spylib 0.1.1

A library to facilitate interfacing with Shopify's API

Shopify Python Library - SPyLib

The Shopify python library or SPyLib, simplifies the use of the Shopify services such as the REST and GraphQL APIs as well as the OAuth authentication. All of this is done asynchronously using asyncio.

Features:

• Shopify admin API
  • REST API
    • Rate limit handling per API key
    • Retry mechanism for 5xx status codes from Shopify
  • GraphQL API
• Shopify OAuth
  • Provide FastAPI router ready to handle the OAuth process
  • Handle the whole Oauth process
  • Callback validation without the need to track nonce locally
  • Provide post installation and post login function injection for full control

Installation

pip install spylib

Contributing

If you want to contribute a small change/feature, the best is to just create a PR with your changes. For bigger changes/features it's best to open an issue first and discuss it to agree on the code organization and overall implementation before spending too much time on the code, unless you want to keep it in your own forked repo.

Setting up the development environment

We use the python poetry package to manage this package. Follow the official instructions to install poetry on your system then once you clone this repository just just need to do the following to install the dependencies from the development environment, as well as install spylib in editable mode:

poetry install

Then you can start monitoring the code for changes and run the test suite this way:

poetry shell
scripts/test_watch.sh

Tutorial

Shopify admin API

The Shopify API version is defined here and is typically the latest stable version.

Initialize store instance

All the API calls to Shopify are done using an instance of the Store class. This class provides methods to perform Rest and GraphQL calls.

from spylib import Store

store = Store.load(store_id='mystore1', name='mystore', access_token='shppa_7a1e466ab2a')

Each store instance is defined using a store_id identifying the store uniquely, a name corresponding to the Shopify store subdomain such that https://<name>.myshopify.com, and the access_token or offline token to be used in the calls.

Rest API

The Store instance tracks and handles the rate limit for Rest calls. When the rate limit is hit (status code 429), the Store instance will estimate how quickly the available calls are restored and it will randomly retry one of the calls that were stalled.

Besides the rate limit, the Store will also automatically retry calls failing with a status code 5xx which are due to failures of Shopify or the network. All other status codes are assumed to be errors of the implementation and therefore shouldn't be needlessly retried but instead raise an exception right away.

Example GET request to retrieve all the tags on a customer account:

custid = 1234567890
customerjson = await store.shoprequest(
    goodstatus=200,
    method='get',
    debug=f'Couldn\'t get tags for customer id {custid} from Shopify.',
    endpoint=f'/customers.json?fields=tags&ids={custid}',
)

The goodstatus defines the expected successful status code. Any other status code will be considered to be an error and will raise an exception. In case the Rest call cannot go through, the debug message is used to provide an app specific messaging along with the generic exception message.

The method and endpoint define the Rest API call with the endpoint being the last part of the URL.

Example POST request to update tags on a customer account:

custdata = await store.shoprequest(
    goodstatus=200,
    method='put',
    debug=(f'Couldn\'t update tags for customer id {custid} from Shopify.'),
    endpoint=f'/customers/{custid}.json',
    json={'customer': {'id': custid, 'tags': 'mytag, yourtag, hertag, histag'}},
)

The body of the POST request is given as a dictionary to the json argument. Similarly the headers can be passed as a dictionary to the headers argument. The URL parameters can be passed as part of endpoint although it is preferrable to provide a dictionary to the params argument.

GraphQL API

To perform a GraphQL call, one must simply provide the query string. The query name of the argument matches the GraphQL specification for the field name that corresponds to a string with the query or mutation code in the GraphQL format.

The variables of the GraphQL query are passed as a dictionary to the variables argument.

See this example retrieving the list of webhooks defined for a store with a given callback URL:

query = """query getWebhooks($callbackUrl: URL!) {
  webhookSubscriptions(callbackUrl: $callbackUrl, first: 1) {
    edges {
      node {
        id
        callbackUrl
      }
    }
  }
}"""

res = await store.execute_gql(
    query=QUERY,
    variables={'callbackUrl': 'https://my.app.com/webhooks'},
)
webhook_nodes = res['data']['webhookSubscription']['edges']

Shopify OAuth

Rather than reimplementing for each app the Shopify OAuth authentication one can simple get a FastAPI router that provides the install and callback endpoints ready to handle the whole OAuth process. You just need to call init_oauth_router such that:

from spylib.oauth import init_oauth_router, OfflineToken, OnlineToken

async def my_post_install(storename: str, offline_token: OfflineToken):
    """Function handling the offline token obtained at the end of the installation"""
    # Store to database
    pass

async def my_post_login(storename: str, online_token: OnlineToken):
    """Function handling the online token obtained at the end of the user login"""
    # Store to database
    pass

oauth_router = init_oauth_router(
    app_scopes=['write_orders', 'write_products'],
    user_scopes=['read_orders', 'write_products'],
    public_domain='my.app.com',
    private_key='KEY_FOR_OAUTH_JWT',
    post_install=my_post_install,
    post_login=my_post_login,
)

The app_scopes are for the offline token and the user_scopes for the online token. The public_domain is used to set the callback URL used in the OAuth process.

This library uses a JWT encoded nonce to avoid the need for a database or some other mechanism to track the nonce. This JWT has an expiration time and is unique for each OAuth process making it a valid nonce mechanism. The private_key parameter defines the key used to encode and decode this JWT.

The post_install and post_login provide a way to inject functions handling the result of the installation and the login processes respectivaly. They are meant in particular to record the offline and online tokens in your app's database. They can be synchronous or asynchronous functions taking the storename and the token as arguments.