def testVersionInfo(self):
profile = Profile(PROFILE_URL, profile=self.profile_dict)
self.assertEqual(profile.profile_version_info, (1, 2, 0),
'Bundled: 1.2.0')
del (self.profile_dict['BagIt-Profile-Info']['BagIt-Profile-Version'])
profile = Profile(PROFILE_URL, profile=self.profile_dict)
self.assertEqual(profile.profile_version_info, (1, 1, 0),
'Default profile version 1.1.0')
def test_existing_not_allowed(self):
self.profile_dict["Tag-Files-Allowed"] = []
with open(join(self.bagdir, 'tag-foo'), 'w'): pass
profile = Profile('TEST', self.profile_dict)
result = profile.validate(Bag(self.bagdir))
self.assertFalse(result)
self.assertEqual(len(profile.report.errors), 1)
self.assertTrue("Existing tag file" in profile.report.errors[0].value)
def test_profile_kwarg(self):
profile_url = Profile(PROFILE_URL)
profile_dict = Profile(PROFILE_URL, profile=self.profile_dict)
profile_str = Profile(PROFILE_URL, profile=self.profile_str)
self.assertEqual(json.dumps(profile_str.profile),
json.dumps(profile_dict.profile),
'Loaded from string')
self.assertEqual(json.dumps(profile_url.profile),
json.dumps(profile_dict.profile), 'Loaded from URL')
def test_existing_not_allowed(self):
self.profile_dict["Tag-Files-Allowed"] = []
with open(join(self.bagdir, 'tag-foo'), 'w'):
pass
profile = Profile('TEST', self.profile_dict)
result = profile.validate(Bag(self.bagdir))
self.assertFalse(result)
self.assertEqual(len(profile.report.errors), 1)
self.assertTrue("Existing tag file" in profile.report.errors[0].value)
def test_validate_serialization(self):
# Test on unzipped Bag.
self.assertTrue(
self.profile.validate_serialization(
os.path.abspath("fixtures/test-bar")))
# Test on zipped Bag.
self.profile = Profile(PROFILE_URL)
self.assertTrue(
self.profile.validate_serialization(
os.path.abspath("fixtures/test-foo.zip")))
def __init__(self, resolver, path_to_zip):
"""
Arguments:
resolver (Resolver): resolver
path_to_zip (string): Path to the OCRD-ZIP file
"""
self.resolver = resolver
self.path_to_zip = path_to_zip
self.report = ValidationReport()
self.profile_validator = Profile(OCRD_BAGIT_PROFILE_URL,
profile=OCRD_BAGIT_PROFILE)
class OcrdZipValidator():
"""
Validate conformance with BagIt and OCR-D bagit profile.
See:
- https://ocr-d.github.io/ocrd_zip
- https://ocr-d.github.io/bagit-profile.json
- https://ocr-d.github.io/bagit-profile.yml
"""
def __init__(self, resolver, path_to_zip):
"""
Arguments:
resolver (Resolver): resolver
path_to_zip (string): Path to the OCRD-ZIP file
"""
self.resolver = resolver
self.path_to_zip = path_to_zip
self.report = ValidationReport()
self.profile_validator = Profile(OCRD_BAGIT_PROFILE_URL,
profile=OCRD_BAGIT_PROFILE)
def _validate_profile(self, bag):
"""
Validate against OCRD BagIt profile (bag-info fields, algos etc)
"""
if not self.profile_validator.validate(bag):
raise Exception(str(self.profile_validator.report))
def _validate_bag(self, bag, **kwargs):
"""
Validate BagIt (checksums, payload.oxum etc)
"""
failed = None
try:
bag.validate(**kwargs)
except BagValidationError as e:
failed = e
# for d in e.details:
# log = getLogger('ocrd.ocrd_zip_validator')
# if isinstance(d, ChecksumMismatch):
# log.error("Validation Error: expected %s to have %s checksum of %s but found %s", d.path, d.algorithm, d.expected, d.found)
# else:
# log.error("Validation Error: %s", d)
if failed:
raise BagValidationError("%s" % failed)
def validate(self,
skip_checksums=False,
skip_bag=False,
skip_unzip=False,
skip_delete=False,
processes=2):
"""
Validate an OCRD-ZIP file for profile, bag and workspace conformance
Arguments:
skip_bag (boolean): Whether to skip all checks of manifests and files
skip_checksums (boolean): Whether to omit checksum checks but still check basic BagIt conformance
skip_unzip (boolean): Whether the OCRD-ZIP is unzipped, i.e. a directory
skip_delete (boolean): Whether to skip deleting the unpacked OCRD-ZIP dir after valdiation
processes (integer): Number of processes used for checksum validation
"""
if skip_unzip:
bagdir = self.path_to_zip
skip_delete = True
else:
# try:
self.profile_validator.validate_serialization(self.path_to_zip)
