[Bigquery insert data error] table.insert_data() issue

I got this error while using table.insert_data(rows)

--> 246     if not value.tzinfo:
    247         value = value.replace(tzinfo=UTC)
    248     # Regardless of what timezone is on the value, convert it to UTC.

AttributeError: 'str' object has no attribute 'tzinfo'

The doc of table.insert_data() said rows is list of tuple, so this is my schema

sql_schema = [('viewer_id', 'INTEGER'),
              ('ip','STRING'),
              ('author_id', 'INTEGER'),
              ('post_id', 'STRING'),
              ('time', 'TIMESTAMP')]

And this is a test data:

[['25512',
  '126.107.92.215',
  '6524',
  '9f4047d3-7bf8-406c-823c-9803b4730fb9',
  '2015-12-14 18:29:01.245142'],
 ['24534',
  '87.13.200.79',
  '29096',
  'e09baadc-9647-4cde-9019-7c6846dfd662',
  '2015-12-14 18:29:01.245237']]

I know '2015-12-14 18:29:01.245237' is a string which doesn’t have tzinfo field. Do I have to convert it to Datetime object every time I try to load time data into bigquery? Do I have better options?

Thanks

[dhermes]

Sorry the doc is vague:

:type rows: list of tuples
:param rows: row data to be inserted

As an example, our system test uses the schema

full_name = bigquery.SchemaField('full_name', 'STRING',
                                 mode='REQUIRED')
age = bigquery.SchemaField('age', 'INTEGER', mode='REQUIRED')
now = bigquery.SchemaField('now', 'TIMESTAMP')

to create the table

table = dataset.table(TABLE_NAME, schema=[full_name, age, now])

and then uses rows as a list of triples

ROWS = [
    ('Phred Phlyntstone', 32, NOW),
    ('Bharney Rhubble', 33, NOW + datetime.timedelta(seconds=10)),
    ('Wylma Phlyntstone', 29, NOW + datetime.timedelta(seconds=20)),
    ('Bhettye Rhubble', 27, None),
]

that correspond to the schema (i.e. a required string, a required integer and an optional timestamp). After does this, they are inserted with a set of row IDS:

ROW_IDS = [1, 2, 3, 4]
errors = table.insert_data(ROWS, ROW_IDS)

(For the record, I wouldn’t have expected you to have looked here, just wanted to provide an example.)

---

1. As of right now, we expect a datetime.datetime object for a TIMESTAMP schema field.
2. What situation are you in that you have lots of datetime’s as strings rather than native? (So I can understand the use-case)
3. Could you provide the code snippets you used to generate the error? It’s unclear how sql_schema and your “test data” were used.

[dhermes]

Thanks