pyclean.modern.normalize() - Code Metrics - Inspection of "Fix --ignore flag to handle complex directory name..." - bittner/pyclean - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — main (#93)

by Peter

created 2025-10-16 01:10 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
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
            # Check if the pattern matches anywhere in the path hierarchy
            # Path must have at least as many parts as the pattern
            if len(path.parts) < len(pattern_parts):
                continue
            # Check if pattern matches at any position in the path
            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)

    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',
        )

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


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)


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.',
        )


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 normalize(path_pattern: str) -> str:
94			"""
95			Normalize path separators in a pattern for cross-platform support.
96
97			On Windows, both forward slash and backslash are valid path separators.
98			On Unix/Posix, only forward slash is valid (backslash can be part of filename).
99			"""
100			return path_pattern.replace(os.sep, os.altsep or os.sep)
101
102
103			def should_ignore(path: Path, ignore_patterns: list[str]) -> bool:
104			"""
105			Check if a path should be ignored based on ignore patterns.
106
107			Patterns can be:
108			- Simple names like 'bar': matches any directory with that name
109			- Paths like 'foo/bar': matches 'bar' directory inside 'foo' directory
110			and also ignores everything inside that directory
111			"""
112			if not ignore_patterns:
113			return False
114
115			for pattern in ignore_patterns:
116			# Check if pattern has multiple components (is a path with separators)
117			pattern_parts = Path(normalize(pattern)).parts
118			if len(pattern_parts) > 1:
119			# Pattern contains path separator - match relative path
120			# Check if the pattern matches anywhere in the path hierarchy
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 at any position in the path
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			for dir_name in args.directory:
174			dir_path = Path(dir_name)
175
176			log.info('Cleaning directory %s', dir_path)
177			descend_and_clean(dir_path, BYTECODE_FILES, BYTECODE_DIRS)
178
179			for topic in args.debris:
180			remove_debris_for(topic, dir_path)
181
182			remove_freeform_targets(args.erase, args.yes, dir_path)
183
184			log.info(
185			'Total %d files, %d directories %s.',
186			Runner.unlink_count,
187			Runner.rmdir_count,
188			'would be removed' if args.dry_run else 'removed',
189			)
190
191			if Runner.unlink_failed or Runner.rmdir_failed:
192			log.debug(
193			'%d files, %d directories %s not be removed.',
194			Runner.unlink_failed,
195			Runner.rmdir_failed,
196			'would' if args.dry_run else 'could',
197			)
198
199			# Suggest --debris option if it wasn't used
200			if not args.debris:
201			suggest_debris_option(args)
202
203
204			def descend_and_clean(directory, file_types, dir_names):
205			"""
206			Walk and descend a directory tree, cleaning up files of a certain type
207			along the way. Only delete directories if they are empty, in the end.
208			"""
209			for child in sorted(directory.iterdir()):
210			if child.is_file():
211			if child.suffix in file_types:
212			Runner.unlink(child)
213			elif child.is_dir():
214			if should_ignore(child, Runner.ignore):
215			log.debug('Skipping %s', child)
216			else:
217			descend_and_clean(child, file_types, dir_names)
218
219			if child.name in dir_names:
220			Runner.rmdir(child)
221			else:
222			log.debug('Ignoring %s (neither a file nor a folder)', child)
223
224
225			def remove_debris_for(topic, directory):
226			"""
227			Clean up debris for a specific topic.
228			"""
229			log.debug('Scanning for debris of %s ...', topic.title())
230
231			for path_glob in DEBRIS_TOPICS[topic]:
232			delete_filesystem_objects(directory, path_glob, recursive=True)
233
234
235			def remove_freeform_targets(glob_patterns, yes, directory):
236			"""
237			Remove free-form targets using globbing.
238
239			This is potentially dangerous since users can delete everything
