pyclean.modern.should_ignore() - Code Metrics - Inspection of "[WIP] Fix handling of complex directory names with..." - bittner/pyclean - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — main (#93)

unknown

created 2025-10-15 15:52 UTC

pyclean.modern.should_ignore() B

↳ Parent: pyclean.modern

Complexity

Conditions

Size

Total Lines	39
Code Lines	17

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	17
dl	0
loc	39
rs	7.3333
c	0
b	0
f	0
cc	8
nop	2

# 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
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 should_ignore(path, ignore_patterns):
    """
    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
    
    Args:
        path: Path object to check
        ignore_patterns: List of ignore patterns
        
    Returns:
        True if the path should be ignored, False otherwise
    """
    if not ignore_patterns:
        return False
        
    for pattern in ignore_patterns:
        if '/' in pattern:
            # Pattern contains path separator - match relative path
            # We need to check if the path ends with this pattern
            try:
                # Get parts from the pattern
                pattern_parts = Path(pattern).parts
                # Path must have at least as many parts as the pattern
                if len(path.parts) < len(pattern_parts):
                    continue
                # Get the trailing parts of the path that match the pattern length
                path_parts = path.parts[-len(pattern_parts):]
                if path_parts == pattern_parts:
                    return True
            except (ValueError, IndexError):
                continue
        else:
            # Simple name - match the directory name anywhere
            if 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)

    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(args.erase, args.yes, 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',
        )


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())

    for path_glob in DEBRIS_TOPICS[topic]:
        delete_filesystem_objects(directory, path_glob, recursive=True)


def remove_freeform_targets(glob_patterns, yes, directory):
    """
    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)


def delete_filesystem_objects(directory, path_glob, prompt=False, recursive=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 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 prompt and not confirm('Remove empty directory %s' % dir_object):
            Runner.rmdir_failed += 1
            continue
        Runner.rmdir(dir_object)

    if recursive:
        subdirs = (Path(name.path) for name in os.scandir(directory) if name.is_dir())
        for subdir in subdirs:
            if should_ignore(subdir, Runner.ignore):
                log.debug('Skipping %s', subdir)
            else:
                delete_filesystem_objects(subdir, path_glob, prompt, recursive)


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)


