Add hydrate-localize composite GitHub Action

.github/actions/hydrate-localize/action.yml

@@ -0,0 +1,26 @@
+name: 'Hydrate Localize'
+description: 'Crawl a site with Chrome to trigger Localize phrase detection'
+inputs:
+  target_base_url:
+    description: 'Base URL of the docs site to crawl'
+    required: true
+  sitemap_path:
+    description: 'Path to the sitemap file relative to the base URL'
+    required: false
+    default: 'sitemap.xml'
+runs:
+  using: "composite"
+  steps:
+    - name: Install dependencies
+      run: cd ${{ github.action_path }} && npm ci
+      shell: bash
+
+    - name: Set up headless Chrome
+      run: cd ${{ github.action_path }} && npx playwright install --with-deps chromium
+      shell: bash
+
+    - name: Run crawler script
+      run: node ${{ github.action_path }}/index.js ${{ inputs.target_base_url }} ${{ inputs.sitemap_path }}
+      shell: bash
+      env:
+        CRAWLEE_MEMORY_MBYTES: 8000

.github/actions/hydrate-localize/index.js

@@ -0,0 +1,31 @@
+import { PlaywrightCrawler, Sitemap } from 'crawlee';
+
+const targetHost = process.argv[2];
+const sitemapPath = process.argv[3] || 'sitemap.xml';
+
+const crawler = new PlaywrightCrawler({
+    async requestHandler({ request, page, enqueueLinks }) {
+        const requestHost = 'https://global.localizecdn.com';
+        const requestMethod = 'POST';
+        if (request.loadedUrl.indexOf('api-catalog') >= 0) {
+            // Localize is currently disabled on API catalog pages, so we
+            // can't wait for a request on those pages
+            console.log(`Not waiting for Localize on ${request.loadedUrl}`);
+        } else {
+            await page.waitForRequest(
+                req => new URL(req.url()).origin === requestHost && req.method() === requestMethod
+            );
+            console.log(`Done waiting for ${requestMethod} request to ${requestHost} on ${request.loadedUrl}`);
+        }
+    }
+});
+
+const { urls } = await Sitemap.load(`${targetHost}/${sitemapPath}`);
+
+await crawler.addRequests(urls.map((url) => {
+    const { pathname, search, hash } = new URL(url);
+    return `${targetHost}${pathname}${search}${hash}`;
+}));
+
+// Run the crawler
+await crawler.run();