It's for a batch job. I'll think about possible workarounds then (perhaps I
can have a prestage process that reads the necessary info from the files
and passes as an argument). Thanks everyone!

On Wed, Aug 23, 2017 at 9:17 PM, Chamikara Jayalath <[email protected]>
wrote:

> Ah, actually for this feature I think we need a BigQuery custom sink. Which
> also doesn't have an ETA currently.
>
> Thanks,
> Cham
>
> On Wed, Aug 23, 2017 at 6:14 PM Chamikara Jayalath <[email protected]>
> wrote:
>
> > We have to develop a BigQuery BoundedSource for this first. So I expect
> > support for Python SDK to be at least few months away.
> >
> > Thanks,
> > Cham
> >
> >
> > On Wed, Aug 23, 2017 at 6:10 PM Asha Rostamianfar
> > <[email protected]> wrote:
> >
> >> ah, ok..is there an existing JIRA entry with some approximate ETA? (i.e.
> >> weeks, months, or quarters).
> >>
> >> Thanks again.
> >>
> >> On Wed, Aug 23, 2017 at 8:58 PM, Reuven Lax <[email protected]>
> >> wrote:
> >>
> >> > Not yet, only in Java so far.
> >> >
> >> > On Wed, Aug 23, 2017 at 5:53 PM, Asha Rostamianfar <
> >> > [email protected]> wrote:
> >> >
> >> > > Thanks, Eugene! Is this also supported in the python SDK?
> >> > >
> >> > > On Wed, Aug 23, 2017 at 8:39 PM, Eugene Kirpichov <
> >> > > [email protected]> wrote:
> >> > >
> >> > > > Yes, this is possible using the BigQueryIO.write().to(
> >> > > DynamicDestinations)
> >> > > > API. It allows you to write different values to different tables
> >> with
> >> > > > different schemas.
> >> > > > On Wed, Aug 23, 2017, 2:51 PM Asha Rostamianfar
> >> > > > <[email protected]>
> >> > > > wrote:
> >> > > >
> >> > > > > Hi,
> >> > > > >
> >> > > > > I'm wondering whether it's possible to dynamically generate a
> >> > BigQuery
> >> > > > > schema based on input. For instance, the fields would be
> >> specified in
> >> > > one
> >> > > > > or more input files that are read and processed as part of the
> >> > > pipeline.
> >> > > > >
> >> > > > > Thanks,
> >> > > > > Asha
> >> > > > >
> >> > > >
> >> > >
> >> >
> >>
> >
>

Reply via email to