Avoid off-by-one error in stripping noqa following multi-byte char (#…

…8979) Closes #8976.
astral-sh · Dec 3, 2023 · 17c8817 · 17c8817
1 parent 1dda669
commit 17c8817
Show file tree

Hide file tree

Showing 3 changed files with 57 additions and 10 deletions.
diff --git a/crates/ruff_linter/resources/test/fixtures/ruff/RUF100_3.py b/crates/ruff_linter/resources/test/fixtures/ruff/RUF100_3.py
@@ -23,3 +23,6 @@
 print(a)  # noqa: E501, F821  # comment
 print(a)  # noqa: E501, F821 comment
 print(a)  # noqa: E501, F821  comment
+
+print(a)  # comment with unicode µ    # noqa: E501
+print(a)  # comment with unicode µ    # noqa: E501, F821
diff --git a/crates/ruff_linter/src/checkers/noqa.rs b/crates/ruff_linter/src/checkers/noqa.rs
@@ -3,10 +3,10 @@
 use std::path::Path;
 
 use itertools::Itertools;
-use ruff_text_size::{Ranged, TextLen, TextRange, TextSize};
+use ruff_text_size::{Ranged, TextLen, TextRange};
 
 use ruff_diagnostics::{Diagnostic, Edit, Fix};
-use ruff_python_trivia::CommentRanges;
+use ruff_python_trivia::{CommentRanges, PythonWhitespace};
 use ruff_source_file::Locator;
 
 use crate::noqa;
@@ -200,17 +200,11 @@ fn delete_noqa(range: TextRange, locator: &Locator) -> Edit {
 
     // Compute the leading space.
     let prefix = locator.slice(TextRange::new(line_range.start(), range.start()));
-    let leading_space = prefix
-        .rfind(|c: char| !c.is_whitespace())
-        .map_or(prefix.len(), |i| prefix.len() - i - 1);
-    let leading_space_len = TextSize::try_from(leading_space).unwrap();
+    let leading_space_len = prefix.text_len() - prefix.trim_whitespace_end().text_len();
 
     // Compute the trailing space.
     let suffix = locator.slice(TextRange::new(range.end(), line_range.end()));
-    let trailing_space = suffix
-        .find(|c: char| !c.is_whitespace())
-        .map_or(suffix.len(), |i| i);
-    let trailing_space_len = TextSize::try_from(trailing_space).unwrap();
+    let trailing_space_len = suffix.text_len() - suffix.trim_whitespace_start().text_len();
 
     // Ex) `# noqa`
     if line_range

diff --git a/crates/ruff_linter/src/rules/ruff/snapshots/ruff_linter__rules__ruff__tests__ruf100_3.snap b/crates/ruff_linter/src/rules/ruff/snapshots/ruff_linter__rules__ruff__tests__ruf100_3.snap
@@ -344,6 +344,7 @@ RUF100_3.py:23:11: RUF100 [*] Unused `noqa` directive (unused: `E501`)
    23 |+print(a)  # noqa: F821  # comment
 24 24 | print(a)  # noqa: E501, F821 comment
 25 25 | print(a)  # noqa: E501, F821  comment
+26 26 | 
 
 RUF100_3.py:24:11: RUF100 [*] Unused `noqa` directive (unused: `E501`)
    |
@@ -362,13 +363,17 @@ RUF100_3.py:24:11: RUF100 [*] Unused `noqa` directive (unused: `E501`)
 24    |-print(a)  # noqa: E501, F821 comment
    24 |+print(a)  # noqa: F821 comment
 25 25 | print(a)  # noqa: E501, F821  comment
+26 26 | 
+27 27 | print(a)  # comment with unicode µ    # noqa: E501
 
 RUF100_3.py:25:11: RUF100 [*] Unused `noqa` directive (unused: `E501`)
    |
 23 | print(a)  # noqa: E501, F821  # comment
 24 | print(a)  # noqa: E501, F821 comment
 25 | print(a)  # noqa: E501, F821  comment
    |           ^^^^^^^^^^^^^^^^^^ RUF100
+26 | 
+27 | print(a)  # comment with unicode µ    # noqa: E501
    |
    = help: Remove unused `noqa` directive
 
@@ -378,5 +383,50 @@ RUF100_3.py:25:11: RUF100 [*] Unused `noqa` directive (unused: `E501`)
 24 24 | print(a)  # noqa: E501, F821 comment
 25    |-print(a)  # noqa: E501, F821  comment
    25 |+print(a)  # noqa: F821  comment
+26 26 | 
+27 27 | print(a)  # comment with unicode µ    # noqa: E501
+28 28 | print(a)  # comment with unicode µ    # noqa: E501, F821
+
+RUF100_3.py:27:7: F821 Undefined name `a`
+   |
+25 | print(a)  # noqa: E501, F821  comment
+26 | 
+27 | print(a)  # comment with unicode µ    # noqa: E501
+   |       ^ F821
+28 | print(a)  # comment with unicode µ    # noqa: E501, F821
+   |
+
+RUF100_3.py:27:39: RUF100 [*] Unused `noqa` directive (unused: `E501`)
+   |
+25 | print(a)  # noqa: E501, F821  comment
+26 | 
+27 | print(a)  # comment with unicode µ    # noqa: E501
+   |                                       ^^^^^^^^^^^^ RUF100
+28 | print(a)  # comment with unicode µ    # noqa: E501, F821
+   |
+   = help: Remove unused `noqa` directive
+
+ℹ Safe fix
+24 24 | print(a)  # noqa: E501, F821 comment
+25 25 | print(a)  # noqa: E501, F821  comment
+26 26 | 
+27    |-print(a)  # comment with unicode µ    # noqa: E501
+   27 |+print(a)  # comment with unicode µ
+28 28 | print(a)  # comment with unicode µ    # noqa: E501, F821
+
+RUF100_3.py:28:39: RUF100 [*] Unused `noqa` directive (unused: `E501`)
+   |
+27 | print(a)  # comment with unicode µ    # noqa: E501
+28 | print(a)  # comment with unicode µ    # noqa: E501, F821
+   |                                       ^^^^^^^^^^^^^^^^^^ RUF100
+   |
+   = help: Remove unused `noqa` directive
+
+ℹ Safe fix
+25 25 | print(a)  # noqa: E501, F821  comment
+26 26 | 
+27 27 | print(a)  # comment with unicode µ    # noqa: E501
+28    |-print(a)  # comment with unicode µ    # noqa: E501, F821
+   28 |+print(a)  # comment with unicode µ    # noqa: F821