feat: add assert_unchecked for memchr

Author: xtqqczze

src/uu/cat/src/cat.rs

@@ -650,6 +650,7 @@ fn write_end<W: Write>(
 fn write_to_end<W: Write>(in_buf: &[u8], writer: &mut W) -> io::Result<usize> {
     // using memchr2 significantly improves performances
     if let Some(p) = memchr2(b'\n', b'\r', in_buf) {
+        unsafe { std::hint::assert_unchecked(p < in_buf.len()) };
         writer.write_all(&in_buf[..p])?;
         Ok(p)
     } else {

src/uu/cut/src/cut.rs

@@ -294,6 +294,7 @@ fn cut_fields_newline_char_delim<R: Read, W: Write>(
                 has_processed_data = true;
 
                 if let Some(pos) = memchr::memchr(newline_char, buf) {
+                    unsafe { std::hint::assert_unchecked(pos < buf.len()) };
                     let amt = pos + 1;
                     line.extend_from_slice(&buf[..amt]);
                     reader.consume(amt);
@@ -315,6 +316,7 @@ fn cut_fields_newline_char_delim<R: Read, W: Write>(
                 has_processed_data = true;
 
                 if let Some(pos) = memchr::memchr(newline_char, buf) {
+                    unsafe { std::hint::assert_unchecked(pos < buf.len()) };
                     let bytes_to_consume = pos + 1;
                     reader.consume(bytes_to_consume);
                     break;

src/uu/cut/src/matcher.rs

@@ -28,6 +28,7 @@ impl Matcher for ExactMatcher<'_> {
         let mut pos = 0usize;
         loop {
             let match_idx = memchr(self.needle[0], &haystack[pos..])?;
+            unsafe { std::hint::assert_unchecked(match_idx < haystack.len()) };
             let match_idx = match_idx + pos; // account for starting from pos
 
             if self.needle.len() == 1 || haystack[match_idx + 1..].starts_with(&self.needle[1..]) {
@@ -45,6 +46,7 @@ pub struct WhitespaceMatcher {}
 impl Matcher for WhitespaceMatcher {
     fn next_match(&self, haystack: &[u8]) -> Option<(usize, usize)> {
         let match_idx = memchr2(b' ', b'\t', haystack)?;
+        unsafe { std::hint::assert_unchecked(match_idx < haystack.len()) };
         let mut skip = match_idx + 1;
 
         while skip < haystack.len() {

src/uu/factor/src/factor.rs

@@ -194,6 +194,7 @@ pub fn uumain(args: impl uucore::Args) -> UResult<()> {
                     let (mut display, mut prev) = (true, 0);
                     for i in memchr3_iter(DELIM_SPACE, DELIM_TAB, DELIM_NULL, &line).chain(once(le))
                     {
+                        unsafe { std::hint::assert_unchecked(i < line.len()) };
                         let has_null = line.get(i) == Some(&DELIM_NULL);
                         if display && (prev != i || has_null) {
                             write_factors_str(&line[prev..i], &mut w, print_exponents)?;

src/uu/numfmt/src/numfmt.rs

@@ -58,7 +58,10 @@ fn format_and_write<W: std::io::Write>(
 ) -> UResult<bool> {
     // GNU truncates at the first embedded null byte.
     let line = match memchr::memchr(b'\0', input_line) {
-        Some(i) => &input_line[..i],
+        Some(i) => {
+            unsafe { std::hint::assert_unchecked(i < input_line.len()) };
+            &input_line[..i]
+        }
         None => input_line,
     };
 

src/uu/pr/src/pr.rs

@@ -934,6 +934,7 @@ fn get_pages(options: &OutputOptions, file_id: usize, buf: &[u8]) -> Vec<(usize,
     // and a page comprises several lines. A form feed character marks
     // the end of a page regardless of how many lines have been read.
     for i in memchr::memchr2_iter(FF, NL, buf) {
+        unsafe { std::hint::assert_unchecked(i < buf.len()) };
         if buf[i] == FF {
             // Treat everything up to (but not including) the form feed
             // character as the last line of the page.

src/uu/sort/src/chunks.rs

@@ -316,6 +316,7 @@ fn parse_lines<'a>(
     let mut start = 0usize;
     let mut index = 0usize;
     for sep_idx in memchr_iter(separator, read) {
+        unsafe { std::hint::assert_unchecked(sep_idx < read.len()) };
         let line = &read[start..sep_idx];
         lines.push(Line::create(line, index, line_data, token_buffer, settings));
         index += 1;

src/uu/split/src/split.rs

@@ -879,6 +879,7 @@ impl Write for LineChunkWriter<'_> {
         let mut total_bytes_written = 0;
         let sep = self.settings.separator;
         for i in memchr::memchr_iter(sep, buf) {
+            unsafe { std::hint::assert_unchecked(i < buf.len()) };
             // If we have exceeded the number of lines to write in the
             // current chunk, then start a new chunk and its
             // corresponding writer.

src/uu/tac/src/tac.rs

@@ -209,6 +209,7 @@ fn buffer_tac(data: &[u8], before: bool, separator: &OsStr) -> std::io::Result<(
     // the end of the line (as in "abc\ndef\n") or at the beginning of
     // the line (as in "/abc/def").
     for i in memmem::rfind_iter(data, separator.as_encoded_bytes()) {
+        unsafe { std::hint::assert_unchecked(i < data.len()) };
         if before {
             out.write_all(&data[i..following_line_start])?;
             following_line_start = i;

src/uucore/src/lib/features/sum.rs

@@ -559,6 +559,7 @@ impl Write for DigestWriter<'_> {
         // the beginning of this "\r\n".
         let mut i_prev = 0;
         for i in memmem::find_iter(buf, b"\r\n") {
+            unsafe { std::hint::assert_unchecked(i < buf.len()) };
             self.digest.hash_update(&buf[i_prev..i]);
             i_prev = i + 1;
         }