def extract_tar_file(archive_path, output_dir, relative_to, extractors=None):
"""
Extract regular or compressed tar archive into the output directory.
archive_path is the pathlib.Path to the archive to unpack
output_dir is a pathlib.Path to the directory to unpack. It must already exist.
relative_to is a pathlib.Path for directories that should be stripped relative to the
root of the archive, or None if no path components should be stripped.
extractors is a dictionary of PlatformEnum to a command or path to the
extractor binary. Defaults to 'tar' for tar, and '_use_registry' for 7-Zip and WinRAR.
Raises ExtractionError if unexpected issues arise during unpacking.
"""
if extractors is None:
extractors = DEFAULT_EXTRACTORS
current_platform = get_running_platform()
if current_platform == PlatformEnum.WINDOWS:
# Try to use 7-zip first
sevenzip_cmd = extractors.get(ExtractorEnum.SEVENZIP)
if sevenzip_cmd == USE_REGISTRY:
sevenzip_cmd = str(_find_7z_by_registry())
sevenzip_bin = _find_extractor_by_cmd(sevenzip_cmd)
if sevenzip_bin is not None:
_extract_tar_with_7z(sevenzip_bin, archive_path, output_dir, relative_to)
return
# Use WinRAR if 7-zip is not found
winrar_cmd = extractors.get(ExtractorEnum.WINRAR)
if winrar_cmd == USE_REGISTRY:
winrar_cmd = str(_find_winrar_by_registry())
winrar_bin = _find_extractor_by_cmd(winrar_cmd)
if winrar_bin is not None:
_extract_tar_with_winrar(winrar_bin, archive_path, output_dir, relative_to)
return
get_logger().warning(
'Neither 7-zip nor WinRAR were found. Falling back to Python extractor...')
elif current_platform == PlatformEnum.UNIX:
# NOTE: 7-zip isn't an option because it doesn't preserve file permissions
tar_bin = _find_extractor_by_cmd(extractors.get(ExtractorEnum.TAR))
if not tar_bin is None:
_extract_tar_with_tar(tar_bin, archive_path, output_dir, relative_to)
return
else:
# This is not a normal code path, so make it clear.
raise NotImplementedError(current_platform)
# Fallback to Python-based extractor on all platforms
_extract_tar_with_python(archive_path, output_dir, relative_to)
def extract_with_7z(
archive_path,
output_dir,
relative_to, #pylint: disable=too-many-arguments
extractors=None):
"""
Extract archives with 7-zip into the output directory.
Only supports archives with one layer of unpacking, so compressed tar archives don't work.
archive_path is the pathlib.Path to the archive to unpack
output_dir is a pathlib.Path to the directory to unpack. It must already exist.
relative_to is a pathlib.Path for directories that should be stripped relative to the
root of the archive.
extractors is a dictionary of PlatformEnum to a command or path to the
extractor binary. Defaults to 'tar' for tar, and '_use_registry' for 7-Zip.
Raises ExtractionError if unexpected issues arise during unpacking.
"""
# TODO: It would be nice to extend this to support arbitrary standard IO chaining of 7z
# instances, so _extract_tar_with_7z and other future formats could use this.
if extractors is None:
extractors = DEFAULT_EXTRACTORS
sevenzip_cmd = extractors.get(ExtractorEnum.SEVENZIP)
if sevenzip_cmd == USE_REGISTRY:
if not get_running_platform() == PlatformEnum.WINDOWS:
get_logger().error('"%s" for 7-zip is only available on Windows', sevenzip_cmd)
raise ExtractionError()
sevenzip_cmd = str(_find_7z_by_registry())
sevenzip_bin = _find_extractor_by_cmd(sevenzip_cmd)
if not relative_to is None and (output_dir / relative_to).exists():
get_logger().error('Temporary unpacking directory already exists: %s',
output_dir / relative_to)
raise ExtractionError()
cmd = (sevenzip_bin, 'x', str(archive_path), '-aoa', '-o{}'.format(str(output_dir)))
get_logger().debug('7z command line: %s', ' '.join(cmd))
result = subprocess.run(cmd)
if result.returncode != 0:
get_logger().error('7z command returned %s', result.returncode)
raise ExtractionError()
_process_relative_to(output_dir, relative_to)
def extract_with_7z(
archive_path,
output_dir,
relative_to, #pylint: disable=too-many-arguments
extractors=None):
"""
Extract archives with 7-zip into the output directory.
Only supports archives with one layer of unpacking, so compressed tar archives don't work.
archive_path is the pathlib.Path to the archive to unpack
output_dir is a pathlib.Path to the directory to unpack. It must already exist.
relative_to is a pathlib.Path for directories that should be stripped relative to the
root of the archive.
extractors is a dictionary of PlatformEnum to a command or path to the
extractor binary. Defaults to 'tar' for tar, and '_use_registry' for 7-Zip.
