Merge pull request #3735 from resistor/main

Implement a fast path for character counting in wc.

src/uu/wc/src/count_fast.rs

@@ -134,3 +134,33 @@ pub(crate) fn count_bytes_and_lines_fast<R: Read>(
         }
     }
 }
+
+/// Returns a WordCount that counts the number of Unicode characters encoded in UTF-8 read via a Reader.
+///
+/// This corresponds to the `-m` command line flag to wc.
+///
+/// # Arguments
+///
+/// * `R` - A Reader from which the UTF-8 stream will be read.
+pub(crate) fn count_chars_fast<R: Read>(handle: &mut R) -> (WordCount, Option<io::Error>) {
+    /// Mask of the value bits of a continuation byte
+    const CONT_MASK: u8 = 0b0011_1111u8;
+    /// Value of the tag bits (tag mask is !CONT_MASK) of a continuation byte
+    const TAG_CONT_U8: u8 = 0b1000_0000u8;
+
+    let mut total = WordCount::default();
+    let mut buf = [0; BUF_SIZE];
+    loop {
+        match handle.read(&mut buf) {
+            Ok(0) => return (total, None),
+            Ok(n) => {
+                total.chars += buf[..n]
+                    .iter()
+                    .filter(|&&byte| (byte & !CONT_MASK) != TAG_CONT_U8)
+                    .count();
+            }
+            Err(ref e) if e.kind() == ErrorKind::Interrupted => continue,
+            Err(e) => return (total, Some(e)),
+        }
+    }
+}

src/uu/wc/src/wc.rs

@@ -13,7 +13,7 @@ extern crate uucore;
 mod count_fast;
 mod countable;
 mod word_count;
-use count_fast::{count_bytes_and_lines_fast, count_bytes_fast};
+use count_fast::{count_bytes_and_lines_fast, count_bytes_fast, count_chars_fast};
 use countable::WordCountable;
 use unicode_width::UnicodeWidthChar;
 use utf8::{BufReadDecoder, BufReadDecoderError};
@@ -315,6 +315,7 @@ fn word_count_from_reader<T: WordCountable>(
     ) {
         // Specialize scanning loop to improve the performance.
         (false, false, false, false, false) => unreachable!(),
+        (false, true, false, false, false) => count_chars_fast(&mut reader),
         (true, false, false, false, false) => {
             // Fast path when only show_bytes is true.
             let (bytes, error) = count_bytes_fast(&mut reader);