pyclean.modern.normalize() - Code Metrics - Inspection of "[WIP] Add -g, --git-clean flag to execute git clea..." - bittner/pyclean - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — main (#105)

unknown

created 2025-11-04 19:09 UTC

pyclean.modern.normalize() A

↳ Parent: pyclean.modern

Complexity

Conditions

Size

Total Lines	8
Code Lines	2

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	1
eloc	2
nop	1
dl	0
loc	8
rs	10
c	0
b	0
f	0

# SPDX-FileCopyrightText: 2020 Peter Bittner <[email protected]>
#
# SPDX-License-Identifier: GPL-3.0-or-later

"""
Modern, cross-platform, pure-Python pyclean implementation.
"""

import logging
import os
import subprocess
from pathlib import Path

BYTECODE_FILES = ['.pyc', '.pyo']
BYTECODE_DIRS = ['__pycache__']
DEBRIS_TOPICS = {
    'cache': [
        '.cache/**/*',
        '.cache/',
    ],
    'coverage': [
        '.coverage',
        'coverage.json',
        'coverage.lcov',
        'coverage.xml',
        'htmlcov/**/*',
        'htmlcov/',
    ],
    'jupyter': [
        '.ipynb_checkpoints/**/*',
        '.ipynb_checkpoints/',
    ],
    'mypy': [
        '.mypy_cache/**/*',
        '.mypy_cache/',
    ],
    'package': [
        'build/bdist.*/**/*',
        'build/bdist.*/',
        'build/lib/**/*',
        'build/lib/',
        'build/',
        'dist/**/*',
        'dist/',
        'sdist/**/*',
        'sdist/',
        '*.egg-info/**/*',
        '*.egg-info/',
    ],
    'pytest': [
        '.pytest_cache/**/*',
        '.pytest_cache/',
        'pytestdebug.log',
    ],
    'ruff': [
        '.ruff_cache/**/*',
        '.ruff_cache/',
    ],
    'tox': [
        '.tox/**/*',
        '.tox/',
    ],
}


class CleanupRunner:
    """Module-level configuration and value store."""

    def __init__(self):
        """Cleanup runner with optional dry-run behavior."""
        self.unlink = None
        self.rmdir = None
        self.ignore = None
        self.unlink_count = None
        self.unlink_failed = None
        self.rmdir_count = None
        self.rmdir_failed = None

    def configure(self, args):
        """Set up runner according to command line options."""
        self.unlink = print_filename if args.dry_run else remove_file
        self.rmdir = print_dirname if args.dry_run else remove_directory
        self.ignore = args.ignore
        self.unlink_count = 0
        self.unlink_failed = 0
        self.rmdir_count = 0
        self.rmdir_failed = 0


log = logging.getLogger(__name__)
Runner = CleanupRunner()


def normalize(path_pattern: str) -> str:
    """
    Normalize path separators in a pattern for cross-platform support.

    On Windows, both forward slash and backslash are valid path separators.
    On Unix/Posix, only forward slash is valid (backslash can be part of filename).
    """
    return path_pattern.replace(os.sep, os.altsep or os.sep)


def should_ignore(path: Path, ignore_patterns: list[str]) -> bool:
    """
    Check if a path should be ignored based on ignore patterns.

    Patterns can be:
    - Simple names like 'bar': matches any directory with that name
    - Paths like 'foo/bar': matches 'bar' directory inside 'foo' directory
      and also ignores everything inside that directory
    """
    if not ignore_patterns:
        return False

    for pattern in ignore_patterns:
        # Check if pattern has multiple components (is a path with separators)
        pattern_parts = Path(normalize(pattern)).parts
        if len(pattern_parts) > 1:
            # Pattern contains path separator - match relative path
            # Path must have at least as many parts as the pattern
            if len(path.parts) < len(pattern_parts):
                continue
            # Check if pattern matches anywhere in the path hierarchy
            for i in range(len(path.parts) - len(pattern_parts) + 1):
                path_slice = path.parts[i : i + len(pattern_parts)]
                if path_slice == pattern_parts:
                    return True
        # Simple name - match the directory name anywhere
        elif path.name == pattern:
            return True
    return False


