icase: bool,
    exclude: bool,
    /// `:(top)` / short `:/` — paths are relative to repo root.
    top: bool,
    prefix: Option<String>,
    /// `:(attr:...)` requirements.
    attr_requirements: Vec<AttrRequirement>,
}

#[derive(Debug, Clone, PartialEq, Eq)]
enum AttrRequirement {
    Set(String),
    Unset(String),
    Unspecified(String),
    Value(String, String),
}

impl AttrRequirement {
    fn name(&self) -> &str {
        match self {
            AttrRequirement::Set(name)
            | AttrRequirement::Unset(name)
            | AttrRequirement::Unspecified(name)
            | AttrRequirement::Value(name, _) => name,
        }
    }
}

/// The library never reads the process environment, so Git's `GIT_*_PATHSPECS` globals are
/// always unset here and every global flag takes its default. This runs on every pathspec
/// comparison, so it must stay allocation-free.
fn git_env_bool(_key: &str, default: bool) -> bool {
    default
}

fn literal_global() -> bool {
    git_env_bool("GIT_LITERAL_PATHSPECS", false)
}

/// Whether `GIT_LITERAL_PATHSPECS` is enabled (shell `*` and `?` are literal, not globs).
#[must_use]
pub fn literal_pathspecs_enabled() -> bool {
    literal_global()
}

fn glob_global() -> bool {
    git_env_bool("GIT_GLOB_PATHSPECS", false)
}

fn noglob_global() -> bool {
    git_env_bool("GIT_NOGLOB_PATHSPECS", false)
}

fn icase_global() -> bool {
    git_env_bool("GIT_ICASE_PATHSPECS", false)
}

/// Validates global pathspec environment flags the same way Git does.
///
/// Returns an error message suitable for `bail!` when flags are incompatible.
pub fn validate_global_pathspec_flags() -> Result<(), String> {
    let lit = literal_global();
    let glob = glob_global();
    let noglob = noglob_global();
    let icase = icase_global();

    if glob && noglob {
        return Err("global 'glob' and 'noglob' pathspec settings are incompatible".to_string());
    }
    if lit && (glob || noglob || icase) {
        return Err(
            "global 'literal' pathspec setting is incompatible with all other global pathspec settings"
                .to_string(),
        );
    }
    Ok(())
}

fn is_valid_attr_name(name: &str) -> bool {
    !name.is_empty()
        && name
            .bytes()
            .all(|b| b.is_ascii_alphanumeric() || matches!(b, b'_' | b'-' | b'.'))
}

fn split_attr_expr(expr: &str) -> Result<Vec<String>, AttrPathspecError> {
    let mut parts = Vec::new();
    let mut cur = String::new();
    let mut in_value = false;
    let mut escaped = false;

    for ch in expr.chars() {
        if escaped {
            if ch.is_ascii_whitespace() {
                return Err(AttrPathspecError::DanglingEscape);
            }
            if ch != ',' {
                return Err(AttrPathspecError::UnsupportedEscape);
            }
            cur.push(ch);
            escaped = false;
            continue;
        }
        if in_value && ch == '\\' {
            escaped = true;
            continue;
        }
        if ch == '=' {
            in_value = true;
            cur.push(ch);
            continue;
        }
        if ch.is_ascii_whitespace() {
            if !cur.is_empty() {
                parts.push(cur);
                cur = String::new();
            }
            in_value = false;
            continue;
        }
        cur.push(ch);
    }

    if escaped {
        return Err(AttrPathspecError::DanglingEscape);
    }
    if !cur.is_empty() {
        parts.push(cur);
    }
    Ok(parts)
}

fn parse_attr_requirements(expr: &str) -> Result<Vec<AttrRequirement>, AttrPathspecError> {
    let name_only = |name: &str| {
        if name.contains('=') || !is_valid_attr_name(name) {
            Err(AttrPathspecError::InvalidName(name.to_owned()))
        } else {
            Ok(name.to_owned())
        }
    };
    let mut out = Vec::new();
    for token in split_attr_expr(expr)? {
        let req = if let Some(name) = token.strip_prefix('-') {
            AttrRequirement::Unset(name_only(name)?)
        } else if let Some(name) = token.strip_prefix('!') {
            AttrRequirement::Unspecified(name_only(name)?)
        } else if let Some((name, value)) = token.split_once('=') {
            AttrRequirement::Value(name_only(name)?, value.to_owned())
        } else {
            AttrRequirement::Set(name_only(&token)?)
        };
        out.push(req);
    }
    if out.is_empty() {
        return Err(AttrPathspecError::Empty);
    }
    Ok(out)
}

/// The `attr:` clauses inside a long-form `:(...)` magic prefix (none for other specs).
fn attr_clauses(spec: &str) -> Vec<String> {
    if literal_global() {
        return Vec::new();
    }
    let Some(rest) = spec.strip_prefix(":(") else {
        return Vec::new();
    };
    let Some(close) = rest.find(')') else {
        return Vec::new();
    };
    split_long_magic_tokens_raw(&rest[..close])
        .into_iter()
        .filter_map(|token| token.trim_start().strip_prefix("attr:").map(str::to_owned))
        .collect()
}

/// Validate the `:(attr:...)` magic of every spec in `specs`.
///
/// Specs without attr magic are always accepted. A spec is rejected when it has more than one
/// `attr:` clause or a clause that does not parse (empty clause, invalid attribute name, bad
/// escape). A clause of only whitespace has no requirements and is accepted.
///
/// # Errors
///
/// [`crate::error::Error::InvalidAttrPathspec`] naming the first rejected spec and its
/// [`AttrPathspecError`].
pub fn validate_attr_pathspecs(specs: &[String]) -> crate::error::Result<()> {
    for spec in specs {
        let invalid = |reason| crate::error::Error::InvalidAttrPathspec {
            spec: spec.clone(),
            reason,
        };
        let clauses = attr_clauses(spec);
        if clauses.len() > 1 {
            return Err(invalid(AttrPathspecError::MultipleAttrClauses));
        }
        match clauses.first() {
            Some(expr) if expr.is_empty() => return Err(invalid(AttrPathspecError::Empty)),
            // Whitespace alone is an attr clause with no requirements, which Git accepts.
            Some(expr) if expr.trim().is_empty() => {}
            Some(expr) => {
                parse_attr_requirements(expr).map_err(invalid)?;
            }
            None => {}
        }
    }
    Ok(())
}

fn split_long_magic_tokens(magic_part: &str) -> Vec<String> {
    split_long_magic_tokens_raw(magic_part)
        .into_iter()
        .map(|token| token.trim().to_string())
        .collect()
}

/// Split long magic on unescaped `,`, keeping surrounding whitespace.
fn split_long_magic_tokens_raw(magic_part: &str) -> Vec<String> {
    let mut tokens = Vec::new();
    let mut cur = String::new();
    let mut escaped = false;
    for ch in magic_part.chars() {
        if escaped {
            cur.push('\\');
            cur.push(ch);
            escaped = false;
            continue;
        }
        if ch == '\\' {
            escaped = true;
            continue;
        }
        if ch == ',' {
            tokens.push(std::mem::take(&mut cur));
            continue;
        }
        cur.push(ch);
    }
    if escaped {
        cur.push('\\');
    }
    tokens.push(cur);
    tokens
}

fn parse_long_magic(rest_after_paren: &str) -> Option<(PathspecMagic, &str)> {
    let close = rest_after_paren.find(')')?;
    let magic_part = &rest_after_paren[..close];
    let tail = &rest_after_paren[close + 1..];
    let mut magic = PathspecMagic::default();
    for raw in split_long_magic_tokens(magic_part) {
        let token = raw.trim();
        if token.is_empty() {
            continue;
        }
        if let Some(p) = token.strip_prefix("prefix:") {
            magic.prefix = Some(p.to_string());
            continue;
        }
        if let Some(expr) = token.strip_prefix("attr:") {
            if let Ok(reqs) = parse_attr_requirements(expr) {
                magic.attr_requirements = reqs;
            }
            continue;
        }
        if token.eq_ignore_ascii_case("literal") {
            magic.literal = true;
        } else if token.eq_ignore_ascii_case("glob") {
            magic.glob = true;
        } else if token.eq_ignore_ascii_case("icase") {
            magic.icase = true;
        } else if token.eq_ignore_ascii_case("exclude") {
            magic.exclude = true;
        } else if token.eq_ignore_ascii_case("top") {
            magic.top = true;
        }
    }
    Some((magic, tail))
}

/// `elem` is the full pathspec beginning with `:` (short magic form, not `:(...)`).
fn parse_short_magic(elem: &str) -> (PathspecMagic, &str) {
    let bytes = elem.as_bytes();
    let mut i = 1usize;
    let mut magic = PathspecMagic::default();
    while i < bytes.len() && bytes[i] != b':' {
        let ch = bytes[i];
        if ch == b'^' {
            magic.exclude = true;
            i += 1;
            continue;
        }
        let is_magic = match ch {
            b'!' => {
                magic.exclude = true;
                true
            }
            b'/' => {
                magic.top = true;
                true
            } // short `:/` = top
            _ => false,
        };
        if is_magic {
            i += 1;
            continue;
        }
        break;
    }
    if i < bytes.len() && bytes[i] == b':' {
        i += 1;
    }
    (magic, &elem[i..])
}

/// Strip `:(magic)` / `:magic` prefix when not in literal-global mode.
fn parse_element_magic(elem: &str) -> (PathspecMagic, &str) {
    if !elem.starts_with(':') || literal_global() {
        return (PathspecMagic::default(), elem);
    }
    if let Some(rest) = elem.strip_prefix(":(") {
        return parse_long_magic(rest).unwrap_or((PathspecMagic::default(), elem));
    }
    parse_short_magic(elem)
}

fn combine_magic(element: PathspecMagic) -> PathspecMagic {
    let mut m = element;
    if literal_global() {
        m.literal = true;
    }
    if glob_global() && !m.literal {
        m.glob = true;
    }
    if icase_global() {
        m.icase = true;
    }
    if noglob_global() && !m.glob {
        m.literal = true;
    }
    m
}

fn strip_top_magic(mut pattern: &str) -> &str {
    if let Some(r) = pattern.strip_prefix(":/") {
        pattern = r;
    }
    pattern
}

/// Path prefix used for Bloom-filter lookups (`revision.c` `convert_pathspec_to_bloom_keyvec`).
///
/// `cwd_from_repo_root` is the path from the repository work tree to the process cwd, using `/`
/// separators and no leading slash (empty string at repo root). Used for `:(top)` / `:/`.
#[must_use]
pub fn bloom_lookup_prefix_with_cwd(
    spec: &str,
    cwd_from_repo_root: Option<&str>,
) -> Option<String> {
    let (elem_magic, raw_pattern) = parse_element_magic(spec);
    let magic = combine_magic(elem_magic);
    if magic.exclude || magic.icase {
        return None;
    }
    let pattern = strip_top_magic(raw_pattern);
    if pattern.is_empty() {
        return None;
    }
    let combined = if magic.top {
        let cwd = cwd_from_repo_root.unwrap_or("").trim_end_matches('/');
        if cwd.is_empty() {
            pattern.to_string()
        } else {
            format!("{cwd}/{pattern}")
        }
    } else {
        pattern.to_string()
    };
    let pattern = combined.as_str();
    let mut len = simple_length(pattern);
    if len != pattern.len() {
        while len > 0 && pattern.as_bytes()[len - 1] != b'/' {
            len -= 1;
        }
    }
    while len > 0 && pattern.as_bytes()[len - 1] == b'/' {
        len -= 1;
    }
    if len == 0 {
        return None;
    }
    Some(combined[..len].to_string())
}

