def test_7zextract_bytesio(self):
"""Test 7z file extraction from object"""
from fuglu.extensions.filearchives import Archivehandle, SEVENZIP_AVAILABLE
if not SEVENZIP_AVAILABLE > 0:
print(
"============================================================="
)
print(
"== WARNING =="
)
print(
"== Skipping 7z extract test since library is not installed =="
)
print(
"============================================================="
)
return
archive_filename = join(unittestsetup.TESTDATADIR, "test.7z")
# --
# use BytesIO as done in attachment manager
# --
with open(archive_filename, 'rb') as f:
buffer = f.read()
buffer = BytesIO(buffer)
print("Type of buffer sent to Archivehandler is: %s" % type(buffer))
handle = Archivehandle('7z', buffer)
self.runArchiveChecks(handle)
handle.close()
def test_7zextract_filename(self):
"""Test 7z file extraction from filename"""
from fuglu.extensions.filearchives import Archivehandle, SEVENZIP_AVAILABLE
if not SEVENZIP_AVAILABLE > 0:
print(
"============================================================="
)
print(
"== WARNING =="
)
print(
"== Skipping 7z extract test since library is not installed =="
)
print(
"============================================================="
)
return
# --
# use filename
# --
archive_filename = join(unittestsetup.TESTDATADIR, "test.7z")
handle = Archivehandle('7z', archive_filename)
self.runArchiveChecks(handle)
handle.close()
def test_7zextract_fileobject(self):
"""Test 7z file extraction from object"""
from fuglu.extensions.filearchives import Archivehandle, SEVENZIP_AVAILABLE
if not SEVENZIP_AVAILABLE > 0:
print(
"============================================================="
)
print(
"== WARNING =="
)
print(
"== Skipping 7z extract test since library is not installed =="
)
print(
"============================================================="
)
return
archive_filename = join(unittestsetup.TESTDATADIR, "test.7z")
# --
# use file descriptor
# --
f = open(archive_filename, 'rb')
try:
handle = Archivehandle('7z', f)
self.runArchiveChecks(handle)
handle.close()
finally:
f.close()
def test_rarfileextract_unicode_password2(self):
"""Test rar file extraction for encrypted (files and filelist) rar"""
from fuglu.extensions.filearchives import Archivehandle, RARFILE_AVAILABLE
if not RARFILE_AVAILABLE > 0:
print(
"=============================================================="
)
print(
"== WARNING =="
)
print(
"== Skipping rar extract test since library is not installed =="
)
print(
"=============================================================="
)
return
archive_filename = join(unittestsetup.TESTDATADIR, "password2.rar")
# --
# use filename
# --
handle = Archivehandle('rar', archive_filename)
archive_flist = handle.namelist()
self.assertEqual([], archive_flist)
handle.close()
def lint_archivetypes(self):
if not Archivehandle.avail('rar'):
print("rarfile library not found, RAR support disabled")
if not Archivehandle.avail('7z'):
print("pylzma/py7zlip library not found, 7z support disabled")
print("Archive scan, available file extensions: %s" %
(",".join(sorted(Archivehandle.avail_archive_extensions_list))))
print("Archive scan, active file extensions: %s" %
(",".join(sorted(self.active_archive_extensions.keys()))))
return True
def test_7zextract_bytesio_fullyencrypted(self):
"""Test 7z file extraction from object (encrypted including filenames)"""
from fuglu.extensions.filearchives import Archivehandle, SEVENZIP_AVAILABLE
if not SEVENZIP_AVAILABLE > 0:
print("================================================")
print("== WARNING ==")
print("== Skipping 7z bytesio encrypted extract test ==")
print("== since library is not installed ==")
print("================================================")
return
from py7zlib import NoPasswordGivenError
archive_filename = join(unittestsetup.TESTDATADIR,
"test.enc_filenames.p_is_secret.7z")
# --
# use BytesIO as done in attachment manager
# --
with open(archive_filename, 'rb') as f:
buffer = f.read()
buffer = BytesIO(buffer)
# 7z a test.enc_filenames.p_is_secret.7z -psecret -mhe test.txt
# compresses *.txt files to archive .7 z using password "secret". It also encrypts archive headers(-mhe switch),
# so filenames will be encrypted. This raises an exception already when creating the handle.
with self.assertRaises(NoPasswordGivenError):
_ = Archivehandle('7z', buffer)
def archive_handle(self):
"""
(Cached Property-Getter)
Create an archive handle to check, extract, ... files in the buffered archive.