def remove_file(fileobj):
    """Attempt to delete a file object for real."""
    log.debug('Deleting file: %s', fileobj)
    try:
        fileobj.unlink()
        Runner.unlink_count += 1
    except OSError as err:
        log.debug('File not deleted. %s', err)
        Runner.unlink_failed += 1


def remove_directory(dirobj):
    """Attempt to remove a directory object for real."""
    log.debug('Removing directory: %s', dirobj)
    try:
        dirobj.rmdir()
        Runner.rmdir_count += 1
    except OSError as err:
        log.debug('Directory not removed. %s', err)
        Runner.rmdir_failed += 1


def print_filename(fileobj):
    """Only display the file name, used with --dry-run."""
    log.debug('Would delete file: %s', fileobj)
    Runner.unlink_count += 1


def print_dirname(dirobj):
    """Only display the directory name, used with --dry-run."""
    log.debug('Would delete directory: %s', dirobj)
    Runner.rmdir_count += 1


def pyclean(args):
    """Cross-platform cleaning of Python bytecode."""
    Runner.configure(args)
    
    # Ensure git_clean attribute exists (for tests that create args manually)
    if not hasattr(args, 'git_clean'):
        args.git_clean = False

    for dir_name in args.directory:
        dir_path = Path(dir_name)

        log.info('Cleaning directory %s', dir_path)
        descend_and_clean(dir_path, BYTECODE_FILES, BYTECODE_DIRS)

        for topic in args.debris:
            remove_debris_for(topic, dir_path)

        remove_freeform_targets(dir_path, args.erase, args.yes, args.dry_run)

        if args.folders:
            log.debug('Removing empty directories...')
            remove_empty_directories(dir_path)

    log.info(
        'Total %d files, %d directories %s.',
        Runner.unlink_count,
        Runner.rmdir_count,
        'would be removed' if args.dry_run else 'removed',
    )

    if Runner.unlink_failed or Runner.rmdir_failed:
        log.debug(
            '%d files, %d directories %s not be removed.',
            Runner.unlink_failed,
            Runner.rmdir_failed,
            'would' if args.dry_run else 'could',
        )

    # Suggest --debris option if it wasn't used
    if not args.debris:
        suggest_debris_option(args)

    # Run git clean as the very last step if requested
    if args.git_clean:
        log.info('Running git clean...')
        for dir_name in args.directory:
            dir_path = Path(dir_name)
            exit_code = run_git_clean(
                dir_path, 
                args.ignore, 
                dry_run=args.dry_run,
                force=args.yes
            )
            
            if exit_code == 128:
                # Directory is not under version control
                log.warning(
                    'Directory %s is not under version control. Skipping git clean.',
                    dir_path,
                )
            elif exit_code != 0:
                # Other git errors should cause immediate exit
                raise SystemExit(exit_code)


def descend_and_clean(directory, file_types, dir_names):
    """
    Walk and descend a directory tree, cleaning up files of a certain type
    along the way. Only delete directories if they are empty, in the end.
    """
    for child in sorted(directory.iterdir()):
        if child.is_file():
            if child.suffix in file_types:
                Runner.unlink(child)
        elif child.is_dir():
            if should_ignore(child, Runner.ignore):
                log.debug('Skipping %s', child)
            else:
                descend_and_clean(child, file_types, dir_names)

            if child.name in dir_names:
                Runner.rmdir(child)
        else:
            log.debug('Ignoring %s (neither a file nor a folder)', child)


def remove_debris_for(topic, directory):
    """
    Clean up debris for a specific topic.
    """
    log.debug('Scanning for debris of %s ...', topic.title())

    patterns = DEBRIS_TOPICS[topic]
    recursive_delete_debris(directory, patterns)


def remove_empty_directories(directory):
    """
    Recursively remove empty directories in the given directory tree.

    This walks the directory tree in post-order (bottom-up), attempting to
    remove directories that are empty.
    """
    try:
        subdirs = [
            Path(entry.path) for entry in os.scandir(directory) if entry.is_dir()
        ]
    except (OSError, PermissionError) as err:
        log.warning('Cannot access directory %s: %s', directory, err)
        return

    for subdir in subdirs:
        if should_ignore(subdir, Runner.ignore):
            log.debug('Skipping %s', subdir)
        else:
            remove_empty_directories(subdir)  # recurse down the hierarchy
            try:
                if next(subdir.iterdir(), None) is None:
                    Runner.rmdir(subdir)
            except (OSError, PermissionError) as err:
                log.debug('Cannot check or remove directory %s: %s', subdir, err)


def remove_freeform_targets(directory, glob_patterns, yes, dry_run=False):
    """
    Remove free-form targets using globbing.

