def google_sheets_to_csv_notebook(
work_policy,
google_sheets_export_policy,
csv_transform_policy,
csv_load_policy,
flowitem_model,
):
from flowmaster.operators.etl.loaders import Loaders
from flowmaster.operators.etl.policy import ETLNotebook
from flowmaster.operators.etl.providers import Providers
csv_transform_policy.column_schema["date_col"] = {
"name": "date",
"dtype": "date",
"errors": "default",
"dt_format": "%Y-%m-%d",
"allow_null": True,
}
name = "__test_google_sheets_to_csv__"
flowitem_model.clear(name)
yield ETLNotebook(
name=name,
provider=Providers.GoogleSheetsProvider.name,
storage=Loaders.CSVLoader.name,
work=work_policy,
export=google_sheets_export_policy,
transform=csv_transform_policy,
load=csv_load_policy,
)
def ymm_goals_to_csv_notebook(
flowitem_model,
work_policy,
yandex_metrika_management_credentials,
csv_transform_policy,
csv_load_policy,
):
from flowmaster.operators.etl.loaders import Loaders
from flowmaster.operators.etl.policy import ETLNotebook
from flowmaster.operators.etl.providers import Providers
name = "__test_ya_metrika_goals_to_csv__"
flowitem_model.clear(name)
yield ETLNotebook(
name=name,
provider=Providers.YandexMetrikaManagementProvider.name,
storage=Loaders.CSVLoader.name,
work=work_policy,
export=Providers.YandexMetrikaManagementProvider.policy_model(
resource=Providers.YandexMetrikaManagementProvider.policy_model.ResourceNames.goals,
credentials=Providers.YandexMetrikaManagementProvider.policy_model.CredentialsPolicy(
**yandex_metrika_management_credentials
),
),
transform=csv_transform_policy,
load=csv_load_policy,
)
def seconds_interval_work_policy(pendulum_utcnow):
from flowmaster.operators.etl.policy import ETLNotebook
return ETLNotebook.WorkPolicy(
triggers=ETLNotebook.WorkPolicy.TriggersPolicy(
schedule=ETLNotebook.WorkPolicy.TriggersPolicy.SchedulePolicy(
timezone="Europe/Moscow",
start_time="00:00:00",
from_date=None,
interval=60,
)))
def work_policy(pendulum_utcnow):
from flowmaster.operators.etl.policy import ETLNotebook
return ETLNotebook.WorkPolicy(
triggers=ETLNotebook.WorkPolicy.TriggersPolicy(
schedule=ETLNotebook.WorkPolicy.TriggersPolicy.SchedulePolicy(
timezone="Europe/Moscow",
start_time="00:00:00",
from_date=pendulum.today() - dt.timedelta(5),
interval="daily",
)))
def fakedata_to_csv_notebook(flowitem_model, work_policy, csv_transform_policy,
csv_load_policy):
from flowmaster.operators.etl.providers import Providers
from flowmaster.operators.etl.loaders import Loaders
from flowmaster.operators.etl.policy import ETLNotebook
name = "__test_ya_metrika_stats_to_csv__"
flowitem_model.clear(name)
yield ETLNotebook(
name=name,
provider=Providers.FakeDataProvider.name,
storage=Loaders.CSVLoader.name,
work=work_policy,
export=Providers.FakeDataProvider.policy_model(rows=1),
transform=csv_transform_policy,
load=csv_load_policy,
)
def ya_direct_report_to_csv_notebook(
flowitem_model,
work_policy,
csv_transform_policy,
csv_load_policy,
yandex_direct_credentials,
):
from flowmaster.operators.etl.providers import Providers
from flowmaster.operators.etl.loaders import Loaders
from flowmaster.operators.etl.policy import ETLNotebook
name = "__test2_ya_direct_report_to_csv__"
flowitem_model.clear(name)
yield ETLNotebook(
name=name,
provider=Providers.YandexDirectProvider.name,
storage=Loaders.CSVLoader.name,
work=work_policy,
export=Providers.YandexDirectProvider.policy_model(
credentials=Providers.YandexDirectProvider.policy_model.CredentialsPolicy(
**yandex_direct_credentials
),
resource="reports",
headers=Providers.YandexDirectProvider.policy_model.HeadersPolicy(
return_money_in_micros=True
),
body=Providers.YandexDirectProvider.policy_model.ReportBodyPolicy(
params=Providers.YandexDirectProvider.policy_model.ReportBodyPolicy.ReportParamsPolicy(
ReportType="ACCOUNT_PERFORMANCE_REPORT",
DateRangeType="AUTO",
FieldNames=["CampaignType", "Cost"],
IncludeVAT="NO",
Page=Providers.YandexDirectProvider.policy_model.ReportBodyPolicy.ReportParamsPolicy.PagePolicy(
Limit=10
),
),
),
),
transform=csv_transform_policy,
load=csv_load_policy,
)
flowitem_model.clear(name)
def test_codex_telegram():
def export_func(start_period,
end_period) -> Iterator[tuple[dict, list, list]]:
yield ({}, ["date"], [[start_period]])
ya_metrika_logs_to_csv_notebook.work.notifications = ETLNotebook.WorkPolicy.NotificationsPolicy(
codex_telegram=ETLNotebook.WorkPolicy.NotificationsPolicy.
