SchemaField cannot initialize from __repr__

[WesRoach]

Environment details

• OS type and version: macOS Big Sur 11.6.1
• Python version: 3.9.7
• pip version: 21.3.1
• google-cloud-bigquery version: 2.29.0

Description

Prior versions of google-cloud-bigquery (2.3.1) allowed SchemaField instantiation based on an existing SchemaField __repr__.

Use Case

1. System generates list of SchemaField and passes it through a Jinja template as part of building hundreds of Airflow DAGs.
2. Jinja prints __repr__ from the SchemaField into the templated .py.
3. Airflow picks up the .py files and executes them.
4. Execution results in the below error.

Code example

In version 2.3.1

from google.cloud.bigquery import SchemaField
from google.cloud import bigquery
bigquery.__version__
'2.3.1'

# This works
field = SchemaField('name', 'STRING', 'REQUIRED', 'Name') 
field.__repr__()
"SchemaField('name', 'STRING', 'REQUIRED', 'Name', (), ())"
# Jinja prints ^^ to a .py file
# When the file is executed the SchemaField is created without issue

Version 2.29.0:

from google.cloud.bigquery import SchemaField
from google.cloud import bigquery
bigquery.__version__
'2.29.0'

# This raises an error for the tuple in the last field
field = SchemaField('name', 'STRING', 'REQUIRED', 'Name') 
field.__repr__()
"SchemaField('name', 'STRING', 'REQUIRED', 'Name', (), ())"
# Jinja prints ^^ to a .py file
# When the file is executed, an error is raised:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/wes.roach/.pyenv/versions/upgr-dbt/lib/python3.9/site-packages/google/cloud/bigquery/schema.py", line 125, in __init__
    policy_tags.to_api_repr() if policy_tags is not None else None
AttributeError: 'tuple' object has no attribute 'to_api_repr'

Setting the policy_tags to None still results in ().

field = SchemaField('name', 'STRING', 'REQUIRED', 'Name', (), None)
field.__repr__()
"SchemaField('name', 'STRING', 'REQUIRED', 'Name', (), ())"

[tswast]

@WesRoach Thanks for the report! This does appear to be a bug introduced when policy tags were added.

[tswast]

@plamut Potential project for you if you have cycles? It'd be good to get this in v2.

[plamut]

@tswast I'll take a look, yes.

[plamut]

This does appear to be a bug introduced when policy tags were added.

Confirmed, the regression was caused by #983.