def test_translate_str():
assert translate_str(b"\xf1\x90\x80\x80\xc2\x90")['encoding'] == "utf-8"
assert translate_str(
b"fran\xc3\xa7ais \xc3\xa9l\xc3\xa8ve")['encoding'] == "utf-8"
assert translate_str(b'\x83G\x83\x93\x83R\x81[\x83f\x83B\x83\x93\x83O\x82'
b'\xcd\x93\xef\x82\xb5\x82\xad\x82\xc8\x82\xa2'
)['language'] == "Japanese"
def test_translate_str():
teststr = 'Стамболийски'
encoded_test_str = teststr.encode('ISO-8859-5')
result = str_utils.translate_str(encoded_test_str)
assert result['language'] == 'Bulgarian'
assert result['encoding'] == 'ISO-8859-5'
result = str_utils.translate_str('abcdéfg')
assert result['language'] == 'unknown'
assert result['encoding'] == 'utf-8'
def parse_comp_obj(self, field, res):
try:
self.cache_fields(field, res)
user_type = field["user_type"]
user_type_value = user_type.value.encode(user_type.charset)
char_enc_guessed = translate_str(user_type_value)
res.body['user_type'] = char_enc_guessed['converted']
res.body['user_type_encoding'] = char_enc_guessed['encoding']
except MissingField:
pass
try:
res.body['prog_id'] = field['prog_id'].value
except MissingField:
pass
def jar_extract(self, filename, dest_dir):
zf = None
try:
zf = zipfile.ZipFile(filename, "r")
# Make sure this is actually a JAR
unknown_charset_counter = 0
for zfname in zf.namelist():
uni_zfname = ""
o = None
try:
zf_info = zf.getinfo(zfname)
if not zf_info.orig_filename.endswith('\\') and not zf_info.orig_filename.endswith('/'):
char_enc_guessed = translate_str(zfname)
uni_zfname = char_enc_guessed['converted']
if char_enc_guessed['encoding'] == 'unknown':
uni_zfname = f"unknown_charset_filename_{unknown_charset_counter}"
unknown_charset_counter += 1
# creating the directory as problems if the filename
# starts with a /, strip it off.
if uni_zfname.startswith("/"):
uni_zfname = uni_zfname[1:]
unzipped_filename = os.path.normpath(os.path.join(dest_dir, uni_zfname))
zf_content = zf.read(zfname)
if not os.path.exists(os.path.dirname(unzipped_filename)):
os.makedirs(os.path.dirname(unzipped_filename))
try:
o = open(unzipped_filename, 'wb')
except Exception:
# just in case there was invalid char ...
uni_zfname = f"unknown_charset_filename_{unknown_charset_counter}"
unknown_charset_counter += 1
unzipped_filename = os.path.normpath(os.path.join(dest_dir, uni_zfname))
o = open(unzipped_filename, 'wb')
o.write(zf_content)
except Exception as e:
self.log.exception(f"Failed at extracting files from the JAR "
f"({filename.encode('utf-8')} :: + {uni_zfname}). Error: {str(e)}")
return False
finally:
if o is not None:
try:
o.close()
except Exception:
pass
except (IOError, zipfile.BadZipfile):
self.log.info(f"Not a ZIP File or Corrupt ZIP File: {filename}")
return False
except Exception as e:
if type(e) == NotJARException:
self.log.info(f"Not a JAR File: {filename}")
raise
self.log.exception(f"Caught an exception while analysing the file {filename}. [{e}]")
return False
finally:
if zf is not None:
try:
zf.close()
except Exception:
pass
return True
