def each(self, target):
tmpdir = tempdir()
rf = RarFile(target)
namelist = rf.namelist()
for name in namelist:
try:
rf.extract(name, tmpdir)
filepath = os.path.join(
tmpdir,
name,
)
if os.path.isfile(filepath):
self.add_extracted_file(filepath)
except RuntimeError:
for password in ['virus', 'infected']:
try:
filepath = rf.extract(name, tmpdir, pwd=password)
if os.path.isfile(filepath):
self.add_extracted_file(filepath)
break
except RuntimeError:
pass
else:
self.log('error', 'Could not extract {}'.format(name))
return True
def extract_rar(j):
"""
Extract a rar archive and update the job so that the extracted
file is organized instead of the archive.
:param dict j: organization job
:return: extraction success
:rtype: bool
"""
rf = RarFile(j['s'])
media = None
for f in rf.infolist():
_, ext = os.path.splitext(f.filename)
if ext in MEDIA_EXTS:
media = f
break
if not media:
return False
d, _ = os.path.split(j['s'])
extract_path = os.path.join(d, media.filename)
LOG.info(f"extract {j['s']} -> {extract_path}")
rf.extract(rf.infolist()[0], path=extract_path)
j['s'] = extract_path
target_dir, _ = os.path.split(j['t'])
j['t'] = get_target_path(extract_path, target_dir)
return True
def unrar(self, file_name):
zf = RarFile(join(self.open_path, file_name), 'r')
to_extract = zf.namelist()[0]
zf.extract(
to_extract,
path=self.open_path,
)
return to_extract
def unRarFull(fias_object):
"""Распаковка из полной базы ФИАС"""
rf = RarFile(fiases.fias_data.WORK_DIR + fiases.fias_data.FIAS_XML_RAR)
objectMatcher = re.compile(fias_object.FILE)
print('')
for f in rf.infolist():
if objectMatcher.match(f.filename):
fias_object.xml_file = f.filename
fias_object.xml_file_size = f.file_size
if (fias_object.xml_file_size > 0):
rf.extract(fias_object.xml_file, fiases.fias_data.WORK_DIR)
def unrarUpdate(fias_object):
"""Распаковка обновления """
rf = RarFile(fiases.fias_data.WORK_DIR +
fiases.fias_data.FIAS_DELTA_XML_RAR)
fias_objectMatcher = re.compile(fias_object.FILE)
for f in rf.infolist():
if fias_objectMatcher.match(f.filename):
fias_object.xml_delta_file = f.filename
fias_object.xml_delta_file_size = f.file_size
if (fias_object.xml_delta_file_size > 0):
rf.extract(fias_object.xml_delta_file, fiases.fias_data.WORK_DIR)
def __extract_cbr(self, file):
""" extract first image in cbr file, cache it in local cache folder"""
try:
archive_rar = RarFile(file, 'r')
extract_path = self.cache + path.basename(file)
if self.mode is 'all':
archive_rar.extractall(path=extract_path)
else:
first_file = archive_rar.namelist()[0]
archive_rar.extract(member=first_file, path=extract_path)
except BadRarFile as e:
raise e
finally:
archive_rar.close()
return extract_path
def run(self):
for idx, file in enumerate(self.input()):
ext = os.path.splitext(file.path)[-1]
if ext == '.rar':
arch = RarFile(file.path)
elif ext == '.zip':
arch = zipfile.ZipFile(file.path)
f = arch.namelist()[0]
target = os.path.join(os.getenv('TEMP_DIR'), f)
arch.extract(f, os.path.join(os.getenv('TEMP_DIR')))
copyfile(
target,
os.path.join(os.getenv('TEMP_DIR'),
os.path.basename(self.output()[idx].path)))
def extract_rar(complete_folder, rar_file, rar_name, data_name):
"""
Extract rar.gz file and get the data.