    This is **potentially dangerous** since users can delete everything
    anywhere in their file system, including the entire project they're
    working on. For this reason, the implementation imposes the following
    (user experience-related) restrictions:

    - Deleting (directories) is not recursive, directory contents must be
      explicitly specified using globbing (e.g. ``dirname/**/*``).
    - The user is responsible for the deletion order, so that a directory
      is empty when it is attempted to be deleted.
    - A confirmation prompt for the deletion of every single file system
      object is shown (unless the ``--yes`` option is used, in addition).
    """
    for path_glob in glob_patterns:
        log.debug('Erase file system objects matching: %s', path_glob)
        delete_filesystem_objects(directory, path_glob, prompt=not yes, dry_run=dry_run)


def recursive_delete_debris(directory, patterns):
    """
    Recursively delete debris matching any of the given patterns.

    This function walks the directory tree once and applies all patterns
    at each level, avoiding redundant directory scans.
    """
    for pattern in patterns:
        delete_filesystem_objects(directory, pattern)

    try:
        subdirs = (
            Path(entry.path) for entry in os.scandir(directory) if entry.is_dir()

        )
    except (OSError, PermissionError) as err:
        log.warning('Cannot access directory %s: %s', directory, err)
        return

    for subdir in subdirs:
        if should_ignore(subdir, Runner.ignore):
            log.debug('Skipping %s', subdir)
        else:
            recursive_delete_debris(subdir, patterns)


def delete_filesystem_objects(directory, path_glob, prompt=False, dry_run=False):
    """
    Identifies all pathnames matching a specific glob pattern, and attempts
    to delete them in the proper order, optionally asking for confirmation.

    Implementation Note: We sort the file system objects in *reverse order*
    and first delete *all files* before removing directories. This way we
    make sure that the directories that are deepest down in the hierarchy
    are empty (for both files & directories) when we attempt to remove them.
    """
    all_names = sorted(directory.glob(path_glob), reverse=True)
    dirs = (name for name in all_names if name.is_dir() and not name.is_symlink())

    files = (name for name in all_names if not name.is_dir() or name.is_symlink())

    for file_object in files:
        file_type = 'symlink' if file_object.is_symlink() else 'file'
        if (
            not dry_run
            and prompt
            and not confirm('Delete %s %s' % (file_type, file_object))
        ):
            Runner.unlink_failed += 1
            continue
        Runner.unlink(file_object)

    for dir_object in dirs:
        if (
            not dry_run
            and prompt
            and not confirm('Remove empty directory %s' % dir_object)
        ):
            Runner.rmdir_failed += 1
            continue
        Runner.rmdir(dir_object)


def confirm(message):
    """An interactive confirmation prompt."""
    try:
        answer = input('%s? ' % message)
        return answer.strip().lower() in ['y', 'yes']
    except KeyboardInterrupt:
        msg = 'Aborted by user.'
        raise SystemExit(msg)


def detect_debris_in_directory(directory):
    """
    Scan a directory for debris artifacts and return a list of detected topics.
    """
    detected_topics = []

    for topic, patterns in DEBRIS_TOPICS.items():
        for pattern in patterns:
            # Skip patterns that are for recursive cleanup (contain **)
            if '**' in pattern:
                continue
            # Check if the pattern matches anything in the directory
            matches = list(directory.glob(pattern))
            if matches:
                detected_topics.append(topic)
                break  # Found at least one match for this topic, move to next

    return detected_topics


def suggest_debris_option(args):
    """
    Suggest using the --debris option when it wasn't used.
    Optionally provide targeted suggestions based on detected artifacts.
    """
    # Collect all detected debris topics across all directories
    all_detected = set()
    for dir_name in args.directory:
        dir_path = Path(dir_name)
        if dir_path.exists():
            detected = detect_debris_in_directory(dir_path)
            all_detected.update(detected)

    if all_detected:
        # Provide targeted suggestion
        topics_str = ' '.join(sorted(all_detected))
        log.info(
            'Hint: Use --debris to also clean up build artifacts. Detected: %s',
            topics_str,
        )
    else:
        # Provide general suggestion
        log.info(
            'Hint: Use --debris to also clean up build artifacts '
            'from common Python development tools.',
        )


def run_git_clean(directory, ignore_patterns, dry_run=False, force=False):
    """
    Run git clean in the specified directory with appropriate flags.