/// Bloom filter lookup prefix for a pathspec at the repository root.
#[must_use]
pub fn bloom_lookup_prefix(spec: &str) -> Option<String> {
    bloom_lookup_prefix_with_cwd(spec, None)
}

/// Whether every pathspec can participate in Bloom precomputation (Git `forbid_bloom_filters`).
#[must_use]
pub fn pathspecs_allow_bloom(specs: &[String]) -> bool {
    specs.iter().all(|s| {
        !s.is_empty() && !pathspec_is_exclude(s) && bloom_lookup_prefix_with_cwd(s, None).is_some()
    })
}

/// Whether `path` is included when Git applies a pathspec list with optional `:(exclude)` entries.
///
/// A path is rejected if any exclude pathspec matches it. When at least one non-exclude pathspec is
/// present, the path must also match one of those positives (`OR` semantics).
#[must_use]
pub fn path_allowed_by_pathspec_list(specs: &[String], path: &str) -> bool {
    let mut has_positive = false;
    let mut positive_match = false;
    for s in specs {
        let (elem, raw_pattern) = parse_element_magic(s);
        let magic = combine_magic(elem);
        if magic.exclude {
            if path_matches_pathspec_tail(raw_pattern, path, magic) {
                return false;
            }
            continue;
        }
        has_positive = true;
        if pathspec_matches(s, path) {
            positive_match = true;
        }
    }
    !has_positive || positive_match
}

/// True when `spec` matches `path` for pathspec bookkeeping (positive match or exclude hit).
#[must_use]
pub fn pathspec_contributes_match(spec: &str, path: &str) -> bool {
    pathspec_matches(spec, path) || pathspec_exclude_matches(spec, path)
}

fn path_matches_pathspec_tail(raw_pattern: &str, path: &str, magic: PathspecMagic) -> bool {
    if magic.literal && magic.glob {
        return false;
    }
    let pattern = strip_top_magic(raw_pattern);
    let path_for_match = if let Some(prefix) = magic.prefix.as_deref() {
        if !path.starts_with(prefix) {
            return false;
        }
        &path[prefix.len()..]
    } else {
        path
    };
    pathspec_matches_tail(pattern, path_for_match, magic)
}

/// True if `path` is matched by `spec` (Git pathspec syntax, including magic and globals).
///
/// Same as [`matches_pathspec`] (default file context; exclude specs never match positively here).
/// See [`matches_pathspec_list`].
#[must_use]
pub fn pathspec_matches(spec: &str, path: &str) -> bool {
    matches_pathspec(spec, path)
}

/// Returns whether `spec` uses Git's exclude magic (`:(exclude)`, `:!`, `:^`, etc.).
#[must_use]
pub fn pathspec_is_exclude(spec: &str) -> bool {
    let (elem_magic, _) = parse_element_magic(spec);
    combine_magic(elem_magic).exclude
}

/// Whether a path-limited revision tree walk should recurse into directory `path`.
///
/// Uses [`pathspec_wants_descent_into_tree`] for non-exclude specs. Exclude-only lists never
/// trigger that helper alone, but Git's implicit match-all still requires walking non-excluded
/// directories so nested paths can match [`matches_pathspec_list`].
#[must_use]
pub fn pathspec_wants_descent_for_rev_tree(path: &str, specs: &[String]) -> bool {
    if specs.is_empty() {
        return true;
    }
    for spec in specs {
        if !pathspec_is_exclude(spec) && pathspec_wants_descent_into_tree(spec, path) {
            return true;
        }
    }
    if !specs.iter().any(|s| pathspec_is_exclude(s)) {
        return false;
    }
    let ctx = PathspecMatchContext {
        is_directory: true,
        is_git_submodule: false,
        precompose_paths: false,
    };
    for spec in specs {
        if pathspec_is_exclude(spec)
            && pathspec_exclude_element_matches_with_context(spec, path, ctx)
        {
            return false;
        }
    }
    !specs.iter().any(|s| !pathspec_is_exclude(s))
}

/// Whether tree-walking should recurse into directory `full_name` for pathspec `spec` without
/// `-r` (Git `read_tree` / `show_recursive` “interesting” descent).
///
/// Exclude-only patterns never trigger descent alone.
#[must_use]
pub fn pathspec_wants_descent_into_tree(spec: &str, full_name: &str) -> bool {
    if pathspec_is_exclude(spec) {
        return false;
    }
    let (elem_magic, raw_pattern) = parse_element_magic(spec);
    let magic = combine_magic(elem_magic);
    if magic.exclude {
        return false;
    }
    let pattern = strip_top_magic(raw_pattern);
    let pattern = pattern.strip_prefix("./").unwrap_or(pattern);
    if pattern.is_empty() || pattern == "." {
        return true;
    }
    let dir_prefix = format!("{full_name}/");
    if pattern.starts_with(&dir_prefix) {
        return true;
    }
    let probe = format!("{full_name}/.__grit_ls_tree_probe__");
    matches_ls_tree_pathspec(spec, &probe, 0o100644, &[])
}

/// Like [`matches_pathspec_set_for_object`], but uses [`matches_ls_tree_pathspec`] for each
/// element so `ls-files` / index filtering agrees with `ls-tree` on patterns such as `a[a]`.
#[must_use]
pub fn matches_pathspec_set_for_object_ls_tree(
    specs: &[String],
    path: &str,
    mode: u32,
    attr_rules: &[AttrRule],
) -> bool {
    if specs.is_empty() {
        return true;
    }
    let mut positives: Vec<&str> = Vec::new();
    let mut excludes: Vec<&str> = Vec::new();
    for s in specs {
        if pathspec_is_exclude(s) {
            excludes.push(s.as_str());
        } else {
            positives.push(s.as_str());
        }
    }
    let positive_ok = if positives.is_empty() {
        true
    } else {
        positives
            .iter()
            .any(|s| matches_ls_tree_pathspec(s, path, mode, attr_rules))
    };
    if !positive_ok {
        return false;
    }
    for ex in excludes {
        if matches_ls_tree_pathspec(ex, path, mode, attr_rules) {
            return false;
        }
    }
    true
}

/// True if `path` matches the combined pathspec list: any positive spec (or all paths when there
/// are only excludes, matching Git `parse_pathspec`), and not matched by any exclude spec.
#[must_use]
pub fn matches_pathspec_set_for_object(
    specs: &[String],
    path: &str,
    mode: u32,
    attr_rules: &[AttrRule],
) -> bool {
    if specs.is_empty() {
        return true;
    }
    let mut positives: Vec<&str> = Vec::new();
    let mut excludes: Vec<&str> = Vec::new();
    for s in specs {
        if pathspec_is_exclude(s) {
            excludes.push(s.as_str());
        } else {
            positives.push(s.as_str());
        }
    }
    let positive_ok = if positives.is_empty() {
        true
    } else {
        positives
            .iter()
            .any(|s| matches_pathspec_for_object(s, path, mode, attr_rules))
    };
    if !positive_ok {
        return false;
    }
    for ex in excludes {
        if matches_pathspec_for_object(ex, path, mode, attr_rules) {
            return false;
        }
    }
    true
}

/// True if `spec` uses `:(top)` or short `:/` (repo-root-relative) magic.
#[must_use]
pub fn pathspec_has_top(spec: &str) -> bool {
    let (elem_magic, _) = parse_element_magic(spec);
    combine_magic(elem_magic).top
}

/// True if `spec` carries an `attr:` clause in long-form `:(...)` magic, so matching it needs
/// `.gitattributes` rules. Malformed clauses count too; [`validate_attr_pathspecs`] rejects them.
///
/// Such specs only match through the attribute-aware entry points:
/// [`crate::repo::Repository::pathspec_matcher`], which loads the repository's attributes and
/// validates the clauses, or [`matches_pathspec_list_for_object`] with caller-supplied rules.
#[must_use]
pub fn pathspec_has_attr_magic(spec: &str) -> bool {
    // Attr magic only exists in long form; skip the allocating parse for every other spec.
    spec.starts_with(":(") && spec.contains("attr:") && !attr_clauses(spec).is_empty()
}

/// A resolved pathspec list bound to one repository's matching rules, built by
/// [`crate::repo::Repository::pathspec_matcher`].
///
/// Holds the repository's attributes when a spec carries `:(attr:...)` magic, so testing many
/// paths reads them once. Attr requirements see `[attr]` macro expansion and match attribute
/// patterns case-insensitively when `core.ignorecase` is set, as Git does. Paths are treated as regular files: a spec ending in `/` does not match
/// a path equal to its directory, and `builtin_objectmode` reads as `100644`. Use
/// [`matches_pathspec_list_for_object`] when the entry mode matters.
#[derive(Debug, Clone)]
pub struct PathspecMatcher<'s> {
    specs: Cow<'s, [String]>,
    precompose: bool,
    attrs: Option<std::sync::Arc<crate::attributes::ParsedGitAttributes>>,
    /// Match attribute patterns case-insensitively (`core.ignorecase`).
    attr_icase: bool,
}

impl<'s> PathspecMatcher<'s> {
    pub(crate) fn new(
        specs: &'s [String],
        precompose: bool,
        attrs: Option<std::sync::Arc<crate::attributes::ParsedGitAttributes>>,
        attr_icase: bool,
    ) -> Self {
        // The attribute-aware matcher has no precompose context, so normalize the specs up front.
        let specs = if precompose && attrs.is_some() {
            Cow::Owned(
                specs
                    .iter()
                    .map(|spec| precompose_utf8_path(spec).into_owned())
                    .collect(),
            )
        } else {
            Cow::Borrowed(specs)
        };
        Self {
            specs,
            precompose,
            attrs,
            attr_icase,
        }
    }

    /// Whether `path` is selected: it matches at least one positive spec (any path when there
    /// are none or the list is empty) and no `:(exclude)` spec.
    #[must_use]
    pub fn matches(&self, path: &str) -> bool {
        const REGULAR_FILE: u32 = 0o100644;
        let Some(attrs) = &self.attrs else {
            return matches_pathspec_list_with_context(
                path,
                &self.specs,
                pathspec_match_context_with_precompose(self.precompose),
            );
        };
        let path = if self.precompose {
            precompose_utf8_path(path)
        } else {
            Cow::Borrowed(path)
        };
        let lookup = AttrLookup::new(&attrs.rules, &attrs.macros, self.attr_icase);
        matches_pathspec_list_with_attrs(&path, REGULAR_FILE, lookup, &self.specs)
    }

