refactor to allow for multiple code locations
This commit is contained in:
58
apps/vinyl/src/jobs.py
Normal file
58
apps/vinyl/src/jobs.py
Normal file
@@ -0,0 +1,58 @@
|
||||
from assets import deals, new_deals, works
|
||||
|
||||
from dagster import (
|
||||
AssetKey,
|
||||
AssetMaterialization,
|
||||
OpExecutionContext,
|
||||
define_asset_job,
|
||||
job,
|
||||
op,
|
||||
)
|
||||
|
||||
deals_job = define_asset_job(
|
||||
"deals_job", selection=[deals], partitions_def=deals.partitions_def
|
||||
)
|
||||
|
||||
|
||||
@op
|
||||
def check_partititions(context: OpExecutionContext):
|
||||
# Replace with your asset/job name
|
||||
asset_key = "deals"
|
||||
|
||||
context.log_event(
|
||||
AssetMaterialization(asset_key=asset_key, partition="2024-09-30|sounds")
|
||||
)
|
||||
|
||||
# Fetch the materializations for the asset key
|
||||
materializations = context.instance.get_materialized_partitions(
|
||||
asset_key=AssetKey(asset_key)
|
||||
)
|
||||
context.log.info("Existing partitions", extra=dict(partitions=materializations))
|
||||
|
||||
import polars as pl
|
||||
|
||||
storage_dir = context.instance.storage_directory()
|
||||
ic(storage_dir)
|
||||
for row in (
|
||||
pl.scan_parquet(f"{storage_dir}/{asset_key}/*/*.parquet")
|
||||
.select(["date", "source"])
|
||||
.unique()
|
||||
.collect()
|
||||
.iter_rows()
|
||||
):
|
||||
partition = "|".join(row)
|
||||
if partition not in materializations:
|
||||
context.log.info(f"Missing partition: {partition}")
|
||||
context.log_event(
|
||||
AssetMaterialization(asset_key=asset_key, partition=partition)
|
||||
)
|
||||
|
||||
|
||||
@job
|
||||
def check_partititions_job():
|
||||
check_partititions()
|
||||
|
||||
|
||||
musicbrainz_lookup_job = define_asset_job(
|
||||
"musicbrainz_lookup_job", selection=[works, new_deals]
|
||||
)
|
||||
Reference in New Issue
Block a user