[morganchen12]

I will untangle the git history soon.

[gemini-code-assist]

Code Review

This pull request introduces a comprehensive set of pipeline snippets for Cloud Firestore in Python, covering joins, aggregations, DML operations, and document insertions. The feedback highlights a few key improvements: correcting an invalid argument usage in upsert_in_place by using .add_fields() before .raw_stage(), adding a clarifying comment in pipeline_force_table_scan regarding the lack of forceIndex support in the Python SDK, and refactoring the non-idiomatic magic indexing [1] in load_test_data to use tuple destructuring.

[gemini-code-assist]

In upsert_in_place, passing Constant.of(880000).as_("population") directly as a positional argument to raw_stage is incorrect because the second parameter of raw_stage is options (which expects a dictionary), not a pipeline expression. To upsert specific fields, you should use .add_fields(...) before calling .raw_stage("upsert").

Suggested change

	results = (
	db.pipeline()
	.collection("cities")
	.where(Field.of("name").equal("San Francisco"))
	.raw_stage("upsert", Constant.of(880000).as_("population"))
	.execute()
	)
	results = (
	db.pipeline()
	.collection("cities")
	.where(Field.of("name").equal("San Francisco"))
	.add_fields(Constant.of(880000).as_("population"))
	.raw_stage("upsert")
	.execute()
	)

[gemini-code-assist]

The pipeline_force_table_scan snippet claims to force a full-table scan, but the code is just a standard query with no parameters or options to force a scan. Since forceIndex is not currently supported in the Python SDK, it is highly recommended to add an explanatory comment to avoid confusing users, similar to other snippets in the repository.

Suggested change

	def pipeline_force_table_scan(db):
	# [START pipeline_force_table_scan]
	# Force Planner to only do a Full-Table Scan
	results = (
	db.pipeline()
	.collection_group("customers")
	.limit(100)
	.execute()
	)
	# [END pipeline_force_table_scan]
	return results
	def pipeline_force_table_scan(db):
	# [START pipeline_force_table_scan]
	# Force Planner to only do a Full-Table Scan
	# Note: forceIndex="primary" is not currently supported in the Python SDK.
	results = (
	db.pipeline()
	.collection_group("customers")
	.limit(100)
	.execute()
	)
	# [END pipeline_force_table_scan]
	return results

[gemini-code-assist]

Using a magic index [1] to extract the DocumentReference from the tuple returned by CollectionReference.add() is less readable and non-idiomatic in Python. Consider using tuple destructuring (e.g., _, rest1 = ...) to improve code clarity. This pattern can be applied to all other restaurant additions in this function.

Suggested change

	rest1 = sf_restaurants.add(
	{
	"name": "Golden Gate Pizza",
	"type": "pizza",
	"owner_id": "Mario Rossi",
	}
	)[1]
	_, rest1 = sf_restaurants.add(
	{
	"name": "Golden Gate Pizza",
	"type": "pizza",
	"owner_id": "Mario Rossi",
	}
	)