    Args:
        directory: Path to the directory to clean
        ignore_patterns: List of patterns to exclude from git clean
        dry_run: If True, only show what would be deleted
        force: If True, delete without prompting (otherwise interactive)

    Returns:
        Exit code from git clean
    """
    # Build git clean command
    cmd = ['git', 'clean', '-dx']

    # Add exclude patterns for ignored directories
    for pattern in ignore_patterns:
        cmd.extend(['-e', pattern])

    # Add mode flag: -n for dry-run, -f for force, -i for interactive
    if dry_run:
        cmd.append('-n')
    elif force:
        cmd.append('-f')
    else:
        cmd.append('-i')

    log.debug('Running: %s in %s', ' '.join(cmd), directory)

    try:
        result = subprocess.run(
            cmd,
            cwd=directory,
            capture_output=True,
            text=True,
            check=False,
        )

        # Print output from git clean
        if result.stdout:
            print(result.stdout, end='')
        if result.stderr:
            print(result.stderr, end='')

        return result.returncode

    except FileNotFoundError:
        log.error('Git command not found. Cannot execute git clean.')
        return 127


1			# SPDX-FileCopyrightText: 2020 Peter Bittner <[email protected]>
2			#
3			# SPDX-License-Identifier: GPL-3.0-or-later
4
5			"""
6			Modern, cross-platform, pure-Python pyclean implementation.
7			"""
8
9			import logging
10			import os
11			import subprocess
12			from pathlib import Path
13
14			BYTECODE_FILES = ['.pyc', '.pyo']
15			BYTECODE_DIRS = ['__pycache__']
16			DEBRIS_TOPICS = {
17			'cache': [
18			'.cache/*/',
19			'.cache/',
20			],
21			'coverage': [
22			'.coverage',
23			'coverage.json',
24			'coverage.lcov',
25			'coverage.xml',
26			'htmlcov/*/',
27			'htmlcov/',
28			],
29			'jupyter': [
30			'.ipynb_checkpoints/*/',
31			'.ipynb_checkpoints/',
32			],
33			'mypy': [
34			'.mypy_cache/*/',
35			'.mypy_cache/',
36			],
37			'package': [
38			'build/bdist.//',
39			'build/bdist.*/',
40			'build/lib/*/',
41			'build/lib/',
42			'build/',
43			'dist/*/',
44			'dist/',
45			'sdist/*/',
46			'sdist/',
47			'.egg-info//',
48			'*.egg-info/',
49			],
50			'pytest': [
51			'.pytest_cache/*/',
52			'.pytest_cache/',
53			'pytestdebug.log',
54			],
55			'ruff': [
56			'.ruff_cache/*/',
57			'.ruff_cache/',
58			],
59			'tox': [
60			'.tox/*/',
61			'.tox/',
62			],
63			}
64
65
66			class CleanupRunner:
67			"""Module-level configuration and value store."""
68
69			def __init__(self):
70			"""Cleanup runner with optional dry-run behavior."""
71			self.unlink = None
72			self.rmdir = None
73			self.ignore = None
74			self.unlink_count = None
75			self.unlink_failed = None
76			self.rmdir_count = None
77			self.rmdir_failed = None
78
79			def configure(self, args):
80			"""Set up runner according to command line options."""
81			self.unlink = print_filename if args.dry_run else remove_file
82			self.rmdir = print_dirname if args.dry_run else remove_directory
83			self.ignore = args.ignore
84			self.unlink_count = 0
85			self.unlink_failed = 0
86			self.rmdir_count = 0
87			self.rmdir_failed = 0
88
89
90			log = logging.getLogger(__name__)
91			Runner = CleanupRunner()
92
93
94			def normalize(path_pattern: str) -> str:
95			"""
96			Normalize path separators in a pattern for cross-platform support.
97
98			On Windows, both forward slash and backslash are valid path separators.
99			On Unix/Posix, only forward slash is valid (backslash can be part of filename).
100			"""
101			return path_pattern.replace(os.sep, os.altsep or os.sep)
102
103
104			def should_ignore(path: Path, ignore_patterns: list[str]) -> bool:
105			"""
106			Check if a path should be ignored based on ignore patterns.
107
108			Patterns can be:
109			- Simple names like 'bar': matches any directory with that name
110			- Paths like 'foo/bar': matches 'bar' directory inside 'foo' directory
111			and also ignores everything inside that directory
112			"""
113			if not ignore_patterns:
114			return False
115
116			for pattern in ignore_patterns:
