dircolors: skip invalid UTF-8 lines instead of truncating the config

src/uu/dircolors/src/dircolors.rs

@@ -176,7 +176,7 @@ pub fn uumain(args: impl uucore::Args) -> UResult<()> {
             let result = if *file_arg == "-" {
                 let fin = BufReader::new(std::io::stdin());
                 // For example, for echo "owt 40;33"|dircolors -b -
-                parse(fin.lines().map_while(Result::ok), &out_format, "-")
+                parse(config_lines(fin), &out_format, "-")
             } else {
                 let path = Path::new(&file_arg);
                 if path.is_dir() {
@@ -188,11 +188,7 @@ pub fn uumain(args: impl uucore::Args) -> UResult<()> {
                 let file = File::open(path)
                     .map_err(|e| USimpleError::new(1, format!("{}: {e}", path.maybe_quote())))?;
                 let fin = BufReader::new(file);
-                parse(
-                    fin.lines().map_while(Result::ok),
-                    &out_format,
-                    &path.to_string_lossy(),
-                )
+                parse(config_lines(fin), &out_format, &path.to_string_lossy())
             };
 
             let string = result.map_err(|s| USimpleError::new(1, s))?;
@@ -313,6 +309,18 @@ enum ParseState {
     Pass,
 }
 
+/// Iterate over the lines of a config file.
+/// GNU dircolors is byte-oriented; invalid UTF-8 maps to an empty line,
+/// which the parser skips but still counts — line numbers stay aligned.
+fn config_lines(fin: impl BufRead) -> impl Iterator<Item = String> {
+    fin.split(b'\n').map_while(Result::ok).map(|mut line| {
+        if line.last() == Some(&b'\r') {
+            line.pop();
+        }
+        String::from_utf8(line).unwrap_or_default()
+    })
+}
+
 fn parse<T>(user_input: T, fmt: &OutputFmt, fp: &str) -> Result<String, String>
 where
     T: IntoIterator,

tests/by-util/test_dircolors.rs

@@ -5,7 +5,6 @@
 
 // spell-checker:ignore overridable colorterm
 
-#[cfg(target_os = "linux")]
 use uutests::at_and_ucmd;
 use uutests::new_ucmd;
 
@@ -249,3 +248,58 @@ fn test_invalid_term_glob() {
         .succeeds()
         .stdout_only("LS_COLORS='';\nexport LS_COLORS\n");
 }
+
+#[test]
+fn test_invalid_utf8_line_file() {
+    let (at, mut ucmd) = at_and_ucmd!();
+
+    let filename = "invalid-utf8-line";
+    std::fs::write(
+        at.plus(filename),
+        b"DIR 01;31\nBAD\xffLINE 99\n*.txt 00;32\n",
+    )
+    .unwrap();
+
+    // a line that fails to decode is skipped without truncating the rest
+    ucmd.args(&["-b", filename])
+        .succeeds()
+        .stdout_contains("di=01;31")
+        .stdout_contains("*.txt=00;32");
+}
+
+#[test]
+fn test_invalid_utf8_line_stdin() {
+    // the bad line must be skipped, not truncate the rest of the config
+    new_ucmd!()
+        .pipe_in(b"DIR 01;31\nBAD\xffLINE 99\n*.txt 00;32\n".to_vec())
+        .args(&["-b", "-"])
+        .succeeds()
+        .stdout_contains("di=01;31")
+        .stdout_contains("*.txt=00;32");
+}
+
+#[test]
+fn test_invalid_utf8_only_bad() {
+    let (at, mut ucmd) = at_and_ucmd!();
+
+    let filename = "invalid-utf8-only";
+    std::fs::write(at.plus(filename), b"BAD\xffLINE 99\n").unwrap();
+
+    // the config holds nothing parseable: empty LS_COLORS, exit 0
+    ucmd.args(&["-b", filename])
+        .succeeds()
+        .stdout_only("LS_COLORS='';\nexport LS_COLORS\n");
+}
+
+#[test]
+fn test_invalid_utf8_line_preserves_line_numbers() {
+    let (at, mut ucmd) = at_and_ucmd!();
+
+    let filename = "invalid-utf8-line-number";
+    std::fs::write(at.plus(filename), b"BAD\xffLINE 99\nONLYKEY\n").unwrap();
+
+    // GNU counts the skipped line, so the error points at the real line 2
+    ucmd.args(&["-b", filename])
+        .fails_with_code(1)
+        .stderr_contains(":2: invalid line");
+}