pygqlc

Python client for graphql APIs

Scope

This is an open source project, please feel free to fork it and PR to contribute with the community! Repo for the project: https://github.com/valiot/pygqlc

Installation

Requirements:

• Python 3.9+
• Pipenv

Install directly from pypi:

$ cd <my-project-dir>
$ pipenv --python 3.7 # or 3.6
$ pipenv shell
$ pipenv install pygqlc
$ python
$ >> import pygqlc
$ >> print(pygqlc.name)

If you get "pygqlc" printed in the python repl, the installation succeded!

Usage

import os
from pygqlc import GraphQLClient
gql = GraphQLClient()
gql.addEnvironment(
    'dev',
    url=os.environ.get('API'), # should be an https url
    wss=os.environ.get('WSS'), # should be an ws/wss url
    headers={'Authorization': f"Bearer {os.environ.get('TOKEN')}"},
    default=True)

From now on, you can access to the main API:

gql.query, gql.mutate, gql.subscribe

For queries:

query = '''
query{
  authors{
    name
  }
}
'''
data, errors = gql.query( query )

For mutations:

create_author = '''
mutation {
  createAuthor(){
    successful
    messages{field message}
    result{id insertedAt}
  }
}
'''
data, errors = gql.mutate( create_author )

For subscriptions:

def on_auth_created(message):
  print(message)

sub_author_created = '''
subscription{
  authorCreated{
    successful
    messages{field message}
    result{id insertedAt}
  }
}
'''
# unsub may be None if subscription fails (no internet connection, host unreachable, bad subscription doc, etc)
unsub = gql.subscribe(sub_author_created, callback=on_auth_created)
...
# when finishing the subscription:
unsub()
# when finishing all subscriptions:
gql.close()

The subscribe method, returns an unsubscribe function, this allows to stop subscriptions whenever needed.

After finishing all subscriptions, the method GraphQLClient.close() should be called to close correctly the open GQL/websocket connections.

To reset all subscriptions and websocket connection use the method GraphQLClient.resetSubsConnection().

To be noted:

All main methods from the API accept a variables param. it is a dictionary type and may include variables from your queries or mutations:

query_with_vars = '''
query CommentsFromAuthor(
  $authorName: String!
  $limit: Int
){
  author(
    findBy:{ name: $authorName }
  ){
    id
    name
    comments(
      orderBy:{desc: ID}
      limit: $limit
    ){
      id
      blogPost{name}
      body
    }
  }
}
'''

data, errors = gql.query(
  query=query_with_vars,
  variables={
    "authorName": "Baruc",
    "limit": 10
  }
)

There is also an optional parameter flatten that simplifies the response format:

# From this:
response = {
  'data': {
    'authors': [
      { 'name': 'Baruc' },
      { 'name': 'Juan' },
      { 'name': 'Gerardo' }
    ]
  }
}
# To this:
authors = [
  { 'name': 'Baruc' },
  { 'name': 'Juan' },
  { 'name': 'Gerardo' }
]

Simplifying the data access from this:

response['data']['authors'][0]['name']

to this:

authors[0]['name']

It is query(query, variables, flatten=True) by default, to avoid writing it down everytime

The (_, errors) part of the response, is the combination of GraphQL errors, and communication errors, simplifying validations, it has this form:

errors = [
  {"field": <value>, "message":<msg>},
  {"field": <value>, "message":<msg>},
  {"field": <value>, "message":<msg>},
  ...
]

The field Attribute it's only available for GraphQL errors, when it is included in the response, so it's suggested that every mutation has at least this parameters in the response:

mutation{
  myMutation(<mutationParams>){
    successful
    messages{
      field
      message
    }
    result{
      <data of interest>
    }
  }
}

Post timeout:

You can set a post timeout to avoid an inactive process.

Use gql.setPostTimeout(seconds), or directly in the environment gql.addEnvironment(post_timeout=seconds). Default port_timeout is 60 seconds

Websocket timeout:

You can set a websocket timeout to keep subscriptions alive.

Use gql.setTimeoutWebsocket(seconds), or directly in the environment gql.addEnvironment(timeoutWebsocket=seconds). Default timeoutWebsocket is 60 seconds

for mantainers:

Initial configuration

first of all, ensure you have configured poetry repositories correctly: poetry config repositories.valiot https://pypi.valiot.io/

and their credentials:

For private valiot's pypi:

poetry config http-basic.valiot <username> <password>

(ask adrian to send you the proper username and password for this step)

And for public pypi:

poetry config pypi-token.pypi <pypi-token>

(ask adrian or baruc to generate a token for you)

then,

regular publish steps (after initial configuration)

deploy using:

poetry version <patch | minor | major>

then publish to valiot's private pypi:

poetry publish --build -r valiot # build and PUBLISH TO PRIVATE VALIOTs PYPI

or:

poetry publish -r valiot

(if you already built the package)

then publish to public pypi:

poetry publish

After release, publish to github:

cat pygqlc/__version__.py

gh release create

gh release upload v<#.#.#> ./dist/pygqlc-<#.#.#>-py3-none-any.whl

and don't forget to keep the CHANGELOG.md updated!