cat: Fix reporting "input file is output file" error when outputting to an input file

src/uu/cat/Cargo.toml

@@ -26,6 +26,13 @@ uucore = { workspace = true, features = ["fast-inc", "fs", "pipes"] }
 [target.'cfg(unix)'.dependencies]
 nix = { workspace = true }
 
+[target.'cfg(windows)'.dependencies]
+winapi-util = { workspace = true }
+windows-sys = { workspace = true, features = ["Win32_Storage_FileSystem"] }
+
+[dev-dependencies]
+tempfile = { workspace = true }
+
 [[bin]]
 name = "cat"
 path = "src/main.rs"

src/uu/cat/src/cat.rs

@@ -4,6 +4,10 @@
 // file that was distributed with this source code.
 
 // spell-checker:ignore (ToDO) nonprint nonblank nonprinting ELOOP
+
+mod platform;
+
+use crate::platform::is_unsafe_overwrite;
 use std::fs::{File, metadata};
 use std::io::{self, BufWriter, IsTerminal, Read, Write};
 /// Unix domain socket support
@@ -18,12 +22,9 @@ use std::os::unix::net::UnixStream;
 
 use clap::{Arg, ArgAction, Command};
 use memchr::memchr2;
-#[cfg(unix)]
-use nix::fcntl::{FcntlArg, fcntl};
 use thiserror::Error;
 use uucore::display::Quotable;
 use uucore::error::UResult;
-use uucore::fs::FileInformation;
 use uucore::locale::get_message;
 use uucore::{fast_inc::fast_inc_one, format_usage};
 
@@ -366,42 +367,17 @@ fn cat_handle<R: FdReadable>(
     }
 }
 
-/// Whether this process is appending to stdout.
-#[cfg(unix)]
-fn is_appending() -> bool {
-    let stdout = io::stdout();
-    let Ok(flags) = fcntl(stdout.as_fd(), FcntlArg::F_GETFL) else {
-        return false;
-    };
-    // TODO Replace `1 << 10` with `nix::fcntl::Oflag::O_APPEND`.
-    let o_append = 1 << 10;
-    (flags & o_append) > 0
-}
-
-#[cfg(not(unix))]
-fn is_appending() -> bool {
-    false
-}
-
-fn cat_path(
-    path: &str,
-    options: &OutputOptions,
-    state: &mut OutputState,
-    out_info: Option<&FileInformation>,
-) -> CatResult<()> {
+fn cat_path(path: &str, options: &OutputOptions, state: &mut OutputState) -> CatResult<()> {
     match get_input_type(path)? {
         InputType::StdIn => {
             let stdin = io::stdin();
-            let in_info = FileInformation::from_file(&stdin)?;
+            if is_unsafe_overwrite(&stdin, &io::stdout()) {
+                return Err(CatError::OutputIsInput);
+            }
             let mut handle = InputHandle {
                 reader: stdin,
                 is_interactive: io::stdin().is_terminal(),
             };
-            if let Some(out_info) = out_info {
-                if in_info == *out_info && is_appending() {
-                    return Err(CatError::OutputIsInput);
-                }
-            }
             cat_handle(&mut handle, options, state)
         }
         InputType::Directory => Err(CatError::IsDirectory),
@@ -417,15 +393,9 @@ fn cat_path(
         }
         _ => {
             let file = File::open(path)?;
-
-            if let Some(out_info) = out_info {
-                if out_info.file_size() != 0
-                    && FileInformation::from_file(&file).ok().as_ref() == Some(out_info)
-                {
-                    return Err(CatError::OutputIsInput);
-                }
+            if is_unsafe_overwrite(&file, &io::stdout()) {
+                return Err(CatError::OutputIsInput);
             }
-
             let mut handle = InputHandle {
                 reader: file,
                 is_interactive: false,
@@ -436,8 +406,6 @@ fn cat_path(
 }
 
 fn cat_files(files: &[String], options: &OutputOptions) -> UResult<()> {
-    let out_info = FileInformation::from_file(&io::stdout()).ok();
-
     let mut state = OutputState {
         line_number: LineNumber::new(),
         at_line_start: true,
@@ -447,7 +415,7 @@ fn cat_files(files: &[String], options: &OutputOptions) -> UResult<()> {
     let mut error_messages: Vec<String> = Vec::new();
 
     for path in files {
-        if let Err(err) = cat_path(path, options, &mut state, out_info.as_ref()) {
+        if let Err(err) = cat_path(path, options, &mut state) {
             error_messages.push(format!("{}: {err}", path.maybe_quote()));
         }
     }

src/uu/cat/src/platform/mod.rs

@@ -0,0 +1,16 @@
+// This file is part of the uutils coreutils package.
+//
+// For the full copyright and license information, please view the LICENSE
+// file that was distributed with this source code.
+
+#[cfg(unix)]
+pub use self::unix::is_unsafe_overwrite;
+
+#[cfg(windows)]
+pub use self::windows::is_unsafe_overwrite;
+
+#[cfg(unix)]
+mod unix;
+
+#[cfg(windows)]
+mod windows;

