Error when running BiqQuery query with APPROX_QUANTILES

[tebeka]

1. OS type and version - Ubuntu 16.04.1 LTS
2. Python version and virtual environment information python --version - 2.7.12
3. google-cloud-python version pip show google-cloud, pip show google-<service> or pip freeze - 0.21.0
4. Stacktrace if available

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-23-1528bd7c380f> in <module>()
----> 1 job.rows()

/home/mtebeka/python-docs-samples/appengine/flexible/hello_world/venv/local/lib/python2.7/site-packages/google/cloud/bigquery/query.pyc in rows(self)
    225         :returns: fields describing the schema (None until set by the server).
    226         """
--> 227         return _rows_from_json(self._properties.get('rows', ()), self.schema)
    228 
    229     @property

/home/mtebeka/python-docs-samples/appengine/flexible/hello_world/venv/local/lib/python2.7/site-packages/google/cloud/bigquery/_helpers.pyc in _rows_from_json(rows, schema)
    114 def _rows_from_json(rows, schema):
    115     """Convert JSON row data to rows with appropriate types."""
--> 116     return [_row_from_json(row, schema) for row in rows]
    117 
    118 

/home/mtebeka/python-docs-samples/appengine/flexible/hello_world/venv/local/lib/python2.7/site-packages/google/cloud/bigquery/_helpers.pyc in _row_from_json(row, schema)
    105         if field.mode == 'REPEATED':
    106             row_data.append([converter(item, field)
--> 107                              for item in cell['v']])
    108         else:
    109             row_data.append(converter(cell['v'], field))

/home/mtebeka/python-docs-samples/appengine/flexible/hello_world/venv/local/lib/python2.7/site-packages/google/cloud/bigquery/_helpers.pyc in _float_from_json(value, field)
     33     """Coerce 'value' to a float, if set or not nullable."""
     34     if _not_null(value, field):
---> 35         return float(value)
     36 
     37 

TypeError: float() argument must be a string or a number

5. Steps to reproduce - Run a query with APPROX_QUANTILES.
6. Code example

from google.cloud import bigquery

client = bigquery.Client(project='PROJECT-ID')

sql = '''
SELECT
APPROX_QUANTILES(
  CASE
     WHEN ctr_before = 0 THEN NULL
     ELSE (ctr_after - ctr_before) / ctr_before
  END,
  100) quantile
FROM
 `ltv.user_ctr`
 '''
job = client.run_sync_query(sql)
job.use_legacy_sql = False
job.run()
job.rows

[daspecster]

Thanks for reporting @tebeka!

@tseaver @dhermes I think this needs to support array types in the response right?
See: https://cloud.google.com/bigquery/docs/reference/standard-sql/functions-and-operators#approx_quantiles

[tseaver]

@tebeka Can you print the value of job._properties['rows'] in the failing case?

[tebeka]

In [8]: job._properties['rows']
Out[8]: 
[{u'f': [{u'v': [{u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-1.0'},
     {u'v': u'-0.8864864864864864'},
     {u'v': u'-0.8303571428571428'},
     {u'v': u'-0.7902330743618202'},
     {u'v': u'-0.7543859649122807'},
     {u'v': u'-0.7235023041474654'},
     {u'v': u'-0.693069306930693'},
     {u'v': u'-0.6666666666666666'},
     {u'v': u'-0.6375'},
     {u'v': u'-0.6117216117216117'},
     {u'v': u'-0.5855263157894737'},
     {u'v': u'-0.5608856088560885'},
     {u'v': u'-0.5353535353535354'},
     {u'v': u'-0.5114155251141552'},
     {u'v': u'-0.4864864864864864'},
     {u'v': u'-0.4625'},
     {u'v': u'-0.43750000000000006'},
     {u'v': u'-0.41442715700141447'},
     {u'v': u'-0.3902308105206656'},
     {u'v': u'-0.3653846153846154'},
     {u'v': u'-0.34146341463414626'},
     {u'v': u'-0.31677018633540377'},
     {u'v': u'-0.2923766816143497'},
     {u'v': u'-0.2666666666666667'},
     {u'v': u'-0.24242424242424238'},
     {u'v': u'-0.21590383726306067'},
     {u'v': u'-0.19047619047619052'},
     {u'v': u'-0.16396903589021816'},
     {u'v': u'-0.13746630727762796'},
     {u'v': u'-0.10954235637779944'},
     {u'v': u'-0.08196721311475408'},
     {u'v': u'-0.05263157894736842'},
     {u'v': u'-0.02325581395348833'},
     {u'v': u'0.0038560411311053186'},
     {u'v': u'0.036920659858601754'},
     {u'v': u'0.06928066037735833'},
     {u'v': u'0.10179640718562884'},
     {u'v': u'0.1366666666666667'},
     {u'v': u'0.17073170731707327'},
     {u'v': u'0.20923076923076936'},
     {u'v': u'0.2484848484848484'},
     {u'v': u'0.28813559322033905'},
     {u'v': u'0.33333333333333326'},
     {u'v': u'0.3749999999999999'},
     {u'v': u'0.42051282051282063'},
     {u'v': u'0.46875000000000006'},
     {u'v': u'0.5217391304347826'},
     {u'v': u'0.5766423357664235'},
     {u'v': u'0.6363636363636365'},
     {u'v': u'0.7'},
     {u'v': u'0.7692307692307692'},
     {u'v': u'0.8400000000000001'},
     {u'v': u'0.9230769230769231'},
     {u'v': u'1.0'},
     {u'v': u'1.103723404255319'},
     {u'v': u'1.206896551724138'},
     {u'v': u'1.333333333333333'},
     {u'v': u'1.456140350877193'},
     {u'v': u'1.6085043988269796'},
     {u'v': u'1.7777777777777775'},
     {u'v': u'1.9889502762430937'},
     {u'v': u'2.210526315789473'},
     {u'v': u'2.5000000000000004'},
     {u'v': u'2.857142857142857'},
     {u'v': u'3.352941176470588'},
     {u'v': u'4.0'},
     {u'v': u'5.083333333333333'},
     {u'v': u'7.2'},
     {u'v': u'1987.9999999999998'}]}]}]

[tseaver]

I believe #2787 addressed this issue: it fixed marshalling arrays.

The imminent 0.22.0 release of google-cloud-bigquery will contain the fix. @tebeka are you set up to test with the current git master branch?

[lukesneeringer]

Since we believe this has already been fixed, I am going to close this. Please re-open if it is still an issue on the current version of the library.