    /// Whether a changed file is selected, checking its current `path` and, for a rename or
    /// copy, its `old_path` too (Git limits diffs on either side of a rename).
    #[must_use]
    pub fn matches_change(&self, path: &str, old_path: Option<&str>) -> bool {
        self.specs.is_empty()
            || self.matches(path)
            || old_path.is_some_and(|old| old != path && self.matches(old))
    }

    /// Whether the list is empty (every path is selected).
    #[must_use]
    pub fn is_empty(&self) -> bool {
        self.specs.is_empty()
    }
}

fn pathspec_match_one_positive(path: &str, magic: PathspecMagic, raw_pattern: &str) -> bool {
    if magic.literal && magic.glob {
        return false;
    }
    let pattern = strip_top_magic(raw_pattern);
    let path_for_match = if let Some(prefix) = magic.prefix.as_deref() {
        if !path.starts_with(prefix) {
            return false;
        }
        &path[prefix.len()..]
    } else {
        path
    };
    pathspec_matches_tail(pattern, path_for_match, magic)
}

/// Attribute state that `:(attr:...)` requirements are evaluated against.
#[derive(Clone, Copy)]
pub(crate) struct AttrLookup<'a> {
    rules: &'a [AttrRule],
    /// `[attr]` macro definitions; `None` means no macros are defined.
    macros: Option<&'a MacroTable>,
    /// Match attribute patterns case-insensitively (`core.ignorecase`).
    icase: bool,
}

impl<'a> AttrLookup<'a> {
    pub(crate) fn new(rules: &'a [AttrRule], macros: &'a MacroTable, icase: bool) -> Self {
        Self {
            rules,
            macros: Some(macros),
            icase,
        }
    }

    fn rules_only(rules: &'a [AttrRule]) -> Self {
        Self {
            rules,
            macros: None,
            icase: false,
        }
    }
}

fn attr_requirements_match(
    requirements: &[AttrRequirement],
    attrs: AttrLookup<'_>,
    path: &str,
    is_dir: bool,
    mode: u32,
) -> bool {
    let no_macros = MacroTable::default();
    let assigned = path_gitattribute_values(
        attrs.rules,
        attrs.macros.unwrap_or(&no_macros),
        attrs.icase,
        path,
        is_dir,
    );
    requirements.iter().all(|req| {
        let value: Option<Cow<'_, str>> = if req.name() == "builtin_objectmode" {
            (mode != 0).then(|| Cow::Owned(format!("{mode:06o}")))
        } else {
            assigned
                .get(req.name())
                .map(|v| Cow::Borrowed(v.display()))
                .filter(|v| v != "unspecified")
        };
        match req {
            AttrRequirement::Set(_) => value.as_deref() == Some("set"),
            AttrRequirement::Unset(_) => value.as_deref() == Some("unset"),
            AttrRequirement::Unspecified(_) => value.is_none(),
            AttrRequirement::Value(_, expected) => value.as_deref() == Some(expected.as_str()),
        }
    })
}

fn matches_pathspec_element_with_context(
    spec: &str,
    path: &str,
    ctx: PathspecMatchContext,
) -> bool {
    let (elem_magic, raw_pattern) = parse_element_magic(spec);
    let magic = combine_magic(elem_magic);
    if magic.exclude {
        return false;
    }
    if magic.literal && magic.glob {
        return false;
    }
    if !magic.attr_requirements.is_empty() {
        return false;
    }
    if magic.literal || magic.glob || magic.icase {
        return pathspec_matches(spec, path);
    }
    let pattern = strip_top_magic(raw_pattern);
    let path_for_match = if let Some(prefix) = magic.prefix.as_deref() {
        if !path.starts_with(prefix) {
            return false;
        }
        &path[prefix.len()..]
    } else {
        path
    };
    matches_pathspec_with_context(pattern, path_for_match, ctx)
}

fn pathspec_exclude_element_matches_with_context(
    spec: &str,
    path: &str,
    ctx: PathspecMatchContext,
) -> bool {
    let (elem_magic, raw_pattern) = parse_element_magic(spec);
    let mut magic = combine_magic(elem_magic);
    if !magic.exclude {
        return false;
    }
    magic.exclude = false;
    if magic.literal && magic.glob {
        return false;
    }
    if !magic.attr_requirements.is_empty() {
        // Attribute pathspecs need `.gitattributes` context; use
        // [`matches_pathspec_list_for_object`] for those.
        return false;
    }
    if magic.literal || magic.glob || magic.icase {
        return pathspec_match_one_positive(path, magic, raw_pattern);
    }
    let pattern = strip_top_magic(raw_pattern);
    let path_for_match = if let Some(prefix) = magic.prefix.as_deref() {
        if !path.starts_with(prefix) {
            return false;
        }
        &path[prefix.len()..]
    } else {
        path
    };
    matches_pathspec_with_context(pattern, path_for_match, ctx)
}

/// True if `path` is matched by an exclude pathspec's pattern. Returns `false` if `spec` is not
/// an exclude pathspec.
#[must_use]
pub fn pathspec_exclude_matches(spec: &str, path: &str) -> bool {
    pathspec_exclude_element_matches_with_context(spec, path, PathspecMatchContext::default())
}

/// When every pathspec is an exclude and none use `:(top)` / `:/`, Git prepends an implicit
/// positive that matches only under the process cwd (relative to the work tree), not the whole
/// repository (`PATHSPEC_PREFER_CWD` in `pathspec.c`). `cwd_from_repo_root` is that prefix
/// without a trailing slash, or empty at the work tree root.
#[must_use]
pub fn extend_pathspec_list_implicit_cwd(
    specs: &[String],
    cwd_from_repo_root: Option<&str>,
) -> Vec<String> {
    if specs.is_empty() {
        return specs.to_vec();
    }
    if !specs.iter().all(|s| pathspec_is_exclude(s)) {
        return specs.to_vec();
    }
    let any_top = specs.iter().any(|s| pathspec_has_top(s));
    if any_top {
        return specs.to_vec();
    }
    let Some(cwd) = cwd_from_repo_root.map(str::trim).filter(|s| !s.is_empty()) else {
        return specs.to_vec();
    };
    let cwd = cwd.trim_end_matches('/');
    if cwd.is_empty() {
        return specs.to_vec();
    }
    let mut out = Vec::with_capacity(specs.len() + 1);
    out.push(format!("{cwd}/"));
    out.extend_from_slice(specs);
    out
}

/// Git `match_pathspec` semantics over a pathspec list: OR of positive specs minus OR of exclude
/// specs. If every element is exclude-only, Git implicitly prepends `.` (match all); this
/// function does the same.
///
/// No `.gitattributes` rules are available here, so a positive `:(attr:...)` spec never matches
/// and an exclude `:(exclude,attr:...)` spec never excludes. Use
/// [`crate::repo::Repository::paths_match_pathspecs`] or [`matches_pathspec_list_for_object`]
/// when specs may carry attr magic (see [`pathspec_has_attr_magic`]).
#[must_use]
pub fn matches_pathspec_list(path: &str, specs: &[String]) -> bool {
    matches_pathspec_list_with_context(path, specs, PathspecMatchContext::default())
}

/// Like [`matches_pathspec_list`], but uses `ctx` for non-magic pathspec elements (trailing `/`).
///
/// Like [`matches_pathspec_list`], this does not evaluate `:(attr:...)` magic: positive attr specs
/// never match and attr excludes never exclude.
#[must_use]
pub fn matches_pathspec_list_with_context(
    path: &str,
    specs: &[String],
    ctx: PathspecMatchContext,
) -> bool {
    if specs.is_empty() {
        return true;
    }
    let mut has_exclude = false;
    let mut has_positive = false;
    let mut positive = false;
    for spec in specs {
        if pathspec_is_exclude(spec) {
            has_exclude = true;
        } else {
            has_positive = true;
            positive = positive || matches_pathspec_element_with_context(spec, path, ctx);
        }
    }
    if has_positive && !positive {
        return false;
    }
    if !has_exclude {
        return true;
    }
    let excluded = specs.iter().any(|s| {
        pathspec_is_exclude(s) && pathspec_exclude_element_matches_with_context(s, path, ctx)
    });
    !excluded
}

/// Whether any of `sorted_paths` matches the single pathspec `spec`.
///
/// Same result as testing every path with [`matches_pathspec_list_with_context`] and `[spec]`,
/// but a plain literal pathspec (no magic, no wildcards, no `./`) is answered with a binary
/// search instead of a scan, so checking many pathspecs against a large index stays cheap.
///
/// `sorted_paths` must be sorted in byte order (the [`str`] ordering) for the result to be
/// correct; duplicates are allowed.
pub(crate) fn sorted_paths_match_pathspec<S: AsRef<str>>(
    spec: &str,
    sorted_paths: &[S],
    ctx: PathspecMatchContext,
) -> bool {
    let list = [spec.to_owned()];
    let matches_linearly = |p: &S| matches_pathspec_list_with_context(p.as_ref(), &list, ctx);
    if !is_plain_literal_pathspec(spec, ctx) {
        return sorted_paths.iter().any(matches_linearly);
    }
    let spec_nfc = literal_spec_for_lookup(spec, ctx);
    if sorted_literal_lookup(&spec_nfc, sorted_paths) {
        return true;
    }
    // A hit on the NFC spelling is always a match; a path stored in another normalization form
    // can only match through a non-ASCII segment, so a miss rechecks just those paths.
    ctx.precompose_paths
        && sorted_paths
            .iter()
            .filter(|p| !p.as_ref().is_ascii())
            .any(matches_linearly)
}

/// Binary-search `sorted_paths` for the plain literal `spec` (exact path or directory prefix).
fn sorted_literal_lookup<S: AsRef<str>>(spec: &str, sorted_paths: &[S]) -> bool {
    let has_path_with_prefix = |prefix: &str| {
        let at = sorted_paths.partition_point(|p| p.as_ref() < prefix);
        sorted_paths
            .get(at)
            .is_some_and(|p| p.as_ref().starts_with(prefix))
    };
    if spec.ends_with('/') {
        return has_path_with_prefix(spec);
    }
    let at = sorted_paths.partition_point(|p| p.as_ref() < spec);
    if sorted_paths.get(at).is_some_and(|p| p.as_ref() == spec) {
        return true;
    }
    has_path_with_prefix(&format!("{spec}/"))
}

/// The spelling of a plain literal `spec` that [`matches_pathspec_with_context`] compares paths
/// against: NFC when the context precomposes paths, else `spec` itself.
fn literal_spec_for_lookup(spec: &str, ctx: PathspecMatchContext) -> Cow<'_, str> {
    if ctx.precompose_paths {
        precompose_utf8_path(spec)
    } else {
        Cow::Borrowed(spec)
    }
}

/// Repository-relative paths when every spec is a plain literal file path (no trailing `/`).
///
/// Used to narrow index/worktree diffs to a small path set without scanning the full index.
/// Returns `None` when `ctx` precomposes paths: another normalization form of a spec could also
/// match, and only a scan of the candidate paths can find it.
pub(crate) fn exact_literal_pathspec_set(
    specs: &[String],
    ctx: PathspecMatchContext,
) -> Option<std::collections::BTreeSet<String>> {
    if specs.is_empty() || ctx.precompose_paths {
        return None;
    }
    if !specs
        .iter()
        .all(|s| is_plain_literal_pathspec(s, ctx) && !s.ends_with('/'))
    {
        return None;
    }
    Some(specs.iter().cloned().collect())
}

