feat(source): wrap PathBuf with PathEntry

This gives us more room to store file metadata.
For example, knowing a source file is a symlink
and resolving it when packaging.

Author: weihanglo

src/cargo/ops/cargo_package/mod.rs

@@ -16,6 +16,7 @@ use crate::core::Workspace;
 use crate::core::{Package, PackageId, PackageSet, Resolve, SourceId};
 use crate::ops::lockfile::LOCKFILE_NAME;
 use crate::ops::registry::{infer_registry, RegistryOrIndex};
+use crate::sources::path::PathEntry;
 use crate::sources::registry::index::{IndexPackage, RegistryDependency};
 use crate::sources::{PathSource, CRATES_IO_REGISTRY};
 use crate::util::cache_lock::CacheLockMode;
@@ -396,7 +397,7 @@ fn prepare_archive(
 fn build_ar_list(
     ws: &Workspace<'_>,
     pkg: &Package,
-    src_files: Vec<PathBuf>,
+    src_files: Vec<PathEntry>,
     vcs_info: Option<vcs::VcsInfo>,
 ) -> CargoResult<Vec<ArchiveFile>> {
     let mut result = HashMap::new();
@@ -420,7 +421,7 @@ fn build_ar_list(
                     .push(ArchiveFile {
                         rel_path: rel_path.to_owned(),
                         rel_str: rel_str.to_owned(),
-                        contents: FileContents::OnDisk(src_file.clone()),
+                        contents: FileContents::OnDisk(src_file.to_path_buf()),
                     });
             }
         }

src/cargo/ops/cargo_package/vcs.rs

@@ -9,6 +9,7 @@ use serde::Serialize;
 use tracing::debug;
 
 use crate::core::Package;
+use crate::sources::PathEntry;
 use crate::CargoResult;
 use crate::GlobalContext;
 
@@ -41,7 +42,7 @@ pub struct GitVcsInfo {
 #[tracing::instrument(skip_all)]
 pub fn check_repo_state(
     p: &Package,
-    src_files: &[PathBuf],
+    src_files: &[PathEntry],
     gctx: &GlobalContext,
     opts: &PackageOpts<'_>,
 ) -> CargoResult<Option<VcsInfo>> {
@@ -114,7 +115,7 @@ pub fn check_repo_state(
 fn git(
     pkg: &Package,
     gctx: &GlobalContext,
-    src_files: &[PathBuf],
+    src_files: &[PathEntry],
     repo: &git2::Repository,
     opts: &PackageOpts<'_>,
 ) -> CargoResult<Option<GitVcsInfo>> {
@@ -136,6 +137,7 @@ fn git(
     let mut dirty_src_files: Vec<_> = src_files
         .iter()
         .filter(|src_file| dirty_files.iter().any(|path| src_file.starts_with(path)))
+        .map(|p| p.as_path_buf())
         .chain(dirty_metadata_paths(pkg, repo)?.iter())
         .map(|path| {
             pathdiff::diff_paths(path, cwd)

src/cargo/ops/vendor.rs

@@ -2,6 +2,7 @@ use crate::core::shell::Verbosity;
 use crate::core::{GitReference, Package, Workspace};
 use crate::ops;
 use crate::sources::path::PathSource;
+use crate::sources::PathEntry;
 use crate::sources::CRATES_IO_REGISTRY;
 use crate::util::cache_lock::CacheLockMode;
 use crate::util::{try_canonicalize, CargoResult, GlobalContext};
@@ -315,13 +316,14 @@ fn sync(
 fn cp_sources(
     pkg: &Package,
     src: &Path,
-    paths: &[PathBuf],
+    paths: &[PathEntry],
     dst: &Path,
     cksums: &mut BTreeMap<String, String>,
     tmp_buf: &mut [u8],
     gctx: &GlobalContext,
 ) -> CargoResult<()> {
     for p in paths {
+        let p = p.as_path();
         let relative = p.strip_prefix(&src).unwrap();
 
