Fixes #10192 - fix(comm): improve stdout handling and add test for lossy UTF-8 output

src/uu/comm/src/comm.rs

@@ -8,7 +8,7 @@
 use std::cmp::Ordering;
 use std::ffi::OsString;
 use std::fs::{File, metadata};
-use std::io::{self, BufRead, BufReader, Read, StdinLock, stdin};
+use std::io::{self, BufRead, BufReader, BufWriter, Read, StdinLock, Write, stdin};
 use std::path::Path;
 use uucore::display::Quotable;
 use uucore::error::{FromIo, UResult, USimpleError};
@@ -184,13 +184,25 @@ pub fn are_files_identical(path1: &Path, path2: &Path) -> io::Result<bool> {
     }
 }
 
+fn write_line_with_delimiter<W: Write>(writer: &mut W, delim: &[u8], line: &[u8]) -> UResult<()> {
+    writer
+        .write_all(delim)
+        .map_err_context(|| translate!("comm-error-write"))?;
+    writer
+        .write_all(line)
+        .map_err_context(|| translate!("comm-error-write"))?;
+    Ok(())
+}
+
 fn comm(a: &mut LineReader, b: &mut LineReader, delim: &str, opts: &ArgMatches) -> UResult<()> {
     let width_col_1 = usize::from(!opts.get_flag(options::COLUMN_1));
     let width_col_2 = usize::from(!opts.get_flag(options::COLUMN_2));
 
     let delim_col_2 = delim.repeat(width_col_1);
     let delim_col_3 = delim.repeat(width_col_1 + width_col_2);
 
+    let mut writer = BufWriter::new(io::stdout().lock());
+
     let ra = &mut Vec::new();
     let mut na = a.read_line(ra);
     let rb = &mut Vec::new();
@@ -239,7 +251,9 @@ fn comm(a: &mut LineReader, b: &mut LineReader, delim: &str, opts: &ArgMatches)
                     break;
                 }
                 if !opts.get_flag(options::COLUMN_1) {
-                    print!("{}", String::from_utf8_lossy(ra));
+                    writer
+                        .write_all(ra)
+                        .map_err_context(|| translate!("comm-error-write"))?;
                 }
                 ra.clear();
                 na = a.read_line(ra);
@@ -250,7 +264,7 @@ fn comm(a: &mut LineReader, b: &mut LineReader, delim: &str, opts: &ArgMatches)
                     break;
                 }
                 if !opts.get_flag(options::COLUMN_2) {
-                    print!("{delim_col_2}{}", String::from_utf8_lossy(rb));
+                    write_line_with_delimiter(&mut writer, delim_col_2.as_bytes(), rb)?;
                 }
                 rb.clear();
                 nb = b.read_line(rb);
@@ -262,7 +276,7 @@ fn comm(a: &mut LineReader, b: &mut LineReader, delim: &str, opts: &ArgMatches)
                     break;
                 }
                 if !opts.get_flag(options::COLUMN_3) {
-                    print!("{delim_col_3}{}", String::from_utf8_lossy(ra));
+                    write_line_with_delimiter(&mut writer, delim_col_3.as_bytes(), ra)?;
                 }
                 ra.clear();
                 rb.clear();
@@ -280,12 +294,16 @@ fn comm(a: &mut LineReader, b: &mut LineReader, delim: &str, opts: &ArgMatches)
 
     if opts.get_flag(options::TOTAL) {
         let line_ending = LineEnding::from_zero_flag(opts.get_flag(options::ZERO_TERMINATED));
-        print!(
+        write!(
+            writer,
             "{total_col_1}{delim}{total_col_2}{delim}{total_col_3}{delim}{}{line_ending}",
             translate!("comm-total")
-        );
+        )
+        .map_err_context(|| translate!("comm-error-write"))?;
     }
 
+    writer.flush().ok();
+
     if should_check_order && (checker1.has_error || checker2.has_error) {
         // Print the input error message once at the end
         if input_error {

tests/by-util/test_comm.rs

@@ -649,6 +649,30 @@ fn test_comm_eintr_handling() {
         .stdout_contains("line3");
 }
 
+#[test]
+fn test_output_lossy_utf8() {
+    let scene = TestScenario::new(util_name!());
+    let at = &scene.fixtures;
+
+    // Create files with invalid UTF-8
+    // A: \xfe\n\xff\n
+    // B: \xff\n\xfe\n
+    at.write_bytes("a", b"\xfe\n\xff\n");
+    at.write_bytes("b", b"\xff\n\xfe\n");
+
+    // GNU comm output (and uutils with fix):
+    // \xfe\n (col 1)
+    // \t\t\xff\n (col 3)
+    // \t\xfe\n (col 2)
+    // Hex: fe 0a 09 09 ff 0a 09 fe 0a
+
+    scene
+        .ucmd()
+        .args(&["a", "b"])
+        .fails() // Fails because of unsorted input
+        .stdout_is_bytes(b"\xfe\n\t\t\xff\n\t\xfe\n");
+}
+
 #[test]
 #[cfg(any(target_os = "linux", target_os = "android"))]
 fn test_comm_anonymous_pipes() {