Add utility for building GQL queries from Python dicts/sequences

[jlowin]

Building GQL as strings is fragile and difficult to trap errors.

In lieu of building a tool that validates GQL against a schema, I made a simple helper for building up a GQL query from simple Python objects. The query can be parsed into a GQL string, which itself can be validated against a schema using one of many tools.

This tools expects a combination of dicts and lists (or other sequences) and stitches them together into a query. The parser is useable with strings, but also has a more advanced GQLObject

If you use the GQLObject, you get three advantages:

1. GQLObjects can be treated as "lite" schemas, with class attributes corresponding to available fields
2. GQLObjects can change their string representation without affecting the object itself, allowing schemas to change without code changing
3. GQLObjects can be called, allowing a nice syntax for creating arguments.

For example, using strings:

q = {
    'query': {
        'authors': [
            'id',
            'name'
        ],
        'books(sort: ASC)': [
            'id',
            'title',
            {'author': {
                'name'
            }}
        ]
    }
}
parsed = parse_graphql(q)

assert parsed == """
query {
    authors {
        id
        name
    }
    books(sort: ASC) {
        id
        title
        author {
            name
        }
    }
}
""".strip()

[cicdw]

f-strings messin' up the tests

[jlowin]

🤦‍♂️

[jlowin]

and then an unsorted dict issue on Py < 3.6... this is why we can't have nice things.

[cicdw]

LGTM! I'll incorporate this into the Client tomorrow on the plane 👍

[jlowin]

👍

[joshmeek]

I like this very much, really clean!