Merge pull request #278 from sei-protocol/checks/enforce-links-check

Enforce link checks per PR

Author: alexander-sei

.github/workflows/link_checker.yml

@@ -1,6 +1,7 @@
 name: Link checker
 
 on:
+  pull_request:
   workflow_dispatch:
   schedule:
     - cron: '0 12 * * *'
@@ -27,8 +28,20 @@ jobs:
 
       - name: Run crawler
         id: crawler
-        run: yarn tsx scripts/checkLinks.ts
         continue-on-error: true
+        run: |
+          if [[ "${{ github.event_name }}" == "pull_request" ]]; then
+            echo "Running PR workflow: Building and starting local server..."
+            yarn build
+            sleep 5
+            yarn start & 
+            sleep 15 
+            export BASE_URL="http://localhost:3000"
+          else
+            export BASE_URL="https://www.docs.sei.io/"
+          fi
+          echo "Running link checker for ${BASE_URL}"
+          yarn tsx scripts/checkLinks.ts
 
       - name: Upload broken links artifact
         if: steps.crawler.outcome == 'failure'

scripts/checkLinks.ts

@@ -7,9 +7,8 @@ const visitedLinks = new Set<string>();
 
 async function main() {
 	const browserInstance = await chromium.launch();
-	const baseUrl = 'https://www.docs.sei.io/';
+	const baseUrl = process.env.BASE_URL || 'https://www.docs.sei.io/';
 	await crawlPages(baseUrl, browserInstance, 'main');
-
 	fs.writeFileSync('brokenLinks.json', JSON.stringify([...brokenLinks], null, 2));
 
 	if (brokenLinks.size > 0) {
@@ -31,7 +30,7 @@ async function crawlPages(url: string, browser: Browser, path: string) {
 }
 
 function isInternal(url: string) {
-	return url.includes('docs.sei');
+	return url.includes('docs.sei') || url.includes('localhost:3000');
 }
 
 async function checkInternalLinks(url: string, page: Page, path: string, browser: Browser) {
@@ -58,7 +57,7 @@ async function checkExternalLinks(url: string, page: Page, path: string, browser
 
 async function isLinkBroken(page: Page, url: string, path: string) {
 	if (visitedLinks.has(url)) return false;
-	if (url.includes('localhost') || url.includes('.tar.gz')) return false;
+	if ((url.includes('localhost') && !url.includes(':3000')) || url.includes('.tar.gz')) return false;
 
 	let pageResponse: Response;
 	try {