def test_deserialize_wrong_collection_set(self):
# This is testing that deserialization is not permitted when collection is placed
# in incorrect collection set directory.
serializer = serialize.RecordSerializer(data_dir=self.data_dir)
serializer.serialize_collection_set(self.collection_set)
# Partially clean the database in preparation for deserializing
self.collection1.delete()
self.collection_set.delete()
self.assertTrue(os.path.exists(self.collection1_records_path))
new_collection_set_path = "{}x".format(self.collection_set_path)
shutil.move(self.collection_set_path, new_collection_set_path)
self.assertFalse(os.path.exists(self.collection1_records_path))
deserializer = serialize.RecordDeserializer(data_dir=self.data_dir)
caught_error = False
try:
deserializer.deserialize_collection_set(new_collection_set_path)
except DeserializationError:
caught_error = True
self.assertTrue(caught_error)
def test_serialize_by_collection(self):
serializer = serialize.RecordSerializer(data_dir=self.data_dir)
serializer.serialize_collection_set(self.collection_set)
deserializer = serialize.RecordDeserializer(data_dir=self.data_dir)
deserializer.deserialize_collection(self.collection2_path)
deserializer.deserialize_collection(self.collection1_path)
# Nothing should change
self.assertEqual(2, Group.objects.count())
self.assertEqual(1, CollectionSet.objects.count())
self.assertEqual(2, CollectionSet.history.count())
self.assertEqual(2, Collection.objects.count())
self.assertEqual(3, Collection.history.count())
self.assertEqual(3, Credential.objects.count())
self.assertEqual(4, Credential.history.count())
self.assertEqual(2, User.objects.count())
self.assertEqual(2, Seed.objects.count())
self.assertEqual(3, Seed.history.count())
self.assertEqual(3, Harvest.objects.count())
self.assertEqual(3, HarvestStat.objects.count())
self.assertEqual(2, Warc.objects.count())
# Partially clean the database in preparation for deserializing
Warc.objects.all().delete()
HarvestStat.objects.all().delete()
self.harvest1.delete()
self.harvest2.delete()
self.harvest3.delete()
Seed.objects.all().delete()
Seed.history.all().delete()
self.collection1.delete()
self.collection2.delete()
Collection.history.all().delete()
# Note that credential1 still exists.
self.credential2.delete()
self.credential3.delete()
# This is also deleting credential1's history
Credential.history.all().delete()
# self.group2.delete()
# Note that group1 and group2 still exists.
self.assertEqual(2, Group.objects.count())
# Note that user1 still exists
self.user2.delete()
self.assertEqual(1, User.objects.count())
self.assertEqual(1, Credential.objects.count())
# Note that collection set still exists
self.assertEqual(1, CollectionSet.objects.count())
self.assertEqual(2, CollectionSet.history.count())
# Now deserialize again
deserializer.deserialize_collection(self.collection2_path)
deserializer.deserialize_collection(self.collection1_path)
# And check the deserialization
self.assertEqual(2, Group.objects.count())
self.assertEqual(1, CollectionSet.objects.count())
self.assertEqual(2, CollectionSet.history.count())
self.assertEqual(2, Collection.objects.count())
# +2 for turning off collection after it is deserialized
# +2 for added history note
self.assertEqual(7, Collection.history.count())
self.assertEqual(Collection.history.first().instance.history_note,
"Collection imported.")
self.assertEqual(3, Credential.objects.count())
# This is one less since credential1's history was deleted.
self.assertEqual(3, Credential.history.count())
self.assertEqual(2, User.objects.count())
self.assertEqual(2, Seed.objects.count())
self.assertEqual(3, Seed.history.count())
self.assertEqual(3, Harvest.objects.count())
self.assertEqual(3, HarvestStat.objects.count())
self.assertEqual(2, Warc.objects.count())
# Number of historical records for particular objects
self.assertEqual(
4,
Collection.objects.get(
collection_id=self.collection1.collection_id).history.count())
# Make sure we got the right historical objects
for h_collection in Collection.objects.get(
collection_id=self.collection1.collection_id).history.all():
self.assertEqual("test_collection1", h_collection.name)
self.assertEqual(
2,
CollectionSet.objects.get(collection_set_id=self.collection_set.
