✨ Add ignore patterns (#92)

Co-authored-by: Marcelo Trylesinski <marcelotryle@gmail.com>

Author: lemonyte

bump_pydantic/__main__.py

@@ -1,4 +1,4 @@
-from bump_pydantic.main import app
+from bump_pydantic.main import entrypoint
 
 if __name__ == "__main__":
-    app()
+    entrypoint()

bump_pydantic/glob_helpers.py

@@ -0,0 +1,47 @@
+import fnmatch
+import re
+from pathlib import Path
+from typing import List
+
+MATCH_SEP = r"(?:/|\\)"
+MATCH_SEP_OR_END = r"(?:/|\\|\Z)"
+MATCH_NON_RECURSIVE = r"[^/\\]*"
+MATCH_RECURSIVE = r"(?:.*)"
+
+
+def glob_to_re(pattern: str) -> str:
+    """Translate a glob pattern to a regular expression for matching."""
+    fragments: List[str] = []
+    for segment in re.split(r"/|\\", pattern):
+        if segment == "":
+            continue
+        if segment == "**":
+            # Remove previous separator match, so the recursive match can match zero or more segments.
+            if fragments and fragments[-1] == MATCH_SEP:
+                fragments.pop()
+            fragments.append(MATCH_RECURSIVE)
+        elif "**" in segment:
+            raise ValueError("invalid pattern: '**' can only be an entire path component")
+        else:
+            fragment = fnmatch.translate(segment)
+            fragment = fragment.replace(r"(?s:", r"(?:")
+            fragment = fragment.replace(r".*", MATCH_NON_RECURSIVE)
+            fragment = fragment.replace(r"\Z", r"")
+            fragments.append(fragment)
+        fragments.append(MATCH_SEP)
+    # Remove trailing MATCH_SEP, so it can be replaced with MATCH_SEP_OR_END.
+    if fragments and fragments[-1] == MATCH_SEP:
+        fragments.pop()
+    fragments.append(MATCH_SEP_OR_END)
+    return rf"(?s:{''.join(fragments)})"
+
+
+def match_glob(path: Path, pattern: str) -> bool:
+    """Check if a path matches a glob pattern.
+
+    If the pattern ends with a directory separator, the path must be a directory.
+    """
+    match = bool(re.fullmatch(glob_to_re(pattern), str(path)))
+    if pattern.endswith("/") or pattern.endswith("\\"):
+        return match and path.is_dir()
+    return match

bump_pydantic/main.py

@@ -19,12 +19,17 @@
 from bump_pydantic import __version__
 from bump_pydantic.codemods import Rule, gather_codemods
 from bump_pydantic.codemods.class_def_visitor import ClassDefVisitor
+from bump_pydantic.glob_helpers import match_glob
 
 app = Typer(invoke_without_command=True, add_completion=False)
 
+entrypoint = functools.partial(app, windows_expand_args=False)
+
 P = ParamSpec("P")
 T = TypeVar("T")
 
+DEFAULT_IGNORES = [".venv/**"]
+
 
 def version_callback(value: bool):
     if value:
@@ -37,6 +42,7 @@ def main(
     path: Path = Argument(..., exists=True, dir_okay=True, allow_dash=False),
     disable: List[Rule] = Option(default=[], help="Disable a rule."),
     diff: bool = Option(False, help="Show diff instead of applying changes."),
+    ignore: List[str] = Option(default=DEFAULT_IGNORES, help="Ignore a path glob pattern."),
     log_file: Path = Option("log.txt", help="Log errors to this file."),
     version: bool = Option(
         None,
@@ -57,13 +63,19 @@ def main(
 
     if os.path.isfile(path):
         package = path.parent
-        files = [str(path.relative_to("."))]
+        all_files = [path]
     else:
         package = path
-        files_str = list(package.glob("**/*.py"))
-        files = [str(file.relative_to(".")) for file in files_str]
+        all_files = list(package.glob("**/*.py"))
+
+    filtered_files = [file for file in all_files if not any(match_glob(file, pattern) for pattern in ignore)]
+    files = [str(file.relative_to(".")) for file in filtered_files]
 
-    console.log(f"Found {len(files)} files to process")
+    if files:
+        console.log(f"Found {len(files)} files to process")
+    else:
+        console.log("No files to process.")
+        raise Exit()
 
     providers = {FullyQualifiedNameProvider, ScopeProvider}
     metadata_manager = FullRepoManager(".", files, providers=providers)  # type: ignore[arg-type]

pyproject.toml

@@ -30,7 +30,7 @@ Issues = "https://github.com/pydantic/bump-pydantic/issues"
 Source = "https://github.com/pydantic/bump-pydantic"
 
