This is the GraphQL API that powers the maintainer dashboard for the Gatsby Store and the relevant parts of the docs like the feedback widget. It handles calls to privileged services such as Shopify and MailChimp to avoid exposing API keys in a client-side app.

Services:

• Google Cloud Functions
• Prisma Cloud
• Shopify

API:

• GitHub

Frameworks:

• ExpressJS
• Prisma Client
• Serverless Framework

The way these services and frameworks work together is explained in more detail in the Architecture section below

All GraphQL calls for the store are sent to this endpoint.

For a full schema reference, see the schema.

All GraphQL calls from gatsbyjs.org are sent to this endpoint.

For a full schema reference, see the schema.

# Clone the repo.
git clone [email protected]:gatsbyjs/api.gatsbyjs.org.git

# Move into the newly cloned project.
cd api.gatsbyjs.org

# Install dependencies
yarn # or `npm install`

# Copy the example .env file into a real .env file
cp .env.EXAMPLE .env

The .env.EXAMPLE file contains a list of env variables used in various locations throughout the repository. Set each variable's value per your environment setup and credentials in .env.

The serverless-offline package only works to emulate AWS serverless code, since transitioning to GCP it's less ergonomic to test locally.

You can invoke a function locally with this command, (from to the serverless documentation):

serverless invoke local --function public --data '{ "query": "query { ping }" }'

However there are some issues circulating on the serverless-google-cloudfunctions package around better support for local emulation and development.

If you want to send straight-up POST requests so you can wear sunglasses indoors and pretend you’re a hacker, you can also send cURL requests like so:

curl \
  -H "Content-Type: application/json" \
  --data '{ "query": "query { getFeedback { id, comment } }" }' \
  -X POST http://localhost:3000/public

curl \
  -H "Content-Type: application/json" \
  --data '{
      "query": "query($user:String!) { contributorInformation(githubUsername:$user) { totalContributions } }",
      "variables": {
        "user": "[GITHUB_USERNAME]"
      }
    }' \
  -X POST http://localhost:3000/graphql

Note: These commands need the POST url to match a place where they are running like api.gatsbyjs.org, or to a local port if you're emulating the functions locally.

Before the functions can be deployed, you need to save a keys file to your local machine that the serverless.yml references.

Verify that you have downloaded the keys file from where it's stored (in 1Password for Gatsby employees) and referenced in the serverless.yml:

provider:
  name: google
  stage: dev
  runtime: nodejs10
  region: us-central1
  project: gatsby-core
  credentials: ~/Downloads/gatsby-core.json # <-- this needs to match where you're file is

The serverless deploy command is set up in the package.json to be run by the following yarn command:

yarn deploy

Serverless will zip up all the files and deploy them to Google Cloud Functions over the top of existing functions of the same name and project.

For deployment to multiple environments

This repository is deployed using the Serverless framework to configure all the pieces of the infrastructure required to function. The following things take place on GCP:

• the source code is packaged up uploaded to Google Cloud Storage
• Functions are created for each GraphQL endpoint

Other pieces of the puzzle that fit everything together:

• a DNS record pointing at the deployed functions needs to be set up for the domain that the API is being deployed to
• a prisma service serves as the database for the feedback gathered by the feedback widget on gatsbyjs.org, the endpoint for it is in prisma/prisma.yml