class SevenZipArchive(BaseArchive):
def __init__(self, file):
self._archive = SevenZipFile(file, mode='r')
def list(self, *args, **kwargs):
self._archive.printdir(*args, **kwargs)
def filenames(self):
return self._archive.getnames()
def namelist(self):
return self._archive.getnames()
def close(self):
return self._archive.close()
def extract(self, file):
return self._archive.extract(file)
def is_encrypted(self):
return self._archive.password_protected
def extractall(self, file):
return self._archive.extractall(file)
def test_register_archive_format(tmp_path):
tmp_path.joinpath('src').mkdir()
tmp_path.joinpath('tgt').mkdir()
# Prepare test data
py7zr.unpack_7zarchive(os.path.join(testdata_path, 'test_1.7z'), path=tmp_path.joinpath('src'))
#
shutil.register_archive_format('7zip', pack_7zarchive, description='7zip archive')
shutil.make_archive(str(tmp_path.joinpath('target')), '7zip', str(tmp_path.joinpath('src')))
# check result
archive = SevenZipFile(tmp_path.joinpath('target.7z'), 'r')
archive.extractall(path=tmp_path.joinpath('tgt'))
archive.close()
m = hashlib.sha256()
m.update((tmp_path / 'tgt' / 'setup.py').open('rb').read())
assert m.digest() == binascii.unhexlify('b916eed2a4ee4e48c51a2b51d07d450de0be4dbb83d20e67f6fd166ff7921e49')
m = hashlib.sha256()
m.update((tmp_path / 'tgt' / 'scripts' / 'py7zr').open('rb').read())
assert m.digest() == binascii.unhexlify('b0385e71d6a07eb692f5fb9798e9d33aaf87be7dfff936fd2473eab2a593d4fd')
def test_register_archive_format(tmp_path):
tmp_path.joinpath("src").mkdir()
tmp_path.joinpath("tgt").mkdir()
# Prepare test data
py7zr.unpack_7zarchive(os.path.join(testdata_path, "test_1.7z"), path=tmp_path.joinpath("src"))
#
shutil.register_archive_format("7zip", pack_7zarchive, description="7zip archive")
shutil.make_archive(str(tmp_path.joinpath("target")), "7zip", str(tmp_path.joinpath("src")))
# check result
archive = SevenZipFile(tmp_path.joinpath("target.7z"), "r")
archive.extractall(path=tmp_path.joinpath("tgt"))
archive.close()
m = hashlib.sha256()
m.update((tmp_path / "tgt" / "setup.py").open("rb").read())
assert m.digest() == binascii.unhexlify("b916eed2a4ee4e48c51a2b51d07d450de0be4dbb83d20e67f6fd166ff7921e49")
m = hashlib.sha256()
m.update((tmp_path / "tgt" / "scripts" / "py7zr").open("rb").read())
assert m.digest() == binascii.unhexlify("b0385e71d6a07eb692f5fb9798e9d33aaf87be7dfff936fd2473eab2a593d4fd")
def _download_dataset(self):
if not osp.isdir(self._dataset_root):
os.mkdir(self._dataset_root)
if self._verbose >= 1:
print('Download files for the dataset...')
infos = FILES_INFOS[self._subset]
# Download archives files
for name, info in infos.items():
filename, url, hash_ = info['filename'], info['url'], info['hash']
filepath = osp.join(self._dataset_root, filename)
if not osp.isfile(filepath):
if self._verbose >= 1:
print(f'Download file "{filename}" from url "{url}"...')
if osp.exists(filepath):
raise RuntimeError(
f'Object "{filepath}" already exists but it\'s not a file.'
)
download_url(url,
self._dataset_root,
filename,
hash_value=hash_,
hash_type='md5')
# Extract audio files from archives
for name, info in infos.items():
filename = info['filename']
filepath = osp.join(self._dataset_root, filename)
extension = filename.split('.')[-1]
if extension == '7z':
extracted_path = osp.join(self._dataset_root, self._subset)
if not osp.isdir(extracted_path):
if self._verbose >= 1:
print(f'Extract archive file "{filename}"...')
archive_file = SevenZipFile(filepath)
archive_file.extractall(self._dataset_root)
archive_file.close()