Merge pull request #2209 from jfinkels/head-ring-buffer

head: add abstractions for "all but last n lines"

src/uu/head/Cargo.toml

@@ -16,7 +16,7 @@ path = "src/head.rs"
 
 [dependencies]
 clap = "2.33"
-uucore = { version=">=0.0.8", package="uucore", path="../../uucore" }
+uucore = { version=">=0.0.8", package="uucore", path="../../uucore", features=["ringbuffer"] }
 uucore_procs = { version=">=0.0.5", package="uucore_procs", path="../../uucore_procs" }
 
 [[bin]]

src/uu/head/src/head.rs

@@ -27,8 +27,12 @@ mod options {
     pub const ZERO_NAME: &str = "ZERO";
     pub const FILES_NAME: &str = "FILE";
 }
+mod lines;
 mod parse;
 mod split;
+mod take;
+use lines::zlines;
+use take::take_all_but;
 
 fn app<'a>() -> App<'a, 'a> {
     App::new(executable!())
@@ -293,36 +297,22 @@ fn rbuf_but_last_n_bytes(input: &mut impl std::io::BufRead, n: usize) -> std::io
 }
 
 fn rbuf_but_last_n_lines(
-    input: &mut impl std::io::BufRead,
+    input: impl std::io::BufRead,
     n: usize,
     zero: bool,
 ) -> std::io::Result<()> {
-    if n == 0 {
-        //prints everything
-        return rbuf_n_bytes(input, std::usize::MAX);
+    if zero {
+        let stdout = std::io::stdout();
+        let mut stdout = stdout.lock();
+        for bytes in take_all_but(zlines(input), n) {
+            stdout.write_all(&bytes?)?;
+        }
+    } else {
+        for line in take_all_but(input.lines(), n) {
+            println!("{}", line?);
+        }
     }
-    let mut ringbuf = vec![Vec::new(); n];
-    let stdout = std::io::stdout();
-    let mut stdout = stdout.lock();
-    let mut line = Vec::new();
-    let mut lines = 0usize;
-    split::walk_lines(input, zero, |e| match e {
-        split::Event::Data(dat) => {
-            line.extend_from_slice(dat);
-            Ok(true)
-        }
-        split::Event::Line => {
-            if lines < n {
-                ringbuf[lines] = std::mem::replace(&mut line, Vec::new());
-                lines += 1;
-            } else {
-                stdout.write_all(&ringbuf[0])?;
-                ringbuf.rotate_left(1);
-                ringbuf[n - 1] = std::mem::replace(&mut line, Vec::new());
-            }
-            Ok(true)
-        }
-    })
+    Ok(())
 }
 
 fn head_backwards_file(input: &mut std::fs::File, options: &HeadOptions) -> std::io::Result<()> {

src/uu/head/src/lines.rs

@@ -0,0 +1,73 @@
+//! Iterate over zero-terminated lines.
+use std::io::BufRead;
+
+/// The zero byte, representing the null character.
+const ZERO: u8 = 0;
+
+/// Returns an iterator over the lines of the given reader.
+///
+/// The iterator returned from this function will yield instances of
+/// [`io::Result`]<[`Vec`]<[`u8`]>>, representing the bytes of the line
+/// *including* the null character (with the possible exception of the
+/// last line, which may not have one).
+///
+/// # Examples
+///
+/// ```rust,ignore
+/// use std::io::Cursor;
+///
+/// let cursor = Cursor::new(b"x\0y\0z\0");
+/// let mut iter = zlines(cursor).map(|l| l.unwrap());
+/// assert_eq!(iter.next(), Some(b"x\0".to_vec()));
+/// assert_eq!(iter.next(), Some(b"y\0".to_vec()));
+/// assert_eq!(iter.next(), Some(b"z\0".to_vec()));
+/// assert_eq!(iter.next(), None);
+/// ```
+pub fn zlines<B>(buf: B) -> ZLines<B> {
+    ZLines { buf }
+}
+
+/// An iterator over the zero-terminated lines of an instance of `BufRead`.
+pub struct ZLines<B> {
+    buf: B,
+}
+
+impl<B: BufRead> Iterator for ZLines<B> {
+    type Item = std::io::Result<Vec<u8>>;
+
+    fn next(&mut self) -> Option<std::io::Result<Vec<u8>>> {
+        let mut buf = Vec::new();
+        match self.buf.read_until(ZERO, &mut buf) {
+            Ok(0) => None,
+            Ok(_) => Some(Ok(buf)),
+            Err(e) => Some(Err(e)),
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+
+    use crate::lines::zlines;
+    use std::io::Cursor;
+
+    #[test]
+    fn test_null_terminated() {
+        let cursor = Cursor::new(b"x\0y\0z\0");
+        let mut iter = zlines(cursor).map(|l| l.unwrap());
+        assert_eq!(iter.next(), Some(b"x\0".to_vec()));
+        assert_eq!(iter.next(), Some(b"y\0".to_vec()));
+        assert_eq!(iter.next(), Some(b"z\0".to_vec()));
+        assert_eq!(iter.next(), None);
+    }
+
+    #[test]
+    fn test_not_null_terminated() {
+        let cursor = Cursor::new(b"x\0y\0z");
+        let mut iter = zlines(cursor).map(|l| l.unwrap());
+        assert_eq!(iter.next(), Some(b"x\0".to_vec()));
+        assert_eq!(iter.next(), Some(b"y\0".to_vec()));
+        assert_eq!(iter.next(), Some(b"z".to_vec()));
+        assert_eq!(iter.next(), None);
+    }
+}

