Merge pull request #2552 from miDeb/cat/show-crlf

cat: show `\r\n` as `^M$` when `-E` is enabled

src/uu/cat/src/cat.rs

@@ -128,6 +128,12 @@ struct OutputState {
 
     /// Whether the output cursor is at the beginning of a new line
     at_line_start: bool,
+
+    /// Whether we skipped a \r, which still needs to be printed
+    skipped_carriage_return: bool,
+
+    /// Whether we have already printed a blank line
+    one_blank_kept: bool,
 }
 
 /// Represents an open file handle, stream, or other device
@@ -373,6 +379,8 @@ fn cat_files(files: Vec<String>, options: &OutputOptions) -> UResult<()> {
     let mut state = OutputState {
         line_number: 1,
         at_line_start: true,
+        skipped_carriage_return: false,
+        one_blank_kept: false,
     };
     let mut error_messages: Vec<String> = Vec::new();
 
@@ -381,6 +389,9 @@ fn cat_files(files: Vec<String>, options: &OutputOptions) -> UResult<()> {
             error_messages.push(format!("{}: {}", path, err));
         }
     }
+    if state.skipped_carriage_return {
+        print!("\r");
+    }
     if error_messages.is_empty() {
         Ok(())
     } else {
@@ -458,7 +469,6 @@ fn write_lines<R: Read>(
     let mut in_buf = [0; 1024 * 31];
     let stdout = io::stdout();
     let mut writer = stdout.lock();
-    let mut one_blank_kept = false;
 
     while let Ok(n) = handle.reader.read(&mut in_buf) {
         if n == 0 {
@@ -469,8 +479,13 @@ fn write_lines<R: Read>(
         while pos < n {
             // skip empty line_number enumerating them if needed
             if in_buf[pos] == b'\n' {
-                if !state.at_line_start || !options.squeeze_blank || !one_blank_kept {
+                // \r followed by \n is printed as ^M when show_ends is enabled, so that \r\n prints as ^M$
-                    one_blank_kept = true;
+                if state.skipped_carriage_return && options.show_ends {
+                    writer.write_all(b"^M")?;
+                    state.skipped_carriage_return = false;
+                }
+                if !state.at_line_start || !options.squeeze_blank || !state.one_blank_kept {
+                    state.one_blank_kept = true;
                     if state.at_line_start && options.number == NumberingMode::All {
                         write!(&mut writer, "{0:6}\t", state.line_number)?;
                         state.line_number += 1;
@@ -484,7 +499,12 @@ fn write_lines<R: Read>(
                 pos += 1;
                 continue;
             }
-            one_blank_kept = false;
+            if state.skipped_carriage_return {
+                writer.write_all(b"\r")?;
+                state.skipped_carriage_return = false;
+                state.at_line_start = false;
+            }
+            state.one_blank_kept = false;
             if state.at_line_start && options.number != NumberingMode::None {
                 write!(&mut writer, "{0:6}\t", state.line_number)?;
                 state.line_number += 1;
@@ -499,17 +519,22 @@ fn write_lines<R: Read>(
                 write_to_end(&in_buf[pos..], &mut writer)
             };
             // end of buffer?
-            if offset == 0 {
+            if offset + pos == in_buf.len() {
                 state.at_line_start = false;
                 break;
             }
-            // print suitable end of line
+            if in_buf[pos + offset] == b'\r' {
-            writer.write_all(options.end_of_line().as_bytes())?;
+                state.skipped_carriage_return = true;
-            if handle.is_interactive {
+            } else {
-                writer.flush()?;
+                assert_eq!(in_buf[pos + offset], b'\n');
+                // print suitable end of line
+                writer.write_all(options.end_of_line().as_bytes())?;
+                if handle.is_interactive {
+                    writer.flush()?;
+                }
+                state.at_line_start = true;
             }
-            state.at_line_start = true;
+            pos += offset + 1;
-            pos += offset;
         }
     }
 
@@ -517,17 +542,19 @@ fn write_lines<R: Read>(
 }
 
 // write***_to_end methods
-// Write all symbols till end of line or end of buffer is reached
+// Write all symbols till \n or \r or end of buffer is reached
-// Return the (number of written symbols + 1) or 0 if the end of buffer is reached
+// We need to stop at \r because it may be written as ^M depending on the byte after and settings;
+// however, write_nonprint_to_end doesn't need to stop at \r because it will always write \r as ^M.
+// Return the number of written symbols
 fn write_to_end<W: Write>(in_buf: &[u8], writer: &mut W) -> usize {
-    match in_buf.iter().position(|c| *c == b'\n') {
+    match in_buf.iter().position(|c| *c == b'\n' || *c == b'\r') {
         Some(p) => {
             writer.write_all(&in_buf[..p]).unwrap();
-            p + 1
+            p
         }
         None => {
             writer.write_all(in_buf).unwrap();
-            0
+            in_buf.len()
         }
     }
 }
@@ -535,20 +562,25 @@ fn write_to_end<W: Write>(in_buf: &[u8], writer: &mut W) -> usize {
 fn write_tab_to_end<W: Write>(mut in_buf: &[u8], writer: &mut W) -> usize {
     let mut count = 0;
     loop {
-        match in_buf.iter().position(|c| *c == b'\n' || *c == b'\t') {
+        match in_buf
+            .iter()
+            .position(|c| *c == b'\n' || *c == b'\t' || *c == b'\r')
+        {
             Some(p) => {
                 writer.write_all(&in_buf[..p]).unwrap();
                 if in_buf[p] == b'\n' {
-                    return count + p + 1;
+                    return count + p;
-                } else {
+                } else if in_buf[p] == b'\t' {
                     writer.write_all(b"^I").unwrap();
                     in_buf = &in_buf[p + 1..];
                     count += p + 1;
+                } else {
+                    return count + p;
                 }
             }
             None => {
                 writer.write_all(in_buf).unwrap();
-                return 0;
+                return in_buf.len();
             }
         };
     }
@@ -573,11 +605,7 @@ fn write_nonprint_to_end<W: Write>(in_buf: &[u8], writer: &mut W, tab: &[u8]) ->
         .unwrap();
         count += 1;
     }
-    if count != in_buf.len() {
+    count
-        count + 1
-    } else {
-        0
-    }
 }
 
 #[cfg(test)]

tests/by-util/test_cat.rs

@@ -273,6 +273,26 @@ fn test_stdin_show_ends() {
     }
 }
 
+#[test]
+fn squeeze_all_files() {
+    // empty lines at the end of a file are "squeezed" together with empty lines at the beginning
+    let (at, mut ucmd) = at_and_ucmd!();
+    at.write("input1", "a\n\n");
+    at.write("input2", "\n\nb");
+    ucmd.args(&["input1", "input2", "-s"])
+        .succeeds()
+        .stdout_only("a\n\nb");
+}
+
+#[test]
+fn test_show_ends_crlf() {
+    new_ucmd!()
+        .arg("-E")
+        .pipe_in("a\nb\r\n\rc\n\r\n\r")
+        .succeeds()
+        .stdout_only("a$\nb^M$\n\rc$\n^M$\n\r");
+}
+
 #[test]
 fn test_stdin_show_all() {
     for same_param in &["-A", "--show-all"] {