/// Whether `spec` matches exactly the path `spec` and everything under directory `spec`
/// (no magic, no wildcards, no `./`, and no context that changes trailing-`/` handling).
///
/// With [`PathspecMatchContext::precompose_paths`], "the path `spec`" means equality after NFC
/// normalization of both sides; see [`literal_spec_for_lookup`].
fn is_plain_literal_pathspec(spec: &str, ctx: PathspecMatchContext) -> bool {
    !ctx.is_directory
        && !ctx.is_git_submodule
        && !spec.is_empty()
        && spec != "."
        && !spec.starts_with("./")
        && !spec.starts_with(':')
        && simple_length(spec) == spec.len()
        && !(literal_global() || glob_global() || noglob_global() || icase_global())
}

/// A pathspec list prepared once for matching many paths with the same context.
///
/// Gives the same answers as [`matches_pathspec_list_with_context`] with the list and context
/// it was built from. When every element is a plain literal (see
/// [`sorted_paths_match_pathspec`]), a path is answered with hash lookups of the path and its
/// leading directories instead of comparing it against every element, which keeps matching a
/// large index against many pathspecs linear in the index size.
pub(crate) struct PathspecListMatcher<'a> {
    specs: &'a [String],
    ctx: PathspecMatchContext,
    literals: Option<LiteralPathspecSet>,
}

/// Plain literal pathspecs split by how they match, spelled as [`literal_spec_for_lookup`] does.
struct LiteralPathspecSet {
    /// Specs without a trailing `/`: they match the path itself.
    exact: HashSet<String>,
    /// Directories whose contents match: every spec, with one trailing `/` removed.
    dirs: HashSet<String>,
}

impl<'a> PathspecListMatcher<'a> {
    /// Prepare `specs` for repeated matching under `ctx`.
    pub(crate) fn new(specs: &'a [String], ctx: PathspecMatchContext) -> Self {
        let all_plain =
            !specs.is_empty() && specs.iter().all(|s| is_plain_literal_pathspec(s, ctx));
        let literals = all_plain.then(|| {
            let mut set = LiteralPathspecSet {
                exact: HashSet::with_capacity(specs.len()),
                dirs: HashSet::with_capacity(specs.len()),
            };
            for spec in specs {
                let spec = literal_spec_for_lookup(spec, ctx);
                match spec.strip_suffix('/') {
                    Some(dir) => {
                        set.dirs.insert(dir.to_owned());
                    }
                    None => {
                        set.exact.insert(spec.to_string());
                        set.dirs.insert(spec.into_owned());
                    }
                }
            }
            set
        });
        Self {
            specs,
            ctx,
            literals,
        }
    }

    /// Whether `path` matches the pathspec list.
    pub(crate) fn matches(&self, path: &str) -> bool {
        let Some(set) = &self.literals else {
            return matches_pathspec_list_with_context(path, self.specs, self.ctx);
        };
        let path = literal_spec_for_lookup(path, self.ctx);
        let path = path.as_ref();
        if set.exact.contains(path) {
            return true;
        }
        path.bytes()
            .enumerate()
            .any(|(i, b)| b == b'/' && set.dirs.contains(&path[..i]))
    }
}

/// `matches_pathspec_list` for tree/index objects with mode and `.gitattributes` rules.
///
/// `attr_rules` are applied without `[attr]` macro definitions and with case-sensitive
/// patterns; [`crate::repo::Repository::pathspec_matcher`] also honors macros and
/// `core.ignorecase`.
#[must_use]
pub fn matches_pathspec_list_for_object(
    path: &str,
    mode: u32,
    attr_rules: &[AttrRule],
    specs: &[String],
) -> bool {
    matches_pathspec_list_with_attrs(path, mode, AttrLookup::rules_only(attr_rules), specs)
}

pub(crate) fn matches_pathspec_list_with_attrs(
    path: &str,
    mode: u32,
    attrs: AttrLookup<'_>,
    specs: &[String],
) -> bool {
    if specs.is_empty() {
        return true;
    }
    let mut has_exclude = false;
    let mut has_positive = false;
    let mut positive = false;
    for spec in specs {
        if pathspec_is_exclude(spec) {
            has_exclude = true;
        } else {
            has_positive = true;
            positive = positive || matches_pathspec_for_object_with_attrs(spec, path, mode, attrs);
        }
    }
    if has_positive && !positive {
        return false;
    }
    if !has_exclude {
        return true;
    }
    !specs.iter().any(|s| {
        pathspec_is_exclude(s) && matches_pathspec_exclude_for_object(s, path, mode, attrs)
    })
}

fn matches_pathspec_exclude_for_object(
    spec: &str,
    path: &str,
    mode: u32,
    attrs: AttrLookup<'_>,
) -> bool {
    let (elem_magic, raw_pattern) = parse_element_magic(spec);
    let mut magic = combine_magic(elem_magic);
    if !magic.exclude {
        return false;
    }
    magic.exclude = false;
    if magic.literal && magic.glob {
        return false;
    }
    let ctx = context_from_mode_bits(mode);
    let is_dir_for_attr = path.ends_with('/') || ctx.is_directory || ctx.is_git_submodule;
    if !magic.attr_requirements.is_empty()
        && !attr_requirements_match(&magic.attr_requirements, attrs, path, is_dir_for_attr, mode)
    {
        return false;
    }
    let pattern = strip_top_magic(raw_pattern);
    let path_for_match = if let Some(prefix) = magic.prefix.as_deref() {
        if !path.starts_with(prefix) {
            return false;
        }
        &path[prefix.len()..]
    } else {
        path
    };
    if magic.literal || magic.glob || magic.icase {
        pathspec_matches_tail(pattern, path_for_match, magic)
    } else {
        matches_pathspec_with_context(pattern, path_for_match, ctx)
    }
}

fn pathspec_matches_tail(pattern: &str, path: &str, magic: PathspecMagic) -> bool {
    if pattern.is_empty() {
        return true;
    }

    let flags = if magic.icase { WM_CASEFOLD } else { 0 };

    if magic.literal {
        return literal_prefix_match(pattern, path);
    }

    let wm_flags = if magic.glob {
        flags | WM_PATHNAME
    } else {
        flags
    };

    let pattern_bytes = pattern.as_bytes();
    let path_bytes = path.as_bytes();
    let simple = simple_length(pattern);

    // Git `match_pathspec_item`: exact / directory prefix before `git_fnmatch`.
    // Only when the pattern has no glob metacharacters (`simple_length` spans the whole pattern);
    // otherwise a pattern like `a[a]` must not match children via `a[a]/` prefix (t6130 vs ls-tree).
    if ps_str_eq(pattern, path, magic.icase) {
        return true;
    }
    if simple == pattern.len() {
        if let Some(prefix) = pattern.strip_suffix('/') {
            if ps_str_eq(prefix, path, magic.icase) {
                return true;
            }
            if path_is_under_dir(path, prefix, magic.icase) {
                return true;
            }
        } else if path_is_under_dir(path, pattern, magic.icase) {
            return true;
        }
    }

    // `:(glob)**/*.txt` at repo root: Git matches `untracked.txt` (leading `**/` is optional).
    if magic.glob && !path.contains('/') && pattern.starts_with("**/") {
        if wildmatch(pattern_bytes, path_bytes, wm_flags) {
            return true;
        }
        if let Some(suffix) = pattern.strip_prefix("**/") {
            if wildmatch(suffix.as_bytes(), path_bytes, wm_flags) {
                return true;
            }
        }
    }

    // Wildcard: require literal bytes up to `simple_length`, then wildmatch the tail only.
    if simple < pattern.len() {
        if path_bytes.len() < simple {
            return false;
        }
        let path_lit = &path_bytes[..simple];
        let pat_lit = &pattern_bytes[..simple];
        let same = if magic.icase {
            path_lit.eq_ignore_ascii_case(pat_lit)
        } else {
            path_lit == pat_lit
        };
        if !same {
            return false;
        }
        let pat_rest = &pattern[simple..];
        let path_rest = &path[simple..];
        return wildmatch(pat_rest.as_bytes(), path_rest.as_bytes(), wm_flags);
    }

    ps_str_eq(pattern, path, magic.icase) || path_is_under_dir(path, pattern, magic.icase)
}

fn ps_str_eq(a: &str, b: &str, icase: bool) -> bool {
    if icase {
        a.eq_ignore_ascii_case(b)
    } else {
        a == b
    }
}

fn path_starts_with(path: &str, prefix: &str, icase: bool) -> bool {
    if icase {
        path.get(..prefix.len())
            .is_some_and(|head| head.eq_ignore_ascii_case(prefix))
    } else {
        path.starts_with(prefix)
    }
}

/// Whether `path` starts with `dir` followed by `/` (the path lies inside directory `dir`).
fn path_is_under_dir(path: &str, dir: &str, icase: bool) -> bool {
    path.as_bytes().get(dir.len()) == Some(&b'/') && path_starts_with(path, dir, icase)
}

fn literal_prefix_match(pattern: &str, path: &str) -> bool {
    if let Some(prefix) = pattern.strip_suffix('/') {
        return path == prefix || path_is_under_dir(path, prefix, false);
    }
    path == pattern || path_is_under_dir(path, pattern, false)
}

/// Literal pathspec match for `ls-tree` when the pattern has no `*`/`?` (brackets stay literal).
fn ls_tree_literal_match(pattern: &str, path: &str, ctx: PathspecMatchContext) -> bool {
    if let Some(prefix) = pattern.strip_suffix('/') {
        if path_is_under_dir(path, prefix, false) {
            return true;
        }
        if path == prefix {
            return ctx.is_directory || ctx.is_git_submodule;
        }
        return false;
    }
    path == pattern || path_is_under_dir(path, pattern, false)
}

/// Optional path metadata for literal pathspecs with a trailing `/` (tree-walk / diff-tree).
///
/// Git treats `dir/` as “directory or git submodule only”: a regular file `dir`
/// does not match, but a tree entry `dir` or gitlink `dir` does.
#[derive(Debug, Clone, Copy, Default, PartialEq, Eq)]
pub struct PathspecMatchContext {
    /// The index/tree entry is a directory (mode `040000`).
    pub is_directory: bool,
    /// The entry is a git submodule / gitlink (`160000`).
    pub is_git_submodule: bool,
    /// NFC-normalize pathspec and path when matching (per-repository `core.precomposeunicode`).
    pub precompose_paths: bool,
}

/// Returns whether `path` matches the pathspec `spec` with default (file) context.
///
/// For pathspecs ending in `/`, a path equal to the prefix matches only when
/// [`PathspecMatchContext`] indicates a directory or submodule; see
/// [`matches_pathspec_with_context`].
///
/// `:(attr:...)` requirements are not evaluated here (only the pattern part is matched); use
/// [`matches_pathspec_for_object`] or [`crate::repo::Repository::paths_match_pathspecs`] for
/// specs where [`pathspec_has_attr_magic`] is true.
#[must_use]
pub fn matches_pathspec(spec: &str, path: &str) -> bool {
    matches_pathspec_with_context(spec, path, PathspecMatchContext::default())
}