         match relative.to_str() {

src/cargo/sources/mod.rs

@@ -29,6 +29,7 @@
 pub use self::config::SourceConfigMap;
 pub use self::directory::DirectorySource;
 pub use self::git::GitSource;
+pub use self::path::PathEntry;
 pub use self::path::PathSource;
 pub use self::path::RecursivePathSource;
 pub use self::registry::{

src/cargo/sources/path.rs

@@ -94,7 +94,7 @@ impl<'gctx> PathSource<'gctx> {
     /// use other methods like `.gitignore`, `package.include`, or
     /// `package.exclude` to filter the list of files.
     #[tracing::instrument(skip_all)]
-    pub fn list_files(&self, pkg: &Package) -> CargoResult<Vec<PathBuf>> {
+    pub fn list_files(&self, pkg: &Package) -> CargoResult<Vec<PathEntry>> {
         list_files(pkg, self.gctx)
     }
 
@@ -278,7 +278,7 @@ impl<'gctx> RecursivePathSource<'gctx> {
     /// are relevant for building this package, but it also contains logic to
     /// use other methods like `.gitignore`, `package.include`, or
     /// `package.exclude` to filter the list of files.
-    pub fn list_files(&self, pkg: &Package) -> CargoResult<Vec<PathBuf>> {
+    pub fn list_files(&self, pkg: &Package) -> CargoResult<Vec<PathEntry>> {
         list_files(pkg, self.gctx)
     }
 
@@ -404,6 +404,84 @@ impl<'gctx> Source for RecursivePathSource<'gctx> {
     }
 }
 
+/// Type that abstracts over [`gix::dir::entry::Kind`] and [`fs::FileType`].
+#[derive(Debug, Clone, Copy)]
+enum FileType {
+    File,
+    Dir,
+    Symlink,
+    Other,
+}
+
+impl From<fs::FileType> for FileType {
+    fn from(value: fs::FileType) -> Self {
+        if value.is_file() {
+            FileType::File
+        } else if value.is_dir() {
+            FileType::Dir
+        } else if value.is_symlink() {
+            FileType::Symlink
+        } else {
+            FileType::Other
+        }
+    }
+}
+
+impl From<gix::dir::entry::Kind> for FileType {
+    fn from(value: gix::dir::entry::Kind) -> Self {
+        use gix::dir::entry::Kind;
+        match value {
+            Kind::Untrackable => FileType::Other,
+            Kind::File => FileType::File,
+            Kind::Symlink => FileType::Symlink,
+            Kind::Directory | Kind::Repository => FileType::Dir,
+        }
+    }
+}
+
+/// [`PathBuf`] with extra metadata.
+#[derive(Clone)]
+pub struct PathEntry {
+    path: PathBuf,
+    ty: FileType,
+}
+
+impl PathEntry {
+    pub fn into_path_buf(self) -> PathBuf {
+        self.path
+    }
+
+    pub fn as_path_buf(&self) -> &PathBuf {
+        &self.path
+    }
+
+    pub fn is_file(&self) -> bool {
+        matches!(self.ty, FileType::File)
+    }
+
+    pub fn is_dir(&self) -> bool {
+        matches!(self.ty, FileType::Dir)
+    }
+
+    pub fn is_symlink(&self) -> bool {
+        matches!(self.ty, FileType::Symlink)
+    }
+}
+
+impl fmt::Debug for PathEntry {
+    fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
+        self.path.fmt(f)
+    }
+}
+
+impl std::ops::Deref for PathEntry {
+    type Target = PathBuf;
+
+    fn deref(&self) -> &Self::Target {
+        &self.path
+    }
+}
+
 fn first_package<'p>(
     pkg_id: PackageId,
     pkgs: &'p Vec<Package>,
@@ -446,7 +524,7 @@ fn first_package<'p>(
 /// are relevant for building this package, but it also contains logic to
 /// use other methods like `.gitignore`, `package.include`, or
 /// `package.exclude` to filter the list of files.
-pub fn list_files(pkg: &Package, gctx: &GlobalContext) -> CargoResult<Vec<PathBuf>> {
+pub fn list_files(pkg: &Package, gctx: &GlobalContext) -> CargoResult<Vec<PathEntry>> {
     _list_files(pkg, gctx).with_context(|| {
         format!(
             "failed to determine list of files in {}",
@@ -456,7 +534,7 @@ pub fn list_files(pkg: &Package, gctx: &GlobalContext) -> CargoResult<Vec<PathBu
 }
 
