Use memchr for invalid-escape-sequence (#5994)

crates/ruff/resources/test/fixtures/pycodestyle/W605_0.py

@@ -41,3 +41,5 @@ regex = '\w'  # noqa
 regex = '''
 \w
 '''  # noqa
+
+regex = '\\\_'

crates/ruff/src/checkers/tokens.rs

@@ -126,6 +126,7 @@ pub(crate) fn check_tokens(
             }
         }
     }
+
     // PLE2510, PLE2512, PLE2513
     if enforce_invalid_string_character {
         for (tok, range) in tokens.iter().flatten() {
@@ -177,7 +178,7 @@ pub(crate) fn check_tokens(
     }
 
     // TD001, TD002, TD003, TD004, TD005, TD006, TD007
-    // T001, T002, T003, T004
+    // FIX001, FIX002, FIX003, FIX004
     if enforce_todos {
         let todo_comments: Vec<TodoComment> = indexer
             .comment_ranges()
@@ -188,9 +189,7 @@ pub(crate) fn check_tokens(
                 TodoComment::from_comment(comment, *comment_range, i)
             })
             .collect();
-
         flake8_todos::rules::todos(&mut diagnostics, &todo_comments, locator, indexer, settings);
-
         flake8_fixme::rules::todos(&mut diagnostics, &todo_comments);
     }
 

crates/ruff/src/rules/pycodestyle/rules/invalid_escape_sequence.rs

@@ -1,3 +1,4 @@
+use memchr::memchr_iter;
 use ruff_text_size::{TextLen, TextRange, TextSize};
 
 use ruff_diagnostics::{AlwaysAutofixableViolation, Diagnostic, Edit, Fix};
@@ -60,29 +61,26 @@ pub(crate) fn invalid_escape_sequence(
         return;
     }
 
-    let start_offset = range.start() + TextSize::try_from(leading_quote.len()).unwrap();
-
-    let mut chars_iter = body.char_indices().peekable();
-
     let mut contains_valid_escape_sequence = false;
-
     let mut invalid_escape_sequence = Vec::new();
-    while let Some((i, c)) = chars_iter.next() {
-        if c != '\\' {
-            continue;
-        }
 
+    let mut prev = None;
+    let bytes = body.as_bytes();
+    for i in memchr_iter(b'\\', bytes) {
         // If the previous character was also a backslash, skip.
-        if i > 0 && body.as_bytes()[i - 1] == b'\\' {
+        if prev.map_or(false, |prev| prev == i - 1) {
+            prev = None;
             continue;
         }
 
-        // If we're at the end of the file, skip.
+        prev = Some(i);
-        let Some((_, next_char)) = chars_iter.peek() else {
+
+        let Some(next_char) = body[i + 1..].chars().next() else {
+            // If we're at the end of the file, skip.
             continue;
         };
 
-        // If we're at the end of the line, skip
+        // If we're at the end of line, skip.
         if matches!(next_char, '\n' | '\r') {
             continue;
         }
@@ -120,9 +118,9 @@ pub(crate) fn invalid_escape_sequence(
             continue;
         }
 
-        let location = start_offset + TextSize::try_from(i).unwrap();
+        let location = range.start() + leading_quote.text_len() + TextSize::try_from(i).unwrap();
         let range = TextRange::at(location, next_char.text_len() + TextSize::from(1));
-        invalid_escape_sequence.push(Diagnostic::new(InvalidEscapeSequence(*next_char), range));
+        invalid_escape_sequence.push(Diagnostic::new(InvalidEscapeSequence(next_char), range));
     }
 
     if autofix {

crates/ruff/src/rules/pycodestyle/snapshots/ruff__rules__pycodestyle__tests__W605_W605_0.py.snap

@@ -119,4 +119,20 @@ W605_0.py:28:12: W605 [*] Invalid escape sequence: `\.`
 30 30 | #: Okay
 31 31 | regex = r'\.png$'
 
+W605_0.py:45:12: W605 [*] Invalid escape sequence: `\_`
+   |
+43 | '''  # noqa
+44 | 
+45 | regex = '\\\_'
+   |            ^^ W605
+   |
+   = help: Add backslash to escape sequence
+
+ℹ Fix
+42 42 | \w
+43 43 | '''  # noqa
+44 44 | 
+45    |-regex = '\\\_'
+   45 |+regex = '\\\\_'
+