xtask: assets audit no longer panics on Unicode names and reads real paths from git (#424)

xtask/src/assets.rs

@@ -67,18 +67,24 @@ fn glob(pat: &str, s: &str) -> bool {
         None => pat == s,
         Some((head, tail)) => {
             let Some(rest) = s.strip_prefix(head) else { return false };
-            (0..=rest.len()).take_while(|&i| i == 0 || !rest[..i].contains('/')).any(|i| rest.is_char_boundary(i) && glob(tail, &rest[i..]))
+            let ends = rest.char_indices().map(|(i, _)| i).chain(std::iter::once(rest.len()));
+            ends.take_while(|&i| !rest[..i].contains('/')).any(|i| glob(tail, &rest[i..]))
         }
     }
 }
 
+/// Splits `git ls-files -z` output into real file names (no display quoting of Unicode or special characters).
+fn parse_ls_files(out: &[u8]) -> Vec<String> {
+    out.split(|&b| b == 0).filter(|n| !n.is_empty()).map(|n| String::from_utf8_lossy(n).into_owned()).collect()
+}
+
 pub fn run(root: &Path) -> Result<(), String> {
     let out = Command::new("git")
         .current_dir(root)
-        .args(["ls-files", "--cached", "--others", "--exclude-standard"])
+        .args(["ls-files", "--cached", "--others", "--exclude-standard", "-z"])
         .output()
         .map_err(|e| format!("git ls-files: {e}"))?;
-    let files: Vec<String> = String::from_utf8_lossy(&out.stdout).lines().filter(|l| root.join(l).exists()).map(str::to_owned).collect();
+    let files: Vec<String> = parse_ls_files(&out.stdout).into_iter().filter(|l| root.join(l).exists()).collect();
     let md = std::fs::read_to_string(root.join("ATTRIBUTION.md")).map_err(|e| format!("ATTRIBUTION.md: {e}"))?;
     let miss = missing(&files, &md);
     if miss.is_empty() {
@@ -104,4 +110,19 @@ mod tests {
         assert!(missing(&["assets/app-icon/hicolor/16x16/apps/x.png".into()], md).is_empty());
         assert_eq!(missing(&["assets/app-icon/hicolor/a/b/apps/x.png".into()], md).len(), 1);
     }
+
+    #[test]
+    fn glob_handles_multibyte_names() {
+        assert!(glob("assets/*/x.png", "assets/żółć/x.png"));
+        assert!(glob("assets/*.png", "assets/日本語.png"));
+        assert!(!glob("assets/*/x.png", "assets/ż/ó/x.png"));
+        assert!(missing(&["assets/żółć/x.png".into()], "`assets/*/x.png`").is_empty());
+        assert_eq!(missing(&["assets/é.png".into()], "`assets/*.svg`"), vec!["assets/é.png".to_string()]);
+    }
+
+    #[test]
+    fn ls_files_output_is_nul_separated_and_unquoted() {
+        let names = parse_ls_files("assets/zażółć.png\0docs/images/a b\nc.png\0".as_bytes());
+        assert_eq!(names, vec!["assets/zażółć.png".to_string(), "docs/images/a b\nc.png".into()]);
+    }
 }