meltanolabs/tap-bigquery

tap-bigquery's Issues

Support Python >=3.10

Support configuring a destination for large result sets

BigQuery queries have a limited response size¹ so syncs may fail when a large response is generated.

The python-bigquery-sqlalchemy² library supports passing a destination query parameter so the fix for this probably involves adding a new setting (e.g. destination_table) and passing that to the SQLAlchemy URL construction in

tap-bigquery/tap_bigquery/connector.py

Lines 45 to 47 in 0e37a0f

	def get_sqlalchemy_url(self, config: dict) -> str:
	"""Concatenate a SQLAlchemy URL for use in connecting to the source."""
	return f"bigquery://{config['project_id']}"

The string in question is a fully qualified table, e.g. different-project.different-dataset.table.

Footnotes

1. https://cloud.google.com/bigquery/docs/writing-results ↩

2. https://github.com/googleapis/python-bigquery-sqlalchemy?tab=readme-ov-file#connection-string-parameters ↩

Prohibitively long discovery times

We're seeing extremely long --discover times (anywhere from 5-25min) running this tap on a BigQuery setup with a large amount of tables. We're running this via meltano invoke <tap name> --discover.

We're using this tap in an embedded ETL context where we want to provide our customers a more structured approach to providing their credential details, then they're able to see their datasets, then they can see the tables within the dataset they select, and so on, and discovery is how we do that with a variety of other taps. However, the timing as it works out here is way too long for that process to be smooth.

Any clue how we might optimize this?