collect unique works from deals dumps

This commit is contained in:
2024-10-14 15:08:07 +02:00
parent 6f2c4d1249
commit 1e0528bdfb
9 changed files with 164 additions and 33 deletions

View File

@@ -1,13 +1,32 @@
from dagster import Definitions
from collections.abc import Sequence
from dagster_duckdb import DuckDBIOManager
from dagster_duckdb.io_manager import DbTypeHandler
from dagster_duckdb_pandas import DuckDBPandasTypeHandler
from dagster_polars import PolarsParquetIOManager
from .assets import deals
from .jobs import check_partititions_job, deals_job
from dagster import Definitions
from .assets import deals, new_deals, works
from .jobs import check_partititions_job, deals_job, musicbrainz_lookup_job
from .schedules import deals_schedule
from .sensors import musicbrainz_lookup_sensor
class PandasDuckDBIOManager(DuckDBIOManager):
@staticmethod
def type_handlers() -> Sequence[DbTypeHandler]:
return [DuckDBPandasTypeHandler()]
vinyl = Definitions(
assets=[deals],
resources={"polars_parquet_io_manager": PolarsParquetIOManager()},
jobs=[deals_job, check_partititions_job],
assets=[deals, new_deals, works],
resources={
"polars_parquet_io_manager": PolarsParquetIOManager(),
"duckdb_io_manager": PandasDuckDBIOManager(database="vinyl.duckdb"),
# "duckdb": DuckDBResource(database="vinyl.duckdb")
},
jobs=[deals_job, check_partititions_job, musicbrainz_lookup_job],
schedules=[deals_schedule],
sensors=[musicbrainz_lookup_sensor],
)