Merge branch 'worktree-stack'

Cargo.toml

@@ -173,9 +173,11 @@ members = [
     "gix-tempfile",
     "gix-lock",
     "gix-attributes",
+    "gix-ignore",
     "gix-pathspec",
     "gix-refspec",
     "gix-path",
+    "gix-utils",
     "gix",
     "gitoxide-core",
     "gix-hashtable",

README.md

@@ -76,6 +76,7 @@ is usable to some extent.
   * [gix-discover](https://github.com/Byron/gitoxide/blob/main/crate-status.md#gix-discover)
   * [gix-path](https://github.com/Byron/gitoxide/blob/main/crate-status.md#gix-path)
   * [gix-attributes](https://github.com/Byron/gitoxide/blob/main/crate-status.md#gix-attributes)
+  * [gix-ignore](https://github.com/Byron/gitoxide/blob/main/crate-status.md#gix-ignore)
   * [gix-pathspec](https://github.com/Byron/gitoxide/blob/main/crate-status.md#gix-pathspec)
   * [gix-index](https://github.com/Byron/gitoxide/blob/main/crate-status.md#gix-index)
   * [gix-revision](https://github.com/Byron/gitoxide/blob/main/crate-status.md#gix-revision)
@@ -84,6 +85,7 @@ is usable to some extent.
   * [gix-refspec](https://github.com/Byron/gitoxide/blob/main/crate-status.md#gix-refspec)
   * `gitoxide-core`
 * **very early**  _(possibly without any documentation and many rough edges)_
+  * [gix-utils](https://github.com/Byron/gitoxide/blob/main/crate-status.md#gix-utils)
   * [gix-worktree](https://github.com/Byron/gitoxide/blob/main/crate-status.md#gix-worktree)
   * [gix-bitmap](https://github.com/Byron/gitoxide/blob/main/crate-status.md#gix-bitmap)
   * [gix-date](https://github.com/Byron/gitoxide/blob/main/crate-status.md#gix-date)

crate-status.md

@@ -101,10 +101,15 @@ and itself relies on all `git-*` crates. It's not meant for consumption, for app
 * [x] write the table of contents
 
 ### gix-hashtable
-
 * [x] hashmap
 * [x] hashset
 
+### gix-utils
+
+* **filesystem**
+   * [x] probe capabilities
+   * [x] symlink creation and removal
+   * [x] file snapshots
 
 ### gix-object
 * *decode (zero-copy)* borrowed objects
@@ -323,11 +328,13 @@ Check out the [performance discussion][gix-traverse-performance] as well.
     * [ ] Some examples
 
 ### gix-attributes
-* [x] parse git-ignore files (aka gix-attributes without the attributes or negation)
-* [x] parse gix-attributes files
-* [ ] create an attributes stack, ideally one that includes 'ignored' status from .gitignore files.
-   * [ ] support for built-in `binary` macro for `-text -diff -merge`
-
+* [x] parse `.gitattribute` files
+* [ ] an attributes stack for matching paths to their attributes, with support for built-in `binary` macro for `-text -diff -merge`
+
+### gix-ignore
+* [x] parse `.gitignore` files 
+* [x] an attributes stack for checking if paths are excluded
+
 ### gix-quote
 * **ansi-c**
   * [x] quote
@@ -440,7 +447,7 @@ Make it the best-performing implementation and the most convenient one.
      - [ ] handle submodules
      - [ ] handle sparse directories
      - [ ] handle sparse index
-     - [ ] linear scaling with multi-threading up to IO saturation
+     - [x] linear scaling with multi-threading up to IO saturation
   - supported attributes to affect working tree and index contents
      - [ ] eol
      - [ ] working-tree-encoding
@@ -450,8 +457,10 @@ Make it the best-performing implementation and the most convenient one.
      - [ ] `ident`
      - [ ] filter processes
      - [ ] single-invocation clean/smudge filters
-* [x] access to all .gitignore/exclude information 
-* [ ] access to all attributes information
+* manage multiple worktrees
+* access to per-path information, like `.gitignore` and `.gitattributes` in a manner well suited for efficient lookups
+  * [x] _exclude_ information
+  * [ ] attributes
 
 ### gix-revision
 * [x] `describe()` (similar to `git name-rev`)
@@ -602,6 +611,8 @@ See its [README.md](https://github.com/Byron/gitoxide/blob/main/gix-lock/README.
         * [x] tree with other tree
            * [ ] respect case-sensitivity of host filesystem.
            * [x] a way to access various diff related settings or use them
+           * [ ] respect `diff.*.textconv`, `diff.*.cachetextconv` and external diff viewers with `diff.*.command`, 
+                 [along with support for reading `diff` gitattributes](https://github.com/git/git/blob/73876f4861cd3d187a4682290ab75c9dccadbc56/Documentation/gitattributes.txt#L699:L699).
            * **rewrite tracking**
               * **deviation** - git keeps up to four candidates whereas we use the first-found candidate that matches the similarity percentage.
                                 This can lead to different sources being found. As such, we also don't consider the filename at all.
@@ -614,7 +625,7 @@ See its [README.md](https://github.com/Byron/gitoxide/blob/main/gix-lock/README.
                  * [x] renames
                  * [x] copies
               * [x] 'find-copies-harder' - find copies with the source being the entire tree.
-        * [ ] tree with working tree
+        * [ ] tree or index with working tree
         * [x] diffs between modified blobs with various algorithms
         * [ ] tree with index
     * [x] initialize
@@ -673,6 +684,8 @@ See its [README.md](https://github.com/Byron/gitoxide/blob/main/gix-lock/README.
           * [ ] obtain 'prunable' information
        * [x] proper handling of worktree related refs
        * [ ] create, move, remove, and repair
+       * [x] access exclude information
+       * [ ]  access attribute information
        * [x] respect `core.worktree` configuration
           - **deviation**
              * The delicate interplay between `GIT_COMMON_DIR` and `GIT_WORK_TREE` isn't implemented.

