def test_dump_log(self):
log_dir = 'log_dir/'
ndb.put_multi(self.fixtures[:2])
datastore_to_gcs.dump_log(TestModel, self.bucket, log_dir)
log_files = gcs.list_objects(self.bucket, log_dir)
self.assertEqual(len(log_files), 1)
isoformat_regex = '\d\d\d\d-\d\d-\d\dT\d\d:\d\d:\d\d\.json'
self.assertRegexpMatches(log_files[0], isoformat_regex)
log_files = gcs.list_objects(self.bucket, log_dir)
pprint(log_files)
log_items = gcs.download_object(self.bucket, log_dir + log_files[0])
pprint(log_items)
self.assertEqual(len(log_items), 2)
# Add more items, expect no overlap
sleep(1)
self.fixtures[2].put()
self.fixtures[3].put()
datastore_to_gcs.dump_log(TestModel, self.bucket, log_dir)
log_files = gcs.list_objects(self.bucket, log_dir)
self.assertEqual(len(log_files), 2)
log_files = gcs.list_objects(self.bucket, log_dir)
log_items = list(
itertools.chain(*[
gcs.download_object(self.bucket, log_dir + lf)
for lf in log_files
]))
pprint(log_items)
self.assertEqual(len(log_items), 4)
def test_nothing_new_to_update(self):
self.initial_dump()
initial_result = gcs.download_object(self.bucket, self.object_name)
self.assertIn(self.object_name, gcs.list_objects(self.bucket))
# Don't change anything but run update -- should leave file as is
self.run_datastore_to_gcs_update()
updated_result = gcs.download_object(self.bucket, self.object_name)
self.assertTrue(initial_result == updated_result,
msg='{}\ndoes not match{}'.format(
pformat(updated_result), pformat(initial_result)))
def test_initial_dump(self):
self.initial_dump()
# Assert that the transferred object exists
self.assertIn(self.object_name, gcs.list_objects(self.bucket))
transferred_items = gcs.download_object(self.bucket, self.object_name)
# Assert that it has the right number of items
self.assertEqual(len(transferred_items), 2)
# Assert that field names are right
self.assertTrue('email' in transferred_items[0])
self.assertTrue('id' in transferred_items[0])
self.assertTrue('last_modified' in transferred_items[0])
def test_download_object(self):
filename = self.upload_json()
downloaded_object = gcs.download_object(self.bucket, filename)
# Check that it's a DataDict
print "DOWNLOADED"
pprint(downloaded_object)
self.assertIsInstance(downloaded_object[0], util.DataDict)
print "\nFIXTURE"
data_dict_fixture = [
util.DataDict(fixture) for fixture in self.dict_fixtures
]
pprint(data_dict_fixture)
self.assertTrue(downloaded_object == data_dict_fixture)
def test_download_object_filter_fields(self):
filename = self.upload_json()
fields = ['id', 'nested.inner']
downloaded_object = gcs.download_object(self.bucket, filename, fields)
self.assertEqual(len(downloaded_object), len(self.dict_fixtures))
self.assertIsInstance(downloaded_object[0]['id'], unicode)
self.assertIsInstance(downloaded_object[0]['nested']['inner'], int)
first_fixture_filtered = util.DataDict({
'id':
self.dict_fixtures[0]['id'],
'nested': {
'inner': self.dict_fixtures[0]['nested']['inner']
}
})
self.assertDictEqual(downloaded_object[0], first_fixture_filtered)
def test_download_object_model(self):
filename = self.upload_model_json()
downloaded_object = gcs.download_object(self.bucket,
filename,
object_class=FixtureModel)
# Check that it's a FixtureModel
self.assertIsInstance(downloaded_object[0], FixtureModel)
# Fudge the dates
for d in downloaded_object:
d.dt = dateutil.parser.parse(d.dt).date()
print "DOWNLOADED"
pprint([d.__dict__ for d in downloaded_object])
print "\nFIXTURE"
pprint([f.__dict__ for f in self.model_fixtures])
self.assertTrue(
all([
d.__dict__ == f.__dict__ for (d, f) in itertools.izip(
downloaded_object, self.model_fixtures)
]))
def test_modify_existing(self):
self.initial_dump()
new_email = '*****@*****.**'
orig_results = TestModel.query(
TestModel.email == self.fixtures[6].email).fetch()
self.assertEqual(len(orig_results), 1)
orig = orig_results[0]
print 'ORIG:'
print orig
orig.email = new_email
orig.put()
self.run_datastore_to_gcs_update()
transferred_items = gcs.download_object(self.bucket, self.object_name)
new_results = [
e for e in transferred_items if e['id'] == orig.key.id()
]
self.assertEqual(1, len(new_results))
modified = new_results[0]
print 'MODIFIED:'
print modified
self.assertEqual(modified['email'], new_email)
def test_add_new_items(self):
self.initial_dump() # Adds 2 items
self.add_new_items() # Adds 3 items
transferred_items = gcs.download_object(self.bucket, self.object_name)
pprint(transferred_items)
self.assertEqual(len(transferred_items), 5)
