Fix paste to handle non-UTF-8 filenames

src/uu/paste/src/paste.rs

@@ -5,9 +5,11 @@
 
 use clap::{Arg, ArgAction, Command};
 use std::cell::{OnceCell, RefCell};
+use std::ffi::OsString;
 use std::fs::File;
 use std::io::{BufRead, BufReader, Stdin, Write, stdin, stdout};
 use std::iter::Cycle;
+use std::path::Path;
 use std::rc::Rc;
 use std::slice::Iter;
 use uucore::LocalizedCommand;
@@ -30,7 +32,7 @@ pub fn uumain(args: impl uucore::Args) -> UResult<()> {
     let serial = matches.get_flag(options::SERIAL);
     let delimiters = matches.get_one::<String>(options::DELIMITER).unwrap();
     let files = matches
-        .get_many::<String>(options::FILE)
+        .get_many::<OsString>(options::FILE)
         .unwrap()
         .cloned()
         .collect();
@@ -67,7 +69,8 @@ pub fn uu_app() -> Command {
                 .value_name("FILE")
                 .action(ArgAction::Append)
                 .default_value("-")
-                .value_hint(clap::ValueHint::FilePath),
+                .value_hint(clap::ValueHint::FilePath)
+                .value_parser(clap::value_parser!(OsString)),
         )
         .arg(
             Arg::new(options::ZERO_TERMINATED)
@@ -80,7 +83,7 @@ pub fn uu_app() -> Command {
 
 #[allow(clippy::cognitive_complexity)]
 fn paste(
-    filenames: Vec<String>,
+    filenames: Vec<OsString>,
     serial: bool,
     delimiters: &str,
     line_ending: LineEnding,
@@ -92,17 +95,16 @@ fn paste(
     let mut input_source_vec = Vec::with_capacity(filenames.len());
 
     for filename in filenames {
-        let input_source = match filename.as_str() {
-            "-" => InputSource::StandardInput(
+        let input_source = if filename == "-" {
+            InputSource::StandardInput(
                 stdin_once_cell
                     .get_or_init(|| Rc::new(RefCell::new(stdin())))
                     .clone(),
-            ),
-            st => {
-                let file = File::open(st)?;
-
-                InputSource::File(BufReader::new(file))
-            }
+            )
+        } else {
+            let path = Path::new(&filename);
+            let file = File::open(path)?;
+            InputSource::File(BufReader::new(file))
         };
 
         input_source_vec.push(input_source);

tests/by-util/test_paste.rs

@@ -4,7 +4,8 @@
 // file that was distributed with this source code.
 
 // spell-checker:ignore bsdutils toybox
-
+#[cfg(target_os = "linux")]
+use std::os::unix::ffi::OsStringExt;
 use uutests::at_and_ucmd;
 use uutests::new_ucmd;
 
@@ -252,6 +253,7 @@ FIRST!SECOND@THIRD#FOURTH!ABCDEFG
 }
 
 #[test]
+#[cfg(unix)]
 fn test_non_utf8_input() {
     // 0xC0 is not valid UTF-8
     const INPUT: &[u8] = b"Non-UTF-8 test: \xC0\x00\xC0.\n";
@@ -375,3 +377,20 @@ fn test_data() {
             .stdout_is(example.out);
     }
 }
+
+#[test]
+#[cfg(target_os = "linux")]
+fn test_paste_non_utf8_paths() {
+    let (at, mut ucmd) = at_and_ucmd!();
+
+    let filename1 = std::ffi::OsString::from_vec(vec![0xFF, 0xFE]);
+    let filename2 = std::ffi::OsString::from_vec(vec![0xF0, 0x90]);
+
+    std::fs::write(at.plus(&filename1), b"line1\nline2\n").unwrap();
+    std::fs::write(at.plus(&filename2), b"col1\ncol2\n").unwrap();
+
+    ucmd.arg(&filename1)
+        .arg(&filename2)
+        .succeeds()
+        .stdout_is("line1\tcol1\nline2\tcol2\n");
+}