graphql-python-archive / graphql-django-view Goto Github PK

Assuming you have these two queries:

First Route makes this query:

{
  viewer {
      items(first: 10) {
        edges {
          node {
           id
           title
          }
        }
      }
  }
}

Second Route makes this query:

{
  viewer {
      items(first: 10) {
        edges {
          node {
           id
           title
           description
          }
        }
      }
  }
}

with the default relay network layer, you will end up with 10 (n) node requests like so:

node(id: *id) {
  description
}

This is bad. So, the wonderful @nodkz wrote a great network layer, that when paired with some express-js middleware, handles batching.

That's wonderful if we (the collective python graphql users) were using express-js, but alas we are not so we need to implement a solution for such.

The JS implementation can be found here:

https://github.com/nodkz/react-relay-network-layer/blob/master/src/express-middleware/graphqlBatchHTTPWrapper.js

I plan to take a first pass at this tomorrow, but figured I would share with the public in case this has been on anyone else's mind.

        elif content_type == 'application/json':
            try:
                request_json = json.loads(request.body.decode())
                assert isinstance(request_json, dict)
                return request_json
            except:
                raise HttpError(HttpResponseBadRequest('POST body sent invalid JSON.'))

i believe this code should be refactored

pls.

This may be an issue with other execution paths in graphene as well.

It seems that schema plugins (such as the DjangoDebugPlugin) only work if you call schema.execute - they are not called if you use graphql-django-view.

https://github.com/graphql-python/graphql-django-view/blob/master/graphql_django_view/__init__.py#L93 will tell the user:

<HttpResponseBadRequest status_code=400, "application/json">
b'{"errors":[{"message":"Must provide query string."}]}'

if the content-type is unexpected. An informative exception should be raised instead, eg:

raise HttpError(HttpResponseBadRequest('Unrecognized content-type: text/plain. Please send one of [application/graphql, application/json, application/x-www-form-urlencoded, multipart/form-data]'))

Currently, query is not parsed if the content_type is multipart/form-data.

Submitting a PR to handle this case. Uploaded files should then be accessible in mutate_and_get_payload in info.request_context.POST[filename]

In this line 96[https://github.com/graphql-python/graphql-django-view/blob/master/graphql_django_view/init.py#L96] it seens that we should get middlewares from self.schema and pass it as parameter to execute (from graphql-core[https://github.com/graphql-python/graphql-core/blob/features/next/graphql/execution/executor.py#L27])

How can i set cookie through HttpReponse? for exmaple:

response = HttpReponse('something')
reponse.set_cookie('my_cookie', 'cookie_value')
return reponse

Either using django's logger or through some hook there needs to be a way that things like sentry can know about these uncaught exceptions.

Issue is primarily in graphql_django_view/__init__.py in dispatch() of GraphQLView

except HttpError as e:
            response = e.response
            response['Content-Type'] = 'application/json'
            response.content = self.json_encode(request, {
                'errors': [self.format_error(e)]
            })
            return response

I am creating a new model within a mutation. I would like to set the model's created_by field to point to the current user (using request.user). However, AFAIK the request object is not available within a mutation's mutate_and_get_payload() method.

Note: It is entirely possible that I'm going about this the wrong way, I'm still new to GraphQL.