Raises ExtractionError if unexpected issues arise during unpacking.
"""
# TODO: It would be nice to extend this to support arbitrary standard IO chaining of 7z
# instances, so _extract_tar_with_7z and other future formats could use this.
if extractors is None:
extractors = DEFAULT_EXTRACTORS
sevenzip_cmd = extractors.get(ExtractorEnum.SEVENZIP)
if sevenzip_cmd == SEVENZIP_USE_REGISTRY:
if not get_running_platform() == PlatformEnum.WINDOWS:
get_logger().error('"%s" for 7-zip is only available on Windows', sevenzip_cmd)
raise ExtractionError()
sevenzip_cmd = str(_find_7z_by_registry())
sevenzip_bin = _find_extractor_by_cmd(sevenzip_cmd)
if not relative_to is None and (output_dir / relative_to).exists():
get_logger().error('Temporary unpacking directory already exists: %s',
output_dir / relative_to)
raise ExtractionError()
cmd = (sevenzip_bin, 'x', str(archive_path), '-aoa', '-o{}'.format(str(output_dir)))
get_logger().debug('7z command line: %s', ' '.join(cmd))
result = subprocess.run(cmd)
if result.returncode != 0:
get_logger().error('7z command returned %s', result.returncode)
raise ExtractionError()
_process_relative_to(output_dir, relative_to)
def extract_with_winrar(
archive_path,
output_dir,
relative_to, #pylint: disable=too-many-arguments
extractors=None):
"""
Extract archives with WinRAR into the output directory.
Only supports archives with one layer of unpacking, so compressed tar archives don't work.
archive_path is the pathlib.Path to the archive to unpack
output_dir is a pathlib.Path to the directory to unpack. It must already exist.
relative_to is a pathlib.Path for directories that should be stripped relative to the
root of the archive.
extractors is a dictionary of PlatformEnum to a command or path to the
extractor binary. Defaults to 'tar' for tar, and '_use_registry' for WinRAR.
Raises ExtractionError if unexpected issues arise during unpacking.
"""
if extractors is None:
extractors = DEFAULT_EXTRACTORS
winrar_cmd = extractors.get(ExtractorEnum.WINRAR)
if winrar_cmd == USE_REGISTRY:
if not get_running_platform() == PlatformEnum.WINDOWS:
get_logger().error('"%s" for WinRAR is only available on Windows', winrar_cmd)
raise ExtractionError()
winrar_cmd = str(_find_winrar_by_registry())
winrar_bin = _find_extractor_by_cmd(winrar_cmd)
if not relative_to is None and (output_dir / relative_to).exists():
get_logger().error('Temporary unpacking directory already exists: %s',
output_dir / relative_to)
raise ExtractionError()
cmd = (winrar_bin, 'x', '-o+', str(archive_path), str(output_dir))
get_logger().debug('WinRAR command line: %s', ' '.join(cmd))
result = subprocess.run(cmd)
if result.returncode != 0:
get_logger().error('WinRAR command returned %s', result.returncode)
raise ExtractionError()
_process_relative_to(output_dir, relative_to)
def extract_tar_file(archive_path, output_dir, relative_to, extractors=None):
"""
Extract regular or compressed tar archive into the output directory.
archive_path is the pathlib.Path to the archive to unpack
output_dir is a pathlib.Path to the directory to unpack. It must already exist.
relative_to is a pathlib.Path for directories that should be stripped relative to the
root of the archive, or None if no path components should be stripped.
extractors is a dictionary of PlatformEnum to a command or path to the
extractor binary. Defaults to 'tar' for tar, and '_use_registry' for 7-Zip.
Raises ExtractionError if unexpected issues arise during unpacking.
"""
if extractors is None:
extractors = DEFAULT_EXTRACTORS
current_platform = get_running_platform()
if current_platform == PlatformEnum.WINDOWS:
sevenzip_cmd = extractors.get(ExtractorEnum.SEVENZIP)
if sevenzip_cmd == SEVENZIP_USE_REGISTRY:
sevenzip_cmd = str(_find_7z_by_registry())
sevenzip_bin = _find_extractor_by_cmd(sevenzip_cmd)
if not sevenzip_bin is None:
_extract_tar_with_7z(sevenzip_bin, archive_path, output_dir, relative_to)
return
elif current_platform == PlatformEnum.UNIX:
# NOTE: 7-zip isn't an option because it doesn't preserve file permissions
tar_bin = _find_extractor_by_cmd(extractors.get(ExtractorEnum.TAR))
if not tar_bin is None:
_extract_tar_with_tar(tar_bin, archive_path, output_dir, relative_to)
return
else:
# This is not a normal code path, so make it clear.
raise NotImplementedError(current_platform)
# Fallback to Python-based extractor on all platforms
_extract_tar_with_python(archive_path, output_dir, relative_to)