fix(normalize_path): remove current dir components, ensure backslashes on windows, and ensure no trailing slash (#15)

Author: dsherret

Cargo.lock

@@ -26,7 +26,7 @@ divan = "0.1.21"
 members = ["."]
 
 [workspace.dependencies]
-sys_traits = "0.1.10"
+sys_traits = "0.1.17"
 
 [[bench]]
 name = "bench"

Cargo.toml

@@ -8,14 +8,18 @@ fn main() {
   divan::main();
 }
 
-#[divan::bench(sample_size = 512000)]
+#[divan::bench(sample_size = 51200)]
 fn bench_normalize_path_changed(bencher: divan::Bencher) {
   let path = PathBuf::from("/testing/../this/./out/testing/../test");
   bencher.bench(|| normalize_path(Cow::Borrowed(&path)))
 }
 
-#[divan::bench(sample_size = 512000)]
+#[divan::bench(sample_size = 51200)]
 fn bench_normalize_path_no_change(bencher: divan::Bencher) {
-  let path = PathBuf::from("/testing/this/out/testing/test");
+  let path = if cfg!(windows) {
+    PathBuf::from("C:\\testing\\this\\out\\testing\\test")
+  } else {
+    PathBuf::from("/testing/this/out/testing/test")
+  };
   bencher.bench(|| normalize_path(Cow::Borrowed(&path)))
 }

benches/bench.rs

@@ -146,16 +146,67 @@ fn url_to_file_path_wasm(url: &Url) -> Result<PathBuf, ()> {
 #[inline]
 pub fn normalize_path(path: Cow<Path>) -> Cow<Path> {
   fn should_normalize(path: &Path) -> bool {
+    if path_has_trailing_separator(path) {
+      return true;
+    }
+
     for component in path.components() {
       match component {
-        Component::Prefix(..) | Component::CurDir | Component::ParentDir => {
-          return true
+        Component::CurDir | Component::ParentDir => {
+          return true;
         }
-        Component::RootDir | Component::Normal(_) => {
+        Component::Prefix(..) | Component::RootDir | Component::Normal(_) => {
           // ok
         }
       }
     }
+
+    path_has_cur_dir_separator(path)
+  }
+
+  fn path_has_trailing_separator(path: &Path) -> bool {
+    #[cfg(unix)]
+    let raw = std::os::unix::ffi::OsStrExt::as_bytes(path.as_os_str());
+    #[cfg(windows)]
+    let raw = path.as_os_str().as_encoded_bytes();
+    #[cfg(target_arch = "wasm32")]
+    let raw = path.to_string_lossy();
+    #[cfg(target_arch = "wasm32")]
+    let raw = raw.as_bytes();
+
+    if sys_traits::impls::is_windows() {
+      raw.contains(&b'/') || raw.ends_with(b"\\")
+    } else {
+      raw.ends_with(b"/")
+    }
+  }
+
+  // Rust normalizes away `Component::CurDir` most of the time
+  // so we need to explicitly check for it in the bytes
+  fn path_has_cur_dir_separator(path: &Path) -> bool {
+    #[cfg(unix)]
+    let raw = std::os::unix::ffi::OsStrExt::as_bytes(path.as_os_str());
+    #[cfg(windows)]
+    let raw = path.as_os_str().as_encoded_bytes();
+    #[cfg(target_arch = "wasm32")]
+    let raw = path.to_string_lossy();
+    #[cfg(target_arch = "wasm32")]
+    let raw = raw.as_bytes();
+
+    if sys_traits::impls::is_windows() {
+      for window in raw.windows(3) {
+        if matches!(window, [b'\\', b'.', b'\\']) {
+          return true;
+        }
+      }
+    } else {
+      for window in raw.windows(3) {
+        if matches!(window, [b'/', b'.', b'/']) {
+          return true;
+        }
+      }
+    }
+
     false
   }
 
@@ -603,22 +654,39 @@ mod tests {
     }
   }
 
+  #[test]
+  fn test_normalize_path_basic() {
+    let run_test = run_normalize_path_test;
+    run_test("a/../b", "b");
+    run_test("a/./b/", &PathBuf::from("a").join("b").to_string_lossy());
+    run_test(
+      "a/./b/../c",
+      &PathBuf::from("a").join("c").to_string_lossy(),
+    );
+  }
+
   #[cfg(windows)]
   #[test]
-  fn test_normalize_path() {
-    use super::*;
+  fn test_normalize_path_win() {
+    let run_test = run_normalize_path_test;
 
     run_test("C:\\test\\file.txt", "C:\\test\\file.txt");
     run_test("C:\\test\\./file.txt", "C:\\test\\file.txt");
     run_test("C:\\test\\../other/file.txt", "C:\\other\\file.txt");
     run_test("C:\\test\\../other\\file.txt", "C:\\other\\file.txt");
+    run_test(
+      "C:\\test\\removes_trailing_slash\\",
+      "C:\\test\\removes_trailing_slash",
+    );
+    run_test("C:\\a\\.\\b\\..\\c", "C:\\a\\c");
+  }
 
-    fn run_test(input: &str, expected: &str) {
-      assert_eq!(
-        normalize_path(Cow::Owned(PathBuf::from(input))),
-        PathBuf::from(expected)
-      );
-    }
+  #[track_caller]
+  fn run_normalize_path_test(input: &str, expected: &str) {
+    assert_eq!(
+      normalize_path(Cow::Owned(PathBuf::from(input))).to_string_lossy(),
+      expected
+    );
   }
 
   #[test]