Added support for filtered aggregator

[se7entyse7en]

Adds support for filtered aggregator. Usage example:

agg = filtered(Filter(dimension='label', value='impression'), count('count'))
query.groupby(datasource="kafka", granularity="minute", intervals="2015-09-01T00:00:00.000/2015-09-30T00:00:00.000", aggregations={'impressions': agg, 'evs': count('events')})

[se7entyse7en]

Addresses issue #27.

[xvrl]

Not sure if this affects anything, but before, if kwarks already contained a name it would use that instead of overriding it, since we passed the name as the first key/value pair to dict(...)

[se7entyse7en]

Ok, I'm gonna check it, maybe I could also write some tests for this module in order to ensure that the change does not brake compatibility. What do you think?

[xvrl]

Of course, tests would be awesome!

[se7entyse7en]

I updated the pull-request by adding some tests for the filtered aggregator. This branch is rebased on pull-request #32, so this should be done after that. I made two different pull-request so that it is possible to see that no previous behaviour is broken.

[se7entyse7en]

Can someone please have a look on this?

[se7entyse7en]

Rebased on master and fixed test.

[xvrl]

Thanks @se7entyse7en, do you mind signing our CLA before I can merge this? You can do so here http://druid.io/community/cla.html – it also applies to metamx/pydruid

[se7entyse7en]

@xvrl ok! done!