# except IOError as err:
# raise err
# except ProfileValidationError as err:
# self.report.add_error(err.value)
bagdir = mkdtemp(prefix=TMP_BAGIT_PREFIX)
unzip_file_to_dir(self.path_to_zip, bagdir)
try:
bag = Bag(bagdir)
self._validate_profile(bag)
if not skip_bag:
self._validate_bag(bag,
fast=skip_checksums,
processes=processes)
finally:
if not skip_delete:
# remove tempdir
rmtree(bagdir)
return self.report
def setUp(self):
self.bag = Bag('fixtures/test-bar')
self.profile = Profile(PROFILE_URL)
self.retrieved_profile = self.profile.get_profile()
class Test_bag_profile(TestCase):
def setUp(self):
self.bag = Bag('fixtures/test-bar')
self.profile = Profile(PROFILE_URL)
self.retrieved_profile = self.profile.get_profile()
def test_validate_bagit_profile_info(self):
self.assertTrue(
self.profile.validate_bagit_profile_info(self.retrieved_profile))
def test_report_after_validate(self):
self.assertIsNone(self.profile.report)
self.profile.validate(self.bag)
self.assertTrue(self.profile.report.is_valid)
def test_validate(self):
self.assertTrue(self.profile.validate(self.bag))
def test_validate_bag_info(self):
self.assertTrue(self.profile.validate_bag_info(self.bag))
def test_validate_manifests_required(self):
self.assertTrue(self.profile.validate_manifests_required(self.bag))
def test_validate_allow_fetch(self):
self.assertTrue(self.profile.validate_allow_fetch(self.bag))
def test_validate_accept_bagit_version(self):
self.assertTrue(self.profile.validate_accept_bagit_version(self.bag))
def test_validate_serialization(self):
# Test on unzipped Bag.
self.assertTrue(
self.profile.validate_serialization(
os.path.abspath("fixtures/test-bar")))
# Test on zipped Bag.
self.profile = Profile(PROFILE_URL)
self.assertTrue(
self.profile.validate_serialization(
os.path.abspath("fixtures/test-foo.zip")))
def test_find_tag_files(self):
expect = [
join(os.getcwd(), 'fixtures/test-bar', f)
for f in ['DPN/dpnFirstNode.txt', 'DPN/dpnRegistry']
]
self.assertEqual(sorted(find_tag_files(self.bag.path)), expect)
def test_not_given(self):
profile = Profile('TEST', self.profile_dict)
bag = Bag(self.bagdir)
result = profile.validate(bag)
self.assertTrue(result)
def setUp(self):
self.bag = Bag('fixtures/test-bar')
self.profile = Profile(PROFILE_URL)
self.retrieved_profile = self.profile.get_profile()
class Test_bag_profile(TestCase):
def setUp(self):
self.bag = Bag('fixtures/test-bar')
self.profile = Profile(PROFILE_URL)
self.retrieved_profile = self.profile.get_profile()
def test_validate_bagit_profile_info(self):
self.assertTrue(self.profile.validate_bagit_profile_info(self.retrieved_profile))
def test_report_after_validate(self):
self.assertIsNone(self.profile.report)
self.profile.validate(self.bag)
self.assertTrue(self.profile.report.is_valid)
def test_validate(self):
self.assertTrue(self.profile.validate(self.bag))
def test_validate_bag_info(self):
self.assertTrue(self.profile.validate_bag_info(self.bag))
def test_validate_manifests_required(self):
self.assertTrue(self.profile.validate_manifests_required(self.bag))
def test_validate_allow_fetch(self):
self.assertTrue(self.profile.validate_allow_fetch(self.bag))
def test_validate_accept_bagit_version(self):
self.assertTrue(self.profile.validate_accept_bagit_version(self.bag))
def test_validate_serialization(self):
# Test on unzipped Bag.
self.assertTrue(self.profile.validate_serialization(os.path.abspath("fixtures/test-bar")))
# Test on zipped Bag.
self.profile = Profile(PROFILE_URL)
self.assertTrue(self.profile.validate_serialization(os.path.abspath("fixtures/test-foo.zip")))
def test_find_tag_files(self):
expect = [join(os.getcwd(), 'fixtures/test-bar', f) for f in ['DPN/dpnFirstNode.txt', 'DPN/dpnRegistry']]
self.assertEqual(sorted(find_tag_files(self.bag.path)), expect)
def test_not_given(self):
profile = Profile('TEST', self.profile_dict)
bag = Bag(self.bagdir)
result = profile.validate(bag)
self.assertTrue(result)
def test_validate_serialization(self):
# Test on unzipped Bag.
self.assertTrue(self.profile.validate_serialization(os.path.abspath("fixtures/test-bar")))
# Test on zipped Bag.
self.profile = Profile(PROFILE_URL)
self.assertTrue(self.profile.validate_serialization(os.path.abspath("fixtures/test-foo.zip")))