Fix #1897 #3300 Add --git-only

Author: mdeweerd

codespell_lib/_codespell.py

@@ -23,6 +23,7 @@
 import itertools
 import os
 import re
+import subprocess
 import sys
 import textwrap
 from typing import (
@@ -620,6 +621,11 @@ def parse_options(
         action="store_true",
         help="output just a single line for each misspelling in stdin mode",
     )
+    parser.add_argument(
+        "--git-only",
+        action="store_true",
+        help="When selected, only check files under git control",
+    )
     parser.add_argument("--config", type=str, help="path to config file.")
     parser.add_argument("--toml", type=str, help="path to a pyproject.toml file.")
     parser.add_argument("files", nargs="*", help="files or directories to check")
@@ -1096,6 +1102,82 @@ def flatten_clean_comma_separated_arguments(
     ]
 
 
+def get_git_tracked_files(
+    root: str, files: Iterable[str], glob_match: GlobMatch, check_hidden: bool
+) -> Iterable[str]:
+    # Flatten the list of files into a single list of arguments for git ls-files
+    file_args = []
+    for filename in files:
+        if os.path.isdir(filename):
+            file_args.append(f"{filename}/**")
+        else:
+            file_args.append(filename)
+
+    # Add the glob patterns to exclude
+    exclude_patterns = [
+        f":(exclude)**/{pattern}" for pattern in glob_match.pattern_list
+    ]
+
+    # Add pattern to exclude hidden files if check_hidden is False
+    if not check_hidden:
+        exclude_patterns.append(":(exclude)**/.*")
+        exclude_patterns.append(":(exclude).*")
+
+    git_executable = "git"  # Could be future option
+
+    try:
+        # ruff: noqa: S603
+        result = subprocess.run(
+            [git_executable, "ls-files", *file_args, *exclude_patterns],
+            cwd=root,
+            capture_output=True,
+            check=True,
+            text=True,
+        )
+        return set(result.stdout.splitlines())
+    except subprocess.CalledProcessError:
+        # If the command fails, assume no files are tracked
+        return set()
+
+
+def build_file_list_with_os_walk(
+    files: Iterable[str], glob_match: GlobMatch, check_hidden: bool
+) -> Iterable[str]:
+    all_files = []
+    for filename in files:
+        # ignore hidden files
+        if is_hidden(filename, check_hidden):
+            continue
+        if os.path.isdir(filename):
+            for root, dirs, dirfiles in os.walk(filename):
+                if glob_match.match(root):  # skip (absolute) directories
+                    dirs.clear()
+                    continue
+                if is_hidden(root, check_hidden):  # dir itself hidden
+                    continue
+                for file_ in dirfiles:
+                    if is_hidden(
+                        file_, check_hidden
+                    ):  # ignore hidden files in directories
+                        continue
+                    if glob_match.match(file_):  # skip files
+                        continue
+                    fname = os.path.join(root, file_)
+                    if glob_match.match(fname):  # skip paths
+                        continue
+                    all_files.append(fname)
+
+                # skip (relative) directories
+                dirs[:] = [
+                    dir_
+                    for dir_ in dirs
+                    if not glob_match.match(dir_) and not is_hidden(dir_, check_hidden)
+                ]
+        elif not glob_match.match(filename) and not is_hidden(filename, check_hidden):
+            all_files.append(filename)
+    return all_files
+
+
 def _script_main() -> int:
     """Wrap to main() for setuptools."""
     try:
@@ -1278,68 +1360,33 @@ def main(*args: str) -> int:
             "try escaping special characters",
         )
 
-    bad_count = 0
-    for filename in sorted(options.files):
-        # ignore hidden files
-        if is_hidden(filename, options.check_hidden):
-            continue
-
-        if os.path.isdir(filename):
-            for root, dirs, files in os.walk(filename):
-                if glob_match.match(root):  # skip (absolute) directories
-                    dirs.clear()
-                    continue
-                if is_hidden(root, options.check_hidden):  # dir itself hidden
-                    continue
-                for file_ in sorted(files):
-                    # ignore hidden files in directories
-                    if is_hidden(file_, options.check_hidden):
-                        continue
-                    if glob_match.match(file_):  # skip files
-                        continue
-                    fname = os.path.join(root, file_)
-                    if glob_match.match(fname):  # skip paths
-                        continue
-                    bad_count += parse_file(
-                        fname,
-                        colors,
-                        summary,
-                        misspellings,
-                        ignore_words_cased,
-                        exclude_lines,
-                        file_opener,
-                        word_regex,
-                        ignore_word_regex,
-                        uri_regex,
-                        uri_ignore_words,
-                        context,
-                        options,
-                    )
-
-                # skip (relative) directories
-                dirs[:] = [
-                    dir_
-                    for dir_ in dirs
-                    if not glob_match.match(dir_)
-                    and not is_hidden(dir_, options.check_hidden)
-                ]
+    # Build the list of all files based on the git_only option
+    if options.git_only:
+        all_files = get_git_tracked_files(
+            os.getcwd(), options.files, glob_match, options.check_hidden
+        )
+    else:
+        all_files = build_file_list_with_os_walk(
+            options.files, glob_match, options.check_hidden
+        )
 