collection_set_id).history.count())
self.assertEqual(
2,
Credential.objects.get(
credential_id=self.credential2.credential_id).history.count())
# Make sure we got the right historical objects
for h_credential in Credential.objects.get(
credential_id=self.credential2.credential_id).history.all():
self.assertEqual("test_platform2", h_credential.platform)
self.assertEqual(
2,
Seed.objects.get(seed_id=self.seed1.seed_id).history.count())
def test_serialize(self):
self.assertTrue(self.collection1.is_active)
serializer = serialize.RecordSerializer(data_dir=self.data_dir)
serializer.serialize_collection_set(self.collection_set)
# Records files exist
self.assertTrue(os.path.exists(self.collection1_records_path))
# collection set, historical collection set, groups
# collection, historical collection, credentials, historical credentials
# users, seed, historical seeds
# harvests, harvest_stats, warcs, info
self.assertEqual(14, len(os.listdir(self.collection1_records_path)))
# Number of historical records for particular objects
self.assertEqual(2, self.collection1.history.count())
self.assertEqual(2, self.collection_set.history.count())
self.assertEqual(2, self.credential2.history.count())
self.assertEqual(2, self.seed1.history.count())
# Deserialize while collection set already exists
self.assertEqual(2, Group.objects.count())
self.assertEqual(1, CollectionSet.objects.count())
self.assertEqual(2, CollectionSet.history.count())
self.assertEqual(2, Collection.objects.count())
self.assertEqual(3, Collection.history.count())
self.assertEqual(3, Credential.objects.count())
self.assertEqual(4, Credential.history.count())
self.assertEqual(2, User.objects.count())
self.assertEqual(2, Seed.objects.count())
self.assertEqual(3, Seed.history.count())
self.assertEqual(3, Harvest.objects.count())
self.assertEqual(3, HarvestStat.objects.count())
self.assertEqual(2, Warc.objects.count())
deserializer = serialize.RecordDeserializer(data_dir=self.data_dir)
deserializer.deserialize_collection_set(self.collection_set_path)
# Nothing should change
self.assertEqual(2, Group.objects.count())
self.assertEqual(1, CollectionSet.objects.count())
self.assertEqual(2, CollectionSet.history.count())
self.assertEqual(2, Collection.objects.count())
self.assertEqual(3, Collection.history.count())
self.assertEqual(3, Credential.objects.count())
self.assertEqual(4, Credential.history.count())
self.assertEqual(2, User.objects.count())
self.assertEqual(2, Seed.objects.count())
self.assertEqual(3, Seed.history.count())
self.assertEqual(3, Harvest.objects.count())
self.assertEqual(3, HarvestStat.objects.count())
self.assertEqual(2, Warc.objects.count())
# Partially clean the database in preparation for deserializing
Warc.objects.all().delete()
self.assertEqual(0, Warc.objects.count())
HarvestStat.objects.all().delete()
self.assertEqual(0, HarvestStat.objects.count())
self.harvest1.delete()
self.harvest2.delete()
self.harvest3.delete()
self.assertEqual(0, Harvest.objects.count())
Seed.objects.all().delete()
self.assertEqual(0, Seed.objects.count())
Seed.history.all().delete()
self.assertEqual(0, Seed.history.count())
self.collection1.delete()
self.collection2.delete()
self.assertEqual(0, Collection.objects.count())
Collection.history.all().delete()
self.assertEqual(0, Collection.history.count())
# Note that credential1 still exists.
self.credential2.delete()
self.credential3.delete()
self.assertEqual(1, Credential.objects.count())
# This is also deleting credential1's history
Credential.history.all().delete()
self.assertEqual(0, Credential.history.count())
self.collection_set.delete()
self.assertEqual(0, CollectionSet.objects.count())
CollectionSet.history.all().delete()
self.assertEqual(0, CollectionSet.history.count())
self.group2.delete()
# Note that group1 still exists.
self.assertEqual(1, Group.objects.count())
CollectionSet.history.all().delete()
self.assertEqual(0, CollectionSet.history.count())
# Note that user1 still exists
self.user2.delete()
self.assertEqual(1, User.objects.count())
self.assertEqual(1, Credential.objects.count())
# Now deserialize again
deserializer.deserialize_collection_set(self.collection_set_path)
# And check the deserialization
self.assertEqual(2, Group.objects.count())
self.assertEqual(1, CollectionSet.objects.count())
# +1 for added history note
self.assertEqual(3, CollectionSet.history.count())
self.assertEqual(CollectionSet.history.first().instance.history_note,
"Collection set imported.")
self.assertEqual(2, Collection.objects.count())
# +2 for turning off collection after it is deserialized
# +2 for added history note
self.assertEqual(7, Collection.history.count())
self.assertEqual(Collection.history.first().instance.history_note,
"Collection imported.")
self.assertEqual(3, Credential.objects.count())
# This is one less since credential1's history was deleted.
self.assertEqual(3, Credential.history.count())
self.assertEqual(2, User.objects.count())
self.assertEqual(2, Seed.objects.count())
self.assertEqual(3, Seed.history.count())
self.assertEqual(3, Harvest.objects.count())
self.assertEqual(3, HarvestStat.objects.count())
self.assertEqual(2, Warc.objects.count())
# Number of historical records for particular objects
self.assertEqual(
4,
Collection.objects.get(
collection_id=self.collection1.collection_id).history.count())
# Make sure we got the right historical objects
for h_collection in Collection.objects.get(
collection_id=self.collection1.collection_id).history.all():
self.assertEqual("test_collection1", h_collection.name)
self.assertEqual(
3,
CollectionSet.objects.get(collection_set_id=self.collection_set.
collection_set_id).history.count())
self.assertEqual(
2,
Credential.objects.get(
credential_id=self.credential2.credential_id).history.count())
# Make sure we got the right historical objects
for h_credential in Credential.objects.get(
credential_id=self.credential2.credential_id).history.all():
self.assertEqual("test_platform2", h_credential.platform)
self.assertEqual(
2,
Seed.objects.get(seed_id=self.seed1.seed_id).history.count())
# Collection turned off
collection = Collection.objects.get_by_natural_key(
self.collection1.collection_id)
self.assertFalse(collection.is_active)
# User2 is inactive
self.assertFalse(User.objects.get(username="******").is_active)
# READMEs
self.assertTrue(
os.path.exists(os.path.join(self.collection_set_path,
"README.txt")))
self.assertTrue(
os.path.exists(os.path.join(self.collection1_path, "README.txt")))
# Info file
self.assertTrue(
os.path.exists(
os.path.join(self.collection1_records_path, "info.json")))