feat(gatsby-source-drupal): Provide proxyUrl in addition to baseUrl to allow using CDN, API gateway, etc. (#36819)

packages/gatsby-source-drupal/README.md

@@ -250,6 +250,29 @@ module.exports = {
 }
 ```
 
+## CDN
+
+You can add an optional CDN or API gateway URL `proxyUrl` param. The URL can be a simple proxy of the Drupal
+`baseUrl`, or another URL (even containing a path) where the Drupal JSON API resources can be retrieved.
+
+This option is required as Drupal doesn't know about the CDN so it returns URLs pointing to the `baseUrl`. With `proxyUrl` set, the plugin will rewrite URLs returned from Drupal to keep pointing at the `proxyUrl`
+
+```javascript
+// In your gatsby-config.js
+module.exports = {
+  plugins: [
+    {
+      resolve: `gatsby-source-drupal`,
+      options: {
+        baseUrl: `https://live-contentacms.pantheonsite.io/`,
+        proxyUrl: `https://xyz.cloudfront.net/`, // optional, defaults to the value of baseUrl
+        apiBase: `api`, // optional, defaults to `jsonapi`
+      },
+    },
+  ],
+}
+```
+
 ## GET Search Params
 
 You can append optional GET request params to the request url using `params` option.

packages/gatsby-source-drupal/src/__tests__/index.js

@@ -1,5 +1,14 @@
+import got from "got"
+
+const baseUrl = `http://fixture`
+const proxyUrl = `http://fixture-proxy`
+
 jest.mock(`got`, () =>
   jest.fn(path => {
+    if (path.includes(proxyUrl)) {
+      path = path.replace(proxyUrl, baseUrl)
+    }
+
     let last = ``
     if (path.includes(`i18n-test`)) {
       last = `i18n-test-`
@@ -59,7 +68,6 @@ const { handleWebhookUpdate } = require(`../utils`)
 describe(`gatsby-source-drupal`, () => {
   let nodes = {}
   const createNodeId = id => `generated-id-${id}`
-  const baseUrl = `http://fixture`
   const createContentDigest = jest.fn().mockReturnValue(`contentDigest`)
   const { objectContaining } = expect
   const actions = {
@@ -450,6 +458,38 @@ describe(`gatsby-source-drupal`, () => {
     expect(nodes[createNodeId(`und.article-3`)]).toBeDefined()
   })
 
+  it(`Can use the proxyUrl plugin option to use a different API url for sourcing`, async () => {
+    got.mockClear()
+    nodes = {}
+    await sourceNodes(args, { baseUrl, proxyUrl })
+
+    let callSkipCount = 0
+    for (const [index, call] of got.mock.calls.entries()) {
+      if (call[0] === `http://fixture/jsonapi`) {
+        callSkipCount++
+        continue
+      }
+
+      expect(got).toHaveBeenNthCalledWith(
+        index + 1,
+        expect.stringContaining(proxyUrl),
+        expect.anything()
+      )
+    }
+
+    expect(callSkipCount).toBe(1)
+    expect(got).toBeCalledTimes(8)
+
+    expect(Object.keys(nodes).length).not.toEqual(0)
+    expect(nodes[createNodeId(`und.file-1`)]).toBeDefined()
+    expect(nodes[createNodeId(`und.file-2`)]).toBeDefined()
+    expect(nodes[createNodeId(`und.tag-1`)]).toBeDefined()
+    expect(nodes[createNodeId(`und.tag-2`)]).toBeDefined()
+    expect(nodes[createNodeId(`und.article-1`)]).toBeDefined()
+    expect(nodes[createNodeId(`und.article-2`)]).toBeDefined()
+    expect(nodes[createNodeId(`und.article-3`)]).toBeDefined()
+  })
+
   it(`Verify JSON:API includes relationships`, async () => {
     // Reset nodes and test includes relationships.
     Object.keys(nodes).forEach(key => delete nodes[key])

packages/gatsby-source-drupal/src/gatsby-node.js

@@ -155,6 +155,7 @@ exports.sourceNodes = async (
   globalReporter = reporter
   const {
     baseUrl,
+    proxyUrl = baseUrl,
     apiBase = `jsonapi`,
     basicAuth = {},
     filters,
@@ -534,6 +535,11 @@ ${JSON.stringify(webhookBody, null, 4)}`
             }
           }
 
+          // If proxyUrl is defined, use it instead of baseUrl to get the content.
+          if (proxyUrl !== baseUrl) {
+            url = url.replace(baseUrl, proxyUrl)
+          }
+
           let d
           try {
             d = await requestQueue.push([
@@ -833,6 +839,9 @@ exports.pluginOptionsSchema = ({ Joi }) =>
     baseUrl: Joi.string()
       .required()
       .description(`The URL to root of your Drupal instance`),
+    proxyUrl: Joi.string().description(
+      `The CDN URL equivalent to your baseUrl`
+    ),
     apiBase: Joi.string().description(
       `The path to the root of the JSONAPI — defaults to "jsonapi"`
     ),