-        elif not glob_match.match(filename):  # skip files
-            bad_count += parse_file(
-                filename,
-                colors,
-                summary,
-                misspellings,
-                ignore_words_cased,
-                exclude_lines,
-                file_opener,
-                word_regex,
-                ignore_word_regex,
-                uri_regex,
-                uri_ignore_words,
-                context,
-                options,
-            )
+    bad_count = 0
+    for filename in sorted(all_files):
+        bad_count += parse_file(
+            filename,
+            colors,
+            summary,
+            misspellings,
+            ignore_words_cased,
+            exclude_lines,
+            file_opener,
+            word_regex,
+            ignore_word_regex,
+            uri_regex,
+            uri_ignore_words,
+            context,
+            options,
+        )
 
     if summary:
         print("\n-------8<-------\nSUMMARY:")

codespell_lib/tests/test_basic.py

@@ -499,6 +499,7 @@ def test_exclude_file(
     bad_name.write_bytes(
         (combinations + "5 abandonned 5\n6 abandonned 6").encode("utf-8")
     )
+
     assert cs.main(bad_name) == 18
     fname = tmp_path / "tmp.txt"
     fname.write_bytes(
@@ -519,6 +520,77 @@ def test_exclude_file(
     assert cs.main("-x", f"{fname_dummy1},{fname},{fname_dummy2}", bad_name) == 1
 
 
+def run_git(path: Path, *args: Union[Path, str]) -> None:
+    subprocess.run(  # noqa: S603
+        ["git", "-C", path, *list(args)],  # noqa: S607
+        capture_output=False,
+        check=True,
+        text=True,
+    )
+
+
+def test_git_only_exclude_file(
+    tmp_path: Path, capsys: pytest.CaptureFixture[str], monkeypatch: pytest.MonkeyPatch
+) -> None:
+    monkeypatch.chdir(tmp_path)
+    """Test exclude file functionality."""
+    bad_name = tmp_path / "bad.txt"
+    # check all possible combinations of lines to ignore and ignores
+    combinations = "".join(
+        f"{n} abandonned {n}\n"
+        f"{n} abandonned {n}\r\n"
+        f"{n} abandonned {n} \n"
+        f"{n} abandonned {n} \r\n"
+        for n in range(1, 5)
+    )
+    bad_name.write_bytes(
+        (combinations + "5 abandonned 5\n6 abandonned 6").encode("utf-8")
+    )
+
+    run_git(tmp_path, "init")
+    run_git(tmp_path, "add", bad_name)
+
+    assert cs.main(bad_name) == 18
+    fname = tmp_path / "tmp.txt"
+    fname.write_bytes(
+        b"1 abandonned 1\n"
+        b"2 abandonned 2\r\n"
+        b"3 abandonned 3 \n"
+        b"4 abandonned 4 \r\n"
+        b"6 abandonned 6\n"
+    )
+
+    # Not adding fname to git to exclude it
+
+    # Should have 23 total errors (bad_name + fname)
+    assert cs.main(tmp_path) == 23
+
+    # Before adding to git, should not report on fname, only 18 error in bad.txt
+    assert cs.main("--git-only", tmp_path) == 18
+    run_git(tmp_path, "add", fname)
+    assert cs.main(tmp_path) == 23
+    # After adding to git, should report on fname
+    assert cs.main("--git-only", tmp_path) == 23
+    # After adding to git, should not report on excluded file
+    assert cs.main("--git-only", "-x", fname, tmp_path) == 1
+    # comma-separated list of files
+    fname_dummy1 = tmp_path / "dummy1.txt"
+    fname_dummy1.touch()
+    fname_dummy2 = tmp_path / "dummy2.txt"
+    fname_dummy2.touch()
+    run_git(tmp_path, "add", fname_dummy1, fname_dummy2)
+    assert (
+        cs.main(
+            "--git-only", "-x", fname_dummy1, "-x", fname, "-x", fname_dummy2, bad_name
+        )
+        == 1
+    )
+    assert (
+        cs.main("--git-only", "-x", f"{fname_dummy1},{fname},{fname_dummy2}", bad_name)
+        == 1
+    )
+
+
 def test_encoding(
     tmp_path: Path,
     capsys: pytest.CaptureFixture[str],
@@ -636,6 +708,108 @@ def test_check_filename_irregular_file(
     assert cs.main("-f", tmp_path) == 1
 