"""
rar_file_path = os.path.join(complete_folder, rar_file)
rf = RarFile(rar_file_path)
rar_inside = rf.namelist()
for t_i in rar_inside:
if rar_name in t_i:
rf.extract(t_i, path=complete_folder)
shutil.move(os.path.join(complete_folder, t_i), os.path.join(complete_folder, data_name))
break
try:
shutil.rmtree(os.path.join(complete_folder, t_i.split('/')[0]))
except:
pass
def run(self):
ext = os.path.splitext(self.input().path)[-1]
if ext == '.rar':
arch = RarFile(self.input().path)
elif ext == '.zip':
arch = zipfile.ZipFile(self.input().path)
for f in arch.namelist():
if os.path.basename(f) in [
os.path.basename(out.path) for out in self.output()
]:
target = os.path.join(os.getenv('TEMP_DIR'), f)
arch.extract(f, os.path.join(os.getenv('TEMP_DIR')))
copyfile(
target,
os.path.join(os.getenv('TEMP_DIR'), os.path.basename(f)))
def extract(entry, name):
rfile = RarFile(localsettings.BASEDIR + name)
start = datetime.now().replace(microsecond=0)
try:
extdir = localsettings.OUTDIR
if entry[-4:] == ".rar":
extdir = localsettings.BASEDIR + RARTEMP
rfile.extract(entry, path=extdir)
print name, "extracted"
except:
status = 'failed'
print name, "failed extract"
response = jsonify(status=status)
response.status_code = 400
return response
time = str(datetime.now().replace(microsecond=0) - start)
return jsonify(time=time, file=entry, status='success')
def extract_by_wildcard(arch_fpath: str,
directory: str = None,
wildcard: str = '*.xlsx',
names=None):
""" Extract files from archive. Supports only zip and rar formats. """
frmt = identify_file_format(arch_fpath)
# detect archive format
if not frmt:
raise ExternalSourceError("Not supported format")
else:
if frmt == 'rar':
arch_obj = RarFile(arch_fpath)
else:
arch_obj = ZipFile(arch_fpath)
if directory:
_dir = directory
else:
_dir = os.path.abspath(os.path.dirname(arch_fpath))
# filter by wildcard
_flist = fnmatch.filter(arch_obj.namelist(), wildcard)
if names:
_flist = _flist[:len(names)]
extracted_files_list = []
# extracting
for i, f in enumerate(_flist):
_fname = os.path.basename(f)
for _f in arch_obj.namelist():
if os.path.basename(_f) == _fname:
arch_obj.extract(_f, _dir)
src = os.path.join(_dir, _f).replace('/', os.sep)
dest = os.path.join(_dir, _fname)
if names:
dest = os.path.join(_dir, names[i])
if _fname:
move(src, dest)
extracted_files_list.append(dest)
return extracted_files_list
class ArchivedFile(object):
def __init__(self,filename):
if is_zipfile(filename):
self._zfile = ZipFile(filename)
elif is_rarfile(filename):
self._zfile = RarFile(filename)
def filelist(self):
return self._zfile.namelist()
def extract(self,srcfile,targetpath):
"extract srcfile in archive to targetpath"
for fullfilename in self.filelist():
if srcfile in fullfilename:
fpath,fname = os.path.split(fullfilename)
self._zfile.extract(fname.encode('gbk'),targetpath+fname)
return True
return None
def extractall(self,targetpath):
self._zfile.extractall(targetpath)
from rarfile import RarFile
path = '/Users/username/Downloads/'
filename = 'py_20200326_174202.rar'
filepath = 'temp/'
rf = RarFile(path + filename, mode='r') # mode的值只能为'r'
rf_list = rf.namelist() # 得到压缩包里所有的文件
print('rar文件内容', rf_list)
for f in rf_list:
rf.extract(f, path + filepath) # 循环解压，将文件解压到指定路径
# 一次性解压所有文件到指定目录
# rf.extractall(path) # 不传path，默认为当前目录