CodexTelegramPolicy(
links=[credentials["codex_telegram"]],
on_success=True,
))
notebook = ETLNotebook(**dict(ya_metrika_logs_to_csv_notebook))
Providers.YandexMetrikaLogsProvider.export_class.__call__ = Mock(
side_effect=export_func)
etl_flow = ETLOperator(notebook)
list(
etl_flow(start_period=dt.datetime(2021, 1, 1),
end_period=dt.datetime(2021, 1, 1)))
def criteo_to_csv_notebook(
work_policy,
criteo_export_policy,
csv_transform_policy,
csv_load_policy,
flowitem_model,
):
from flowmaster.operators.etl.loaders import Loaders
from flowmaster.operators.etl.policy import ETLNotebook
from flowmaster.operators.etl.providers import Providers
name = "__test_criteo_to_csv__"
flowitem_model.clear(name)
yield ETLNotebook(
name=name,
provider=Providers.CriteoProvider.name,
storage=Loaders.CSVLoader.name,
work=work_policy,
export=criteo_export_policy,
transform=csv_transform_policy,
load=csv_load_policy,
)
def csv_to_clickhouse_notebook(flowitem_model, work_policy, clickhouse_credentials):
from flowmaster.operators.etl.loaders import Loaders
from flowmaster.operators.etl.policy import ETLNotebook
from flowmaster.operators.etl.providers import Providers
name = "__test_csv_to_clickhouse__"
flowitem_model.clear(name)
yield ETLNotebook(
name=name,
provider=Providers.CSVProvider.name,
storage=Loaders.ClickhouseLoader.name,
work=work_policy,
export=Providers.CSVProvider.policy_model(
file_path="", with_columns=False, columns=["date"]
),
transform=Loaders.ClickhouseLoader.transform_policy_model(
error_policy="default",
partition_columns=["Date"],
column_map={"date": "Date"},
),
load=Loaders.ClickhouseLoader.policy_model(
credentials=Loaders.ClickhouseLoader.policy_model.CredentialsPolicy(
**clickhouse_credentials
),
table_schema=Loaders.ClickhouseLoader.policy_model.TableSchemaPolicy(
db="default",
table=name,
columns=[
"Date Date",
],
orders=["Date"],
partition=["Date"],
),
data_cleaning_mode=Loaders.ClickhouseLoader.DataCleaningMode.off,
sql_before=["SELECT 1"],
sql_after=["SELECT 2"],
),
)
def ya_metrika_logs_to_csv_notebook2(
tmp_path,
work_policy,
ya_metrika_logs_export_policy,
csv_transform_policy,
csv_load_policy,
flowitem_model,
):
from flowmaster.operators.etl.loaders import Loaders
from flowmaster.operators.etl.policy import ETLNotebook
from flowmaster.operators.etl.providers import Providers
name = "__test_ya_metrika_logs_to_csv2__"
flowitem_model.clear(name)
yield ETLNotebook(
name=name,
provider=Providers.YandexMetrikaLogsProvider.name,
storage=Loaders.CSVLoader.name,
work=work_policy,
export=ya_metrika_logs_export_policy,
transform=csv_transform_policy,
load=csv_load_policy,
)
def ymstats_to_csv_notebook(
flowitem_model,
work_policy,
yandex_metrika_stats_credentials,
csv_transform_policy,
csv_load_policy,
):
from flowmaster.operators.etl.providers import Providers
from flowmaster.operators.etl.loaders import Loaders
from flowmaster.operators.etl.policy import ETLNotebook
name = "__test_ya_metrika_stats_to_csv__"
flowitem_model.clear(name)
yield ETLNotebook(
name=name,
provider=Providers.YandexMetrikaStatsProvider.name,
storage=Loaders.CSVLoader.name,
work=work_policy,
export=Providers.YandexMetrikaStatsProvider.policy_model(
**yandex_metrika_stats_credentials
),
transform=csv_transform_policy,
load=csv_load_policy,
)
def flowmasterdata_items_to_csv_notebook(
tmp_path,
seconds_interval_work_policy,
flowmasterdata_items_export_policy,
csv_transform_policy,
csv_load_policy,
flowitem_model,
):
from flowmaster.operators.etl.loaders.csv.service import CSVLoader
from flowmaster.operators.etl.policy import ETLNotebook
from flowmaster.operators.etl.providers.flowmaster_data import (
FlowmasterDataProvider, )
name = "__test_flowmasterdata_items_to_csv__"
flowitem_model.clear(name)
yield ETLNotebook(
name=name,
provider=FlowmasterDataProvider.name,
storage=CSVLoader.name,
work=seconds_interval_work_policy,
export=flowmasterdata_items_export_policy,
transform=csv_transform_policy,
load=csv_load_policy,
)
