Fix join to handle non-UTF-8 filenames

src/uu/join/src/join.rs

@@ -413,7 +413,7 @@ impl Line {
 
 struct State<'a> {
     key: usize,
-    file_name: &'a str,
+    file_name: &'a OsString,
     file_num: FileNum,
     print_unpaired: bool,
     lines: Split<Box<dyn BufRead + 'a>>,
@@ -427,7 +427,7 @@ struct State<'a> {
 impl<'a> State<'a> {
     fn new(
         file_num: FileNum,
-        name: &'a str,
+        name: &'a OsString,
         stdin: &'a Stdin,
         key: usize,
         line_ending: LineEnding,
@@ -436,7 +436,8 @@ impl<'a> State<'a> {
         let file_buf = if name == "-" {
             Box::new(stdin.lock()) as Box<dyn BufRead>
         } else {
-            let file = File::open(name).map_err_context(|| format!("{}", name.maybe_quote()))?;
+            let file = File::open(name)
+                .map_err_context(|| format!("{}", name.to_string_lossy().maybe_quote()))?;
             Box::new(BufReader::new(file)) as Box<dyn BufRead>
         };
 
@@ -639,7 +640,7 @@ impl<'a> State<'a> {
                 && (input.check_order == CheckOrder::Enabled
                     || (self.has_unpaired && !self.has_failed))
             {
-                let err_msg = translate!("join-error-not-sorted", "file" => self.file_name.maybe_quote(), "line_num" => self.line_num, "content" => String::from_utf8_lossy(&line.string));
+                let err_msg = translate!("join-error-not-sorted", "file" => self.file_name.to_string_lossy().maybe_quote(), "line_num" => self.line_num, "content" => String::from_utf8_lossy(&line.string));
                 // This is fatal if the check is enabled.
                 if input.check_order == CheckOrder::Enabled {
                     return Err(JoinError::UnorderedInput(err_msg));
@@ -826,8 +827,8 @@ pub fn uumain(args: impl uucore::Args) -> UResult<()> {
 
     let settings = parse_settings(&matches)?;
 
-    let file1 = matches.get_one::<String>("file1").unwrap();
-    let file2 = matches.get_one::<String>("file2").unwrap();
+    let file1 = matches.get_one::<OsString>("file1").unwrap();
+    let file2 = matches.get_one::<OsString>("file2").unwrap();
 
     if file1 == "-" && file2 == "-" {
         return Err(USimpleError::new(
@@ -951,6 +952,7 @@ pub fn uu_app() -> Command {
                 .required(true)
                 .value_name("FILE1")
                 .value_hint(clap::ValueHint::FilePath)
+                .value_parser(clap::value_parser!(OsString))
                 .hide(true),
         )
         .arg(
@@ -958,11 +960,17 @@ pub fn uu_app() -> Command {
                 .required(true)
                 .value_name("FILE2")
                 .value_hint(clap::ValueHint::FilePath)
+                .value_parser(clap::value_parser!(OsString))
                 .hide(true),
         )
 }
 
-fn exec<Sep: Separator>(file1: &str, file2: &str, settings: Settings, sep: Sep) -> UResult<()> {
+fn exec<Sep: Separator>(
+    file1: &OsString,
+    file2: &OsString,
+    settings: Settings,
+    sep: Sep,
+) -> UResult<()> {
     let stdin = stdin();
 
     let mut state1 = State::new(

tests/by-util/test_join.rs

@@ -533,3 +533,41 @@ fn test_full() {
         .fails()
         .stderr_contains("No space left on device");
 }
+
+#[test]
+#[cfg(target_os = "linux")]
+fn test_join_non_utf8_paths() {
+    use std::fs;
+
+    let ts = TestScenario::new(util_name!());
+    let at = &ts.fixtures;
+
+    // Create files with non-UTF-8 names using shell commands
+    // since the test framework doesn't support OsStr for file names
+    let test_dir = at.subdir.as_path();
+
+    // Create temporary files with valid names first
+    at.write("temp1.txt", "a 1\n");
+    at.write("temp2.txt", "a 2\n");
+
+    // Rename them to non-UTF-8 names using std::fs
+    let file1_bytes = b"test_\xFF\xFE_1.txt";
+    let file2_bytes = b"test_\xFF\xFE_2.txt";
+
+    #[cfg(unix)]
+    {
+        use std::os::unix::ffi::OsStrExt;
+        let file1_name = std::ffi::OsStr::from_bytes(file1_bytes);
+        let file2_name = std::ffi::OsStr::from_bytes(file2_bytes);
+
+        fs::rename(test_dir.join("temp1.txt"), test_dir.join(file1_name)).unwrap();
+        fs::rename(test_dir.join("temp2.txt"), test_dir.join(file2_name)).unwrap();
+
+        // Test that join can handle non-UTF-8 filenames
+        ts.ucmd()
+            .arg(file1_name)
+            .arg(file2_name)
+            .succeeds()
+            .stdout_only("a 1 2\n");
+    }
+}