src/uu/head/src/take.rs

@@ -0,0 +1,93 @@
+//! Take all but the last elements of an iterator.
+use uucore::ringbuffer::RingBuffer;
+
+/// Create an iterator over all but the last `n` elements of `iter`.
+///
+/// # Examples
+///
+/// ```rust,ignore
+/// let data = [1, 2, 3, 4, 5];
+/// let n = 2;
+/// let mut iter = take_all_but(data.iter(), n);
+/// assert_eq!(Some(4), iter.next());
+/// assert_eq!(Some(5), iter.next());
+/// assert_eq!(None, iter.next());
+/// ```
+pub fn take_all_but<I: Iterator>(iter: I, n: usize) -> TakeAllBut<I> {
+    TakeAllBut::new(iter, n)
+}
+
+/// An iterator that only iterates over the last elements of another iterator.
+pub struct TakeAllBut<I: Iterator> {
+    iter: I,
+    buf: RingBuffer<<I as Iterator>::Item>,
+}
+
+impl<I: Iterator> TakeAllBut<I> {
+    pub fn new(mut iter: I, n: usize) -> TakeAllBut<I> {
+        // Create a new ring buffer and fill it up.
+        //
+        // If there are fewer than `n` elements in `iter`, then we
+        // exhaust the iterator so that whenever `TakeAllBut::next()` is
+        // called, it will return `None`, as expected.
+        let mut buf = RingBuffer::new(n);
+        for _ in 0..n {
+            let value = match iter.next() {
+                None => {
+                    break;
+                }
+                Some(x) => x,
+            };
+            buf.push_back(value);
+        }
+        TakeAllBut { iter, buf }
+    }
+}
+
+impl<I: Iterator> Iterator for TakeAllBut<I>
+where
+    I: Iterator,
+{
+    type Item = <I as Iterator>::Item;
+
+    fn next(&mut self) -> Option<<I as Iterator>::Item> {
+        match self.iter.next() {
+            Some(value) => self.buf.push_back(value),
+            None => None,
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+
+    use crate::take::take_all_but;
+
+    #[test]
+    fn test_fewer_elements() {
+        let mut iter = take_all_but([0, 1, 2].iter(), 2);
+        assert_eq!(Some(&0), iter.next());
+        assert_eq!(None, iter.next());
+    }
+
+    #[test]
+    fn test_same_number_of_elements() {
+        let mut iter = take_all_but([0, 1].iter(), 2);
+        assert_eq!(None, iter.next());
+    }
+
+    #[test]
+    fn test_more_elements() {
+        let mut iter = take_all_but([0].iter(), 2);
+        assert_eq!(None, iter.next());
+    }
+
+    #[test]
+    fn test_zero_elements() {
+        let mut iter = take_all_but([0, 1, 2].iter(), 0);
+        assert_eq!(Some(&0), iter.next());
+        assert_eq!(Some(&1), iter.next());
+        assert_eq!(Some(&2), iter.next());
+        assert_eq!(None, iter.next());
+    }
+}

src/uu/tail/Cargo.toml

@@ -17,7 +17,7 @@ path = "src/tail.rs"
 [dependencies]
 clap = "2.33"
 libc = "0.2.42"
-uucore = { version=">=0.0.8", package="uucore", path="../../uucore" }
+uucore = { version=">=0.0.8", package="uucore", path="../../uucore", features=["ringbuffer"] }
 uucore_procs = { version=">=0.0.5", package="uucore_procs", path="../../uucore_procs" }
 winapi = { version="0.3", features=["fileapi", "handleapi", "processthreadsapi", "synchapi", "winbase"] }
 

