def setUp(self):
self.resolver = Resolver()
self.bagger = WorkspaceBagger(self.resolver)
self.tempdir = mkdtemp()
self.bagdir = join(self.tempdir, 'kant_aufklaerung_1784')
copytree(assets.path_to('kant_aufklaerung_1784'), self.bagdir)
self.workspace_dir = join(self.bagdir, 'data')
self.workspace = Workspace(self.resolver,
directory=join(self.workspace_dir))
def kant_ocrdzip(ocrd_identifier):
resolver = Resolver()
bagger = WorkspaceBagger(resolver, strict=True)
dest = join(gettempdir(),
'olahd-test-bag-%d.ocrd.zip' % int(round((time() * 1000))))
ws = resolver.workspace_from_url(
assets.path_to('kant_aufklaerung_1784/data/mets.xml'))
bagger.bag(ws, ocrd_identifier, dest=dest)
yield dest
unlink(dest)
def setUp(self):
super().setUp()
pass
if exists(BACKUPDIR):
rmtree(BACKUPDIR)
self.resolver = Resolver()
self.bagger = WorkspaceBagger(self.resolver)
self.tempdir = mkdtemp()
self.bagdir = join(self.tempdir, 'bag')
copytree(assets.path_to('kant_aufklaerung_1784'), self.bagdir)
self.workspace_dir = join(self.bagdir, 'data')
self.workspace = Workspace(self.resolver, directory=self.workspace_dir)
def process(self):
client = OlaHdClient(self.parameter['endpoint'],
self.parameter['username'],
self.parameter['password'])
bagger = WorkspaceBagger(Resolver(), strict=True)
# TODO
dest = join(gettempdir(),
'bag-%d.ocrd.zip' % int(round((time() * 1000))))
# TODO
ocrd_identifier = self.workspace.mets.unique_identifier
bagger.bag(self.workspace, ocrd_identifier, dest=dest)
client.login()
client.post(dest, prev_pid=ocrd_identifier)
class TestOcrdZipValidator(TestCase):
def setUp(self):
self.resolver = Resolver()
self.bagger = WorkspaceBagger(self.resolver)
self.tempdir = mkdtemp()
self.bagdir = join(self.tempdir, 'kant_aufklaerung_1784')
copytree(assets.path_to('kant_aufklaerung_1784'), self.bagdir)
self.workspace_dir = join(self.bagdir, 'data')
self.workspace = Workspace(self.resolver,
directory=join(self.workspace_dir))
def tearDown(self):
rmtree(self.tempdir)
def test_validation(self):
ocrdzip = self.bagger.bag(self.workspace,
'SBB0000F29300010000',
ocrd_manifestation_depth='partial')
report = OcrdZipValidator(self.resolver, ocrdzip).validate()
self.assertEqual(report.is_valid, True)
def test_validation_unzipped(self):
validator = OcrdZipValidator(self.resolver, self.bagdir)
report = validator.validate(skip_unzip=True)
self.assertEqual(report.is_valid, True)
print(report)
def test_validation_unzipped_skip_bag(self):
validator = OcrdZipValidator(self.resolver, self.bagdir)
report = validator.validate(skip_unzip=True, skip_bag=True)
self.assertEqual(report.is_valid, True)
print(report)
def test_fail_validation_no_such_file(self):
validator = OcrdZipValidator(self.resolver, '/does/not/exist.ocrd.zip')
with self.assertRaisesRegex(IOError, "Can't find file"):
validator.validate(skip_unzip=False)
def test_fail_validation_unzipped_extra_files(self):
validator = OcrdZipValidator(self.resolver, self.bagdir)
extrapath = join(self.bagdir, 'data', 'EXTRA')
with open(extrapath, 'w') as f:
f.write('FAIL')
with self.assertRaisesRegex(Exception,
"Payload-Oxum validation failed"):
validator.validate(skip_unzip=True)
def test_fail_validation_extra_tagfile(self):
extrapath = join(self.bagdir, 'NOT-ALLOWED')
with open(extrapath, 'w') as f:
f.write('FAIL')
validator = OcrdZipValidator(self.resolver, self.bagdir)
with self.assertRaisesRegex(
Exception,
"Existing tag file 'NOT-ALLOWED' is not listed in Tag-Files-Allowed."