117			# Check if pattern has multiple components (is a path with separators)
118			pattern_parts = Path(normalize(pattern)).parts
119			if len(pattern_parts) > 1:
120			# Pattern contains path separator - match relative path
121			# Path must have at least as many parts as the pattern
122			if len(path.parts) < len(pattern_parts):
123			continue
124			# Check if pattern matches anywhere in the path hierarchy
125			for i in range(len(path.parts) - len(pattern_parts) + 1):
126			path_slice = path.parts[i : i + len(pattern_parts)]
127			if path_slice == pattern_parts:
128			return True
129			# Simple name - match the directory name anywhere
130			elif path.name == pattern:
131			return True
132			return False
133
134
135			def remove_file(fileobj):
136			"""Attempt to delete a file object for real."""
137			log.debug('Deleting file: %s', fileobj)
138			try:
139			fileobj.unlink()
140			Runner.unlink_count += 1
141			except OSError as err:
142			log.debug('File not deleted. %s', err)
143			Runner.unlink_failed += 1
144
145
146			def remove_directory(dirobj):
147			"""Attempt to remove a directory object for real."""
148			log.debug('Removing directory: %s', dirobj)
149			try:
150			dirobj.rmdir()
151			Runner.rmdir_count += 1
152			except OSError as err:
153			log.debug('Directory not removed. %s', err)
154			Runner.rmdir_failed += 1
155
156
157			def print_filename(fileobj):
158			"""Only display the file name, used with --dry-run."""
159			log.debug('Would delete file: %s', fileobj)
160			Runner.unlink_count += 1
161
162
163			def print_dirname(dirobj):
164			"""Only display the directory name, used with --dry-run."""
165			log.debug('Would delete directory: %s', dirobj)
166			Runner.rmdir_count += 1
167
168
169			def pyclean(args):
170			"""Cross-platform cleaning of Python bytecode."""
171			Runner.configure(args)
172
173			# Ensure git_clean attribute exists (for tests that create args manually)
174			if not hasattr(args, 'git_clean'):
175			args.git_clean = False
176
177			for dir_name in args.directory:
178			dir_path = Path(dir_name)
179
180			log.info('Cleaning directory %s', dir_path)
181			descend_and_clean(dir_path, BYTECODE_FILES, BYTECODE_DIRS)
182
183			for topic in args.debris:
184			remove_debris_for(topic, dir_path)
185
186			remove_freeform_targets(dir_path, args.erase, args.yes, args.dry_run)
187
188			if args.folders:
189			log.debug('Removing empty directories...')
190			remove_empty_directories(dir_path)
191
192			log.info(
193			'Total %d files, %d directories %s.',
194			Runner.unlink_count,
195			Runner.rmdir_count,
196			'would be removed' if args.dry_run else 'removed',
197			)
198
199			if Runner.unlink_failed or Runner.rmdir_failed:
200			log.debug(
201			'%d files, %d directories %s not be removed.',
202			Runner.unlink_failed,
203			Runner.rmdir_failed,
204			'would' if args.dry_run else 'could',
205			)
206
207			# Suggest --debris option if it wasn't used
208			if not args.debris:
209			suggest_debris_option(args)
210
211			# Run git clean as the very last step if requested
212			if args.git_clean:
213			log.info('Running git clean...')
214			for dir_name in args.directory:
215			dir_path = Path(dir_name)
216			exit_code = run_git_clean(
217			dir_path,
218			args.ignore,
219			dry_run=args.dry_run,
220			force=args.yes
221			)
222
223			if exit_code == 128:
224			# Directory is not under version control
225			log.warning(
226			'Directory %s is not under version control. Skipping git clean.',
227			dir_path,
228			)
229			elif exit_code != 0:
230			# Other git errors should cause immediate exit
231			raise SystemExit(exit_code)
232
233
234			def descend_and_clean(directory, file_types, dir_names):
235			"""
236			Walk and descend a directory tree, cleaning up files of a certain type
237			along the way. Only delete directories if they are empty, in the end.
238			"""
239			for child in sorted(directory.iterdir()):
240			if child.is_file():
241			if child.suffix in file_types:
242			Runner.unlink(child)
243			elif child.is_dir():
244			if should_ignore(child, Runner.ignore):
245			log.debug('Skipping %s', child)
246			else:
247			descend_and_clean(child, file_types, dir_names)
