sed: write the newline a file's last line lacks before the next file's output, as GNU sed does

src/sed/fast_io.rs

@@ -594,7 +594,7 @@ pub struct OutputBuffer {
     mmap_chunk: Option<MmapOutput>, // Chunk to write
     // True when the last write didn't end with \n; the \n is deferred so
     // that commands like `p` don't emit a spurious newline under -n.
-    pending_newline: bool,
+    pub(crate) pending_newline: bool,
     #[cfg(test)]
     low_level_flushes: usize, // Number of system call flushes
 }
@@ -687,11 +687,17 @@ impl OutputBuffer {
         self.out.write_all(bytes)
     }
 
-    /// Copy the specified file to the output, after any deferred newline.
-    pub fn copy_file(&mut self, path: &PathBuf) -> io::Result<()> {
-        // As in GNU sed, write the deferred newline even when the file is
-        // empty or cannot be read.
-        self.write_raw(b"")?;
+    /// Copy the specified file to the output. With `after_newline`, write
+    /// any deferred newline first, even when the file is empty or cannot be
+    /// read, as GNU sed does; otherwise leave it for the next output, as GNU
+    /// sed does for `0r`.
+    pub fn copy_file(&mut self, path: &PathBuf, after_newline: bool) -> io::Result<()> {
+        if after_newline {
+            self.flush_pending_newline()?;
+        }
+        // Flush mmap writes, if any.
+        #[cfg(unix)]
+        self.flush_mmap(WriteRange::Complete)?;
 
         let Ok(file) = File::open(path) else {
             // Per POSIX, if the file can't be read treat it as empty.

src/sed/in_place.rs

@@ -67,7 +67,10 @@ impl InPlace {
     /// to the context settings.
     fn begin_resolved(&mut self, file_name: &Path) -> UResult<&mut OutputBuffer> {
         if !self.in_place {
+            // Carry over a newline deferred at the end of the previous file.
+            let pending_newline = self.output.pending_newline;
             self.output = OutputBuffer::new(Box::new(stdout()));
+            self.output.pending_newline = pending_newline;
             return Ok(&mut self.output);
         }
 

src/sed/named_writer.rs

@@ -12,7 +12,7 @@ use crate::sed::error_handling::{ScriptLocation, runtime_error};
 
 use std::cell::RefCell;
 use std::fs::{File, OpenOptions};
-use std::io::{BufWriter, Write};
+use std::io::{self, BufWriter, Write};
 use std::path::PathBuf;
 use std::rc::Rc;
 
@@ -30,6 +30,7 @@ pub struct NamedWriter {
     pub path: PathBuf,
     writer: BufWriter<File>,
     location: ScriptLocation,
+    pending_newline: bool, // The last line written lacked its newline
 }
 
 impl NamedWriter {
@@ -49,6 +50,7 @@ impl NamedWriter {
             path,
             writer: BufWriter::new(file),
             location,
+            pending_newline: false,
         }));
 
         FLUSH_LIST.with(|list| list.borrow_mut().push(Rc::clone(&writer)));
@@ -61,23 +63,26 @@ impl NamedWriter {
     }
 
     /// Write bytes to the file, possibly with a newline, returning errors.
+    /// A missing newline is written before any further line.
     pub fn write_line_bytes(&mut self, line: &[u8], newline: bool) -> UResult<()> {
-        self.writer
-            .write_all(line)
-            .and_then(|()| {
-                if newline {
-                    self.writer.write_all(b"\n")
-                } else {
-                    Ok(())
-                }
-            })
-            .map_err(|e| {
-                runtime_error::<()>(
-                    &self.location,
-                    format!("writing to file {}: {e}", self.path.quote()),
-                )
-                .unwrap_err()
-            })
+        let pending = std::mem::replace(&mut self.pending_newline, !newline);
+        let mut write = || -> io::Result<()> {
+            if pending {
+                self.writer.write_all(b"\n")?;
+            }
+            self.writer.write_all(line)?;
+            if newline {
+                self.writer.write_all(b"\n")?;
+            }
+            Ok(())
+        };
+        write().map_err(|e| {
+            runtime_error::<()>(
+                &self.location,
+                format!("writing to file {}: {e}", self.path.quote()),
+            )
+            .unwrap_err()
+        })
     }
 
     /// Flush the writer, returning a descriptive error.
@@ -138,4 +143,31 @@ mod tests {
 
         assert_eq!(fs::read(path).unwrap(), b"a\xE9");
     }
+
+    #[test]
+    fn test_write_line_bytes_reports_deferred_newline_error() {
+        let file = NamedTempFile::new().unwrap();
+        let path = file.path().to_path_buf();
+        let mut writer = NamedWriter {
+            path: path.clone(),
+            // Unbuffered writes to a read-only handle fail immediately.
+            writer: BufWriter::with_capacity(0, File::open(&path).unwrap()),
+            location: ScriptLocation {
+                input_name: Rc::from("script.sed"),
+                line_number: 3,
+                column_number: 7,
+                ..ScriptLocation::default()
+            },
+            pending_newline: false,
+        };
+
+        writer.write_line_bytes(b"", false).unwrap();
+        // Only the deferred newline has bytes to write.
+        let error = writer.write_line_bytes(b"", false).unwrap_err();
+        assert_eq!(error.code(), 2);
+        let message = error.to_string();
+        assert!(message.starts_with("script.sed:3:7: error: "));
+        assert!(message.contains("writing to file"));
+        assert!(message.contains(path.to_str().unwrap()));
+    }
 }