src/uu/cat/src/platform/unix.rs

@@ -0,0 +1,108 @@
+// This file is part of the uutils coreutils package.
+//
+// For the full copyright and license information, please view the LICENSE
+// file that was distributed with this source code.
+
+// spell-checker:ignore lseek seekable
+
+use nix::fcntl::{FcntlArg, OFlag, fcntl};
+use nix::unistd::{Whence, lseek};
+use std::os::fd::AsFd;
+use uucore::fs::FileInformation;
+
+/// An unsafe overwrite occurs when the same nonempty file is used as both stdin and stdout,
+/// and the file offset of stdin is positioned earlier than that of stdout.
+/// In this scenario, bytes read from stdin are written to a later part of the file
+/// via stdout, which can then be read again by stdin and written again by stdout,
+/// causing an infinite loop and potential file corruption.
+pub fn is_unsafe_overwrite<I: AsFd, O: AsFd>(input: &I, output: &O) -> bool {
+    // `FileInformation::from_file` returns an error if the file descriptor is closed, invalid,
+    // or refers to a non-regular file (e.g., socket, pipe, or special device).
+    let Ok(input_info) = FileInformation::from_file(input) else {
+        return false;
+    };
+    let Ok(output_info) = FileInformation::from_file(output) else {
+        return false;
+    };
+    if input_info != output_info || output_info.file_size() == 0 {
+        return false;
+    }
+    if is_appending(output) {
+        return true;
+    }
+    // `lseek` returns an error if the file descriptor is closed or it refers to
+    // a non-seekable resource (e.g., pipe, socket, or some devices).
+    let Ok(input_pos) = lseek(input.as_fd(), 0, Whence::SeekCur) else {
+        return false;
+    };
+    let Ok(output_pos) = lseek(output.as_fd(), 0, Whence::SeekCur) else {
+        return false;
+    };
+    input_pos < output_pos
+}
+
+/// Whether the file is opened with the `O_APPEND` flag
+fn is_appending<F: AsFd>(file: &F) -> bool {
+    let flags_raw = fcntl(file.as_fd(), FcntlArg::F_GETFL).unwrap_or_default();
+    let flags = OFlag::from_bits_truncate(flags_raw);
+    flags.contains(OFlag::O_APPEND)
+}
+
+#[cfg(test)]
+mod tests {
+    use crate::platform::unix::{is_appending, is_unsafe_overwrite};
+    use std::fs::OpenOptions;
+    use std::io::{Seek, SeekFrom, Write};
+    use tempfile::NamedTempFile;
+
+    #[test]
+    fn test_is_appending() {
+        let temp_file = NamedTempFile::new().unwrap();
+        assert!(!is_appending(&temp_file));
+
+        let read_file = OpenOptions::new().read(true).open(&temp_file).unwrap();
+        assert!(!is_appending(&read_file));
+
+        let write_file = OpenOptions::new().write(true).open(&temp_file).unwrap();
+        assert!(!is_appending(&write_file));
+
+        let append_file = OpenOptions::new().append(true).open(&temp_file).unwrap();
+        assert!(is_appending(&append_file));
+    }
+
+    #[test]
+    fn test_is_unsafe_overwrite() {
+        // Create two temp files one of which is empty
+        let empty = NamedTempFile::new().unwrap();
+        let mut nonempty = NamedTempFile::new().unwrap();
+        nonempty.write_all(b"anything").unwrap();
+        nonempty.seek(SeekFrom::Start(0)).unwrap();
+
+        // Using a different file as input and output does not result in an overwrite
+        assert!(!is_unsafe_overwrite(&empty, &nonempty));
+
+        // Overwriting an empty file is always safe
+        assert!(!is_unsafe_overwrite(&empty, &empty));
+
+        // Overwriting a nonempty file with itself is safe
+        assert!(!is_unsafe_overwrite(&nonempty, &nonempty));
+
+        // Overwriting an empty file opened in append mode is safe
+        let empty_append = OpenOptions::new().append(true).open(&empty).unwrap();
+        assert!(!is_unsafe_overwrite(&empty, &empty_append));
+
+        // Overwriting a nonempty file opened in append mode is unsafe
+        let nonempty_append = OpenOptions::new().append(true).open(&nonempty).unwrap();
+        assert!(is_unsafe_overwrite(&nonempty, &nonempty_append));
+
+        // Overwriting a file opened in write mode is safe
+        let mut nonempty_write = OpenOptions::new().write(true).open(&nonempty).unwrap();
+        assert!(!is_unsafe_overwrite(&nonempty, &nonempty_write));
+
+        // Overwriting a file when the input and output file descriptors are pointing to
+        // different offsets is safe if the input offset is further than the output offset
+        nonempty_write.seek(SeekFrom::Start(1)).unwrap();
+        assert!(!is_unsafe_overwrite(&nonempty_write, &nonempty));
+        assert!(is_unsafe_overwrite(&nonempty, &nonempty_write));
+    }
+}

