ENH: Save BigQuery account credentials in a hidden user folder #83
Conversation
Codecov Report
@@ Coverage Diff @@
## master #83 +/- ##
===========================================
- Coverage 73.38% 28.01% -45.38%
===========================================
Files 4 4
Lines 1563 1578 +15
===========================================
- Hits 1147 442 -705
- Misses 416 1136 +720
Continue to review full report at Codecov.
|
pandas_gbq/gbq.py
Outdated
| # Create a pandas_gbq directory in an application-specific hidden | ||
| # user folder on the operating system. | ||
| if not os.path.exists(config_path): | ||
| os.mkdir(config_path) |
There was a problem hiding this comment.
Is this recursive if .config doesn't exist?
There was a problem hiding this comment.
Great catch! os.mkdir is not recursive. Fixed.
| else: | ||
| config_path = os.path.join(os.path.expanduser('~'), '.config') | ||
|
|
||
| config_path = os.path.join(config_path, 'pandas_gbq') |
There was a problem hiding this comment.
You know this better than I do, but is there a case for having the default path be the default gcloud path, rather than anything specific to pandas?
There was a problem hiding this comment.
This is a great question! I've given this some thought and I don't feel comfortable writing default credentials into the gcloud path in case some users don't want to have their default gcloud credentials set (and I'm also worried about colliding with gcloud). Alternatively, users could run gcloud auth application-default login if they want to configure default credentials and the default gcloud credentials should be picked up. In that case, we shouldn't hit this code path. https://github.com/pydata/pandas-gbq/blob/master/pandas_gbq/gbq.py#L222
@tswast Thoughts?
There was a problem hiding this comment.
+1 to not clobbering gcloud credentials. The token we save doesn't have the full Google Cloud Platform scope so that would break other applications that depend on those credentials.
If folks set up application default credentials it won't hit this code anyway.
There was a problem hiding this comment.
users could run gcloud auth application-default login
If folks set up application default credentials it won't hit this code anyway.
👍
|
I've also made the changes required for pandas.io.gbq in the following branch: I'll hold off on creating a PR in pandas until this is merged. |
parthea
force-pushed
the
store-credentials-in-user-directory
branch
3 times, most recently
from
e455f18 to
0c713b6
Compare
parthea
force-pushed
the
store-credentials-in-user-directory
branch
from
0c713b6 to
d0cd4f2
Compare
docs/source/changelog.rst
Outdated
| @@ -1,12 +1,12 @@ | |||
| Changelog | |||
| ========= | |||
|
|
|||
| 0.2.1 / 2017-??-?? | |||
| 0.3.0 / 2017-??-?? | |||
There was a problem hiding this comment.
I wonder if we should do an incremental release before 0.3.0 if we plan to have that be the one where we swap to the google-cloud-python libraries? Seems like that PR has stalled and I don't have time to rebase it either.
Closes #41
This PR adds ability for users to choose the location where BigQuery user account credentials will be stored via the
credentials_pathparameter.This PR also changes the location of the file
bigquery_credentials.datthat contains the BigQuery user account credentials. Previously this file was saved in the current working directory. The credentials file will now be stored in an application-specific hidden user folder on the operating system. On windows, this path is specified in theAPPDATAenvironment variable as defined here which is typically'C:\Documents and Settings\username\Application Data\pandas-gbq\'. Otherwise, credentials will be stored in'~/.config/pandas-gbq/'. If the filebigquery_credentials.datis found in the current working directory, the file will be moved to the new location.@tswast @jreback Please could you take a look at your earliest convenience?