gitoxide-core/src/index/checkout.rs

@@ -56,7 +56,7 @@ pub fn checkout_exclusive(
     }
 
     let opts = gix::worktree::index::checkout::Options {
-        fs: gix::worktree::fs::Capabilities::probe(dest_directory),
+        fs: gix::utils::FilesystemCapabilities::probe(dest_directory),
 
         destination_is_initially_empty: true,
         overwrite_existing: false,

gitoxide-core/src/repository/exclude.rs

@@ -35,10 +35,7 @@ pub fn query(
         .worktree()
         .with_context(|| "Cannot check excludes without a current worktree")?;
     let index = worktree.index()?;
-    let mut cache = worktree.excludes(
-        &index,
-        Some(gix::attrs::MatchGroup::<gix::attrs::Ignore>::from_overrides(overrides)),
-    )?;
+    let mut cache = worktree.excludes(&index, Some(gix::ignore::Search::from_overrides(overrides)))?;
 
     let prefix = repo.prefix().expect("worktree - we have an index by now")?;
 

gix-attributes/Cargo.toml

@@ -14,23 +14,26 @@ doctest = false
 
 [features]
 ## Data structures implement `serde::Serialize` and `serde::Deserialize`.
-serde1 = ["serde", "bstr/serde", "gix-glob/serde1"]
+serde1 = ["serde", "bstr/serde", "gix-glob/serde1", "kstring/serde"]
 
 [dependencies]
-gix-features = { version = "^0.28.0", path = "../gix-features" }
-gix-path = { version = "^0.7.2", path = "../gix-path" }
+gix-path = { version = "^0.7.3", path = "../gix-path" }
 gix-quote = { version = "^0.4.3", path = "../gix-quote" }
 gix-glob = { version = "^0.5.5", path = "../gix-glob" }
 
 bstr = { version = "1.3.0", default-features = false, features = ["std", "unicode"]}
+smallvec = "1.10.0"
+kstring = "2.0.0"
 unicode-bom = "2.0.2"
 thiserror = "1.0.26"
 serde = { version = "1.0.114", optional = true, default-features = false, features = ["derive"]}
+log = "0.4.17"
 
 document-features = { version = "0.2.1", optional = true }
 
 [dev-dependencies]
 gix-testtools = { path = "../tests/tools"}
+gix-utils = { path = "../gix-utils" }
 
 [package.metadata.docs.rs]
 all-features = true

gix-attributes/src/lib.rs

@@ -1,30 +1,31 @@
-//! Parse `.gitattribute` and `.gitignore` files and provide utilities to match against them.
+//! Parse `.gitattribute` files and provide utilities to match against them.
 //!
 //! ## Feature Flags
 #![cfg_attr(
     feature = "document-features",
     cfg_attr(doc, doc = ::document_features::document_features!())
 )]
 #![cfg_attr(docsrs, feature(doc_cfg, doc_auto_cfg))]
-#![deny(missing_docs, rust_2018_idioms)]
-#![forbid(unsafe_code)]
+#![deny(missing_docs, rust_2018_idioms, unsafe_code)]
 
-use std::path::PathBuf;
-
-use bstr::{BStr, BString};
 pub use gix_glob as glob;
+use kstring::{KString, KStringRef};
 
 mod assignment;
 ///
 pub mod name;
-mod state;
+///
+pub mod state;
 
-mod match_group;
-pub use match_group::{Attributes, Ignore, Match, Pattern};
+///
+pub mod search;
 
 ///
 pub mod parse;
-/// Parse attribute assignments line by line from `bytes`.
+
+/// Parse attribute assignments line by line from `bytes`, and fail the operation on error.
+///
+/// For leniency, ignore errors using `filter_map(Result::ok)` for example.
 pub fn parse(bytes: &[u8]) -> parse::Lines<'_> {
     parse::Lines::new(bytes)
 }
@@ -42,7 +43,7 @@ pub enum StateRef<'a> {
     /// The attribute is set to the given value, which followed the `=` sign.
     /// Note that values can be empty.
     #[cfg_attr(feature = "serde1", serde(borrow))]
-    Value(&'a BStr),
+    Value(state::ValueRef<'a>),
     /// The attribute isn't mentioned with a given path or is explicitly set to `Unspecified` using the `!` sign.
     Unspecified,
 }
@@ -59,19 +60,19 @@ pub enum State {
     Unset,
     /// The attribute is set to the given value, which followed the `=` sign.
     /// Note that values can be empty.
-    Value(BString), // TODO(performance): Is there a non-utf8 compact_str/KBString crate? See https://github.com/cobalt-org/kstring/issues/37#issuecomment-1446777265 .
+    Value(state::Value),
     /// The attribute isn't mentioned with a given path or is explicitly set to `Unspecified` using the `!` sign.
     Unspecified,
 }
 