/// Like [`matches_pathspec`], but uses `ctx` for trailing-`/` literal pathspecs and for
/// wildcard pathspecs where the pattern continues after a directory boundary (Git
/// `matches_pathspec` + directory semantics).
#[must_use]
pub fn matches_pathspec_with_context(spec: &str, path: &str, ctx: PathspecMatchContext) -> bool {
    let spec_nfc: Cow<'_, str> = if ctx.precompose_paths {
        precompose_utf8_path(spec)
    } else {
        Cow::Borrowed(spec)
    };
    let path_nfc: Cow<'_, str> = if ctx.precompose_paths {
        precompose_utf8_path(path)
    } else {
        Cow::Borrowed(path)
    };
    let spec = spec_nfc.as_ref();
    let path = path_nfc.as_ref();

    let trimmed = spec.strip_prefix("./").unwrap_or(spec);
    if trimmed == "." || trimmed.is_empty() {
        return true;
    }

    let (elem_magic, raw_pattern) = parse_element_magic(trimmed);
    let magic = combine_magic(elem_magic);

    if magic.literal && magic.glob {
        return false;
    }
    if magic.exclude {
        return false;
    }

    let pattern = strip_top_magic(raw_pattern);
    let path_for_match = if let Some(prefix) = magic.prefix.as_deref() {
        if !path.starts_with(prefix) {
            return false;
        }
        &path[prefix.len()..]
    } else {
        path
    };

    if magic.literal {
        if let Some(prefix) = pattern.strip_suffix('/') {
            if path_is_under_dir(path_for_match, prefix, false) {
                return true;
            }
            if path_for_match == prefix {
                return ctx.is_directory || ctx.is_git_submodule;
            }
            return false;
        }
        return path_for_match == pattern || path_is_under_dir(path_for_match, pattern, false);
    }

    // No wildcards and trailing `/`: directory-only semantics (Git `matches_pathspec`).
    if let Some(prefix) = pattern.strip_suffix('/') {
        if simple_length(pattern) == pattern.len() {
            if path_is_under_dir(path_for_match, prefix, false) {
                return true;
            }
            if path_for_match == prefix {
                return ctx.is_directory || ctx.is_git_submodule;
            }
            return false;
        }
    }

    if pathspec_matches_tail(pattern, path_for_match, magic) {
        return true;
    }

    if (ctx.is_directory || ctx.is_git_submodule)
        && !path_for_match.is_empty()
        && pattern.len() > path_for_match.len()
        && pattern.as_bytes().get(path_for_match.len()) == Some(&b'/')
        && pattern.starts_with(path_for_match)
        && simple_length(pattern) < pattern.len()
    {
        return true;
    }

    false
}

/// Parse a Git mode string (e.g. `100644`, `040000`) into a [`PathspecMatchContext`].
#[must_use]
pub fn context_from_mode_octal(mode: &str) -> PathspecMatchContext {
    let Ok(bits) = u32::from_str_radix(mode, 8) else {
        return PathspecMatchContext::default();
    };
    context_from_mode_bits(bits)
}

/// Classify a raw Git mode (e.g. from an index or tree entry) for pathspec matching.
#[must_use]
pub fn context_from_mode_bits(mode: u32) -> PathspecMatchContext {
    let ty = mode & 0o170000;
    PathspecMatchContext {
        is_directory: ty == 0o040000,
        is_git_submodule: ty == 0o160000,
        precompose_paths: false,
    }
}

/// Pathspec matching for `ls-tree` after Git forces `pathspec.has_wildcard = 0` (`ls-tree.c`).
///
/// Metacharacters `*` / `?` still participate in [`wildmatch`]; `[` and `\\` are **not** glob
/// starters unless a `*` or `?` appears — so `a[a]` matches the literal directory `a[a]` (t3102),
/// while `a*` matches `a/one`, `aa/two`, `a[a]/three`, …
#[must_use]
pub fn matches_ls_tree_pathspec(
    spec: &str,
    path: &str,
    mode: u32,
    attr_rules: &[AttrRule],
) -> bool {
    let (elem_magic, raw_pattern) = parse_element_magic(spec);
    let mut magic = combine_magic(elem_magic);
    magic.exclude = false;

    if magic.literal && magic.glob {
        return false;
    }

    let ctx = context_from_mode_bits(mode);
    let is_dir_for_attr = path.ends_with('/') || ctx.is_directory || ctx.is_git_submodule;

    if !magic.attr_requirements.is_empty()
        && !attr_requirements_match(
            &magic.attr_requirements,
            AttrLookup::rules_only(attr_rules),
            path,
            is_dir_for_attr,
            mode,
        )
    {
        return false;
    }

    let pattern = strip_top_magic(raw_pattern);
    let path_for_match = if let Some(prefix) = magic.prefix.as_deref() {
        if !path.starts_with(prefix) {
            return false;
        }
        &path[prefix.len()..]
    } else {
        path
    };

    if magic.literal || magic.glob || magic.icase {
        return pathspec_matches_tail(pattern, path_for_match, magic);
    }

    let spec_nfc: Cow<'_, str> = Cow::Borrowed(pattern);
    let path_nfc: Cow<'_, str> = Cow::Borrowed(path_for_match);
    let pattern = spec_nfc.as_ref();
    let path = path_nfc.as_ref();

    let trimmed = pattern.strip_prefix("./").unwrap_or(pattern);
    if trimmed == "." || trimmed.is_empty() {
        return true;
    }

    let uses_star_or_question = trimmed.contains('*') || trimmed.contains('?');
    if !uses_star_or_question {
        return ls_tree_literal_match(trimmed, path, ctx);
    }

    let nwl = simple_length(trimmed);
    let flags = 0u32;
    if nwl == trimmed.len() {
        return wildmatch(trimmed.as_bytes(), path.as_bytes(), flags);
    }
    let lit = trimmed.as_bytes().get(..nwl).unwrap_or_default();
    let path_b = path.as_bytes();
    if path_b.len() < nwl {
        return false;
    }
    if &path_b[..nwl] != lit {
        return false;
    }
    let pat_rest = &trimmed[nwl..];
    let path_rest = &path[nwl..];
    wildmatch(pat_rest.as_bytes(), path_rest.as_bytes(), flags)
}

/// Match a pathspec against a tree path, using `.gitattributes` for `:(attr:...)`.
///
/// `mode` supplies directory/gitlink context for literal pathspecs ending in `/`.
#[must_use]
pub fn matches_pathspec_for_object(
    spec: &str,
    path: &str,
    mode: u32,
    attr_rules: &[AttrRule],
) -> bool {
    matches_pathspec_for_object_with_attrs(spec, path, mode, AttrLookup::rules_only(attr_rules))
}

fn matches_pathspec_for_object_with_attrs(
    spec: &str,
    path: &str,
    mode: u32,
    attrs: AttrLookup<'_>,
) -> bool {
    let (elem_magic, raw_pattern) = parse_element_magic(spec);
    let mut magic = combine_magic(elem_magic);
    magic.exclude = false;

    if magic.literal && magic.glob {
        return false;
    }

    let ctx = context_from_mode_bits(mode);
    let is_dir_for_attr = path.ends_with('/') || ctx.is_directory || ctx.is_git_submodule;

    if !magic.attr_requirements.is_empty()
        && !attr_requirements_match(&magic.attr_requirements, attrs, path, is_dir_for_attr, mode)
    {
        return false;
    }

    let pattern = strip_top_magic(raw_pattern);
    let path_for_match = if let Some(prefix) = magic.prefix.as_deref() {
        if !path.starts_with(prefix) {
            return false;
        }
        &path[prefix.len()..]
    } else {
        path
    };
    if magic.literal || magic.glob || magic.icase {
        pathspec_matches_tail(pattern, path_for_match, magic)
    } else {
        matches_pathspec_with_context(pattern, path_for_match, ctx)
    }
}

/// Resolved path lies outside the repository work tree (Git `prefix_path_gently` failure).
#[derive(Debug, Clone)]
pub struct PathOutsideRepository {
    /// User-facing pathspec token (argv element).
    pub elt: String,
    /// Resolved absolute path outside the work tree.
    pub path: String,
    /// Canonical work tree root.
    pub work_tree: PathBuf,
}

impl std::fmt::Display for PathOutsideRepository {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        write!(
            f,
            "{}: '{}' is outside repository at '{}'",
            self.elt,
            self.path,
            self.work_tree.display()
        )
    }
}

impl From<PathOutsideRepository> for PathspecResolveError {
    fn from(outside: PathOutsideRepository) -> Self {
        Self::OutsideRepository {
            elt: outside.elt,
            path: outside.path,
            work_tree: outside.work_tree.display().to_string(),
        }
    }
}

/// Resolve a magic pathspec relative to a current-directory prefix.
///
/// This keeps the `cwd` prefix case-sensitive (via an internal `prefix:` magic
/// token) while still honoring magic options like `icase` for the tail.
/// Returns `None` when `spec` is not a parseable magic pathspec.
pub fn resolve_magic_pathspec(spec: &str, cwd_prefix: &str) -> Option<String> {
    if !spec.starts_with(":(") {
        return None;
    }
    let close_idx = spec.find(')')?;
    let magic_prefix = &spec[..=close_idx];
    let tail = &spec[close_idx + 1..];
    Some(resolve_magic_pathspec_parts(magic_prefix, tail, cwd_prefix))
}

fn resolve_magic_pathspec_parts(magic_prefix: &str, tail: &str, cwd_prefix: &str) -> String {
    if has_magic_prefix_token(magic_prefix) {
        return format!("{magic_prefix}{tail}");
    }

    if let Some(rooted_tail) = tail.strip_prefix('/') {
        return format!("{magic_prefix}{}", normalize_relative_path_str(rooted_tail));
    }

    let combined = if cwd_prefix.is_empty() {
        normalize_relative_path_str(tail)
    } else {
        normalize_relative_path_str(&format!("{cwd_prefix}{tail}"))
    };

    let cwd_base = normalize_relative_path_str(cwd_prefix.trim_end_matches('/'));
    if !cwd_base.is_empty()
        && (combined == cwd_base || combined.starts_with(&format!("{cwd_base}/")))
    {
        let after_base = combined
            .strip_prefix(&cwd_base)
            .unwrap_or(combined.as_str());
        let remainder = after_base.strip_prefix('/').unwrap_or(after_base);
        let magic_with_prefix = inject_magic_prefix_token(magic_prefix, &format!("{cwd_base}/"));
        return format!("{magic_with_prefix}{remainder}");
    }

    format!("{magic_prefix}{combined}")
}

fn has_magic_prefix_token(magic_prefix: &str) -> bool {
    let Some(inner) = magic_prefix
        .strip_prefix(":(")
        .and_then(|s| s.strip_suffix(')'))
    else {
        return false;
    };
    inner
        .split(',')
        .map(str::trim)
        .any(|token| token.starts_with("prefix:"))
}

fn inject_magic_prefix_token(magic_prefix: &str, prefix: &str) -> String {
    let Some(inner) = magic_prefix
        .strip_prefix(":(")
        .and_then(|s| s.strip_suffix(')'))
    else {
        return magic_prefix.to_string();
    };
    if inner.trim().is_empty() {
        format!(":(prefix:{prefix})")
    } else {
        format!(":({inner},prefix:{prefix})")
    }
}

