Merge pull request #15 from damiondoesthings/master

Add support for linting Jupyter notebooks through nbQA

Author: orsinium

docs/sync.md

@@ -9,3 +9,12 @@ mypy | mypy-baseline sync
 The baseline file (`mypy-baseline.txt` by default) contains all errors that mypy spits out with line numbers replaced with zeros and colors removed. All errors recorded in the baseline will be filtered out from the future mypy runs by `mypy-baseline filter`.
 
 The lines in the baseline come in the same order as mypy produeces them. This order is fragile, and the order of files analyzed (and so the lines in the baseline) might change as the dependency graph between modules changes and even be different on different machines. When syncing the changes with an existing baseline, we try to preserve the lines order there, but sometimes it cannot be done, and the lines may jump around. If that causes merge conflicts, simply re-run `sync` instead of trying to fix conflicts manually.
+
+## Jupyter notebooks
+
+All mypy-baseline commands support [nbQA](https://github.com/nbQA-dev/nbQA) for checking [Jupyter notebooks](https://jupyter.org/):
+
+```bash
+python3 -m pip install nbqa
+nbqa mypy . | mypy-baseline sync
+```

mypy_baseline/__init__.py

@@ -4,5 +4,5 @@
 from ._cli import entrypoint, main
 
 
-__version__ = '0.4.5'
+__version__ = '0.5.0'
 __all__ = ['entrypoint', 'main']

mypy_baseline/_error.py

@@ -10,27 +10,41 @@
 if TYPE_CHECKING:
     from ._config import Config
 
-COLOR_PATTERN = '(\x1b\\[\\d*m?|\x0f)*'
-REX_COLOR = re.compile(COLOR_PATTERN)
-REX_LINE = re.compile(rf"""
+REX_COLOR = re.compile('(\x1b\\[\\d*m?|\x0f)*')
+REX_COLOR_NBQA = re.compile(r'\[\d*m|\x1b|\(B')
+REX_LINE = re.compile(r"""
     (?P<path>.+\.py):
     (?P<lineno>[0-9]+):\s
-    {COLOR_PATTERN}(?P<severity>[a-z]+):{COLOR_PATTERN}\s
+    (?P<severity>[a-z]+):\s
     (?P<message>.+?)
-    (?:\s\s{COLOR_PATTERN}\[(?P<category>[a-z-]+)\]{COLOR_PATTERN})?
+    (?:\s\s\[(?P<category>[a-z-]+)\])?
+    \s*
+""", re.VERBOSE | re.MULTILINE)
+REX_LINE_NBQA = re.compile(r"""
+    (?P<path>.+\.ipynb:cell_[0-9]+):
+    (?P<lineno>[0-9]+):\s
+    (?P<severity>[a-z]+):\s
+    (?P<message>.+?)
+    (?:\s\s\[(?P<category>[a-z-]+)\])?
     \s*
 """, re.VERBOSE | re.MULTILINE)
 REX_LINE_IN_MSG = re.compile(r'defined on line \d+')
 
 
+def _remove_color_codes(line: str) -> str:
+    line = REX_COLOR.sub('', line)
+    return REX_COLOR_NBQA.sub('', line)
+
+
 @dataclass
 class Error:
     raw_line: str
     _match: re.Match[str]
 
     @classmethod
     def new(self, line: str) -> Error | None:
-        match = REX_LINE.fullmatch(line)
+        line = _remove_color_codes(line)
+        match = REX_LINE.fullmatch(line) or REX_LINE_NBQA.fullmatch(line)
         if match is None:
             return None
         return Error(line, match)
@@ -61,6 +75,7 @@ def get_clean_line(self, config: Config) -> str:
         path = Path(*self.path.parts[:config.depth])
         pos = self.line_number if config.preserve_position else 0
         msg = REX_COLOR.sub('', self.message).strip()
+        msg = REX_COLOR_NBQA.sub('', self.message).strip()
         msg = REX_LINE_IN_MSG.sub('defined on line 0', msg)
         line = f'{path}:{pos}: {self.severity}: {msg}'
         if self.category != 'note':

tests/test_commands/helpers.py

@@ -8,6 +8,8 @@
 LINE1 = 'views.py:69: error: Hello world  [assignment]\r\n'
 LINE2 = 'settings.py:42: error: How are you?  [union-attr]\r\n'
 