 /// Represents a validated attribute name
 #[derive(PartialEq, Eq, Debug, Hash, Ord, PartialOrd, Clone)]
 #[cfg_attr(feature = "serde1", derive(serde::Serialize, serde::Deserialize))]
-pub struct Name(pub(crate) String); // TODO(performance): See if `KBString` or `compact_string` could be meaningful here.
+pub struct Name(pub(crate) KString);
 
 /// Holds a validated attribute name as a reference
-#[derive(PartialEq, Eq, Debug, Hash, Ord, PartialOrd)]
-pub struct NameRef<'a>(&'a str);
+#[derive(Copy, Clone, PartialEq, Eq, Debug, Hash, Ord, PartialOrd)]
+pub struct NameRef<'a>(KStringRef<'a>);
 
 /// Name an attribute and describe it's assigned state.
 #[derive(PartialEq, Eq, Debug, Hash, Ord, PartialOrd, Clone)]
@@ -84,54 +85,42 @@ pub struct Assignment {
 }
 
 /// Holds validated attribute data as a reference
-#[derive(PartialEq, Eq, Debug, Hash, Ord, PartialOrd)]
+#[derive(Copy, Clone, PartialEq, Eq, Debug, Hash, Ord, PartialOrd)]
 pub struct AssignmentRef<'a> {
     /// The name of the attribute.
     pub name: NameRef<'a>,
     /// The state of the attribute.
     pub state: StateRef<'a>,
 }
 
-/// A grouping of lists of patterns while possibly keeping associated to their base path.
+/// A grouping of lists of patterns while possibly keeping associated to their base path in order to find matches.
 ///
 /// Pattern lists with base path are queryable relative to that base, otherwise they are relative to the repository root.
 #[derive(PartialEq, Eq, Debug, Hash, Ord, PartialOrd, Clone, Default)]
-pub struct MatchGroup<T: Pattern = Attributes> {
+pub struct Search {
     /// A list of pattern lists, each representing a patterns from a file or specified by hand, in the order they were
     /// specified in.
     ///
-    /// During matching, this order is reversed.
-    pub patterns: Vec<PatternList<T>>,
+    /// When matching, this order is reversed.
+    patterns: Vec<gix_glob::search::pattern::List<search::Attributes>>,
 }
 
-/// A list of patterns which optionally know where they were loaded from and what their base is.
+/// A list of known global sources for git attribute files in order of ascending precedence.
 ///
-/// Knowing their base which is relative to a source directory, it will ignore all path to match against
-/// that don't also start with said base.
-#[derive(PartialEq, Eq, Debug, Hash, Ord, PartialOrd, Clone, Default)]
-pub struct PatternList<T: Pattern> {
-    /// Patterns and their associated data in the order they were loaded in or specified,
-    /// the line number in its source file or its sequence number (_`(pattern, value, line_number)`_).
+/// This means that values from the first variant will be returned first.
+#[derive(Clone, Copy, Debug, Eq, PartialEq, Hash, Ord, PartialOrd)]
+pub enum Source {
+    /// The attribute file that the installation itself ships with.
+    GitInstallation,
+    /// System-wide attributes file. This is typically defined as
+    /// `$(prefix)/etc/gitattributes` (where prefix is the git-installation directory).
+    System,
+    /// This is `<xdg-config-home>/git/attributes` and is git application configuration per user.
     ///
-    /// During matching, this order is reversed.
-    pub patterns: Vec<PatternMapping<T::Value>>,
-
-    /// The path from which the patterns were read, or `None` if the patterns
-    /// don't originate in a file on disk.
-    pub source: Option<PathBuf>,
-
-    /// The parent directory of source, or `None` if the patterns are _global_ to match against the repository root.
-    /// It's processed to contain slashes only and to end with a trailing slash, and is relative to the repository root.
-    pub base: Option<BString>,
+    /// Note that there is no `~/.gitattributes` file.
+    Git,
+    /// The configuration of the repository itself, located in `$GIT_DIR/info/attributes`.
+    Local,
 }
 
-/// An association of a pattern with its value, along with a sequence number providing a sort order in relation to its peers.
-#[derive(PartialEq, Eq, Debug, Hash, Ord, PartialOrd, Clone)]
-pub struct PatternMapping<T> {
-    /// The pattern itself, like `/target/*`
-    pub pattern: gix_glob::Pattern,
-    /// The value associated with the pattern.
-    pub value: T,
-    /// Typically the line number in the file the pattern was parsed from.
-    pub sequence_number: usize,
-}
+mod source;