 /// See [`PathSource::list_files`].
-fn _list_files(pkg: &Package, gctx: &GlobalContext) -> CargoResult<Vec<PathBuf>> {
+fn _list_files(pkg: &Package, gctx: &GlobalContext) -> CargoResult<Vec<PathEntry>> {
     let root = pkg.root();
     let no_include_option = pkg.manifest().include().is_empty();
     let git_repo = if no_include_option {
@@ -580,7 +658,7 @@ fn list_files_gix(
     repo: &gix::Repository,
     filter: &dyn Fn(&Path, bool) -> bool,
     gctx: &GlobalContext,
-) -> CargoResult<Vec<PathBuf>> {
+) -> CargoResult<Vec<PathEntry>> {
     debug!("list_files_gix {}", pkg.package_id());
     let options = repo
         .dirwalk_options()?
@@ -619,7 +697,7 @@ fn list_files_gix(
         vec![include, exclude]
     };
 
-    let mut files = Vec::<PathBuf>::new();
+    let mut files = Vec::<PathEntry>::new();
     let mut subpackages_found = Vec::new();
     for item in repo
         .dirwalk_iter(index.clone(), pathspec, Default::default(), options)?
@@ -701,7 +779,10 @@ fn list_files_gix(
         } else if (filter)(&file_path, is_dir) {
             assert!(!is_dir);
             trace!("  found {}", file_path.display());
-            files.push(file_path);
+            files.push(PathEntry {
+                path: file_path,
+                ty: kind.map(Into::into).unwrap_or(FileType::Other),
+            });
         }
     }
 
@@ -715,7 +796,7 @@ fn list_files_gix(
 /// is not tracked under a Git repository.
 fn list_files_walk(
     path: &Path,
-    ret: &mut Vec<PathBuf>,
+    ret: &mut Vec<PathEntry>,
     is_root: bool,
     filter: &dyn Fn(&Path, bool) -> bool,
     gctx: &GlobalContext,
@@ -756,7 +837,14 @@ fn list_files_walk(
             Ok(entry) => {
                 let file_type = entry.file_type();
                 if file_type.is_file() || file_type.is_symlink() {
-                    ret.push(entry.into_path());
+                    // We follow_links(true) here so check if entry was created from a symlink
+                    let ty = if entry.path_is_symlink() {
+                        FileType::Symlink
+                    } else {
+                        entry.file_type().into()
+                    };
+                    let path = entry.into_path();
+                    ret.push(PathEntry { path, ty });
                 }
             }
             Err(err) if err.loop_ancestor().is_some() => {
@@ -770,7 +858,10 @@ fn list_files_walk(
                 // Otherwise, simply recover from it.
                 // Don't worry about error skipping here, the callers would
                 // still hit the IO error if they do access it thereafter.
-                Some(path) => ret.push(path.to_path_buf()),
+                Some(path) => ret.push(PathEntry {
+                    path: path.to_path_buf(),
+                    ty: FileType::Other,
+                }),
                 None => return Err(err.into()),
             },
         }
@@ -801,7 +892,7 @@ fn last_modified_file(
         let mtime = paths::mtime(&file).unwrap_or_else(|_| FileTime::zero());
         if mtime > max {
             max = mtime;
-            max_path = file;
+            max_path = file.into_path_buf();
         }
     }
     trace!("last modified file {}: {}", path.display(), max);