From a3ebaaa730b5d6a8845533e01cdda58ef2081190 Mon Sep 17 00:00:00 2001
From: Sylvestre Ledru <sylvestre@debian.org>
Date: Fri, 8 Aug 2025 10:51:28 +0200
Subject: [PATCH] cat: fix handling of non-UTF-8 filenames

---
 src/uu/cat/src/cat.rs     | 14 ++++++++------
 tests/by-util/test_cat.rs | 26 +++++++++++++++++++++++++-
 2 files changed, 33 insertions(+), 7 deletions(-)

diff --git a/src/uu/cat/src/cat.rs b/src/uu/cat/src/cat.rs
index f03c7c3b9..d19c39f56 100644
--- a/src/uu/cat/src/cat.rs
+++ b/src/uu/cat/src/cat.rs
@@ -10,6 +10,7 @@ mod platform;
 use crate::platform::is_unsafe_overwrite;
 use clap::{Arg, ArgAction, Command};
 use memchr::memchr2;
+use std::ffi::OsString;
 use std::fs::{File, metadata};
 use std::io::{self, BufWriter, ErrorKind, IsTerminal, Read, Write};
 /// Unix domain socket support
@@ -267,9 +268,9 @@ pub fn uumain(args: impl uucore::Args) -> UResult<()> {
     .any(|v| matches.get_flag(v));
 
     let squeeze_blank = matches.get_flag(options::SQUEEZE_BLANK);
-    let files: Vec<String> = match matches.get_many::<String>(options::FILE) {
+    let files: Vec<OsString> = match matches.get_many::<OsString>(options::FILE) {
         Some(v) => v.cloned().collect(),
-        None => vec!["-".to_owned()],
+        None => vec![OsString::from("-")],
     };
 
     let options = OutputOptions {
@@ -294,6 +295,7 @@ pub fn uu_app() -> Command {
             Arg::new(options::FILE)
                 .hide(true)
                 .action(ArgAction::Append)
+                .value_parser(clap::value_parser!(OsString))
                 .value_hint(clap::ValueHint::FilePath),
         )
         .arg(
@@ -379,7 +381,7 @@ fn cat_handle<R: FdReadable>(
     }
 }
 
-fn cat_path(path: &str, options: &OutputOptions, state: &mut OutputState) -> CatResult<()> {
+fn cat_path(path: &OsString, options: &OutputOptions, state: &mut OutputState) -> CatResult<()> {
     match get_input_type(path)? {
         InputType::StdIn => {
             let stdin = io::stdin();
@@ -417,7 +419,7 @@ fn cat_path(path: &str, options: &OutputOptions, state: &mut OutputState) -> Cat
     }
 }
 
-fn cat_files(files: &[String], options: &OutputOptions) -> UResult<()> {
+fn cat_files(files: &[OsString], options: &OutputOptions) -> UResult<()> {
     let mut state = OutputState {
         line_number: LineNumber::new(),
         at_line_start: true,
@@ -452,8 +454,8 @@ fn cat_files(files: &[String], options: &OutputOptions) -> UResult<()> {
 /// # Arguments
 ///
 /// * `path` - Path on a file system to classify metadata
-fn get_input_type(path: &str) -> CatResult<InputType> {
-    if path == "-" {
+fn get_input_type(path: &OsString) -> CatResult<InputType> {
+    if path.to_str() == Some("-") {
         return Ok(InputType::StdIn);
     }
 
diff --git a/tests/by-util/test_cat.rs b/tests/by-util/test_cat.rs
index c3e25c6d0..647775a3a 100644
--- a/tests/by-util/test_cat.rs
+++ b/tests/by-util/test_cat.rs
@@ -14,9 +14,9 @@ use std::process::Stdio;
 use uutests::at_and_ucmd;
 use uutests::new_ucmd;
 use uutests::util::TestScenario;
+use uutests::util_name;
 #[cfg(not(windows))]
 use uutests::util::vec_of_size;
-use uutests::util_name;
 
 #[test]
 fn test_output_simple() {
@@ -747,6 +747,30 @@ fn test_write_fast_read_error() {
     ucmd.arg("foo").fails().stderr_contains("Permission denied");
 }
 
+#[test]
+#[cfg(target_os = "linux")]
+fn test_cat_non_utf8_paths() {
+    use std::ffi::OsStr;
+    use std::os::unix::ffi::OsStrExt;
+
+    let scene = TestScenario::new(util_name!());
+    let at = &scene.fixtures;
+    
+    // Create a test file with non-UTF-8 bytes in the name
+    let non_utf8_bytes = b"test_\xFF\xFE.txt";
+    let non_utf8_name = OsStr::from_bytes(non_utf8_bytes);
+    
+    // Create the actual file with some content
+    std::fs::write(at.plus(non_utf8_name), "Hello, non-UTF-8 world!\n").unwrap();
+    
+    // Test that cat handles non-UTF-8 file names without crashing
+    let result = scene.ucmd().arg(non_utf8_name).succeeds();
+    
+    // The result should contain the file content
+    let output = result.stdout_str_lossy();
+    assert_eq!(output, "Hello, non-UTF-8 world!\n");
+}
+
 #[test]
 #[cfg(target_os = "linux")]
 fn test_appending_same_input_output() {