Internal:
- archive_type: The archive type (already detected)
- buffer: The file buffer containing the archive
Returns:
(Archivehandle) : The handle to work with the archive
"""
# make sure there's no buffered archive object when
# the archive handle is created (or overwritten)
self._buffer_archobj = {}
handle = None
if self.buffer is not None:
try:
handle = Archivehandle(self.archive_type,
BytesIO(self.buffer),
archivename=self.filename)
except Exception as e:
self.logger.error(
"%s, Problem creating Archivehandle for file: "
"%s using archive handler %s (message: %s) -> ignore" %
(self.fugluid, self.filename, str(
self.archive_type), force_uString(e)))
return handle
def archive_type(self):
"""
(Cached Property-Getter)
Stores the archive type stored in this object.
Internal member dependencies:
- contenttype: File content type
- filename: Filename (Extension might be used to detect archive type)
Returns:
(str): Archive type if object is an archive, None otherwise
"""
self._arext = None
# try guessing the archive type based on magic content type first
archive_type = Archivehandle.archive_type_from_content_type(
self.contenttype)
# if it didn't work, try to guess by the filename extension, if it is enabled
if archive_type is None:
# sort by length, so tar.gz is checked before .gz
for arext in Archivehandle.avail_archive_extensions_list:
if self.filename.lower().endswith('.%s' % arext):
archive_type = Archivehandle.avail_archive_extensions[
arext]
# store archive extension for internal use
self._arext = arext
break
return archive_type
def test_7z_unavailable(self):
"""Tests what happens if 7z is not available"""
# patch system modules import dict to make sure importing py7zlib module fails
with patch.dict('sys.modules', {'py7zlib': None}):
from fuglu.extensions.filearchives import Archivehandle
self.assertFalse(
Archivehandle.avail('7z'),
"7z archive should be unavailable!%s" % importMockingMessage)
#-----------#
# extension #
#-----------#
extensionsHandles = ["7z"]
# archive extensions
for ext in extensionsHandles:
self.assertFalse(
ext in Archivehandle.avail_archive_extensions_list,
"archives with ending '%s' can not be handled (%s)!" %
(ext, Archivehandle.avail_archive_extensions_list))
#---------#
# content #
#---------#
contentHandles = ['^application\/x-7z-compressed']
# mail content regex as used in attachment
for cnt in contentHandles:
self.assertFalse(
cnt in Archivehandle.avail_archive_ctypes_list,
"content regex '%s' can not be handled (%s)" %
(cnt, Archivehandle.avail_archive_ctypes_list))
def test_rarfileextract_emptydir(self):
"""Test rar file extraction for encrypted (only files and not filelist) rar"""
from fuglu.extensions.filearchives import Archivehandle, RARFILE_AVAILABLE
if not RARFILE_AVAILABLE > 0:
print(
"=============================================================="
)
print(
"== WARNING =="
)
print(
"== Skipping rar extract test since library is not installed =="
)
print(
"=============================================================="
)
return
archive_filename = join(unittestsetup.TESTDATADIR, "EmptyDir.rar")
# --
# use filename
# --
handle = Archivehandle('rar', archive_filename)
archive_flist = handle.namelist()
self.assertEqual([u"EmptyDir"], archive_flist)
with self.assertRaises(TypeError):
handle.extract(archive_flist[0], None)
handle.close()
def test_rarfileextract_unicode_password(self):
"""Test rar file extraction for encrypted (only files and not filelist) rar"""
from fuglu.extensions.filearchives import Archivehandle, RARFILE_AVAILABLE
if not RARFILE_AVAILABLE > 0:
print(
"=============================================================="
)
print(
"== WARNING =="
)
print(
"== Skipping rar extract test since library is not installed =="
)
print(
"=============================================================="
)
return
import rarfile
archive_filename = join(unittestsetup.TESTDATADIR, "password.rar")
# --
# use filename
# --
handle = Archivehandle('rar', archive_filename)
archive_flist = handle.namelist()
self.assertEqual([u"One Földer/Hélö Wörld.txt", u"One Földer"],
archive_flist)
with self.assertRaises(rarfile.PasswordRequired):
handle.extract(archive_flist[0], None)
handle.close()
def test_7zextract_bytesio_encrypted(self):
"""Test 7z file extraction from object (encrypted)"""
from fuglu.extensions.filearchives import Archivehandle, SEVENZIP_AVAILABLE
if not SEVENZIP_AVAILABLE > 0:
print("================================================")
print("== WARNING ==")
print("== Skipping 7z bytesio encrypted extract test ==")
print("== since library is not installed ==")
print("================================================")
return
from py7zlib import NoPasswordGivenError
archive_filename = join(unittestsetup.TESTDATADIR,
"test.p_is_secret.7z")
# --
# use BytesIO as done in attachment manager
# --
with open(archive_filename, 'rb') as f:
buffer = f.read()
buffer = BytesIO(buffer)
handle = Archivehandle('7z', buffer)
# even though the file is password protected, the filelist is not
# so it's possible to extract the file list without an exception
archive_flist = handle.namelist()
self.assertEqual(["test.txt"], archive_flist)
# it is however not possible to extract the file
with self.assertRaises(NoPasswordGivenError):
extracted = handle.extract(archive_flist[0], 500000)
handle.close()
def test_7z_available(self):
"""Tests if '7z' archive is correctly detected and assigned to file extension/content"""
