def empty_creds_db() -> CloudFirestoreDatabase:
with mock.patch("cdptools.databases.cloud_firestore_database.CloudFirestoreDatabase._initialize_creds_db"):
db = CloudFirestoreDatabase("/fake/path/to/creds.json")
db._credentials_path = "/fake/path/to/creds.json"
db._root = mock.Mock(firestore.Client)
db._root.collection.return_value = MockedCollection([])
return db
def creds_db() -> CloudFirestoreDatabase:
with mock.patch(
"cdptools.databases.cloud_firestore_database.CloudFirestoreDatabase._initialize_creds_db" # noqa: E501
):
db = CloudFirestoreDatabase("/fake/path/to/creds.json")
db._credentials_path = "/fake/path/to/creds.json"
db._root = mock.Mock(firestore.Client)
db._root.collection.return_value = MockedCollection(
[MockedDocument("0e3bd59c-3f07-452c-83cf-e9eebeb73af2", EVENT_VALUES)]
)
return db
def test_convert_base_where_operator_to_cloud_firestore_where_operator(op, expected):
assert (
CloudFirestoreDatabase._convert_base_where_operator_to_cloud_firestore_where_operator( # noqa: E501
op
)
== expected
)
def test_download_transcripts(example_transcript, order_by_field):
with tempfile.TemporaryDirectory() as tmpdir:
# Patch select rows as list
with mock.patch.object(
CloudFirestoreDatabase,
"select_rows_as_list",
side_effect=
[[{
"event_id":
"1",
"legistar_event_id":
4023,
"event_datetime":
datetime(2019, 7, 15, 9, 30),
"agenda_file_uri":
"doesnt-matter",
"minutes_file_uri":
None,
"video_uri":
"doesnt-matter",
"created":
datetime(2019, 7, 20, 1, 53, 14, 77790),
"body_id":
"1",
"legistar_event_link":
"doesnt-matter",
"source_uri":
"http://www.seattlechannel.org/CouncilBriefings?videoid=x105823"
}],
[{
"transcript_id": "1",
"confidence": 0.9498944201984921,
"event_id": "1",
"created": datetime(2019, 7, 20, 1, 53, 18, 611107),
"file_id": "1"
}],
[{
"body_id": "1",
"name": "Council Briefing",
"created": datetime(2019, 7, 20, 1, 53, 13, 821791),
"description": None
}]]):
# Patch select row by id
with mock.patch.object(CloudFirestoreDatabase,
"select_row_by_id",
return_value={
"file_id":
"1",
"content_type":
None,
"filename":
"example_transcript_sentences.json",
"created":
datetime(2019, 7, 20, 1, 53, 10,
726978),
"description":
None,
"uri":
example_transcript
}):
# Interrupt the request to open the file stream
with mock.patch("requests.get") as mocked_request:
mocked_request.return_value = MockedStreamedRead(
example_transcript)
# Initialize objects
db = CloudFirestoreDatabase("fake-cdp-instance")
fs = GCSFileStore("fake-cdp-instance.appspot.com")
# Get the event corpus map
event_corpus_map = transcript_tools.download_transcripts(
db=db,
fs=fs,
order_by_field=order_by_field,
save_dir=tmpdir)
# Assert structure
assert len(event_corpus_map) == 1
# It should have one transcript and the manifest CSV
assert len(list(Path(tmpdir).iterdir())) == 2
# Assert that the transcript path in the manifest is also correct
manifest = pd.read_csv(
Path(tmpdir) / "transcript_manifest.csv")
with open(manifest.local_path[0], "r") as copied_file:
copied_transcript = json.load(copied_file)
with open(example_transcript, "r") as original_file:
original_transcript = json.load(original_file)
assert copied_transcript == original_transcript
def test_download_most_recent_transcripts(example_transcript):
with tempfile.TemporaryDirectory() as tmpdir:
db = CloudFirestoreDatabase("stg-cdp-seattle")
fs = GCSFileStore("stg-cdp-seattle.appspot.com")
# Mock interactions
with mock.patch(
"cdptools.databases.cloud_firestore_database.CloudFirestoreDatabase.select_rows_as_list"
) as mocked_db_select:
mocked_db_select.side_effect = [
[{
"transcript_id": "9183055d-300d-4204-8741-57cebbb280a9",
"confidence": 0.9498944201984921,
"event_id": "0a8fcd28-b920-4088-bd73-ceacb304db0f",
"created": datetime(2019, 7, 20, 1, 53, 18, 611107),
"file_id": "76b7b54d-2f9b-4cad-b0b8-039a51937c15"
}],
[{
"event_id":
"0a8fcd28-b920-4088-bd73-ceacb304db0f",
"legistar_event_id":
4023,
"event_datetime":
datetime(2019, 7, 15, 9, 30),
"agenda_file_uri":
"http://legistar2.granicus.com/seattle/meetings/2019/7/4023_A_Council_Briefing_19-07-15_Council_Briefing.pdf", # noqa: E501
"minutes_file_uri":
None,
"video_uri":
"http://video.seattle.gov:8080/media/council/brief_071519_2011955V.mp4",
"created":
datetime(2019, 7, 20, 1, 53, 14, 77790),
"body_id":
"f0867cf1-7bb0-4f28-83c9-a8cac6152ea4",
"legistar_event_link":
"https://seattle.legistar.com/MeetingDetail.aspx?LEGID=4023&GID=393&G=FFE3B678-CEF6-4197-84AC-5204EA4CFC0C", # noqa: E501
"source_uri":
"http://www.seattlechannel.org/CouncilBriefings?videoid=x105823"
}],
[{
"body_id": "f0867cf1-7bb0-4f28-83c9-a8cac6152ea4",
"name": "Council Briefing",
"created": datetime(2019, 7, 20, 1, 53, 13, 821791),
"description": None
}],
[{
"file_id":
"76b7b54d-2f9b-4cad-b0b8-039a51937c15",
"content_type":
None,
"filename":
"fc52ca9f9febd50ece14f46170014936f76f3d0227688ff96fcf7e369404eee7_ts_sentences_transcript_0.json", # noqa: E501
"created":
datetime(2019, 7, 20, 1, 53, 10, 726978),
"description":
None,
"uri":
"gs://stg-cdp-seattle.appspot.com/fc52ca9f9febd50ece14f46170014936f76f3d0227688ff96fcf7e369404eee7_ts_sentences_transcript_0.json" # noqa: E501
}]
]
with mock.patch(
"cdptools.file_stores.gcs_file_store.GCSFileStore.download_file"
) as mocked_download:
mocked_download.return_value = example_transcript
# Get the event corpus map
event_corpus_map = transcript_tools.download_most_recent_transcripts(
db, fs, tmpdir)
# Assert structure
assert len(event_corpus_map) == 1
def test_get_cloud_firestore_value_type(val, expected):
assert CloudFirestoreDatabase._get_cloud_firestore_value_type(val) == expected
def test_cloud_firestore_database_init(project_id, credentials_path):
CloudFirestoreDatabase(project_id, credentials_path)
def no_creds_db() -> CloudFirestoreDatabase:
return CloudFirestoreDatabase("this-is-a-fake-database-hi")
