fix: avoid coverage scanner suffix copies (Refs #534)

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Author: karthiknadig

scripts/coverage_detail.py

@@ -15,6 +15,10 @@
 
 from quality_snapshot import SnapshotError, parse_lcov
 
+RAW_STRING_START = re.compile(r'(?:br|cr|r)(#*)"')
+CHARACTER_LITERAL = re.compile(r"'(?:[^'\\\n]|\\(?:u\{[0-9a-fA-F_]+\}|x[0-9a-fA-F]{2}|.))'")
+ITEM_ATTRIBUTE = re.compile(r'\s*#\s*\[')
+
 
 def run(*args: str, cwd: Path) -> str:
     return subprocess.check_output(args, cwd=cwd, text=True, encoding='utf-8')
@@ -112,10 +116,10 @@ def code_mask(source: str) -> str:
             if depth:
                 raise SnapshotError('Unterminated Rust block comment')
         else:
-            raw = re.match(r'(?:br|cr|r)(#*)"', source[i:])
+            raw = RAW_STRING_START.match(source, i)
             if raw:
                 close = '"' + raw[1]
-                end = source.find(close, i + raw.end())
+                end = source.find(close, raw.end())
                 if end == -1:
                     raise SnapshotError('Unterminated Rust raw string')
                 end += len(close)
@@ -132,9 +136,9 @@ def code_mask(source: str) -> str:
                 else:
                     raise SnapshotError('Unterminated Rust string')
             elif source[i] == "'":
-                char = re.match(r"'(?:[^'\\\n]|\\(?:u\{[0-9a-fA-F_]+\}|x[0-9a-fA-F]{2}|.))'", source[i:])
+                char = CHARACTER_LITERAL.match(source, i)
                 if char:
-                    end = i + char.end()
+                    end = char.end()
         if end > i:
             for n in range(i, end):
                 if chars[n] != '\n':
@@ -153,10 +157,10 @@ def test_lines(source: str) -> set[int]:
         end = match.end()
         # Other attributes belong to this same item, not its body.
         while True:
-            attr = re.match(r'\s*#\s*\[', masked[end:])
+            attr = ITEM_ATTRIBUTE.match(masked, end)
             if not attr:
                 break
-            end += attr.end()
+            end = attr.end()
             depth = 1
             while depth and end < len(masked):
                 depth += (masked[end] == '[') - (masked[end] == ']')

scripts/tests/test_coverage_detail.py

@@ -137,6 +137,22 @@ def test_literals_nested_comments_and_attributes_do_not_move_boundaries(self):
         self.assertEqual(detail.test_lines(source), set(range(3, 10)))
         self.assertEqual(detail.code_mask(source).count('\n'), source.count('\n'))
 
+    def test_literal_scanning_matches_at_offsets_without_copying_source_suffixes(self):
+        class NoSlices(str):
+            def __getitem__(self, key):
+                if isinstance(key, slice):
+                    raise AssertionError('scanner copied a source suffix')
+                return super().__getitem__(key)
+
+        prefix = "fn real<'a>(x: &'a str) { "
+        literals = ['r"text"', 'br##"}\\n{"##', 'cr#"text"#',
+                    "'{'", r"'\u{7b}'", r"'\x7b'", r"'\''"]
+        source = prefix + '; '.join(literals) + '; }\n'
+        expected = prefix + '; '.join(' ' * len(value) for value in literals) + '; }\n'
+        for copies in [1, 1000]:
+            with self.subTest(copies=copies):
+                self.assertEqual(detail.code_mask(NoSlices(source * copies)), expected * copies)
+
     def test_cfg_test_function_and_external_module_are_test_code(self):
         source = '#[cfg(test)]\nfn helper() {}\n#[cfg(test)]\nmod fixtures;\nfn real() {}\n'
         self.assertEqual(detail.test_lines(source), {1, 2, 3, 4})