# patch system modules import dict to make sure importing "py7zlib" module is success
# ("py7zlib" is just a mock, but only successful import is needed for this test...)
mock = MagicMock()
with patch.dict('sys.modules', {'py7zlib': mock}):
from fuglu.extensions.filearchives import Archivehandle
self.assertTrue(
Archivehandle.avail('7z'),
"7z archive should unavailable!%s" % importMockingMessage)
#-----------#
# extension #
#-----------#
extensionsHandles = ["7z"]
# archive extensions
for ext in extensionsHandles:
self.assertTrue(
ext in Archivehandle.avail_archive_extensions_list,
"archives with ending '%s' has to be handled (%s)!" %
(ext, Archivehandle.avail_archive_extensions_list))
# archive extension assignment to '7z'
for ext in extensionsHandles:
self.assertEqual(
"7z", Archivehandle.avail_archive_extensions[ext],
"'%s' file endings has to be handled by '7z' archive" %
ext)
#---------#
# content #
#---------#
contentHandles = ['^application\/x-7z-compressed']
# mail content regex as used in attachment
for cnt in contentHandles:
self.assertTrue(
cnt in Archivehandle.avail_archive_ctypes_list,
"content regex '%s' has to be handled (%s)" %
(cnt, Archivehandle.avail_archive_ctypes_list))
# mail content regex assigned to 'rar' as used in attachment
for cnt in contentHandles:
self.assertEqual(
"7z", Archivehandle.avail_archive_ctypes[cnt],
"'%s' content regex has to be handled by 'rar' archive" %
cnt)
def test_rarfileextract(self):
"""Test rar file extraction"""
from fuglu.extensions.filearchives import Archivehandle, RARFILE_AVAILABLE
if not RARFILE_AVAILABLE > 0:
print(
"=============================================================="
)
print(
"== WARNING =="
)
print(
"== Skipping rar extract test since library is not installed =="
)
print(
"=============================================================="
)
return
archive_filename = join(unittestsetup.TESTDATADIR, "test.rar")
# --
# use filename
# --
handle = Archivehandle('rar', archive_filename)
self.runArchiveChecks(handle)
handle.close()
# --
# use file descriptor
# --
f = open(archive_filename, 'rb')
try:
handle = Archivehandle('rar', f)
self.runArchiveChecks(handle)
handle.close()
finally:
f.close()
def test_tarfileextract_bz2(self):
"""Test rar file extraction"""
from fuglu.extensions.filearchives import Archivehandle
# --
# use filename
# --
archive_filename = join(unittestsetup.TESTDATADIR, "test.tar.bz2")
handle = Archivehandle('tar', archive_filename)
self.runArchiveChecks(handle)
handle.close()
# --
# use file descriptor
# --
f = open(archive_filename, 'rb')
try:
handle = Archivehandle('tar', f)
self.runArchiveChecks(handle)
handle.close()
finally:
f.close()
def test_rarfileextract_unicode(self):
"""Test rar file extraction"""
from fuglu.extensions.filearchives import Archivehandle, RARFILE_AVAILABLE
if not RARFILE_AVAILABLE > 0:
print(
"=============================================================="
)
print(
"== WARNING =="
)
print(
"== Skipping rar extract test since library is not installed =="
)
print(
"=============================================================="
)
return
fsystemencoding = sys.getfilesystemencoding().lower()
if not fsystemencoding == "utf-8":
# with gitlab-runner (at least locally) the filesystem is "ascii" and this test
# fails because of the filename being unicode
print(
"=================================================================="
)
print(
"== WARNING =="
)
print(
"== Skipping rar extract unicode test because the file =="
)
print(
"== is not unicode: %s =="
% fsystemencoding)
print(
"== Test will only run if: =="
)
print(
"== \"python -c \"import sys; print(sys.getfilesystemencoding())\"\" =="
)
print(
"== returns utf-8 =="
)
print(
"=================================================================="
)
return
archive_filename = force_uString(
join(unittestsetup.TESTDATADIR, u"One Földer.rar"))
# --
# use filename
# --
print(u"filesystem encoding support: %s" % sys.getfilesystemencoding())
print(u"file %s exists: %s" %
(archive_filename, exists(archive_filename)))
handle = Archivehandle('rar', archive_filename)
archive_flist = handle.namelist()
self.assertEqual([u"One Földer/Hélö Wörld.txt", u"One Földer"],
archive_flist)
extracted = handle.extract(archive_flist[0], None)
self.assertEqual(u"bla bla bla\n", force_uString(extracted))
handle.close()