add util::GlobMatcherExt

Author: aawsome

crates/core/Cargo.toml

@@ -100,6 +100,7 @@ enum-map = { workspace = true }
 enum-map-derive = "0.17.0"
 enumset = { version = "1.1.5", features = ["serde"] }
 gethostname = "0.5.0"
+globset = "0.4.15"
 humantime = "2.1.0"
 itertools = "0.13.0"
 quick_cache = "0.6.9"
@@ -124,7 +125,6 @@ xattr = "1"
 anyhow = { workspace = true }
 expect-test = "1.5.0"
 flate2 = "1.0.35"
-globset = "0.4.15"
 insta = { version = "1.41.1", features = ["redactions", "ron"] }
 mockall = "0.13"
 pretty_assertions = "1.4.1"

crates/core/src/lib.rs

@@ -115,6 +115,7 @@ pub(crate) mod progress;
 /// Structs which are saved in JSON or binary format in the repository
 pub mod repofile;
 pub(crate) mod repository;
+pub mod util;
 /// Virtual File System support - allows to act on the repository like on a file system
 pub mod vfs;
 

crates/core/src/util.rs

@@ -0,0 +1,29 @@
+use globset::GlobMatcher;
+
+/// Extend `globset::GlobMatcher` to allow mathing on unix paths (on every platform)
+pub trait GlobMatcherExt {
+    /// Match on unix paths, i.e. paths which are available as `&[u8]`
+    fn is_unix_match(&self, path: impl AsRef<[u8]>) -> bool;
+}
+
+impl GlobMatcherExt for GlobMatcher {
+    // This is a hacky implementation, espeically for windows where we convert lossily
+    // into an utf8 string and match on the windows path given by that string.
+    // Note: `GlobMatcher` internally converts into a `&[u8]` to perform the matching
+    // TODO: Use https://github.com/BurntSushi/ripgrep/pull/2955 once it is available.
+    #[cfg(not(windows))]
+    fn is_unix_match(&self, path: impl AsRef<[u8]>) -> bool {
+        use std::{ffi::OsStr, os::unix::ffi::OsStrExt, path::PathBuf};
+
+        let path = PathBuf::from(OsStr::from_bytes(path.as_ref()));
+        self.is_match(&path)
+    }
+    #[cfg(windows)]
+    fn is_unix_match(&self, path: impl AsRef<[u8]>) -> bool {
+        use std::{ffi::OsStr, path::Path};
+
+        let string: &str = &String::from_utf8_lossy(path.as_bytes());
+        let path = Path::new(OsStr::new(string));
+        self.is_match(path)
+    }
+}

crates/core/tests/integration/find.rs

@@ -1,10 +1,7 @@
 // don't warn about try_from paths conversion on unix
 #![allow(clippy::unnecessary_fallible_conversions)]
 
-use std::{
-    path::{Path, PathBuf},
-    str::FromStr,
-};
+use std::{path::PathBuf, str::FromStr};
 
 use anyhow::Result;
 use globset::Glob;
@@ -13,6 +10,7 @@ use rstest::rstest;
 
 use rustic_core::{
     repofile::{Node, SnapshotFile},
+    util::GlobMatcherExt,
     BackupOptions, FindMatches, FindNode,
 };
 use typed_path::UnixPath;
@@ -39,9 +37,8 @@ fn test_find(tar_gz_testdata: Result<TestSource>, set_up_repo: Result<RepoOpen>)
     assert_with_win("find-nodes-not-found", not_found);
     // test non-existing match
     let glob = Glob::new("not_existing")?.compile_matcher();
-    let not_found = repo.find_matching_nodes(vec![snapshot.tree], &|path, _| {
-        glob.is_match(PathBuf::try_from(path).unwrap())
-    })?;
+    let not_found =
+        repo.find_matching_nodes(vec![snapshot.tree], &|path, _| glob.is_unix_match(path))?;
     assert_debug_snapshot!("find-matching-nodes-not-found", not_found);
 
     // test existing path
@@ -51,21 +48,15 @@ fn test_find(tar_gz_testdata: Result<TestSource>, set_up_repo: Result<RepoOpen>)
     // test existing match
     let glob = Glob::new("testfile")?.compile_matcher();
     let match_func = |path: &UnixPath, _: &Node| {
-        glob.is_match(PathBuf::try_from(path).unwrap())
-            || path
-                .file_name()
-                .is_some_and(|f| glob.is_match(Path::new(&String::from_utf8(f.to_vec()).unwrap())))
+        glob.is_unix_match(path) || path.file_name().is_some_and(|f| glob.is_unix_match(f))
     };
     let FindMatches { paths, matches, .. } =
         repo.find_matching_nodes(vec![snapshot.tree], &match_func)?;
     assert_debug_snapshot!("find-matching-existing", (paths, matches));
     // test existing match
     let glob = Glob::new("testfile*")?.compile_matcher();
     let match_func = |path: &UnixPath, _: &Node| {
-        glob.is_match(PathBuf::try_from(path).unwrap())
-            || path
-                .file_name()
-                .is_some_and(|f| glob.is_match(Path::new(&String::from_utf8(f.to_vec()).unwrap())))
+        glob.is_unix_match(path) || path.file_name().is_some_and(|f| glob.is_unix_match(f))
     };
     let FindMatches { paths, matches, .. } =
         repo.find_matching_nodes(vec![snapshot.tree], &match_func)?;

examples/find/examples/find.rs

@@ -1,9 +1,9 @@
 //! `ls` example
 use globset::Glob;
 use rustic_backend::BackendOptions;
-use rustic_core::{FindMatches, Repository, RepositoryOptions};
+use rustic_core::{util::GlobMatcherExt, FindMatches, Repository, RepositoryOptions};
 use simplelog::{Config, LevelFilter, SimpleLogger};
-use std::{error::Error, path::PathBuf};
+use std::error::Error;
 
 // don't warn about try_from paths conversion on unix
 #[allow(clippy::unnecessary_fallible_conversions)]
@@ -32,9 +32,7 @@ fn main() -> Result<(), Box<dyn Error>> {
         paths,
         nodes,
         matches,
-    } = repo.find_matching_nodes(tree_ids, &|path, _| {
-        glob.is_match(PathBuf::try_from(path).unwrap())
-    })?;
+    } = repo.find_matching_nodes(tree_ids, &|path, _| glob.is_unix_match(path))?;
     for (snap, matches) in snapshots.iter().zip(matches) {
         println!("results in {snap:?}");
         for (path_idx, node_idx) in matches {