src/uu/tail/src/ringbuffer.rs

@@ -1,61 +0,0 @@
-//! A fixed-size ring buffer.
-use std::collections::VecDeque;
-
-/// A fixed-size ring buffer backed by a `VecDeque`.
-///
-/// If the ring buffer is not full, then calling the [`push_back`]
-/// method appends elements, as in a [`VecDeque`]. If the ring buffer
-/// is full, then calling [`push_back`] removes the element at the
-/// front of the buffer (in a first-in, first-out manner) before
-/// appending the new element to the back of the buffer.
-///
-/// Use [`from_iter`] to take the last `size` elements from an
-/// iterator.
-///
-/// # Examples
-///
-/// After exceeding the size limit, the oldest elements are dropped in
-/// favor of the newest element:
-///
-/// ```rust,ignore
-/// let buffer: RingBuffer<u8> = RingBuffer::new(2);
-/// buffer.push_back(0);
-/// buffer.push_back(1);
-/// buffer.push_back(2);
-/// assert_eq!(vec![1, 2], buffer.data);
-/// ```
-///
-/// Take the last `n` elements from an iterator:
-///
-/// ```rust,ignore
-/// let iter = vec![0, 1, 2, 3].iter();
-/// assert_eq!(vec![2, 3], RingBuffer::from_iter(iter, 2).data);
-/// ```
-pub struct RingBuffer<T> {
-    pub data: VecDeque<T>,
-    size: usize,
-}
-
-impl<T> RingBuffer<T> {
-    pub fn new(size: usize) -> RingBuffer<T> {
-        RingBuffer {
-            data: VecDeque::new(),
-            size,
-        }
-    }
-
-    pub fn from_iter(iter: impl Iterator<Item = T>, size: usize) -> RingBuffer<T> {
-        let mut ringbuf = RingBuffer::new(size);
-        for value in iter {
-            ringbuf.push_back(value);
-        }
-        ringbuf
-    }
-
-    pub fn push_back(&mut self, value: T) {
-        if self.size <= self.data.len() {
-            self.data.pop_front();
-        }
-        self.data.push_back(value)
-    }
-}

src/uu/tail/src/tail.rs

@@ -17,9 +17,7 @@ extern crate uucore;
 
 mod chunks;
 mod platform;
-mod ringbuffer;
 use chunks::ReverseChunks;
-use ringbuffer::RingBuffer;
 
 use clap::{App, Arg};
 use std::collections::VecDeque;
@@ -30,6 +28,7 @@ use std::io::{stdin, stdout, BufRead, BufReader, Read, Seek, SeekFrom, Write};
 use std::path::Path;
 use std::thread::sleep;
 use std::time::Duration;
+use uucore::ringbuffer::RingBuffer;
 
 pub mod options {
     pub mod verbosity {

src/uucore/Cargo.toml

@@ -47,6 +47,7 @@ mode = ["libc"]
 parse_time = []
 perms = ["libc"]
 process = ["libc"]
+ringbuffer = []
 signals = []
 utf8 = []
 utmpx = ["time", "libc"]

src/uucore/src/lib/features.rs

@@ -8,6 +8,8 @@ pub mod fs;
 pub mod fsext;
 #[cfg(feature = "parse_time")]
 pub mod parse_time;
+#[cfg(feature = "ringbuffer")]
+pub mod ringbuffer;
 #[cfg(feature = "zero-copy")]
 pub mod zero_copy;
 