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			from pathlib import Path
12
13			BYTECODE_FILES = ['.pyc', '.pyo']
14			BYTECODE_DIRS = ['__pycache__']
15			DEBRIS_TOPICS = {
16			'cache': [
17			'.cache/*/',
18			'.cache/',
19			],
20			'coverage': [
21			'.coverage',
22			'coverage.json',
23			'coverage.lcov',
24			'coverage.xml',
25			'htmlcov/*/',
26			'htmlcov/',
27			],
28			'jupyter': [
29			'.ipynb_checkpoints/*/',
30			'.ipynb_checkpoints/',
31			],
32			'mypy': [
33			'.mypy_cache/*/',
34			'.mypy_cache/',
35			],
36			'package': [
37			'build/bdist.//',
38			'build/bdist.*/',
39			'build/lib/*/',
40			'build/lib/',
41			'build/',
42			'dist/*/',
43			'dist/',
44			'sdist/*/',
45			'sdist/',
46			'.egg-info//',
47			'*.egg-info/',
48			],
49			'pytest': [
50			'.pytest_cache/*/',
51			'.pytest_cache/',
52			'pytestdebug.log',
53			],
54			'ruff': [
55			'.ruff_cache/*/',
56			'.ruff_cache/',
57			],
58			'tox': [
59			'.tox/*/',
60			'.tox/',
61			],
62			}
63
64
65			class CleanupRunner:
66			"""Module-level configuration and value store."""
67
68			def __init__(self):
69			"""Cleanup runner with optional dry-run behavior."""
70			self.unlink = None
71			self.rmdir = None
72			self.ignore = None
73			self.unlink_count = None
74			self.unlink_failed = None
75			self.rmdir_count = None
76			self.rmdir_failed = None
77
78			def configure(self, args):
79			"""Set up runner according to command line options."""
80			self.unlink = print_filename if args.dry_run else remove_file
81			self.rmdir = print_dirname if args.dry_run else remove_directory
82			self.ignore = args.ignore
83			self.unlink_count = 0
84			self.unlink_failed = 0
85			self.rmdir_count = 0
86			self.rmdir_failed = 0
87
88
89			log = logging.getLogger(__name__)
90			Runner = CleanupRunner()
91
92
93			def should_ignore(path, ignore_patterns):
94			"""
95			Check if a path should be ignored based on ignore patterns.
96
97			Patterns can be:
98			- Simple names like 'bar': matches any directory with that name
99			- Paths like 'foo/bar': matches 'bar' directory inside 'foo' directory
100
101			Args:
102			path: Path object to check
103			ignore_patterns: List of ignore patterns
104
105			Returns:
106			True if the path should be ignored, False otherwise
107			"""
108			if not ignore_patterns:
109			return False
110
111			for pattern in ignore_patterns:
112			if '/' in pattern:
113			# Pattern contains path separator - match relative path
114			# We need to check if the path ends with this pattern
115			try:
116			# Get parts from the pattern
117			pattern_parts = Path(pattern).parts
118			# Path must have at least as many parts as the pattern
119			if len(path.parts) < len(pattern_parts):
120			continue
121			# Get the trailing parts of the path that match the pattern length
122			path_parts = path.parts[-len(pattern_parts):]
123			if path_parts == pattern_parts:
124			return True
125			except (ValueError, IndexError):
126			continue
127			else:
128			# Simple name - match the directory name anywhere
129			if path.name == pattern:
130			return True
131			return False
132
133
134			def remove_file(fileobj):
135			"""Attempt to delete a file object for real."""
136			log.debug('Deleting file: %s', fileobj)
137			try:
138			fileobj.unlink()
139			Runner.unlink_count += 1
140			except OSError as err:
141			log.debug('File not deleted. %s', err)
142			Runner.unlink_failed += 1
143
144
145			def remove_directory(dirobj):
146			"""Attempt to remove a directory object for real."""
147			log.debug('Removing directory: %s', dirobj)
148			try:
149			dirobj.rmdir()
150			Runner.rmdir_count += 1
151			except OSError as err:
152			log.debug('Directory not removed. %s', err)
153			Runner.rmdir_failed += 1
154
155
156			def print_filename(fileobj):
157			"""Only display the file name, used with --dry-run."""
158			log.debug('Would delete file: %s', fileobj)
159			Runner.unlink_count += 1
160
161
162			def print_dirname(dirobj):
163			"""Only display the directory name, used with --dry-run."""
164			log.debug('Would delete directory: %s', dirobj)
165			Runner.rmdir_count += 1
166
167
168			def pyclean(args):
169			"""Cross-platform cleaning of Python bytecode."""
170			Runner.configure(args)
171
172			for dir_name in args.directory:
173			dir_path = Path(dir_name)
174
175			log.info('Cleaning directory %s', dir_path)
176			descend_and_clean(dir_path, BYTECODE_FILES, BYTECODE_DIRS)
177
178			for topic in args.debris:
179			remove_debris_for(topic, dir_path)
180
181			remove_freeform_targets(args.erase, args.yes, dir_path)
182
183			log.info(
184			'Total %d files, %d directories %s.',
185			Runner.unlink_count,
186			Runner.rmdir_count,
187			'would be removed' if args.dry_run else 'removed',
188			)
189
190			if Runner.unlink_failed or Runner.rmdir_failed:
191			log.debug(
192			'%d files, %d directories %s not be removed.',
193			Runner.unlink_failed,
194			Runner.rmdir_failed,
195			'would' if args.dry_run else 'could',
196			)
197
198
199			def descend_and_clean(directory, file_types, dir_names):
200			"""
201			Walk and descend a directory tree, cleaning up files of a certain type
202			along the way. Only delete directories if they are empty, in the end.
203			"""
204			for child in sorted(directory.iterdir()):
205			if child.is_file():
206			if child.suffix in file_types:
207			Runner.unlink(child)
208			elif child.is_dir():
209			if should_ignore(child, Runner.ignore):
210			log.debug('Skipping %s', child)
211			else:
212			descend_and_clean(child, file_types, dir_names)
213
214			if child.name in dir_names:
215			Runner.rmdir(child)
216			else:
217			log.debug('Ignoring %s (neither a file nor a folder)', child)
218
219
220			def remove_debris_for(topic, directory):
221			"""
222			Clean up debris for a specific topic.
223			"""
224			log.debug('Scanning for debris of %s ...', topic.title())
225
226			for path_glob in DEBRIS_TOPICS[topic]:
227			delete_filesystem_objects(directory, path_glob, recursive=True)
228
229
230			def remove_freeform_targets(glob_patterns, yes, directory):
231			"""
232			Remove free-form targets using globbing.
233
234			This is potentially dangerous since users can delete everything
235			anywhere in their file system, including the entire project they're
236			working on. For this reason, the implementation imposes the following
237			(user experience-related) restrictions:
238
239			- Deleting (directories) is not recursive, directory contents must be
240			explicitly specified using globbing (e.g. ``dirname/*/``).
241			- The user is responsible for the deletion order, so that a directory
242			is empty when it is attempted to be deleted.
243			- A confirmation prompt for the deletion of every single file system
244			object is shown (unless the ``--yes`` option is used, in addition).
245			"""
246			for path_glob in glob_patterns:
247			log.debug('Erase file system objects matching: %s', path_glob)
248			delete_filesystem_objects(directory, path_glob, prompt=not yes)
249
250
251			def delete_filesystem_objects(directory, path_glob, prompt=False, recursive=False):
252			"""
253			Identifies all pathnames matching a specific glob pattern, and attempts
254			to delete them in the proper order, optionally asking for confirmation.
255
256			Implementation Note: We sort the file system objects in reverse order
257			and first delete all files before removing directories. This way we
258			make sure that the directories that are deepest down in the hierarchy
259			are empty (for both files & directories) when we attempt to remove them.
260			"""
261			all_names = sorted(directory.glob(path_glob), reverse=True)
262			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...
263			files = (name for name in all_names if not name.is_dir() or name.is_symlink())
264
265			for file_object in files:
266			file_type = 'symlink' if file_object.is_symlink() else 'file'
267			if prompt and not confirm('Delete %s %s' % (file_type, file_object)):
268			Runner.unlink_failed += 1
269			continue
270			Runner.unlink(file_object)
271
272			for dir_object in dirs:
273			if prompt and not confirm('Remove empty directory %s' % dir_object):
274			Runner.rmdir_failed += 1
275			continue
276			Runner.rmdir(dir_object)
277
278			if recursive:
279			subdirs = (Path(name.path) for name in os.scandir(directory) if name.is_dir())
280			for subdir in subdirs:
281			if should_ignore(subdir, Runner.ignore):
282			log.debug('Skipping %s', subdir)
283			else:
284			delete_filesystem_objects(subdir, path_glob, prompt, recursive)
285
286
287			def confirm(message):
288			"""An interactive confirmation prompt."""
289			try:
290			answer = input('%s? ' % message)
291			return answer.strip().lower() in ['y', 'yes']
292			except KeyboardInterrupt:
293			msg = 'Aborted by user.'
294			raise SystemExit(msg)
295

bittner / pyclean

Pull Request — main (#93)

pyclean.modern.should_ignore() B

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like