GraphQL support

[maartenbreddels]

This adds the vaex-graphql package, that adds graphql support to query vaex.

GraphQL is really nice to have a standard way to query data, that is not as general as SQL, but also allows us to implement binby, and anything custom we need. Interfacing with frontends is really simple, a ton of libraries are out there.

e.g.

    result = df.graphql.execute("""
    {
        df {
            min {
                x
                y
            }
            mean {
                x
                y
            }
            max {
                x
                y
            }
        }
    }
    """)
    assert data['df']['min']['x'] == df.x.min()

Although it does not depend on tornado, it adds an option to quickly show this using the graphiql interface, you'd have to install:

$ pip install graphene-tornado

Then run

$ python -m vaex.graphql ~/datasets/nyc_taxi2015.arrow

Which should give you a url, to a webinterface:

All aggregations are done serially for now. Maybe we should fix that after

• graphql-core-next is released
• A solution to Question: how to perform an operation before the futures will be resolved graphql-python/graphql-core#64 is found
• graphene-tornado supports graphql-core-next Plans to support graphql-core-next and graphene >= 3 graphql-python/graphene-tornado#38