 [project.scripts]
-bump-pydantic = "bump_pydantic.main:app"
+bump-pydantic = "bump_pydantic.main:entrypoint"
 
 [tool.hatch.version]
 path = "bump_pydantic/__init__.py"

tests/unit/test_glob_helpers.py

@@ -0,0 +1,70 @@
+from __future__ import annotations
+
+from pathlib import Path
+
+import pytest
+
+from bump_pydantic.glob_helpers import glob_to_re, match_glob
+
+
+class TestGlobHelpers:
+    match_glob_values: list[tuple[str, Path, bool]] = [
+        ("foo", Path("foo"), True),
+        ("foo", Path("bar"), False),
+        ("foo", Path("foo/bar"), False),
+        ("*", Path("foo"), True),
+        ("*", Path("bar"), True),
+        ("*", Path("foo/bar"), False),
+        ("**", Path("foo"), True),
+        ("**", Path("foo/bar"), True),
+        ("**", Path("foo/bar/baz/qux"), True),
+        ("foo/bar", Path("foo/bar"), True),
+        ("foo/bar", Path("foo"), False),
+        ("foo/bar", Path("far"), False),
+        ("foo/bar", Path("foo/foo"), False),
+        ("foo/*", Path("foo/bar"), True),
+        ("foo/*", Path("foo/bar/baz"), False),
+        ("foo/*", Path("foo"), False),
+        ("foo/*", Path("bar"), False),
+        ("foo/**", Path("foo/bar"), True),
+        ("foo/**", Path("foo/bar/baz"), True),
+        ("foo/**", Path("foo/bar/baz/qux"), True),
+        ("foo/**", Path("foo"), True),
+        ("foo/**", Path("bar"), False),
+        ("foo/**/bar", Path("foo/bar"), True),
+        ("foo/**/bar", Path("foo/baz/bar"), True),
+        ("foo/**/bar", Path("foo/baz/qux/bar"), True),
+        ("foo/**/bar", Path("foo/baz/qux"), False),
+        ("foo/**/bar", Path("foo/bar/baz"), False),
+        ("foo/**/bar", Path("foo/bar/bar"), True),
+        ("foo/**/bar", Path("foo"), False),
+        ("foo/**/bar", Path("bar"), False),
+        ("foo/**/*/bar", Path("foo/bar"), False),
+        ("foo/**/*/bar", Path("foo/baz/bar"), True),
+        ("foo/**/*/bar", Path("foo/baz/qux/bar"), True),
+        ("foo/**/*/bar", Path("foo/baz/qux"), False),
+        ("foo/**/*/bar", Path("foo/bar/baz"), False),
+        ("foo/**/*/bar", Path("foo/bar/bar"), True),
+        ("foo/**/*/bar", Path("foo"), False),
+        ("foo/**/*/bar", Path("bar"), False),
+        ("foo/ba*", Path("foo/bar"), True),
+        ("foo/ba*", Path("foo/baz"), True),
+        ("foo/ba*", Path("foo/qux"), False),
+        ("foo/ba*", Path("foo/baz/qux"), False),
+        ("foo/ba*", Path("foo/bar/baz"), False),
+        ("foo/ba*", Path("foo"), False),
+        ("foo/ba*", Path("bar"), False),
+        ("foo/**/ba*/*/qux", Path("foo/a/b/c/bar/a/qux"), True),
+        ("foo/**/ba*/*/qux", Path("foo/a/b/c/baz/a/qux"), True),
+        ("foo/**/ba*/*/qux", Path("foo/a/bar/a/qux"), True),
+        ("foo/**/ba*/*/qux", Path("foo/baz/a/qux"), True),
+        ("foo/**/ba*/*/qux", Path("foo/baz/qux"), False),
+        ("foo/**/ba*/*/qux", Path("foo/a/b/c/qux/a/qux"), False),
+        ("foo/**/ba*/*/qux", Path("foo"), False),
+        ("foo/**/ba*/*/qux", Path("bar"), False),
+    ]
+
+    @pytest.mark.parametrize(("pattern", "path", "expected"), match_glob_values)
+    def test_match_glob(self, pattern: str, path: Path, expected: bool):
+        expr = glob_to_re(pattern)
+        assert match_glob(path, pattern) == expected, f"path: {path}, pattern: {pattern}, expr: {expr}"