Support excluding multiline regexes (fix #996)

coverage/parser.py

@@ -105,14 +105,22 @@ def lines_matching(self, regex: str) -> set[TLineNo]:
         """Find the lines matching a regex.
 
         Returns a set of line numbers, the lines that contain a match for
-        `regex`.  The entire line needn't match, just a part of it.
+        `regex`. The entire line needn't match, just a part of it.
+        Handles multiline regex patterns.
 
         """
-        regex_c = re.compile(regex)
-        matches = set()
-        for i, ltext in enumerate(self.text.split("\n"), start=1):
-            if regex_c.search(ltext):
-                matches.add(self._multiline.get(i, i))
+        regex_c = re.compile(regex, re.MULTILINE)
+        matches: set[TLineNo] = set()
+
+        last_start = 0
+        last_start_line = 0
+        for match in regex_c.finditer(self.text):
+            start, end = match.span()
+            start_line = last_start_line + self.text.count('\n', last_start, start)
+            end_line = last_start_line + self.text.count('\n', last_start, end)
+            matches.update(self._multiline.get(i, i) for i in range(start_line + 1, end_line + 2))
+            last_start = start
+            last_start_line = start_line
         return matches
 
     def _raw_parse(self) -> None:

tests/test_parser.py

@@ -781,6 +781,82 @@ def function() -> int:
         assert parser.raw_statements == {1, 3, 4, 5, 6, 8, 9}
         assert parser.statements == {1, 8, 9}
 
+    def test_multiline_exclusion_single_line(self) -> None:
+        regex = r"print\('.*'\)"
+        parser = self.parse_text("""\
+        def foo():
+            print('Hello, world!')
+        """, regex)
+        assert parser.lines_matching(regex) == {2}
+        assert parser.raw_statements == {1, 2}
+        assert parser.statements == {1}
+
+    def test_multiline_exclusion_suite(self) -> None:
+        regex = r"if T:\n\s+print\('Hello, world!'\)"
+        parser = self.parse_text("""\
+        def foo():
+            if T:
+                print('Hello, world!')
+                print('This is a multiline regex test.')
+        """, regex)
+        assert parser.lines_matching(regex) == {2, 3}
+        assert parser.raw_statements == {1, 2, 3, 4}
+        assert parser.statements == {1}
+
+    def test_multiline_exclusion_no_match(self) -> None:
+        regex = r"nonexistent"
+        parser = self.parse_text("""\
+        def foo():
+            print('Hello, world!')
+        """, regex)
+        assert parser.lines_matching(regex) == set()
+        assert parser.raw_statements == {1, 2}
+        assert parser.statements == {1, 2}
+
+    def test_multiline_exclusion_no_source(self) -> None:
+        regex = r"anything"
+        parser = PythonParser(text="", filename="dummy.py", exclude=regex)
+        assert parser.lines_matching(regex) == set()
+        assert parser.raw_statements == set()
+        assert parser.statements == set()
+
+    def test_multiline_exclusion_multiple_matches(self) -> None:
+        regex = r"print\('.*'\)"
+        parser = self.parse_text("""\
+        def foo():
+            print('Hello, world!')
+        def bar():
+            print('Hello again!')
+        """, regex)
+        assert parser.lines_matching(regex) == {2, 4}
+        assert parser.raw_statements == {1, 2, 3, 4}
+        assert parser.statements == {1, 3}
+
+    def test_multiline_exclusion_suite2(self) -> None:
+        regex = r"print\('Hello, world!'\)\n\s+if T:"
+        parser = self.parse_text("""\
+        def foo():
+            print('Hello, world!')
+            if T:
+                print('This is a test.')
+        """, regex)
+        assert parser.lines_matching(regex) == {2, 3}
+        assert parser.raw_statements == {1, 2, 3, 4}
+        assert parser.statements == {1}
+
+    def test_multiline_exclusion_match_all(self) -> None:
+        regex = (r"def foo\(\):\n\s+print\('Hello, world!'\)\n"
+                 r"\s+if T:\n\s+print\('This is a test\.'\)")
+        parser = self.parse_text("""\
+        def foo():
+            print('Hello, world!')
+            if T:
+                print('This is a test.')
+        """, regex)
+        assert parser.lines_matching(regex) == {1, 2, 3, 4}
+        assert parser.raw_statements == {1, 2, 3, 4}
+        assert parser.statements == set()
+
 
 class ParserMissingArcDescriptionTest(PythonParserTestBase):
     """Tests for PythonParser.missing_arc_description."""