feat: add support for sorting baseline output

mypy_baseline/_config.py

@@ -29,6 +29,7 @@ class Config:
     depth: int = 40
     allow_unsynced: bool = False
     preserve_position: bool = False
+    sort_baseline: bool = False
     hide_stats: bool = False
     no_colors: bool = bool(os.environ.get('NO_COLOR'))
     ignore: list[str] = dataclasses.field(default_factory=list)
@@ -68,6 +69,10 @@ def init_parser(self, parser: ArgumentParser) -> None:
             '--preserve-position', action='store_true',
             help='do not remove line number from the baseline.',
         )
+        add(
+            '--sort-baseline', action='store_true',
+            help='sort the baseline file.',
+        )
         add(
             '--hide-stats', action='store_true',
             help='do not show stats at the end.',

mypy_baseline/_error.py

@@ -79,13 +79,28 @@ def message(self) -> str:
     def category(self) -> str:
         return self._match.group('category') or 'note'
 
-    def get_clean_line(self, config: Config) -> str:
+    def get_clean_line_components(self, config: Config) -> ErrorLineComponents:
         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('', msg).strip()
         msg = REX_LINE_IN_MSG.sub('defined on line 0', msg)
-        line = f'{path}:{pos}: {self.severity}: {msg}'
+        return ErrorLineComponents(path, pos, self.severity, msg, self.category)
+
+    def get_clean_line(self, config: Config) -> str:
+        return self.get_clean_line_components(config).render()
+
+
+@dataclass(order=True)
+class ErrorLineComponents:
+    path: Path
+    pos: int
+    severity: str
+    msg: str
+    category: str
+
+    def render(self) -> str:
+        line = f'{self.path}:{self.pos}: {self.severity}: {self.msg}'
         if self.category != 'note':
             line += f'  [{self.category}]'
         return line

mypy_baseline/commands/_sync.py

@@ -1,6 +1,6 @@
 from __future__ import annotations
 
-from .._error import Error
+from .._error import Error, ErrorLineComponents
 from ._base import Command
 
 
@@ -15,7 +15,7 @@ def run(self) -> int:
             baseline_text = ''
         old_baseline = baseline_text.splitlines()
 
-        new_baseline: list[str] = []
+        new_baseline: list[ErrorLineComponents] = []
         for line in self.stdin:
             error = Error.new(line)
             if error is None:
@@ -25,14 +25,18 @@ def run(self) -> int:
                 continue
             if self.config.is_ignored_category(error.category):
                 continue
-            clean_line = error.get_clean_line(self.config)
+            clean_line = error.get_clean_line_components(self.config)
             new_baseline.append(clean_line)
 
+        if self.config.sort_baseline:
+            new_baseline.sort()
+        new_baseline_lines = [components.render() for components in new_baseline]
+
         synced = False
         if old_baseline:
-            synced = self._stable_sync(old_baseline, new_baseline)
+            synced = self._stable_sync(old_baseline, new_baseline_lines)
         if not synced:
-            self._write_baseline(new_baseline)
+            self._write_baseline(new_baseline_lines)
         return 0
 
     def _stable_sync(self, old_bline: list[str], new_bline: list[str]) -> bool:
@@ -44,9 +48,13 @@ def _stable_sync(self, old_bline: list[str], new_bline: list[str]) -> bool:
         Currently, we can do a stable sync only when there are no new lines added.
         It's hard to insert new lines in the correct positions, and adding them
         at the end of the file will cause merge conflicts.
-        Sorting lines alphabetically would solve the issue, but I want to keep
-        backward compatibility.
+        Sorting lines solves the issue, so we don't use stable sync when the output
+        is sorted.
+        However, sorting is not enabled by default because I want to keep backward
+        compatibility.
         """
+        if not self.config.sort_baseline:
+            return False
         old_set = set(old_bline)
         new_set = set(new_bline)
         removed = old_set - new_set

tests/test_commands/helpers.py

@@ -8,6 +8,7 @@
 LINE1 = 'views.py:69: error: Hello world  [assignment]\r\n'
 LINE2 = 'settings.py:42: error: How are you?  [union-attr]\r\n'
 LINE3 = 'python/utils.py:15: error: Second argument of Enum() must be string  [misc]\n'
+LINE4 = 'python/utils.py:2: note: this line comes before LINE3\n'
 LINE_WITH_NOTE = 'integrations/services.py:0: note: See https://mypy.readthedocs.io/en/stable/running_mypy.html#missing-imports'  # noqa: E501
 SUCCESS_LINE = 'Success: no issues found in 26 source files'
 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

tests/test_commands/test_sync.py

@@ -5,7 +5,7 @@
 
 from mypy_baseline import main
 
-from .helpers import LINE1, LINE2, LINE3, NOTEBOOK_LINE1
+from .helpers import LINE1, LINE2, LINE3, LINE4, NOTEBOOK_LINE1
 
 
 def test_sync(tmp_path: Path):
@@ -34,3 +34,37 @@ def test_sync_notebook(tmp_path: Path):
     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
+
+
+def test_sync_sorted(tmp_path: Path):
+    blpath = tmp_path / 'bline.txt'
+    stdin = StringIO()
+    stdin.write(LINE1)
+    stdin.write(LINE2)
+    stdin.write(LINE3)
+    stdin.seek(0)
+    code = main(['sync', '--sort-baseline', '--baseline-path', str(blpath)], stdin, StringIO())  # noqa: E501
+    assert code == 0
+    actual = blpath.read_text()
+    line1, line2, line3 = actual.splitlines()
+    assert line1 == 'python/utils.py:0: error: Second argument of Enum() must be string  [misc]'  # noqa: E501s
+    assert line2 == 'settings.py:0: error: How are you?  [union-attr]'
+    assert line3 == 'views.py:0: error: Hello world  [assignment]'
+
+
+def test_sync_sorted_line_numbers(tmp_path: Path):
+    blpath = tmp_path / 'bline.txt'
+    stdin = StringIO()
+    stdin.write(LINE1)
+    stdin.write(LINE2)
+    stdin.write(LINE3)
+    stdin.write(LINE4)
+    stdin.seek(0)
+    code = main(['sync', '--preserve-position', '--sort-baseline', '--baseline-path', str(blpath)], stdin, StringIO())  # noqa: E501
+    assert code == 0
+    actual = blpath.read_text()
+    line1, line2, line3, line4 = actual.splitlines()
+    assert line1 == 'python/utils.py:2: note: this line comes before LINE3'
+    assert line2 == 'python/utils.py:15: error: Second argument of Enum() must be string  [misc]'  # noqa: E501s
+    assert line3 == 'settings.py:42: error: How are you?  [union-attr]'
+    assert line4 == 'views.py:69: error: Hello world  [assignment]'