):
validator.validate(skip_unzip=True)
class TestWorkspaceBagger(TestCase):
def setUp(self):
if exists(BACKUPDIR):
rmtree(BACKUPDIR)
self.resolver = Resolver()
self.bagger = WorkspaceBagger(self.resolver)
self.tempdir = mkdtemp()
self.bagdir = join(self.tempdir, 'bag')
copytree(assets.path_to('kant_aufklaerung_1784'), self.bagdir)
self.workspace_dir = join(self.bagdir, 'data')
self.workspace = Workspace(self.resolver, directory=self.workspace_dir)
def tearDown(self):
rmtree(self.tempdir)
def test_bad_manifestation_depth(self):
with self.assertRaisesRegex(Exception, "manifestation_depth must be 'full' or 'partial'"):
self.bagger.bag(self.workspace, 'kant_aufklaerung_1784', ocrd_manifestation_depth='foo')
def test_bad_inplace_and_dest(self):
with self.assertRaisesRegex(Exception, "Setting 'dest' and 'in_place' is a contradiction"):
self.bagger.bag(self.workspace, 'kant_aufklaerung_1784', in_place=True, dest='/x/y/z')
def test_bad_skip_zip_and_dest(self):
with self.assertRaisesRegex(Exception, "Setting 'skip_zip' and not 'in_place' is a contradiction"):
self.bagger.bag(self.workspace, 'kant_aufklaerung_1784', in_place=True, skip_zip=False)
def test_bag_inplace(self):
self.bagger.bag(
self.workspace,
'kant_aufklaerung_1784',
ocrd_manifestation_depth='partial',
skip_zip=True,
in_place=True,
ocrd_base_version_checksum='123',
tag_files=[
README_FILE
],
)
def test_bag_zip_and_spill(self):
self.workspace.mets.find_all_files(ID='INPUT_0017')[0].url = 'bad-scheme://foo'
self.workspace.mets.find_all_files(ID='INPUT_0020')[0].url = 'http://google.com'
self.bagger.bag(self.workspace, 'kant_aufklaerung_1784', ocrd_manifestation_depth='full', skip_zip=False, dest=join(self.tempdir, 'out.ocrd.zip'))
self.bagger.spill(join(self.tempdir, 'out.ocrd.zip'), join(self.tempdir, 'out'))
def test_bag_zip_and_spill_wo_dest(self):
self.bagger.bag(self.workspace, 'kant_aufklaerung_1784', ocrd_manifestation_depth='partial', in_place=False, skip_zip=False, dest=join(self.tempdir, 'out.ocrd.zip'))
self.bagger.spill(join(self.tempdir, 'out.ocrd.zip'), self.tempdir)
def test_bag_wo_dest(self):
makedirs(BACKUPDIR)
self.bagger.bag(self.workspace, 'kant_aufklaerung_1784', ocrd_manifestation_depth='partial', in_place=True, skip_zip=True)
def test_bag_wo_dest_zip(self):
makedirs(BACKUPDIR)
self.bagger.bag(self.workspace, 'kant_aufklaerung_1784', ocrd_manifestation_depth='partial', in_place=False, skip_zip=True)
def test_bag_partial_http_nostrict(self):
self.bagger.strict = False
makedirs(BACKUPDIR)
self.workspace.mets.find_all_files(ID='INPUT_0020')[0].url = 'http://google.com'
self.bagger.bag(self.workspace, 'kant_aufklaerung_1784', ocrd_manifestation_depth='partial', in_place=False)
def test_bag_partial_http_strict(self):
self.bagger.strict = True
makedirs(BACKUPDIR)
self.workspace.mets.find_all_files(ID='INPUT_0020')[0].url = 'http://google.com'
with self.assertRaisesRegex(Exception, "Not fetching non-local files"):
self.bagger.bag(self.workspace, 'kant_aufklaerung_1784', ocrd_manifestation_depth='partial', in_place=False)
def test_bag_full(self):
self.bagger.strict = True
f = self.workspace.mets.find_all_files(ID='INPUT_0017')[0]
f.url = 'bad-scheme://foo'
f.local_filename = None
with self.assertRaisesRegex(Exception, "Not an http URL"):
self.bagger.bag(self.workspace, 'kant_aufklaerung_1784', ocrd_manifestation_depth='full', skip_zip=False)
def test_spill_dest_not_dir(self):
with self.assertRaisesRegex(Exception, "Not a directory: /dev/stdout"):
self.bagger.spill('x', '/dev/stdout')
def test_spill_derived_dest_exists(self):
dest = join(self.bagdir, 'foo')
makedirs(dest)
with self.assertRaisesRegex(Exception, "Directory exists: %s" % dest):
self.bagger.spill('/path/to/foo.ocrd.zip', self.bagdir)
def test_spill_derived_dest(self):
bag_dest = join(self.bagdir, 'foo.ocrd.zip')
spill_dest = join(self.bagdir, 'foo')
self.bagger.bag(self.workspace, 'kant_aufklaerung_1784', ocrd_manifestation_depth='partial', in_place=False, skip_zip=False, dest=bag_dest)
self.bagger.spill(bag_dest, self.bagdir)
self.assertTrue(exists(spill_dest))