/// Drop repeated leading `./` / `.//` from a repo-root pathspec (Git `prefix_path` behavior).
///
/// When stripping exposes a leading `:`, wrap with `:(literal)` so a filesystem path like
/// `:!foo` is not reinterpreted as short exclude magic.
fn normalize_repo_root_relative_pathspec(pathspec: &str) -> String {
    let mut spec = pathspec;
    let mut stripped = false;
    while let Some(rest) = spec.strip_prefix("./") {
        stripped = true;
        spec = rest.trim_start_matches('/');
    }
    if !stripped {
        return pathspec.to_owned();
    }
    if spec.is_empty() {
        return ".".to_owned();
    }
    if spec.starts_with(':') && !literal_pathspecs_enabled() {
        return format!(":(literal){spec}");
    }
    spec.to_owned()
}

/// Normalize user pathspec separators on Windows (`\` → `/`) before resolution.
fn normalize_pathspec_separators(pathspec: &str) -> std::borrow::Cow<'_, str> {
    #[cfg(windows)]
    {
        if !pathspec.starts_with(':') && pathspec.contains('\\') {
            return std::borrow::Cow::Owned(pathspec.replace('\\', "/"));
        }
    }
    std::borrow::Cow::Borrowed(pathspec)
}

fn normalize_relative_path_str(path: &str) -> String {
    let mut parts: Vec<String> = Vec::new();
    for component in std::path::Path::new(path).components() {
        match component {
            std::path::Component::CurDir => {}
            std::path::Component::ParentDir => {
                parts.pop();
            }
            std::path::Component::Normal(seg) => {
                parts.push(seg.to_string_lossy().to_string());
            }
            std::path::Component::RootDir | std::path::Component::Prefix(_) => {}
        }
    }
    parts.join("/")
}

/// Current directory relative to `work_tree`, or `None` if cwd is the work tree root.
#[must_use]
pub fn pathdiff(cwd: &Path, work_tree: &Path) -> Option<String> {
    use crate::git_path::{path_for_disk_compare, path_to_git_slash};

    let cwd_cmp = path_for_disk_compare(cwd);
    let wt_cmp = path_for_disk_compare(work_tree);

    if cwd_cmp == wt_cmp {
        return None;
    }

    let rel = cwd_cmp.strip_prefix(&wt_cmp).ok()?;
    let git = path_to_git_slash(rel);
    if git.is_empty() {
        None
    } else {
        Some(git)
    }
}

/// For exclude (and other cwd-relative) pathspec magic from a subdirectory, Git resolves the
/// pattern against the current directory (`:!sub/` from `repo/sub` → exclude `sub/sub/`).
fn prepend_cwd_to_short_exclude_pathspec(spec: &str, cwd: &str) -> Option<String> {
    let cwd = cwd.trim_end_matches('/');
    if cwd.is_empty() {
        return None;
    }
    let bytes = spec.as_bytes();
    if bytes.first().copied() != Some(b':') {
        return None;
    }
    // `:/path` is `:(top)` short form — exclude is relative to repo root, not cwd (t6132).
    if bytes.get(1).copied() == Some(b'/') {
        return None;
    }
    let mut i = 1usize;
    while i < bytes.len() && bytes[i] != b':' {
        let ch = bytes[i];
        if ch == b'^' {
            i += 1;
            continue;
        }
        let is_magic = matches!(ch, b'!' | b'/');
        if is_magic {
            i += 1;
            continue;
        }
        break;
    }
    if i < bytes.len() && bytes[i] == b':' {
        i += 1;
    }
    let pattern = spec.get(i..)?;
    if pattern.is_empty() || pattern.starts_with('/') {
        return None;
    }
    Some(format!("{}{}/{pattern}", &spec[..i], cwd))
}

/// Build [`PathspecMatchContext`] with optional NFC path normalization for comparisons.
#[must_use]
pub fn pathspec_match_context_with_precompose(precompose_paths: bool) -> PathspecMatchContext {
    PathspecMatchContext {
        precompose_paths,
        ..PathspecMatchContext::default()
    }
}

/// Like [`matches_pathspec_list`], with an explicit `precompose_paths` flag.
#[must_use]
pub fn matches_pathspec_list_precompose(
    path: &str,
    specs: &[String],
    precompose_paths: bool,
) -> bool {
    matches_pathspec_list_with_context(
        path,
        specs,
        pathspec_match_context_with_precompose(precompose_paths),
    )
}

/// True when argv pathspec strings must not be NFC-normalized (magic syntax).
#[must_use]
pub fn argv_pathspec_preserves_spelling(pathspec: &str) -> bool {
    pathspec.starts_with(":(")
        || pathspec.starts_with(":/")
        || pathspec.starts_with(":!")
        || pathspec.starts_with(":^")
}

/// NFC-normalize a literal command-line pathspec (Git `precompose_argv_prefix`).
#[must_use]
pub fn precompose_argv_pathspec(pathspec: &str) -> Cow<'_, str> {
    if argv_pathspec_preserves_spelling(pathspec) {
        Cow::Borrowed(pathspec)
    } else {
        precompose_utf8_path(pathspec)
    }
}

/// Resolve a pathspec string to a path relative to the repository work tree.
///
/// `prefix` is the current directory relative to the work tree (no trailing slash),
/// or `None` when cwd is the work tree root.
#[must_use]
pub fn resolve_pathspec(pathspec: &str, work_tree: &Path, prefix: Option<&str>) -> String {
    resolve_pathspec_with_argv_precompose(pathspec, work_tree, prefix, false)
}

/// Like [`resolve_pathspec`], but NFC-normalizes literal argv path segments when requested.
#[must_use]
pub fn resolve_pathspec_with_argv_precompose(
    pathspec: &str,
    work_tree: &Path,
    prefix: Option<&str>,
    argv_precompose: bool,
) -> String {
    let pathspec = if argv_precompose {
        precompose_argv_pathspec(pathspec)
    } else {
        Cow::Borrowed(pathspec)
    };
    resolve_pathspec_impl(pathspec.as_ref(), work_tree, prefix)
}

fn resolve_pathspec_impl(pathspec: &str, work_tree: &Path, prefix: Option<&str>) -> String {
    let pathspec = normalize_pathspec_separators(pathspec);
    let pathspec = pathspec.as_ref();
    // Git: `.` at repo root means "match the whole tree" (not an empty pathspec).
    // An empty resolved pathspec would match nothing and breaks `grep -- . t` max-depth.
    if pathspec == "." {
        return match prefix {
            Some(p) if !p.is_empty() => p.to_owned(),
            _ => ".".to_owned(),
        };
    }
    if pathspec.contains("../") || pathspec.starts_with("../") {
        let cwd = match prefix.filter(|p| !p.is_empty()) {
            Some(p) => work_tree.join(p),
            None => work_tree.to_path_buf(),
        };
        let abs = cwd.join(pathspec);
        let mut parts: Vec<std::ffi::OsString> = Vec::new();
        for component in abs.components() {
            use std::path::Component;
            match component {
                Component::ParentDir => {
                    parts.pop();
                }
                Component::CurDir => {}
                other => parts.push(other.as_os_str().to_os_string()),
            }
        }
        let abs_norm: PathBuf = parts.iter().collect();
        if let Some(rel) = crate::git_path::strip_worktree_prefix(&abs_norm, work_tree) {
            return rel;
        }
    }
    if Path::new(pathspec).is_absolute() {
        let abs = Path::new(pathspec);
        if let Some(rel) = crate::git_path::strip_worktree_prefix(abs, work_tree) {
            return rel;
        }
        return pathspec.to_owned();
    }

    if pathspec.starts_with(':') {
        if let Some(p) = prefix {
            if !p.is_empty() && !literal_pathspecs_enabled() {
                let cwd_ps = format!("{}/", p.trim_end_matches('/'));
                if pathspec.starts_with(":(") {
                    if let Some(resolved) = resolve_magic_pathspec(pathspec, &cwd_ps) {
                        return resolved;
                    }
                    return pathspec.to_owned();
                }
                if pathspec_is_exclude(pathspec) {
                    if let Some(fixed) = prepend_cwd_to_short_exclude_pathspec(pathspec, p) {
                        return fixed;
                    }
                }
            }
        }
        if let Some(rest) = pathspec.strip_prefix(":/") {
            // `:/!foo` / `:/^bar` — `:/` is `:(top)`; the tail is still short magic, not a literal path.
            if rest.starts_with('!') || rest.starts_with('^') {
                return pathspec.to_owned();
            }
            return rest.to_owned();
        }
        // Long magic `:(...)` must stay intact — `:(exclude)path` is not the same as `path`
        // (t6132-pathspec-exclude, grep --untracked with exclude pathspecs).
        if pathspec.starts_with(":(") {
            return pathspec.to_owned();
        }
        return pathspec.to_owned();
    }

    match prefix {
        Some(p) if !p.is_empty() => {
            normalize_relative_path_str(&PathBuf::from(p).join(pathspec).to_string_lossy())
        }
        _ => normalize_repo_root_relative_pathspec(pathspec),
    }
}

/// Resolve a pathspec and ensure it lies inside `work_tree` (used by `git add`, etc.).
///
/// Returns [`PathOutsideRepository`] when resolution stays absolute, matching Git's
/// `'%s' is outside repository at '%s'` fatal (t7010).
pub fn resolve_pathspec_in_worktree(
    elt: &str,
    pathspec: &str,
    work_tree: &Path,
    prefix: Option<&str>,
) -> Result<String, PathOutsideRepository> {
    resolve_pathspec_in_worktree_with_argv_precompose(elt, pathspec, work_tree, prefix, false)
}

/// Like [`resolve_pathspec_in_worktree`], NFC-normalizing `pathspec` when `argv_precompose` is set.
///
/// Callers holding a [`Repository`] should use [`Repository::resolve_worktree_pathspec`], which
/// computes `argv_precompose` once per handle instead of reloading config per pathspec.
pub(crate) fn resolve_pathspec_in_worktree_with_argv_precompose(
    elt: &str,
    pathspec: &str,
    work_tree: &Path,
    prefix: Option<&str>,
    argv_precompose: bool,
) -> Result<String, PathOutsideRepository> {
    let resolved =
        resolve_pathspec_with_argv_precompose(pathspec, work_tree, prefix, argv_precompose);
    if Path::new(&resolved).is_absolute() {
        let wt = work_tree
            .canonicalize_simplified()
            .unwrap_or_else(|_| work_tree.to_path_buf());
        return Err(PathOutsideRepository {
            elt: elt.to_string(),
            path: resolved,
            work_tree: wt,
        });
    }
    Ok(resolved)
}