240			anywhere in their file system, including the entire project they're
241			working on. For this reason, the implementation imposes the following
242			(user experience-related) restrictions:
243
244			- Deleting (directories) is not recursive, directory contents must be
245			explicitly specified using globbing (e.g. ``dirname/*/``).
246			- The user is responsible for the deletion order, so that a directory
247			is empty when it is attempted to be deleted.
248			- A confirmation prompt for the deletion of every single file system
249			object is shown (unless the ``--yes`` option is used, in addition).
250			"""
251			for path_glob in glob_patterns:
252			log.debug('Erase file system objects matching: %s', path_glob)
253			delete_filesystem_objects(directory, path_glob, prompt=not yes)
254
255
256			def delete_filesystem_objects(directory, path_glob, prompt=False, recursive=False):
257			"""
258			Identifies all pathnames matching a specific glob pattern, and attempts
259			to delete them in the proper order, optionally asking for confirmation.
260
261			Implementation Note: We sort the file system objects in reverse order
262			and first delete all files before removing directories. This way we
263			make sure that the directories that are deepest down in the hierarchy
264			are empty (for both files & directories) when we attempt to remove them.
265			"""
266			all_names = sorted(directory.glob(path_glob), reverse=True)
267			dirs = (name for name in all_names if name.is_dir() and not name.is_symlink())
268			files = (name for name in all_names if not name.is_dir() or name.is_symlink())
269
270			for file_object in files:
271			file_type = 'symlink' if file_object.is_symlink() else 'file'
272			if prompt and not confirm('Delete %s %s' % (file_type, file_object)):
273			Runner.unlink_failed += 1
274			continue
275			Runner.unlink(file_object)
276
277			for dir_object in dirs:
278			if prompt and not confirm('Remove empty directory %s' % dir_object):
279			Runner.rmdir_failed += 1
280			continue
281			Runner.rmdir(dir_object)
282
283			if recursive:
284			subdirs = (Path(name.path) for name in os.scandir(directory) if name.is_dir())
285			for subdir in subdirs:
286			if should_ignore(subdir, Runner.ignore):
287			log.debug('Skipping %s', subdir)
288			else:
289			delete_filesystem_objects(subdir, path_glob, prompt, recursive)
290
291
292			def confirm(message):
293			"""An interactive confirmation prompt."""
294			try:
295			answer = input('%s? ' % message)
296			return answer.strip().lower() in ['y', 'yes']
297			except KeyboardInterrupt:
298			msg = 'Aborted by user.'
299			raise SystemExit(msg)
300
301
302			def detect_debris_in_directory(directory):
303			"""
304			Scan a directory for debris artifacts and return a list of detected topics.
305			"""
306			detected_topics = []
307
308			for topic, patterns in DEBRIS_TOPICS.items():
309			for pattern in patterns:
310			# Skip patterns that are for recursive cleanup (contain **)
311			if '**' in pattern:
312			continue
313			# Check if the pattern matches anything in the directory
314			matches = list(directory.glob(pattern))
315			if matches:
316			detected_topics.append(topic)
317			break # Found at least one match for this topic, move to next
318
319			return detected_topics
320
321
322			def suggest_debris_option(args):
323			"""
324			Suggest using the --debris option when it wasn't used.
325			Optionally provide targeted suggestions based on detected artifacts.
326			"""
327			# Collect all detected debris topics across all directories
328			all_detected = set()
329			for dir_name in args.directory:
330			dir_path = Path(dir_name)
331			if dir_path.exists():
332			detected = detect_debris_in_directory(dir_path)
333			all_detected.update(detected)
334
335			if all_detected:
336			# Provide targeted suggestion
337			topics_str = ' '.join(sorted(all_detected))
338			log.info(
339			'Hint: Use --debris to also clean up build artifacts. Detected: %s',
340			topics_str,
341			)
342			else:
343			# Provide general suggestion
344			log.info(
345			'Hint: Use --debris to also clean up build artifacts '
346			'from common Python development tools.',
347			)
348

bittner / pyclean

Pull Request — main (#93)

pyclean.modern.normalize() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like