how to upload file about gql

[leszekhanusz]

@leahein did you make some progress on this ? I'll have some time in the following weeks so I will maybe start to implement this myself.

[leszekhanusz]

@leahein

Happy to work on a PR for this as well.

I think this would be a great addition to this library. I will accept your pull request in my own PR #70 if it comes with 100% test code coverage (and documentation).

Note: to create temporary files in the tests, you can use this method

You can create a test which will create a temporary text file, send this file in a mutation to the local created aiohttp server, the server will return the first line from the file.

Then another test with a mutation with a list of files which return a list of first lines.

[leahein]

I did start on the feature here, but it's not complete.
Right now there are utils to check whether a request contains files, and I started the transformation for sending them with the requests lib.

[KingDarBoja]

@leahein The link should be https://github.com/leahein/gql/tree/file-support :)

At the end, the base implementation to implement should be https://github.com/jaydenseric/graphql-multipart-request-spec

[leszekhanusz]

See also #179

And to be clear, it should be possible to support file uploads with the RequestsHTTPTransport. I'll accept a PR if it is well tested.

[leszekhanusz]

@maaft Could you please check if the PR #244 works well for you?

[leahein]

I would love to know whether gql supports this, as well.

[leahein]

It doesn't look like this library currently supports files. Is there any plan to support the graphql multipart request spec in the future?

Happy to work on a PR for this as well.

[KingDarBoja]

@leahein the library doesn't support file uploads but there is a repo which implements that for Django and Flask servers, would be worth checking it: graphene-file-upload.

[leahein]

@leahein the library doesn't support file uploads but there is a repo which implements that for Django and Flask servers, would be worth checking it: graphene-file-upload.

I was looking for file upload support for a python client that uses graphene-file-upload server-side.
I found this one that seems to support file uploads well.

[KingDarBoja]

@leahein best idea would be a separated repo to support this feature as this would be a third party integration with gql library, hence making easier to maintain the core features as there aren't many contributors around here 🤔

[leszekhanusz]

Now that the aiohttp transport is implemented in #70 It should be relatively easy to incorporate the implementation from aiogqlc

[leszekhanusz]

@leahein

Now that #70 has been merged, if you are still interested you should create your PR directly from the repo now.

[leahein]

@leszekhanusz Thanks for the details! I will work on adding file support 👍

[sreerajkksd]

@leszekhanusz would you be able to implement this feature in a month or two ? I'm really looking forward to have this feature in this library.

[leszekhanusz]

This is now available in release v3.0.0a3. Tests and feedback is welcome.

[maaft]

Is file-upload support planned for sync transport? I cannot use async io in my application.

[leszekhanusz]

@maaft
No, it is not planned. But you can use AIOHTTPTransport in a sync manner if you want.

[maaft]

The issue with AIOHTTPTransport is that it creates a new session everytime you execute a query/mutation. That sounds very inefficient and also I get random "Transport is already connected" Exceptions from time to time. Might be some race conditions maybe.

[leszekhanusz]

The issue with AIOHTTPTransport is that it creates a new session every time you execute a query/mutation. That sounds very inefficient

Not necessarily, you can execute multiple queries and mutations on the same session (See async usage and async advanced usage). If you want to do it synchronously, it is still possible by running loop.run_until_complete yourself like in Client::execute

and also I get random "Transport is already connected" Exceptions from time to time. Might be some race conditions maybe.

I suspect you try to execute queries from different threads or task with the same transport. If that is not the case, and if you can reproduce a minimal example, please submit an issue.

[maaft]

I suspect you try to execute queries from different threads or task with the same transport. If that is not the case, and if you can reproduce a minimal example, please submit an issue.

Yes, that is indeed the case. I'm also thinking about rewriting all my code to make use of asyncio (instead of threads & processes) but unfortunately I rely on external libraries (pytorch in this case) that make heavy use of threads and processes already. And there is no way around using pytorch sadly. It's the defacto standard in machine learning. But who knows, maybe it will work just fine with asyncio. I'll have to try.

Thanks!

[maaft]

@leszekhanusz yes, it works. Thank you!