/// Normalize a worktree file path for porcelain commands (`blame`, `log`, …).
///
/// Accepts repo-relative or absolute paths under `work_tree`.
#[must_use]
pub fn normalize_worktree_file_path(
    file_path: &str,
    work_tree: &Path,
    prefix: Option<&str>,
) -> String {
    normalize_worktree_file_path_with_git_dir(file_path, work_tree, prefix, None)
}

/// Like [`normalize_worktree_file_path`], honoring `core.precomposeunicode` argv normalization for `git_dir`.
#[must_use]
pub fn normalize_worktree_file_path_with_git_dir(
    file_path: &str,
    work_tree: &Path,
    prefix: Option<&str>,
    git_dir: Option<&Path>,
) -> String {
    let argv_precompose = argv_precompose_enabled(git_dir);
    let resolved =
        resolve_pathspec_with_argv_precompose(file_path, work_tree, prefix, argv_precompose);
    if Path::new(&resolved).is_absolute() {
        file_path.to_string()
    } else {
        resolved
    }
}

#[cfg(test)]
mod tree_entry_pathspec_tests {
    use super::*;

    #[test]
    fn t6130_bracket_filename_matches_pathspec() {
        assert!(matches_pathspec("f[o][o]", "f[o][o]"));
        assert!(matches_pathspec(":(glob)f[o][o]", "f[o][o]"));
    }

    #[test]
    fn literal_prefix_and_exact() {
        assert!(matches_pathspec("path1", "path1/file1"));
        assert!(matches_pathspec_with_context(
            "path1/",
            "path1/file1",
            PathspecMatchContext::default()
        ));
        assert!(matches_pathspec("file0", "file0"));
        assert!(!matches_pathspec("path", "path1/file1"));
    }

    #[test]
    fn dir_prefix_needs_slash_boundary() {
        assert!(matches_pathspec("a/b", "a/b/c"));
        assert!(!matches_pathspec("a/b", "a/bc"));
        assert!(!matches_pathspec("a/b", "a"));
        assert!(matches_pathspec(":(literal)dir", "dir/x"));
        assert!(!matches_pathspec(":(literal)dir", "dirx"));
        assert!(matches_pathspec(":(icase)DIR", "dir/x"));
        assert!(!matches_pathspec(":(icase)DIR", "dirx"));
        assert!(matches_pathspec("dir/", "dir/x"));
        assert!(!matches_pathspec("dir/", "dir"));
    }

    #[test]
    fn sorted_lookup_agrees_with_linear_scan() {
        let mut paths: Vec<&str> = vec![
            "a",
            "a-b",
            "a.txt",
            "a/b",
            "a/b/c.txt",
            "a/bc",
            "a0",
            "b/x[1].txt",
            "dir/sub/file",
            "dir-sub",
            "f[o][o]",
            "z",
            "caf\u{e9}/menu",
            "cafe\u{301}/old",
            "\u{212a}.txt",
        ];
        paths.sort_unstable();
        let specs = [
            "caf\u{e9}",
            "cafe\u{301}/",
            "cafe\u{301}/old",
            "K.txt",
            "\u{212a}.txt",
            "a",
            "a/",
            "a/b",
            "a/b/",
            "a/bc",
            "a-",
            "a-b",
            "a.txt",
            "b",
            "b/",
            "dir",
            "dir/sub",
            "dir/sub/",
            "dir/su",
            "dir-sub",
            "f[o][o]",
            "*.txt",
            "a/*",
            ":(glob)**/*.txt",
            ":(literal)f[o][o]",
            ":(icase)A/B",
            ":!a",
            ".",
            "./a/b",
            "",
            "z",
            "zz",
            "/",
            "a//",
        ];
        for ctx in [
            PathspecMatchContext::default(),
            PathspecMatchContext {
                precompose_paths: true,
                ..PathspecMatchContext::default()
            },
            PathspecMatchContext {
                is_directory: true,
                ..PathspecMatchContext::default()
            },
        ] {
            for spec in specs {
                let list = [spec.to_owned()];
                let linear = paths
                    .iter()
                    .any(|p| matches_pathspec_list_with_context(p, &list, ctx));
                assert_eq!(
                    sorted_paths_match_pathspec(spec, &paths, ctx),
                    linear,
                    "spec {spec:?} ctx {ctx:?}"
                );
            }
        }
        let empty: [&str; 0] = [];
        assert!(!sorted_paths_match_pathspec(
            "a",
            &empty,
            PathspecMatchContext::default()
        ));
    }

    #[test]
    fn list_matcher_agrees_with_linear_matching() {
        let paths = [
            "a",
            "a-b",
            "a.txt",
            "a/",
            "a/b",
            "a/b/c.txt",
            "a/bc",
            "a//x",
            "a0",
            "b/x[1].txt",
            "dir/sub/file",
            "dir-sub",
            "f[o][o]",
            "z",
            "zz/a",
            "é/x",
            "e\u{301}/y",
            "\u{212a}.txt",
            "K.txt",
            "",
        ];
        let lists: &[&[&str]] = &[
            &[],
            &["a"],
            &["a/"],
            &["a//"],
            &["a/b", "dir/sub/", "z"],
            &["a/b/", "a/bc", "dir", "zz"],
            &["dir/su", "dir-sub", "f[o][o]"],
            &["é", "a-"],
            &["e\u{301}", "z"],
            &["e\u{301}/y", "K.txt"],
            &["\u{212a}.txt"],
            &["a", "*.txt"],
            &["dir", ":!dir/sub"],
            &[":!a"],
            &["./a", "z"],
            &[".", "a"],
            &["", "a"],
            &[":(icase)A/B", "z"],
        ];
        for ctx in [
            PathspecMatchContext::default(),
            PathspecMatchContext {
                precompose_paths: true,
                ..PathspecMatchContext::default()
            },
            PathspecMatchContext {
                is_directory: true,
                ..PathspecMatchContext::default()
            },
        ] {
            for list in lists {
                let specs: Vec<String> = list.iter().map(|s| (*s).to_owned()).collect();
                let matcher = PathspecListMatcher::new(&specs, ctx);
                for path in paths {
                    assert_eq!(
                        matcher.matches(path),
                        matches_pathspec_list_with_context(path, &specs, ctx),
                        "path {path:?} specs {specs:?} ctx {ctx:?}"
                    );
                }
            }
        }
    }

    #[test]
    fn list_matcher_uses_literal_lookup_only_for_plain_lists() {
        let ctx = PathspecMatchContext::default();
        let plain = vec!["a/b".to_owned(), "dir/".to_owned()];
        assert!(PathspecListMatcher::new(&plain, ctx).literals.is_some());
        let glob = vec!["a/b".to_owned(), "*.txt".to_owned()];
        assert!(PathspecListMatcher::new(&glob, ctx).literals.is_none());
        let exclude = vec!["a/b".to_owned(), ":!a/b/c".to_owned()];
        assert!(PathspecListMatcher::new(&exclude, ctx).literals.is_none());
        assert!(PathspecListMatcher::new(&[], ctx).literals.is_none());
    }

    #[test]
    fn precomposed_literal_specs_keep_fast_lookups() {
        let ctx = PathspecMatchContext {
            precompose_paths: true,
            ..PathspecMatchContext::default()
        };
        let specs: Vec<String> = (0..100).map(|i| format!("d{i}/file.txt")).collect();
        let matcher = PathspecListMatcher::new(&specs, ctx);
        assert!(
            matcher.literals.is_some(),
            "core.precomposeunicode must not force a per-spec scan of every path"
        );
        assert!(matcher.matches("d42/file.txt"));
        assert!(!matcher.matches("d42/other.txt"));
        let nfd = vec!["cafe\u{301}".to_owned()];
        let matcher = PathspecListMatcher::new(&nfd, ctx);
        assert!(matcher.literals.is_some());
        assert!(matcher.matches("caf\u{e9}/menu"));
        assert!(matcher.matches("cafe\u{301}/menu"));
        assert!(exact_literal_pathspec_set(&specs, ctx).is_none());
    }

    #[test]
    fn ls_tree_bracket_in_name_is_literal_prefix() {
        assert!(matches_ls_tree_pathspec(
            "a[a]",
            "a[a]/three",
            0o100644,
            &[]
        ));
        assert!(!matches_pathspec_with_context(
            "a[a]",
            "a[a]/three",
            PathspecMatchContext::default()
        ));
    }

    #[test]
    fn wildcards_cross_slash_by_default() {
        assert!(matches_pathspec("f*", "file0"));
        assert!(matches_pathspec("*file1", "path1/file1"));
        assert!(matches_pathspec_with_context(
            "path1/f*",
            "path1",
            PathspecMatchContext {
                is_directory: true,
                ..Default::default()
            }
        ));
        assert!(matches_pathspec("path1/*file1", "path1/file1"));
    }

    #[test]
    fn glob_double_star_txt_at_repo_root() {
        assert!(pathspec_matches(":(glob)**/*.txt", "untracked.txt"));
        assert!(pathspec_matches(":(glob)**/*.txt", "d/untracked.txt"));
    }

    #[test]
    fn trailing_slash_directory_only() {
        assert!(!matches_pathspec_with_context(
            "file0/",
            "file0",
            PathspecMatchContext::default()
        ));
        assert!(matches_pathspec_with_context(
            "file0/",
            "file0",
            PathspecMatchContext {
                is_directory: true,
                ..Default::default()
            }
        ));
        assert!(matches_pathspec_with_context(
            "submod/",
            "submod",
            PathspecMatchContext {
                is_git_submodule: true,
                ..Default::default()
            }
        ));
    }

    #[test]
    fn exclude_top_short_magic_subtracts_from_positive() {
        let specs = vec!["*".to_string(), ":/!sub2".to_string()];
        assert!(matches_pathspec_list("sub/file", &specs));
        assert!(!matches_pathspec_list("sub2/file", &specs));
        assert!(pathspec_exclude_matches(":/!sub2", "sub2/file"));
    }
}

#[cfg(test)]
mod exact_literal_pathspec_set_tests {
    use super::*;

    #[test]
    fn exact_literal_pathspec_set_accepts_file_paths() {
        let ctx = PathspecMatchContext::default();
        let set = exact_literal_pathspec_set(&["large.txt".to_string()], ctx).expect("set");
        assert_eq!(set.len(), 1);
        assert!(set.contains("large.txt"));
    }

    #[test]
    fn exact_literal_pathspec_set_rejects_directory_trailing_slash() {
        let ctx = PathspecMatchContext::default();
        assert!(exact_literal_pathspec_set(&["src/".to_string()], ctx).is_none());
    }

    #[test]
    fn exact_literal_pathspec_set_rejects_glob_magic() {
        let ctx = PathspecMatchContext::default();
        assert!(exact_literal_pathspec_set(&[":(glob)*.rs".to_string()], ctx).is_none());
    }
}

#[cfg(test)]
mod pathspec_list_tests {
    use super::*;
    use crate::crlf::parse_gitattributes_content;

    #[test]
    fn exclude_removes_paths_matching_icase_positive() {
        let specs = vec![
            ":(icase)*.txt".to_string(),
            ":(exclude)submodule/subsub/*".to_string(),
        ];
        assert!(path_allowed_by_pathspec_list(&specs, "submodule/g.txt"));
        assert!(!path_allowed_by_pathspec_list(
            &specs,
            "submodule/subsub/e.txt"
        ));
    }