src/uu/cat/src/platform/windows.rs

@@ -0,0 +1,56 @@
+// This file is part of the uutils coreutils package.
+//
+// For the full copyright and license information, please view the LICENSE
+// file that was distributed with this source code.
+
+use std::ffi::OsString;
+use std::os::windows::ffi::OsStringExt;
+use std::path::PathBuf;
+use uucore::fs::FileInformation;
+use winapi_util::AsHandleRef;
+use windows_sys::Win32::Storage::FileSystem::{
+    FILE_NAME_NORMALIZED, GetFinalPathNameByHandleW, VOLUME_NAME_NT,
+};
+
+/// An unsafe overwrite occurs when the same file is used as both stdin and stdout
+/// and the stdout file is not empty.
+pub fn is_unsafe_overwrite<I: AsHandleRef, O: AsHandleRef>(input: &I, output: &O) -> bool {
+    if !is_same_file_by_path(input, output) {
+        return false;
+    }
+
+    // Check if the output file is empty
+    FileInformation::from_file(output)
+        .map(|info| info.file_size() > 0)
+        .unwrap_or(false)
+}
+
+/// Get the file path for a file handle
+fn get_file_path_from_handle<F: AsHandleRef>(file: &F) -> Option<PathBuf> {
+    let handle = file.as_raw();
+    let mut path_buf = vec![0u16; 4096];
+
+    // SAFETY: We should check how many bytes was written to `path_buf`
+    // and only read that many bytes from it.
+    let len = unsafe {
+        GetFinalPathNameByHandleW(
+            handle,
+            path_buf.as_mut_ptr(),
+            path_buf.len() as u32,
+            FILE_NAME_NORMALIZED | VOLUME_NAME_NT,
+        )
+    };
+    if len == 0 {
+        return None;
+    }
+    let path = OsString::from_wide(&path_buf[..len as usize]);
+    Some(PathBuf::from(path))
+}
+
+/// Compare two file handles if they correspond to the same file
+fn is_same_file_by_path<A: AsHandleRef, B: AsHandleRef>(a: &A, b: &B) -> bool {
+    match (get_file_path_from_handle(a), get_file_path_from_handle(b)) {
+        (Some(path1), Some(path2)) => path1 == path2,
+        _ => false,
+    }
+}

tests/by-util/test_cat.rs

@@ -9,6 +9,7 @@ use rlimit::Resource;
 #[cfg(unix)]
 use std::fs::File;
 use std::fs::OpenOptions;
+use std::fs::read_to_string;
 use std::process::Stdio;
 use uutests::at_and_ucmd;
 use uutests::new_ucmd;
@@ -637,6 +638,57 @@ fn test_write_to_self() {
     );
 }
 
+/// Test derived from the following GNU test in `tests/cat/cat-self.sh`:
+///
+/// `cat fxy2 fy 1<>fxy2`
+// TODO: make this work on windows
+#[test]
+#[cfg(unix)]
+fn test_successful_write_to_read_write_self() {
+    let (at, mut ucmd) = at_and_ucmd!();
+    at.write("fy", "y");
+    at.write("fxy2", "x");
+
+    // Open `rw_file` as both stdin and stdout (read/write)
+    let fxy2_file_path = at.plus("fxy2");
+    let fxy2_file = OpenOptions::new()
+        .read(true)
+        .write(true)
+        .open(&fxy2_file_path)
+        .unwrap();
+    ucmd.args(&["fxy2", "fy"]).set_stdout(fxy2_file).succeeds();
+
+    // The contents of `fxy2` and `fy` files should be merged
+    let fxy2_contents = read_to_string(fxy2_file_path).unwrap();
+    assert_eq!(fxy2_contents, "xy");
+}
+
+/// Test derived from the following GNU test in `tests/cat/cat-self.sh`:
+///
+/// `cat fx fx3 1<>fx3`
+#[test]
+fn test_failed_write_to_read_write_self() {
+    let (at, mut ucmd) = at_and_ucmd!();
+    at.write("fx", "g");
+    at.write("fx3", "bold");
+
+    // Open `rw_file` as both stdin and stdout (read/write)
+    let fx3_file_path = at.plus("fx3");
+    let fx3_file = OpenOptions::new()
+        .read(true)
+        .write(true)
+        .open(&fx3_file_path)
+        .unwrap();
+    ucmd.args(&["fx", "fx3"])
+        .set_stdout(fx3_file)
+        .fails_with_code(1)
+        .stderr_only("cat: fx3: input file is output file\n");
+
+    // The contents of `fx` should have overwritten the beginning of `fx3`
+    let fx3_contents = read_to_string(fx3_file_path).unwrap();
+    assert_eq!(fx3_contents, "gold");
+}
+
 #[test]
 #[cfg(unix)]
 #[cfg(not(target_os = "openbsd"))]