Merge pull request #6951 from fuad1502/csplit-pipe-input

csplit: support reading from pipe

src/uu/csplit/src/csplit.rs

@@ -87,7 +87,11 @@ pub fn csplit<T>(options: &CsplitOptions, patterns: &[String], input: T) -> Resu
 where
     T: BufRead,
 {
-    let mut input_iter = InputSplitter::new(input.lines().enumerate());
+    let enumerated_input_lines = input
+        .lines()
+        .map(|line| line.map_err_context(|| "read error".to_string()))
+        .enumerate();
+    let mut input_iter = InputSplitter::new(enumerated_input_lines);
     let mut split_writer = SplitWriter::new(options);
     let patterns: Vec<patterns::Pattern> = patterns::get_patterns(patterns)?;
     let ret = do_csplit(&mut split_writer, patterns, &mut input_iter);
@@ -117,7 +121,7 @@ fn do_csplit<I>(
     input_iter: &mut InputSplitter<I>,
 ) -> Result<(), CsplitError>
 where
-    I: Iterator<Item = (usize, io::Result<String>)>,
+    I: Iterator<Item = (usize, UResult<String>)>,
 {
     // split the file based on patterns
     for pattern in patterns {
@@ -305,7 +309,7 @@ impl SplitWriter<'_> {
         input_iter: &mut InputSplitter<I>,
     ) -> Result<(), CsplitError>
     where
-        I: Iterator<Item = (usize, io::Result<String>)>,
+        I: Iterator<Item = (usize, UResult<String>)>,
     {
         input_iter.rewind_buffer();
         input_iter.set_size_of_buffer(1);
@@ -358,7 +362,7 @@ impl SplitWriter<'_> {
         input_iter: &mut InputSplitter<I>,
     ) -> Result<(), CsplitError>
     where
-        I: Iterator<Item = (usize, io::Result<String>)>,
+        I: Iterator<Item = (usize, UResult<String>)>,
     {
         if offset >= 0 {
             // The offset is zero or positive, no need for a buffer on the lines read.
@@ -470,7 +474,7 @@ impl SplitWriter<'_> {
 /// This is used to pass matching lines to the next split and to support patterns with a negative offset.
 struct InputSplitter<I>
 where
-    I: Iterator<Item = (usize, io::Result<String>)>,
+    I: Iterator<Item = (usize, UResult<String>)>,
 {
     iter: I,
     buffer: Vec<<I as Iterator>::Item>,
@@ -483,7 +487,7 @@ where
 
 impl<I> InputSplitter<I>
 where
-    I: Iterator<Item = (usize, io::Result<String>)>,
+    I: Iterator<Item = (usize, UResult<String>)>,
 {
     fn new(iter: I) -> Self {
         Self {
@@ -547,7 +551,7 @@ where
 
 impl<I> Iterator for InputSplitter<I>
 where
-    I: Iterator<Item = (usize, io::Result<String>)>,
+    I: Iterator<Item = (usize, UResult<String>)>,
 {
     type Item = <I as Iterator>::Item;
 
@@ -581,13 +585,7 @@ pub fn uumain(args: impl uucore::Args) -> UResult<()> {
         Ok(csplit(&options, &patterns, stdin.lock())?)
     } else {
         let file = File::open(file_name)
-            .map_err_context(|| format!("cannot access {}", file_name.quote()))?;
-        let file_metadata = file
-            .metadata()
-            .map_err_context(|| format!("cannot access {}", file_name.quote()))?;
-        if !file_metadata.is_file() {
-            return Err(CsplitError::NotRegularFile(file_name.to_string()).into());
-        }
+            .map_err_context(|| format!("cannot open {} for reading", file_name.quote()))?;
         Ok(csplit(&options, &patterns, BufReader::new(file))?)
     }
 }

src/uu/csplit/src/csplit_error.rs

@@ -35,6 +35,8 @@ pub enum CsplitError {
     SuffixFormatTooManyPercents,
     #[error("{} is not a regular file", ._0.quote())]
     NotRegularFile(String),
+    #[error("{}", _0)]
+    UError(Box<dyn UError>),
 }
 
 impl From<io::Error> for CsplitError {
@@ -43,8 +45,17 @@ impl From<io::Error> for CsplitError {
     }
 }
 
+impl From<Box<dyn UError>> for CsplitError {
+    fn from(error: Box<dyn UError>) -> Self {
+        Self::UError(error)
+    }
+}
+
 impl UError for CsplitError {
     fn code(&self) -> i32 {
-        1
+        match self {
+            Self::UError(e) => e.code(),
+            _ => 1,
+        }
     }
 }

tests/by-util/test_csplit.rs

@@ -1379,7 +1379,7 @@ fn no_such_file() {
     let (_, mut ucmd) = at_and_ucmd!();
     ucmd.args(&["in", "0"])
         .fails()
-        .stderr_contains("cannot access 'in': No such file or directory");
+        .stderr_contains("cannot open 'in' for reading: No such file or directory");
 }
 
 #[test]
@@ -1417,3 +1417,52 @@ fn repeat_everything() {
     assert_eq!(at.read("xxz_004"), generate(37, 44 + 1));
     assert_eq!(at.read("xxz_005"), generate(46, 50 + 1));
 }
+
+#[cfg(unix)]
+#[test]
+fn test_named_pipe_input_file() {
+    let (at, mut ucmd) = at_and_ucmd!();
+
+    let mut fifo_writer =
+        create_named_pipe_with_writer(&at.plus_as_string("fifo"), &generate(1, 51));
+
+    let result = ucmd.args(&["fifo", "10"]).succeeds();
+    fifo_writer.kill().unwrap();
+    fifo_writer.wait().unwrap();
+    result.stdout_only("18\n123\n");
+
+    let count = glob(&at.plus_as_string("xx*"))
+        .expect("there should be splits created")
+        .count();
+    assert_eq!(count, 2);
+    assert_eq!(at.read("xx00"), generate(1, 10));
+    assert_eq!(at.read("xx01"), generate(10, 51));
+}
+
+#[cfg(unix)]
+fn create_named_pipe_with_writer(path: &str, data: &str) -> std::process::Child {
+    // cSpell:ignore IRWXU
+    nix::unistd::mkfifo(path, nix::sys::stat::Mode::S_IRWXU).unwrap();
+    std::process::Command::new("sh")
+        .arg("-c")
+        .arg(format!("printf '{}' > {path}", data))
+        .spawn()
+        .unwrap()
+}
+
+#[test]
+fn test_directory_input_file() {
+    let (at, mut ucmd) = at_and_ucmd!();
+    at.mkdir("test_directory");
+
+    #[cfg(unix)]
+    ucmd.args(&["test_directory", "1"])
+        .fails()
+        .code_is(1)
+        .stderr_only("csplit: read error: Is a directory\n");
+    #[cfg(windows)]
+    ucmd.args(&["test_directory", "1"])
+        .fails()
+        .code_is(1)
+        .stderr_only("csplit: cannot open 'test_directory' for reading: Permission denied\n");
+}