address 2uprn and postcode splitter works locally

This commit is contained in:
Jun-te Kim 2026-02-16 12:13:16 +00:00
parent 8e574c2401
commit c1f784b87f
2 changed files with 5 additions and 7 deletions

View file

@ -504,6 +504,8 @@ def save_results_to_s3(
"""
if bucket_name is None:
bucket_name = os.getenv("S3_BUCKET_NAME")
if bucket_name is None:
bucket_name = "retrofit-data-dev"
if not bucket_name:
logger.error(
@ -544,8 +546,8 @@ def handler(event, context, local=False):
"body": json.dumps(
{
"task_id": "e31f2f21-175b-4a91-a3ec-a6baa325e917",
"sub_task_id": "1c09df07-fd29-4de7-b146-fafb591856a9",
"s3_uri": "s3://retrofit-data-dev/ara_postcode_splitter_batches/e31f2f21-175b-4a91-a3ec-a6baa325e917/8673913b-1a88-42d7-8578-0449123d94b0/2026-02-13T15:54:58.568594_67557923.csv",
"sub_task_id": "6a427b6e-1ece-4983-b1e5-9bffccc53d1d",
"s3_uri": "s3://retrofit-data-dev/ara_postcode_splitter_batches/e31f2f21-175b-4a91-a3ec-a6baa325e917/8673913b-1a88-42d7-8578-0449123d94b0/2026-02-16T12:00:20.257856_7b520c0e.csv",
}
)
}

View file

@ -204,10 +204,6 @@ def handler(event, context, local=False):
csv_data = read_csv_from_s3_dict(bucket, key)
df = pd.DataFrame(csv_data)
# TODO: Change the input to the file you want
# df = df.head(1983)
df = df.head(502)
logger.info(f"CSV loaded: {len(df)} rows, {len(df.columns)} columns")
# Sanitise postcodes
@ -288,7 +284,7 @@ def handler(event, context, local=False):
subtask_interface.update_subtask_status(
subtask_id,
"completed",
outputs={"rows_processed": "todo -> show sensible output"},
outputs={"rows_processed": "completed"},
)
return {