248
249			if child.name in dir_names:
250			Runner.rmdir(child)
251			else:
252			log.debug('Ignoring %s (neither a file nor a folder)', child)
253
254
255			def remove_debris_for(topic, directory):
256			"""
257			Clean up debris for a specific topic.
258			"""
259			log.debug('Scanning for debris of %s ...', topic.title())
260
261			patterns = DEBRIS_TOPICS[topic]
262			recursive_delete_debris(directory, patterns)
263
264
265			def remove_empty_directories(directory):
266			"""
267			Recursively remove empty directories in the given directory tree.
268
269			This walks the directory tree in post-order (bottom-up), attempting to
270			remove directories that are empty.
271			"""
272			try:
273			subdirs = [
274			Path(entry.path) for entry in os.scandir(directory) if entry.is_dir()
275			]
276			except (OSError, PermissionError) as err:
277			log.warning('Cannot access directory %s: %s', directory, err)
278			return
279
280			for subdir in subdirs:
281			if should_ignore(subdir, Runner.ignore):
282			log.debug('Skipping %s', subdir)
283			else:
284			remove_empty_directories(subdir) # recurse down the hierarchy
285			try:
286			if next(subdir.iterdir(), None) is None:
287			Runner.rmdir(subdir)
288			except (OSError, PermissionError) as err:
289			log.debug('Cannot check or remove directory %s: %s', subdir, err)
290
291
292			def remove_freeform_targets(directory, glob_patterns, yes, dry_run=False):
293			"""
294			Remove free-form targets using globbing.
295
296			This is potentially dangerous since users can delete everything
297			anywhere in their file system, including the entire project they're
298			working on. For this reason, the implementation imposes the following
299			(user experience-related) restrictions:
300
301			- Deleting (directories) is not recursive, directory contents must be
302			explicitly specified using globbing (e.g. ``dirname/*/``).
303			- The user is responsible for the deletion order, so that a directory
304			is empty when it is attempted to be deleted.
305			- A confirmation prompt for the deletion of every single file system
306			object is shown (unless the ``--yes`` option is used, in addition).
307			"""
308			for path_glob in glob_patterns:
309			log.debug('Erase file system objects matching: %s', path_glob)
310			delete_filesystem_objects(directory, path_glob, prompt=not yes, dry_run=dry_run)
311
312
313			def recursive_delete_debris(directory, patterns):
314			"""
315			Recursively delete debris matching any of the given patterns.
316
317			This function walks the directory tree once and applies all patterns
318			at each level, avoiding redundant directory scans.
319			"""
320			for pattern in patterns:
321			delete_filesystem_objects(directory, pattern)
322
323			try:
324			subdirs = (
325			Path(entry.path) for entry in os.scandir(directory) if entry.is_dir()
			0 ignored issues – show Comprehensibility Best Practice introduced 2025-10-16 12:00 UTC by Report Bug Copy Issue Report The variable `entry` does not seem to be defined. Loading history...
326			)
327			except (OSError, PermissionError) as err:
328			log.warning('Cannot access directory %s: %s', directory, err)
329			return
330
331			for subdir in subdirs:
332			if should_ignore(subdir, Runner.ignore):
333			log.debug('Skipping %s', subdir)
334			else:
335			recursive_delete_debris(subdir, patterns)
336
337
338			def delete_filesystem_objects(directory, path_glob, prompt=False, dry_run=False):
339			"""
340			Identifies all pathnames matching a specific glob pattern, and attempts
341			to delete them in the proper order, optionally asking for confirmation.
342
343			Implementation Note: We sort the file system objects in reverse order
344			and first delete all files before removing directories. This way we
345			make sure that the directories that are deepest down in the hierarchy
346			are empty (for both files & directories) when we attempt to remove them.
347			"""
348			all_names = sorted(directory.glob(path_glob), reverse=True)
349			dirs = (name for name in all_names if name.is_dir() and not name.is_symlink())