    #[test]
    fn prefixed_attr_exclude_removes_matching_child_path() {
        let specs = vec![
            "sub".to_string(),
            ":(exclude,attr:labelB,prefix:sub/)".to_string(),
        ];
        let exclude_only = vec![":(exclude,attr:labelB,prefix:sub/)".to_string()];
        let attrs = parse_gitattributes_content("fileB labelB\n");
        assert!(!matches_pathspec_list_for_object(
            "sub/fileB",
            0o100644,
            &attrs,
            &specs,
        ));
        assert!(!matches_pathspec_list_for_object(
            "sub/fileB",
            0o100644,
            &attrs,
            &exclude_only,
        ));
    }
}

#[cfg(test)]
mod resolve_pathspec_tests {
    use super::*;

    fn wt() -> PathBuf {
        PathBuf::from("/tmp/grit-pathspec-worktree")
    }

    #[test]
    fn strips_leading_dot_slash_at_repo_root() {
        let wt = wt();
        assert_eq!(resolve_pathspec("./-", &wt, None), "-");
        assert_eq!(resolve_pathspec("./a/b", &wt, None), "a/b");
        assert_eq!(resolve_pathspec("././a", &wt, None), "a");
        assert_eq!(resolve_pathspec(".//a", &wt, None), "a");
        assert_eq!(resolve_pathspec("./", &wt, None), ".");
        assert_eq!(resolve_pathspec(".", &wt, None), ".");
    }

    #[test]
    fn prefixed_cwd_resolution_unchanged() {
        let wt = wt();
        assert_eq!(
            resolve_pathspec("file", &wt, Some("sub/dir")),
            "sub/dir/file"
        );
        assert_eq!(resolve_pathspec(":(glob)*.rs", &wt, None), ":(glob)*.rs");
    }

    #[test]
    fn dot_slash_colon_filename_stays_literal_not_exclude_magic() {
        let wt = wt();
        let resolved = resolve_pathspec("./:!foo", &wt, None);
        assert_eq!(resolved, ":(literal):!foo");
        assert!(!pathspec_is_exclude(&resolved));
        assert!(pathspec_matches(&resolved, ":!foo"));
        assert!(!pathspec_matches(&resolved, "other"));
    }

    #[test]
    fn resolve_absolute_path_under_worktree() {
        let tmp = tempfile::tempdir().expect("tempdir");
        let wt = tmp.path().join("repo");
        std::fs::create_dir_all(wt.join("d")).expect("mkdir");
        let file = wt.join("d/b.txt");
        std::fs::write(&file, b"x").expect("write");
        let wt = wt.canonicalize_simplified().expect("canonicalize");
        let abs = file.canonicalize_simplified().expect("canonicalize file");
        let spec = abs.to_string_lossy();
        assert_eq!(
            resolve_pathspec(&spec, &wt, None),
            "d/b.txt",
            "absolute forward-slash pathspec"
        );
    }

    #[cfg(windows)]
    mod windows {
        use super::*;

        #[test]
        fn resolve_backslash_relative_and_absolute_pathspecs() {
            let tmp = tempfile::tempdir().expect("tempdir");
            let wt = tmp.path().join("repo");
            std::fs::create_dir_all(wt.join("d")).expect("mkdir");
            let file_a = wt.join("d/a.txt");
            let file_b = wt.join("d/b.txt");
            std::fs::write(&file_a, b"a").expect("write");
            std::fs::write(&file_b, b"b").expect("write");
            let wt = wt.canonicalize_simplified().expect("canonicalize worktree");

            assert_eq!(
                resolve_pathspec(r"d\a.txt", &wt, None),
                "d/a.txt",
                "backslash relative pathspec (issue #915 A)"
            );

            let abs_back = file_b.canonicalize_simplified().expect("canonicalize");
            let abs_back_spec = abs_back.display().to_string();
            assert_eq!(
                resolve_pathspec(&abs_back_spec, &wt, None),
                "d/b.txt",
                "absolute backslash pathspec (issue #915 B)"
            );

            let abs_forward = abs_back_spec.replace('\\', "/");
            assert_eq!(
                resolve_pathspec(&abs_forward, &wt, None),
                "d/b.txt",
                "absolute forward-slash pathspec (issue #915 C)"
            );
        }
    }

    #[test]
    fn nfd_pathspec_matches_nfc_index_path_when_precompose_enabled() {
        let nfc_path = "caf\u{00e9}.txt";
        let nfd_spec = format!("cafe\u{0301}.txt");
        assert!(!matches_pathspec_list(&nfc_path, &[nfd_spec.clone()]));
        assert!(matches_pathspec_list_precompose(
            nfc_path,
            &[nfd_spec.clone()],
            true
        ));
    }

    #[test]
    fn argv_precompose_normalizes_nfd_literal_pathspec() {
        let nfd = format!("cafe\u{0301}.txt");
        let nfc = precompose_argv_pathspec(&nfd);
        assert_eq!(nfc.as_ref(), "caf\u{00e9}.txt");
        assert!(argv_pathspec_preserves_spelling(":(glob)*.rs"));
        assert_eq!(
            precompose_argv_pathspec(":(glob)*.rs").as_ref(),
            ":(glob)*.rs"
        );
    }
}

#[cfg(test)]
mod resolved_pathspecs_tests {
    use super::*;
    use crate::repo::{InitOptions, Repository};

    fn repo_with_tree() -> (tempfile::TempDir, Repository) {
        let dir = tempfile::tempdir().expect("tempdir");
        let repo = Repository::init(dir.path(), &InitOptions::default()).expect("init");
        std::fs::write(dir.path().join("top.txt"), b"a").expect("write");
        std::fs::create_dir_all(dir.path().join("sub")).expect("mkdir");
        std::fs::write(dir.path().join("sub/inner.txt"), b"b").expect("write");
        (dir, repo)
    }

    #[test]
    fn resolved_pathspecs_subdir_prefix() {
        let (dir, repo) = repo_with_tree();
        let sub = dir.path().join("sub");
        let resolved =
            ResolvedPathspecs::resolve(&repo, &sub, &["inner.txt".to_string()], None).expect("ok");
        assert_eq!(resolved.specs, ["sub/inner.txt"]);
        assert!(resolved.scoped);
    }

    #[test]
    fn resolved_pathspecs_top_level_magic() {
        let (dir, repo) = repo_with_tree();
        let resolved =
            ResolvedPathspecs::resolve(&repo, dir.path(), &[":/top.txt".to_string()], None)
                .expect("ok");
        assert_eq!(resolved.specs, ["top.txt"]);
    }

    #[test]
    fn resolved_pathspecs_parent_outside_repo() {
        let (dir, repo) = repo_with_tree();
        let outside = dir.path().parent().expect("parent").join("outside.txt");
        std::fs::write(&outside, b"x").expect("write outside");
        let spec = outside.to_string_lossy().into_owned();
        let err =
            ResolvedPathspecs::resolve(&repo, dir.path(), &[spec], None).expect_err("outside");
        assert!(matches!(
            err,
            PathspecResolveError::OutsideRepository { .. }
        ));
    }

    #[test]
    fn resolved_pathspecs_empty_argv() {
        let (dir, repo) = repo_with_tree();
        let err = ResolvedPathspecs::resolve(&repo, dir.path(), &[String::new()], None)
            .expect_err("empty");
        assert_eq!(err, PathspecResolveError::EmptyPathspec);
    }

    #[test]
    fn resolved_pathspecs_from_file_lf_and_quoted() {
        let (dir, repo) = repo_with_tree();
        let input = PathspecInput::new(b"top.txt\n\"sub/inner.txt\"\n".to_vec(), false);
        let resolved = ResolvedPathspecs::resolve(&repo, dir.path(), &[], Some(input)).expect("ok");
        assert_eq!(resolved.specs, ["top.txt", "sub/inner.txt"]);
    }

    #[test]
    fn resolved_pathspecs_from_file_nul() {
        let (dir, repo) = repo_with_tree();
        let input = PathspecInput::new(b"top.txt\0sub/\0".to_vec(), true);
        let resolved = ResolvedPathspecs::resolve(&repo, dir.path(), &[], Some(input)).expect("ok");
        assert_eq!(resolved.specs, ["top.txt", "sub/"]);
    }

    #[test]
    fn resolved_pathspecs_from_file_nul_preserves_literal_bytes() {
        let (dir, repo) = repo_with_tree();
        std::fs::write(dir.path().join("  spaced  "), b"x").expect("write");
        let quoted_dir = dir.path().join("\"double\"");
        std::fs::create_dir_all(&quoted_dir).expect("mkdir");
        std::fs::write(quoted_dir.join("f"), b"y").expect("write");
        let input = PathspecInput::new(b"  spaced  \0\"double\"\0".to_vec(), true);
        let resolved = ResolvedPathspecs::resolve(&repo, dir.path(), &[], Some(input)).expect("ok");
        assert_eq!(resolved.sources, ["  spaced  ", "\"double\""]);
    }

    #[test]
    fn resolved_pathspecs_from_file_blank_line_errors() {
        let (dir, repo) = repo_with_tree();
        let input = PathspecInput::new(b"\n".to_vec(), false);
        let err = ResolvedPathspecs::resolve(&repo, dir.path(), &[], Some(input)).expect_err("err");
        assert_eq!(err, PathspecResolveError::EmptyPathspec);
    }

    #[test]
    fn resolved_pathspecs_from_file_nul_empty_record_errors() {
        let (dir, repo) = repo_with_tree();
        let input = PathspecInput::new(b"a\0\0b".to_vec(), true);
        let err = ResolvedPathspecs::resolve(&repo, dir.path(), &[], Some(input)).expect_err("err");
        assert_eq!(err, PathspecResolveError::EmptyPathspec);
    }

    #[test]
    fn resolved_pathspecs_from_file_invalid_quote() {
        let (dir, repo) = repo_with_tree();
        let input = PathspecInput::new(b"\"bad\\z\"\n".to_vec(), false);
        let err = ResolvedPathspecs::resolve(&repo, dir.path(), &[], Some(input)).expect_err("err");
        assert!(matches!(
            err,
            PathspecResolveError::FromFile(
                PathspecFromFileParseError::InvalidEscapeSequence { .. }
            )
        ));
        let input = PathspecInput::new(b"\"unterminated\n".to_vec(), false);
        let err = ResolvedPathspecs::resolve(&repo, dir.path(), &[], Some(input)).expect_err("err");
        assert!(matches!(
            err,
            PathspecResolveError::FromFile(PathspecFromFileParseError::InvalidCStyleQuoting { .. })
        ));
    }

    #[test]
    fn resolved_pathspecs_entire_tree_when_unscoped() {
        let (dir, repo) = repo_with_tree();
        let resolved = ResolvedPathspecs::resolve(&repo, dir.path(), &[], None).expect("ok");
        assert!(resolved.is_entire_tree());
        assert!(resolved.specs.is_empty());
    }
}