BigQuery: Client.query modifies job_config object #9727
Description
Steps to reproduce
- Create a
QueryJobConfigobject. - Pass the object to the
Client.querymethod. - Observe that the configuration object has changed.
Code example
from google.cloud import bigquery
client = bigquery.Client()
config = bigquery.QueryJobConfig(dry_run=True)
job = client.query(
"""
SELECT name, SUM(number) AS total_people
FROM `bigquery-public-data.usa_names.usa_1910_current`
GROUP BY name
""",
job_config=config
)
print(job.total_bytes_processed)
config.dry_run = False
job = client.query(
"""
SELECT name, SUM(number) AS total_people
FROM `bigquery-public-data.usa_names.usa_1910_current`
GROUP BY name
""",
job_config=config
)
job.result()Stack trace
---------------------------------------------------------------------------
BadRequest Traceback (most recent call last)
<ipython-input-17-97f4f20faa6a> in <module>
----> 1 job.result()
~/miniconda3/envs/scratch/lib/python3.7/site-packages/google/cloud/bigquery/job.py in result(self, timeout, page_size, retry, max_results)
2937 """
2938 try:
-> 2939 super(QueryJob, self).result(timeout=timeout)
2940
2941 # Return an iterator instead of returning the job.
~/miniconda3/envs/scratch/lib/python3.7/site-packages/google/cloud/bigquery/job.py in result(self, timeout, retry)
732 self._begin(retry=retry)
733 # TODO: modify PollingFuture so it can pass a retry argument to done().
--> 734 return super(_AsyncJob, self).result(timeout=timeout)
735
736 def cancelled(self):
~/miniconda3/envs/scratch/lib/python3.7/site-packages/google/api_core/future/polling.py in result(self, timeout)
125 # pylint: disable=raising-bad-type
126 # Pylint doesn't recognize that this is valid in this case.
--> 127 raise self._exception
128
129 return self._result
BadRequest: 400 Cannot explicitly modify anonymous table swast-scratch:_def2aa82a75fc33513bfb65968607e1f49148f83.anon561d4b35377f56c5960a4a49be253d3c90c120db
(job ID: 2692472c-1213-4716-8393-d9c49fd7b678)
-----Query Job SQL Follows-----
| . | . | . | . | . |
1:
2: SELECT name, SUM(number) AS total_people
3: FROM `bigquery-public-data.usa_names.usa_1910_current`
4: GROUP BY name
5:
| . | . | . | . | . |
When you run a job, the configuration object may be modified. In this example, it sets the destination table. It is unexpected that the original Python object would be modified.
I propose that the query (and other "job" methods on client) be updated to make a deep copy of the the job_config argument before passing it on to the job constructor.