Esempio n. 1
0
from recidiviz.common.google_cloud.google_cloud_tasks_client_wrapper import \
    QUEUES_REGION
from recidiviz.ingest.direct.controllers.direct_ingest_gcs_file_system import \
    to_normalized_unprocessed_file_path
from recidiviz.cloud_storage.gcsfs_path import GcsfsFilePath
from recidiviz.ingest.direct.direct_ingest_cloud_task_manager import \
    DirectIngestCloudTaskManagerImpl, _build_task_id, ProcessIngestJobCloudTaskQueueInfo
from recidiviz.ingest.direct.controllers.direct_ingest_types import IngestArgs
from recidiviz.ingest.direct.controllers.gcsfs_direct_ingest_utils import \
    GcsfsIngestArgs, GcsfsDirectIngestFileType, GcsfsRawDataBQImportArgs, GcsfsIngestViewExportArgs
from recidiviz.utils import regions

_REGION = regions.Region(region_code='us_nc',
                         agency_name='agency_name',
                         agency_type='state',
                         base_url='base_url',
                         shared_queue=DIRECT_INGEST_STATE_PROCESS_JOB_QUEUE_V2,
                         timezone='America/New_York',
                         jurisdiction_id='jid',
                         environment='production')


class TestCloudTaskQueueInfo(TestCase):
    """Tests for the CloudTaskQueueInfo."""
    def test_is_task_queued_no_tasks(self):
        # Arrange
        info = ProcessIngestJobCloudTaskQueueInfo(queue_name='queue_name',
                                                  task_names=[])

        file_path = to_normalized_unprocessed_file_path(
            'bucket/file_path.csv', GcsfsDirectIngestFileType.INGEST_VIEW)
        args = IngestArgs(ingest_time=datetime.datetime.now())
    ProcessIngestJobCloudTaskQueueInfo,
)
from recidiviz.ingest.direct.controllers.direct_ingest_types import IngestArgs
from recidiviz.ingest.direct.controllers.gcsfs_direct_ingest_utils import (
    GcsfsIngestArgs,
    GcsfsDirectIngestFileType,
    GcsfsRawDataBQImportArgs,
    GcsfsIngestViewExportArgs,
)
from recidiviz.utils import regions

_REGION = regions.Region(
    region_code="us_nc",
    agency_name="agency_name",
    agency_type="state",
    base_url="base_url",
    shared_queue=DIRECT_INGEST_STATE_PROCESS_JOB_QUEUE_V2,
    timezone="America/New_York",
    jurisdiction_id="jid",
    environment="production",
)


class TestCloudTaskQueueInfo(TestCase):
    """Tests for the CloudTaskQueueInfo."""
    def test_is_task_queued_no_tasks(self) -> None:
        # Arrange
        info = ProcessIngestJobCloudTaskQueueInfo(queue_name="queue_name",
                                                  task_names=[])

        file_path = to_normalized_unprocessed_file_path(
            "bucket/file_path.csv", GcsfsDirectIngestFileType.INGEST_VIEW)
Esempio n. 3
0
from recidiviz.common.constants.county.booking import CustodyStatus
from recidiviz.ingest.scrape import constants, infer_release, scrape_phase
from recidiviz.ingest.scrape.sessions import ScrapeSession
from recidiviz.utils import regions

APP_ID = "recidiviz-infer-release-test"

app = Flask(__name__)
app.register_blueprint(infer_release.infer_release_blueprint)
app.config['TESTING'] = True

_REGIONS = [
    regions.Region(region_code='us_ut',
                   agency_name='agency_name',
                   agency_type='jail',
                   base_url='base_url',
                   queue={'rate': '5/m'},
                   timezone='America/New_York',
                   jurisdiction_id='jid',
                   environment='production'),
    regions.Region(region_code='us_wy',
                   agency_name='agency_name',
                   agency_type='jail',
                   base_url='base_url',
                   timezone='America/New_York',
                   removed_from_website='UNKNOWN_SIGNIFICANCE',
                   jurisdiction_id='jid',
                   environment='production'),
    regions.Region(region_code='us_nc',
                   agency_name='agency_name',
                   agency_type='prison',
                   base_url='base_url',
Esempio n. 4
0
from recidiviz.ingest.scrape import constants, infer_release, scrape_phase
from recidiviz.ingest.scrape.sessions import ScrapeSession
from recidiviz.utils import regions

APP_ID = "recidiviz-infer-release-test"

app = Flask(__name__)
app.register_blueprint(infer_release.infer_release_blueprint)
app.config["TESTING"] = True

_REGIONS = [
    regions.Region(
        region_code="us_ut",
        agency_name="agency_name",
        agency_type="jail",
        base_url="base_url",
        queue={"rate": "5/m"},
        timezone="America/New_York",
        jurisdiction_id="jid",
        environment="production",
    ),
    regions.Region(
        region_code="us_wy",
        agency_name="agency_name",
        agency_type="jail",
        base_url="base_url",
        timezone="America/New_York",
        removed_from_website="UNKNOWN_SIGNIFICANCE",
        jurisdiction_id="jid",
        environment="production",
    ),
    regions.Region(
    GcsfsIngestViewExportArgs,
    GcsfsRawDataBQImportArgs,
    gcsfs_direct_ingest_bucket_for_region,
)
from recidiviz.ingest.direct.direct_ingest_cloud_task_manager import (
    DirectIngestCloudTaskManagerImpl,
    ProcessIngestJobCloudTaskQueueInfo,
    _build_task_id,
)
from recidiviz.utils import regions

_REGION = regions.Region(
    region_code="us_xx",
    agency_name="agency_name",
    agency_type="state",
    base_url="base_url",
    timezone="America/New_York",
    jurisdiction_id="jid",
    environment="production",
    is_direct_ingest=True,
)


class TestProcessIngestJobCloudTaskQueueInfo(TestCase):
    """Tests for the CloudTaskQueueInfo."""
    def setUp(self) -> None:
        bucket = gcsfs_direct_ingest_bucket_for_region(
            project_id="recidiviz-456",
            region_code=_REGION.region_code,
            system_level=SystemLevel.STATE,
            ingest_instance=DirectIngestInstance.PRIMARY,
        )