+NOTEBOOK_LINE1 = 'fail.ipynb:cell_1:2: \x1b[1m\x1b[31merror:\x1b(B\x1b[m Incompatible return value type (got \x1b(B\x1b[m\x1b[1m"int"\x1b(B\x1b[m, expected \x1b(B\x1b[m\x1b[1m"str"\x1b(B\x1b[m)  \x1b(B\x1b[m\x1b[33m[return-value]\x1b(B\x1b[m\n'  # noqa: E501
+
 
 def run(cmd: list[str], exit_code: int = 0) -> str:
     stdout = StringIO()

tests/test_commands/test_filter.py

@@ -4,7 +4,7 @@
 
 from mypy_baseline import main
 
-from .helpers import LINE1, LINE2, run
+from .helpers import LINE1, LINE2, NOTEBOOK_LINE1, run
 
 
 def test_filter():
@@ -25,6 +25,22 @@ def test_filter():
     assert 'Your changes introduced' in actual
 
 
+def test_filter_notebook():
+    stdin = StringIO()
+    stdin.write(NOTEBOOK_LINE1)
+
+    stdin.seek(0)
+    stdout = StringIO()
+    code = main(['filter'], stdin, stdout)
+    assert code == 1
+    stdout.seek(0)
+    actual = stdout.read()
+    assert NOTEBOOK_LINE1 in actual
+    assert '  return-value  ' in actual
+    assert '  unresolved' in actual
+    assert 'Your changes introduced' in actual
+
+
 def test_filter__empty_stdin():
     actual = run(['filter'])
     assert actual == ''

tests/test_commands/test_sync.py

@@ -5,7 +5,7 @@
 
 from mypy_baseline import main
 
-from .helpers import LINE1, LINE2
+from .helpers import LINE1, LINE2, NOTEBOOK_LINE1
 
 
 def test_sync(tmp_path: Path):
@@ -20,3 +20,15 @@ def test_sync(tmp_path: Path):
     line1, line2 = actual.splitlines()
     assert line1 == 'views.py:0: error: Hello world  [assignment]'
     assert line2 == 'settings.py:0: error: How are you?  [union-attr]'
+
+
+def test_sync_notebook(tmp_path: Path):
+    blpath = tmp_path / 'bline.txt'
+    stdin = StringIO()
+    stdin.write(NOTEBOOK_LINE1)
+    stdin.seek(0)
+    code = main(['sync', '--baseline-path', str(blpath)], stdin, StringIO())
+    assert code == 0
+    actual = blpath.read_text()
+    line1 = actual.splitlines()[0]
+    assert line1 == 'fail.ipynb:cell_1:0: error: Incompatible return value type (got "int", expected "str")  [return-value]'  # noqa: E501

tests/test_commands/test_top_files.py

@@ -2,7 +2,7 @@
 
 from pathlib import Path
 
-from .helpers import LINE1, LINE2, run
+from .helpers import LINE1, LINE2, NOTEBOOK_LINE1, run
 
 
 def test_top_files(tmp_path: Path):
@@ -12,3 +12,11 @@ def test_top_files(tmp_path: Path):
     actual = run(cmd)
     assert 'views.py' in actual
     assert 'settings.py' in actual
+
+
+def test_top_notebook(tmp_path: Path):
+    blpath = tmp_path / 'bline.txt'
+    blpath.write_text(f'{NOTEBOOK_LINE1}')
+    cmd = ['top-files', '--baseline-path', str(blpath), '--no-color']
+    actual = run(cmd)
+    assert 'fail.ipynb' in actual