			0 ignored issues – show Comprehensibility Best Practice introduced 2023-01-27 09:22 UTC by Report Bug Copy Issue Report The variable `name` does not seem to be defined. Loading history...
350			files = (name for name in all_names if not name.is_dir() or name.is_symlink())
351
352			for file_object in files:
353			file_type = 'symlink' if file_object.is_symlink() else 'file'
354			if (
355			not dry_run
356			and prompt
357			and not confirm('Delete %s %s' % (file_type, file_object))
358			):
359			Runner.unlink_failed += 1
360			continue
361			Runner.unlink(file_object)
362
363			for dir_object in dirs:
364			if (
365			not dry_run
366			and prompt
367			and not confirm('Remove empty directory %s' % dir_object)
368			):
369			Runner.rmdir_failed += 1
370			continue
371			Runner.rmdir(dir_object)
372
373
374			def confirm(message):
375			"""An interactive confirmation prompt."""
376			try:
377			answer = input('%s? ' % message)
378			return answer.strip().lower() in ['y', 'yes']
379			except KeyboardInterrupt:
380			msg = 'Aborted by user.'
381			raise SystemExit(msg)
382
383
384			def detect_debris_in_directory(directory):
385			"""
386			Scan a directory for debris artifacts and return a list of detected topics.
387			"""
388			detected_topics = []
389
390			for topic, patterns in DEBRIS_TOPICS.items():
391			for pattern in patterns:
392			# Skip patterns that are for recursive cleanup (contain **)
393			if '**' in pattern:
394			continue
395			# Check if the pattern matches anything in the directory
396			matches = list(directory.glob(pattern))
397			if matches:
398			detected_topics.append(topic)
399			break # Found at least one match for this topic, move to next
400
401			return detected_topics
402
403
404			def suggest_debris_option(args):
405			"""
406			Suggest using the --debris option when it wasn't used.
407			Optionally provide targeted suggestions based on detected artifacts.
408			"""
409			# Collect all detected debris topics across all directories
410			all_detected = set()
411			for dir_name in args.directory:
412			dir_path = Path(dir_name)
413			if dir_path.exists():
414			detected = detect_debris_in_directory(dir_path)
415			all_detected.update(detected)
416
417			if all_detected:
418			# Provide targeted suggestion
419			topics_str = ' '.join(sorted(all_detected))
420			log.info(
421			'Hint: Use --debris to also clean up build artifacts. Detected: %s',
422			topics_str,
423			)
424			else:
425			# Provide general suggestion
426			log.info(
427			'Hint: Use --debris to also clean up build artifacts '
428			'from common Python development tools.',
429			)
430
431
432			def run_git_clean(directory, ignore_patterns, dry_run=False, force=False):
433			"""
434			Run git clean in the specified directory with appropriate flags.
435
436			Args:
437			directory: Path to the directory to clean
438			ignore_patterns: List of patterns to exclude from git clean
439			dry_run: If True, only show what would be deleted
440			force: If True, delete without prompting (otherwise interactive)
441
442			Returns:
443			Exit code from git clean
444			"""
445			# Build git clean command
446			cmd = ['git', 'clean', '-dx']
447
448			# Add exclude patterns for ignored directories
449			for pattern in ignore_patterns:
450			cmd.extend(['-e', pattern])
451
452			# Add mode flag: -n for dry-run, -f for force, -i for interactive
453			if dry_run:
454			cmd.append('-n')
455			elif force:
456			cmd.append('-f')
457			else:
458			cmd.append('-i')
459
460			log.debug('Running: %s in %s', ' '.join(cmd), directory)
461
462			try:
463			result = subprocess.run(
464			cmd,
465			cwd=directory,
466			capture_output=True,
467			text=True,
468			check=False,
469			)
470
471			# Print output from git clean
472			if result.stdout:
473			print(result.stdout, end='')
474			if result.stderr:
475			print(result.stderr, end='')
476
477			return result.returncode
478
479			except FileNotFoundError:
480			log.error('Git command not found. Cannot execute git clean.')
481			return 127
482

bittner / pyclean

Pull Request — main (#105)

pyclean.modern.normalize() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like