src/sed/processor.rs

@@ -478,7 +478,7 @@ fn flush_appends(output: &mut OutputBuffer, context: &mut ProcessingContext) ->
                 output.write_raw(text.as_ref())?;
             }
             AppendElement::Path(path) => {
-                output.copy_file(path)?;
+                output.copy_file(path, true)?;
             }
         }
     }
@@ -632,7 +632,7 @@ fn process_address_0(
                     && cmd.addr2.is_none()
                 {
                     let path = extract_variant!(cmd, Path);
-                    output.copy_file(path)?;
+                    output.copy_file(path, false)?;
                 }
 
                 cmd.next.clone()
@@ -902,16 +902,11 @@ fn process_file(
                     // Append only the first line of the pattern space.
                     let writer = extract_variant!(command, NamedWriter);
                     let pattern_bytes = pattern.as_bytes();
-                    let (first_line, found_newline) =
-                        match pattern_bytes.iter().position(|&b| b == b'\n') {
-                            // A slice including the newline
-                            Some(pos) => (&pattern_bytes[..=pos], true),
-                            None => (pattern_bytes, false),
-                        };
-                    writer.borrow_mut().write_line_bytes(
-                        first_line,
-                        !found_newline && pattern.is_newline_terminated(),
-                    )?;
+                    let (first_line, newline) = match memchr(b'\n', pattern_bytes) {
+                        Some(pos) => (&pattern_bytes[..pos], true),
+                        None => (pattern_bytes, pattern.is_newline_terminated()),
+                    };
+                    writer.borrow_mut().write_line_bytes(first_line, newline)?;
                 }
                 'x' => {
                     // Exchange the contents of the pattern and hold spaces.

tests/by-util/test_sed.rs

@@ -2020,12 +2020,12 @@ fn write_first_line_newline() -> std::io::Result<()> {
 
     new_ucmd!()
         .args(&["-n", "-e", &cmd])
-        .pipe_in("abc\ndef\n")
+        .pipe_in("abc\ndef\nghi\njkl\n")
         .succeeds();
 
     let mut actual = String::new();
     temp.reopen()?.read_to_string(&mut actual)?;
-    assert_eq!(actual, "abc\n");
+    assert_eq!(actual, "abc\nghi\n");
 
     Ok(())
 }
@@ -2618,6 +2618,8 @@ fn in_place_edit_separate() -> std::io::Result<()> {
             "b1+b2\n",
         ),
         (&["-n", "-i", "-e", "N;p"], "a1\n", b, "", "b1\nb2\n"),
+        // A newline missing at the end of one file is not written to the next.
+        (&["-i", "-e", "s/a/A/"], "a", "b\n", "A", "b\n"),
     ];
 
     for (args, first, second, expected1, expected2) in cases {
@@ -3408,3 +3410,60 @@ fn test_execute_command_output_is_copied_unchanged() {
         .succeeds()
         .stdout_is("a\n");
 }
+
+/// A newline that a file's last line lacks is written before the next
+/// file's output, both to standard output and to `w` and `W` files.
+#[test]
+fn test_missing_newline_is_written_between_files() -> std::io::Result<()> {
+    type Case = (&'static [&'static str], &'static str, [&'static str; 3]);
+    let cases: &[Case] = &[
+        (&[], "w", ["e\nf", "g\nh", "e\nf\ng\nh"]),
+        (&[], "W", ["e\nf", "g\nh", "e\nf\ng\nh"]),
+        (&["-s"], "w", ["e\nf", "g\nh", "e\nf\ng\nh"]),
+        (&[], "2Q;w", ["x", "y", "x"]),
+    ];
+    for (flags, script, [first, second, expected]) in cases {
+        let (dir, one, two) = two_inputs(first, second)?;
+        let out = dir.path().join("out");
+        new_ucmd!()
+            .args(flags)
+            .arg(format!("{script} {}", out.display()))
+            .arg(&one)
+            .arg(&two)
+            .succeeds()
+            .stdout_is(*expected);
+        assert_eq!(fs::read_to_string(&out)?, *expected, "{flags:?} {script}");
+    }
+    Ok(())
+}
+
+/// `0r` at the start of the next file leaves the missing newline for that
+/// file's first line, as in GNU sed.
+#[test]
+fn test_zero_address_read_keeps_missing_newline_for_next_line() -> std::io::Result<()> {
+    let (dir, one, two) = two_inputs("a", "b\n")?;
+    let text = dir.path().join("text");
+    fs::write(&text, "X\nY")?;
+    new_ucmd!()
+        .args(&["-s", &format!("0r {}", text.display())])
+        .arg(&one)
+        .arg(&two)
+        .succeeds()
+        .stdout_is("X\nYaX\nY\nb\n");
+    Ok(())
+}
+
+/// A `w` file gets the newline that an emptied last line lacks before the
+/// next file's line, as in GNU sed.
+#[test]
+fn test_write_empty_line_without_newline() -> std::io::Result<()> {
+    let (dir, one, two) = two_inputs("a", "b")?;
+    let out = dir.path().join("out");
+    new_ucmd!()
+        .args(&["-n", &format!("s/.*//w {}", out.display())])
+        .arg(&one)
+        .arg(&two)
+        .succeeds();
+    assert_eq!(fs::read_to_string(&out)?, "\n");
+    Ok(())
+}