BigQuery: Converting query results to dataframe fails to convert date columns

[darwinyip]

Environment details

google-api-python-client==1.9.3
google-cloud-bigquery==1.25.0
pyarrow==0.17.1

Steps to reproduce

1. Issuing a select statement that returns a date column and converting the results to dataframe fails to convert the date column and leaves it as string

Code example

select = "select DATE(date) as dt from [project:dataset.table]"
client = bigquery.Client()
job_config = bigquery.QueryJobConfig(use_legacy_sql=True)
select_job = client.query(select, job_config=job_config)
select_job.result()
df = select_job.to_dataframe()
df.dtypes

# dt    object

[HemangChothani]

@darwinyip Could you try to set the dtypes argument to to_dataframe. We don't have any logic to convert from the API response types to pandas dtypes unless the dtypes argument is specified.

Example:

 expected_dtype = {cloumn_name:dtype}    # e.x: dtype = 'datetime64[ns]'
 df = select_job.to_dataframe(dtypes=expected_dtype)

 df.dtypes["column_name"].name   #check dtype of cloumn

The duplicate issue of #11

[darwinyip]

Manually setting the dtype works but it's not a very friendly user experience.

[HemangChothani]

@darwinyip Could you please comment on issue #11 as this is the duplicate issue and discussion is going on that thread so you can get better and faster reply, I have already linked and mentioned your comment there so closing this issue.