Report real line numbers after an --ignore-multiline-regex region

codespell_lib/_codespell.py

@@ -310,22 +310,32 @@ def open_with_internal(
 
     def get_lines(self, f: TextIO) -> list[tuple[bool, int, list[str]]]:
         fragments = []
-        line_number = 0
         if self.ignore_multiline_regex:
             text = f.read()
             pos = 0
             for m in re.finditer(self.ignore_multiline_regex, text):
-                lines = text[pos : m.start()].splitlines(True)
-                fragments.append((False, line_number, lines))
-                line_number += len(lines)
-                lines = m.group().splitlines(True)
-                fragments.append((True, line_number, lines))
-                line_number += len(lines) - 1
+                # A fragment can start mid-line, so its line number has to come
+                # from its offset rather than from counting the lines before it.
+                fragments.append(
+                    (
+                        False,
+                        text.count("\n", 0, pos),
+                        text[pos : m.start()].splitlines(True),
+                    )
+                )
+                fragments.append(
+                    (
+                        True,
+                        text.count("\n", 0, m.start()),
+                        m.group().splitlines(True),
+                    )
+                )
                 pos = m.end()
-            lines = text[pos:].splitlines(True)
-            fragments.append((False, line_number, lines))
+            fragments.append(
+                (False, text.count("\n", 0, pos), text[pos:].splitlines(True))
+            )
         else:
-            fragments.append((False, line_number, f.readlines()))
+            fragments.append((False, 0, f.readlines()))
         return fragments
 
 

codespell_lib/tests/test_basic.py

@@ -1054,6 +1054,46 @@ def test_ignore_regex_option(
     assert cs.main(fname, r"--ignore-regex=\bdonn\b") == 1
 
 
+def test_ignore_multiline_regex_line_numbers(
+    tmp_path: Path,
+    capsys: pytest.CaptureFixture[str],
+) -> None:
+    """Misspellings after an ignored region keep their real line numbers."""
+    fname = tmp_path / "ml.txt"
+    fname.write_text(
+        "line one is fine\n"
+        "# codespell:ignore-begin\n"
+        "abandonned inside the ignored region\n"
+        "# codespell:ignore-end\n"
+        "line five is fine\n"
+        "here is the abandonned we report\n"
+    )
+
+    # The example from --help: the match ends on a line boundary.
+    result = cs.main(
+        fname,
+        "--ignore-multiline-regex",
+        "# codespell:ignore-begin *\n.*# codespell:ignore-end *\n",
+        std=True,
+    )
+    assert isinstance(result, tuple)
+    code, stdout, _ = result
+    assert code == 1
+    assert f"{fname}:6: abandonned" in stdout
+
+    # A match that starts and ends mid-line.
+    result = cs.main(
+        fname,
+        "--ignore-multiline-regex",
+        "codespell:ignore-begin.*codespell:ignore-end",
+        std=True,
+    )
+    assert isinstance(result, tuple)
+    code, stdout, _ = result
+    assert code == 1
+    assert f"{fname}:6: abandonned" in stdout
+
+
 def test_ignore_multiline_regex_option(
     tmp_path: Path,
     capsys: pytest.CaptureFixture[str],