src/uucore/src/lib/features/ringbuffer.rs

@@ -0,0 +1,134 @@
+//! A fixed-size ring buffer.
+use std::collections::VecDeque;
+
+/// A fixed-size ring buffer backed by a `VecDeque`.
+///
+/// If the ring buffer is not full, then calling the [`push_back`]
+/// method appends elements, as in a [`VecDeque`]. If the ring buffer
+/// is full, then calling [`push_back`] removes the element at the
+/// front of the buffer (in a first-in, first-out manner) before
+/// appending the new element to the back of the buffer.
+///
+/// Use [`from_iter`] to take the last `size` elements from an
+/// iterator.
+///
+/// # Examples
+///
+/// After exceeding the size limit, the oldest elements are dropped in
+/// favor of the newest element:
+///
+/// ```rust,ignore
+/// let mut buffer: RingBuffer<u8> = RingBuffer::new(2);
+/// buffer.push_back(0);
+/// buffer.push_back(1);
+/// buffer.push_back(2);
+/// assert_eq!(vec![1, 2], buffer.data);
+/// ```
+///
+/// Take the last `n` elements from an iterator:
+///
+/// ```rust,ignore
+/// let iter = [0, 1, 2].iter();
+/// let actual = RingBuffer::from_iter(iter, 2).data;
+/// let expected = VecDeque::from_iter([1, 2].iter());
+/// assert_eq!(expected, actual);
+/// ```
+pub struct RingBuffer<T> {
+    pub data: VecDeque<T>,
+    size: usize,
+}
+
+impl<T> RingBuffer<T> {
+    pub fn new(size: usize) -> RingBuffer<T> {
+        RingBuffer {
+            data: VecDeque::new(),
+            size,
+        }
+    }
+
+    pub fn from_iter(iter: impl Iterator<Item = T>, size: usize) -> RingBuffer<T> {
+        let mut ringbuf = RingBuffer::new(size);
+        for value in iter {
+            ringbuf.push_back(value);
+        }
+        ringbuf
+    }
+
+    /// Append a value to the end of the ring buffer.
+    ///
+    /// If the ring buffer is not full, this method return [`None`]. If
+    /// the ring buffer is full, appending a new element will cause the
+    /// oldest element to be evicted. In that case this method returns
+    /// that element, or `None`.
+    ///
+    /// In the special case where the size limit is zero, each call to
+    /// this method with input `value` returns `Some(value)`, because
+    /// the input is immediately evicted.
+    ///
+    /// # Examples
+    ///
+    /// Appending an element when the buffer is full returns the oldest
+    /// element:
+    ///
+    /// ```rust,ignore
+    /// let mut buf = RingBuffer::new(3);
+    /// assert_eq!(None, buf.push_back(0));
+    /// assert_eq!(None, buf.push_back(1));
+    /// assert_eq!(None, buf.push_back(2));
+    /// assert_eq!(Some(0), buf.push_back(3));
+    /// ```
+    ///
+    /// If the size limit is zero, then this method always returns the
+    /// input value:
+    ///
+    /// ```rust,ignore
+    /// let mut buf = RingBuffer::new(0);
+    /// assert_eq!(Some(0), buf.push_back(0));
+    /// assert_eq!(Some(1), buf.push_back(1));
+    /// assert_eq!(Some(2), buf.push_back(2));
+    /// ```
+    pub fn push_back(&mut self, value: T) -> Option<T> {
+        if self.size == 0 {
+            return Some(value);
+        }
+        let result = if self.size <= self.data.len() {
+            self.data.pop_front()
+        } else {
+            None
+        };
+        self.data.push_back(value);
+        result
+    }
+}
+
+#[cfg(test)]
+mod tests {
+
+    use crate::ringbuffer::RingBuffer;
+    use std::collections::VecDeque;
+    use std::iter::FromIterator;
+
+    #[test]
+    fn test_size_limit_zero() {
+        let mut buf = RingBuffer::new(0);
+        assert_eq!(Some(0), buf.push_back(0));
+        assert_eq!(Some(1), buf.push_back(1));
+        assert_eq!(Some(2), buf.push_back(2));
+    }
+
+    #[test]
+    fn test_evict_oldest() {
+        let mut buf = RingBuffer::new(2);
+        assert_eq!(None, buf.push_back(0));
+        assert_eq!(None, buf.push_back(1));
+        assert_eq!(Some(0), buf.push_back(2));
+    }
+
+    #[test]
+    fn test_from_iter() {
+        let iter = [0, 1, 2].iter();
+        let actual = RingBuffer::from_iter(iter, 2).data;
+        let expected = VecDeque::from_iter([1, 2].iter());
+        assert_eq!(expected, actual);
+    }
+}

src/uucore/src/lib/lib.rs

@@ -39,6 +39,8 @@ pub use crate::features::fs;
 pub use crate::features::fsext;
 #[cfg(feature = "parse_time")]
 pub use crate::features::parse_time;
+#[cfg(feature = "ringbuffer")]
+pub use crate::features::ringbuffer;
 #[cfg(feature = "zero-copy")]
 pub use crate::features::zero_copy;
 

tests/by-util/test_head.rs

@@ -129,6 +129,15 @@ fn test_zero_terminated_syntax_2() {
         .stdout_is("x\0y");
 }
 
+#[test]
+fn test_zero_terminated_negative_lines() {
+    new_ucmd!()
+        .args(&["-z", "-n", "-1"])
+        .pipe_in("x\0y\0z\0")
+        .run()
+        .stdout_is("x\0y\0");
+}
+
 #[test]
 fn test_negative_byte_syntax() {
     new_ucmd!()