Removing trailing whitespace inside multiline strings is unsafe (#9744)

Fix #8037.

crates/ruff_linter/resources/test/fixtures/pycodestyle/W291.py

@@ -0,0 +1,2 @@
+'''trailing whitespace 
+inside a multiline string'''

crates/ruff_linter/src/rules/pycodestyle/mod.rs

@@ -52,6 +52,7 @@ mod tests {
     #[test_case(Rule::SyntaxError, Path::new("E999.py"))]
     #[test_case(Rule::TabIndentation, Path::new("W19.py"))]
     #[test_case(Rule::TrailingWhitespace, Path::new("W29.py"))]
+    #[test_case(Rule::TrailingWhitespace, Path::new("W291.py"))]
     #[test_case(Rule::TrueFalseComparison, Path::new("E712.py"))]
     #[test_case(Rule::TypeComparison, Path::new("E721.py"))]
     #[test_case(Rule::UselessSemicolon, Path::new("E70.py"))]

crates/ruff_linter/src/rules/pycodestyle/rules/trailing_whitespace.rs

@@ -1,4 +1,4 @@
-use ruff_diagnostics::{AlwaysFixableViolation, Diagnostic, Edit, Fix};
+use ruff_diagnostics::{AlwaysFixableViolation, Applicability, Diagnostic, Edit, Fix};
 use ruff_macros::{derive_message_formats, violation};
 use ruff_python_index::Indexer;
 use ruff_source_file::{Line, Locator};
@@ -102,7 +102,15 @@ pub(crate) fn trailing_whitespace(
             }
         } else if settings.rules.enabled(Rule::TrailingWhitespace) {
             let mut diagnostic = Diagnostic::new(TrailingWhitespace, range);
-            diagnostic.set_fix(Fix::safe_edit(Edit::range_deletion(range)));
+            diagnostic.set_fix(Fix::applicable_edit(
+                Edit::range_deletion(range),
+                // Removing trailing whitespace is not safe inside multiline strings.
+                if indexer.multiline_ranges().intersects(range) {
+                    Applicability::Unsafe
+                } else {
+                    Applicability::Safe
+                },
+            ));
             return Some(diagnostic);
         }
     }

crates/ruff_linter/src/rules/pycodestyle/snapshots/ruff_linter__rules__pycodestyle__tests__W291_W291.py.snap

@@ -0,0 +1,17 @@
+---
+source: crates/ruff_linter/src/rules/pycodestyle/mod.rs
+---
+W291.py:1:23: W291 [*] Trailing whitespace
+  |
+1 | '''trailing whitespace 
+  |                       ^ W291
+2 | inside a multiline string'''
+  |
+  = help: Remove trailing whitespace
+
+ℹ Unsafe fix
+1   |-'''trailing whitespace 
+  1 |+'''trailing whitespace
+2 2 | inside a multiline string'''
+
+

crates/ruff_python_index/src/indexer.rs

@@ -11,6 +11,7 @@ use ruff_source_file::Locator;
 use ruff_text_size::{Ranged, TextRange, TextSize};
 
 use crate::fstring_ranges::{FStringRanges, FStringRangesBuilder};
+use crate::multiline_ranges::{MultilineRanges, MultilineRangesBuilder};
 use crate::CommentRangesBuilder;
 
 pub struct Indexer {
@@ -21,6 +22,9 @@ pub struct Indexer {
 
     /// The range of all f-string in the source document.
     fstring_ranges: FStringRanges,
+
+    /// The range of all multiline strings in the source document.
+    multiline_ranges: MultilineRanges,
 }
 
 impl Indexer {
@@ -29,6 +33,7 @@ impl Indexer {
 
         let mut comment_ranges_builder = CommentRangesBuilder::default();
         let mut fstring_ranges_builder = FStringRangesBuilder::default();
+        let mut multiline_ranges_builder = MultilineRangesBuilder::default();
         let mut continuation_lines = Vec::new();
         // Token, end
         let mut prev_end = TextSize::default();
@@ -61,6 +66,7 @@ impl Indexer {
 
             comment_ranges_builder.visit_token(tok, *range);
             fstring_ranges_builder.visit_token(tok, *range);
+            multiline_ranges_builder.visit_token(tok, *range);
 
             match tok {
                 Tok::Newline | Tok::NonLogicalNewline => {
@@ -82,6 +88,7 @@ impl Indexer {
             comment_ranges: comment_ranges_builder.finish(),
             continuation_lines,
             fstring_ranges: fstring_ranges_builder.finish(),
+            multiline_ranges: multiline_ranges_builder.finish(),
         }
     }
 
@@ -95,6 +102,11 @@ impl Indexer {
         &self.fstring_ranges
     }
 
+    /// Returns the byte offset ranges of multiline strings.
+    pub const fn multiline_ranges(&self) -> &MultilineRanges {
+        &self.multiline_ranges
+    }
+
     /// Returns the line start positions of continuations (backslash).
     pub fn continuation_line_starts(&self) -> &[TextSize] {
         &self.continuation_lines

crates/ruff_python_index/src/lib.rs

@@ -1,6 +1,7 @@
 mod comment_ranges;
 mod fstring_ranges;
 mod indexer;
+mod multiline_ranges;
 
 pub use comment_ranges::{tokens_and_ranges, CommentRangesBuilder};
 pub use indexer::Indexer;

crates/ruff_python_index/src/multiline_ranges.rs

@@ -0,0 +1,46 @@
+use ruff_python_parser::Tok;
+use ruff_text_size::TextRange;
+
+/// Stores the range of all multiline strings in a file sorted by
+/// [`TextRange::start`].
+pub struct MultilineRanges {
+    ranges: Vec<TextRange>,
+}
+
+impl MultilineRanges {
+    /// Returns `true` if the given range is inside a multiline string.
+    pub fn intersects(&self, target: TextRange) -> bool {
+        self.ranges
+            .binary_search_by(|range| {
+                if range.contains_range(target) {
+                    std::cmp::Ordering::Equal
+                } else if range.end() < target.start() {
+                    std::cmp::Ordering::Less
+                } else {
+                    std::cmp::Ordering::Greater
+                }
+            })
+            .is_ok()
+    }
+}
+
+#[derive(Default)]
+pub(crate) struct MultilineRangesBuilder {
+    ranges: Vec<TextRange>,
+}
+
+impl MultilineRangesBuilder {
+    pub(crate) fn visit_token(&mut self, token: &Tok, range: TextRange) {
+        if let Tok::String { triple_quoted, .. } = token {
+            if *triple_quoted {
+                self.ranges.push(range);
+            }
+        }
+    }
+
+    pub(crate) fn finish(self) -> MultilineRanges {
+        MultilineRanges {
+            ranges: self.ranges,
+        }
+    }
+}