 
+def test_check_hidden_git(
+    tmp_path: Path,
+    capsys: pytest.CaptureFixture[str],
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    """Test ignoring of hidden files."""
+    monkeypatch.chdir(tmp_path)
+    run_git(tmp_path, "init")
+    # visible file
+    #
+    #         tmp_path
+    #         └── test.txt
+    #
+    fname = tmp_path / "test.txt"
+    fname.write_text("erorr\n")
+    run_git(tmp_path, "add", ".")
+    assert cs.main("--git-only", fname) == 1
+    assert cs.main("--git-only", tmp_path) == 1
+
+    # hidden file
+    #
+    #         tmp_path
+    #         └── .test.txt
+    #
+    hidden_file = tmp_path / ".test.txt"
+    fname.rename(hidden_file)
+    run_git(tmp_path, "add", ".")
+    assert cs.main("--git-only", hidden_file) == 0
+    assert cs.main("--git-only", tmp_path) == 0
+    assert cs.main("--git-only", "--check-hidden", hidden_file) == 1
+    assert cs.main("--git-only", "--check-hidden", tmp_path) == 1
+
+    # hidden file with typo in name
+    #
+    #         tmp_path
+    #         └── .abandonned.txt
+    #
+    typo_file = tmp_path / ".abandonned.txt"
+    hidden_file.rename(typo_file)
+    run_git(tmp_path, "add", ".")
+    assert cs.main("--git-only", typo_file) == 0
+    assert cs.main("--git-only", tmp_path) == 0
+    assert cs.main("--git-only", "--check-hidden", typo_file) == 1
+    assert cs.main("--git-only", "--check-hidden", tmp_path) == 1
+    assert cs.main("--git-only", "--check-hidden", "--check-filenames", typo_file) == 2
+    assert cs.main("--git-only", "--check-hidden", "--check-filenames", tmp_path) == 2
+
+    # hidden directory
+    #
+    #         tmp_path
+    #         ├── .abandonned
+    #         │   ├── .abandonned.txt
+    #         │   └── subdir
+    #         │       └── .abandonned.txt
+    #         └── .abandonned.txt
+    #
+    assert cs.main("--git-only", tmp_path) == 0
+    assert cs.main("--git-only", "--check-hidden", tmp_path) == 1
+    assert cs.main("--git-only", "--check-hidden", "--check-filenames", tmp_path) == 2
+    hidden = tmp_path / ".abandonned"
+    hidden.mkdir()
+    copyfile(typo_file, hidden / typo_file.name)
+    subdir = hidden / "subdir"
+    subdir.mkdir()
+    copyfile(typo_file, subdir / typo_file.name)
+    run_git(tmp_path, "add", ".")
+    assert cs.main("--git-only", tmp_path) == 0
+    assert cs.main("--git-only", "--check-hidden", tmp_path) == 3
+    assert cs.main("--git-only", "--check-hidden", "--check-filenames", tmp_path) == 8
+    # check again with a relative path
+    try:
+        rel = op.relpath(tmp_path)
+    except ValueError:
+        # Windows: path is on mount 'C:', start on mount 'D:'
+        pass
+    else:
+        assert cs.main("--git-only", rel) == 0
+        assert cs.main("--git-only", "--check-hidden", rel) == 3
+        assert cs.main("--git-only", "--check-hidden", "--check-filenames", rel) == 8
+
+    # hidden subdirectory
+    #
+    #         tmp_path
+    #         ├── .abandonned
+    #         │   ├── .abandonned.txt
+    #         │   └── subdir
+    #         │       └── .abandonned.txt
+    #         ├── .abandonned.txt
+    #         └── subdir
+    #             └── .abandonned
+    #                 └── .abandonned.txt
+    subdir = tmp_path / "subdir"
+    subdir.mkdir()
+    hidden = subdir / ".abandonned"
+    hidden.mkdir()
+    copyfile(typo_file, hidden / typo_file.name)
+    run_git(tmp_path, "add", ".")
+    assert cs.main("--git-only", tmp_path) == 0
+    assert cs.main("--git-only", "--check-hidden", tmp_path) == 4
+    assert cs.main("--git-only", "--check-hidden", "--check-filenames", tmp_path) == 11
+
+
 def test_check_hidden(
     tmp_path: Path,
     capsys: pytest.CaptureFixture[str],