Automate link-checking

cypress.config.js

@@ -1,12 +1,27 @@
 const { defineConfig } = require('cypress');
+const process = require('process');
 
 module.exports = defineConfig({
   e2e: {
     // Automatically prefix cy.visit() and cy.request() commands with a baseUrl.
     baseUrl: 'http://localhost:1313',
+    defaultCommandTimeout: 10000,
+    pageLoadTimeout: 30000,
+    responseTimeout: 30000,
+    experimentalMemoryManagement: true,
+    numTestsKeptInMemory: 5,
     projectId: 'influxdata-docs',
     setupNodeEvents(on, config) {
       // implement node event listeners here
+      on('before:browser:launch', (browser, launchOptions) => {
+        if (browser.name === 'chrome' && browser.isHeadless) {
+          // Force Chrome to use a less memory-intensive approach
+          launchOptions.args.push('--disable-dev-shm-usage');
+          launchOptions.args.push('--disable-gpu');
+          launchOptions.args.push('--disable-extensions');
+          return launchOptions;
+        }
+      });
       on('task', {
         // Fetch the product list configured in /data/products.yml
         getData(filename) {
@@ -24,6 +39,7 @@ module.exports = defineConfig({
           });
         },
       });
+      return config;
     },
   },
 });

cypress/e2e/content/article-links.cy.js

@@ -1,27 +1,122 @@
 /// <reference types="cypress" />
+
 describe('Article links', () => {
   const subjects = Cypress.env('test_subjects').split(',');
+  // Always use HEAD for downloads to avoid timeouts
+  const useHeadForDownloads = true; 
+
+  // Helper function to identify download links - improved
+  function isDownloadLink(href) {
+    // Check for common download file extensions
+    const downloadExtensions = [
+      '.pdf', '.zip', '.tar.gz', '.tgz', '.rar', '.exe', '.dmg', '.pkg', 
+      '.deb', '.rpm', '.xlsx', '.csv', '.doc', '.docx', '.ppt', '.pptx'
+    ];
+
+    // Check for download domains or paths
+    const downloadDomains = [
+      'dl.influxdata.com',
+      'downloads.influxdata.com'
+    ];
+
+    // Check if URL contains a download extension
+    const hasDownloadExtension = downloadExtensions.some(ext => 
+      href.toLowerCase().endsWith(ext)
+    );
+
+    // Check if URL is from a download domain
+    const isFromDownloadDomain = downloadDomains.some(domain => 
+      href.toLowerCase().includes(domain)
+    );
+
+    // Return true if either condition is met
+    return hasDownloadExtension || isFromDownloadDomain;
+  }
+
+  // Helper function to make appropriate request based on link type
+  function testLink(href) {
+    if (useHeadForDownloads && isDownloadLink(href)) {
+      cy.log(`** Testing download link with HEAD: ${href} **`);
+      cy.request({
+        method: 'HEAD',
+        url: href,
+        failOnStatusCode: false,
+        timeout: 10000 // 10 second timeout for download links
+      }).then(response => {
+        expect(response.status).to.be.lt(400);
+      });
+    } else {
+      cy.log(`** Testing link: ${href} **`);
+      cy.request({
+        url: href,
+        failOnStatusCode: false,
+        timeout: 30000 // 30 second timeout for regular links
+      }).then(response => {
+        expect(response.status).to.be.lt(400);
+      });
+    }
+  }
 
   subjects.forEach((subject) => {
-    it('contains valid internal links', function () {
+    it(`contains valid internal links on ${subject}`, function () {
+      cy.visit(`${subject}`);
+
+      // Test internal links (including anchor links)
+      cy.get('article a[href^="/"]').each(($a) => {
+        const href = $a.attr('href');
+        testLink(href);
+      });
+    });
+
+    it(`checks anchor links on ${subject} (with warnings for missing targets)`, function () {
       cy.visit(`${subject}`);
-      cy.get('article a[href^="/"]') //.filter('[href^="/"]')
-        .each(($a) => {
-          cy.log(`** Testing internal link ${$a.attr('href')} **`);
-          // cy.request doesn't show in your browser's Developer Tools
-          // because the request comes from Node, not from the browser.
-          cy.request($a.attr('href')).its('status').should('eq', 200);
-        });
+
+      // Track missing anchors for summary
+      const missingAnchors = [];
+
+      // Process anchor links individually
+      cy.get('article a[href^="#"]').each(($a) => {
+        const href = $a.prop('href'); // Use prop() instead of attr()
+        if (href && href.length > 1) { // Skip empty anchors (#)
+          // Get just the fragment part
+          const url = new URL(href);
+          const anchorId = url.hash.substring(1); // Remove the # character
+
+          if (!anchorId) {
+            cy.log(`Skipping empty anchor in ${href}`);
+            return;
+          }
+
+          // Use DOM to check if the element exists, but don't fail if missing
+          cy.window().then(win => {
+            const element = win.document.getElementById(anchorId);
+            if (element) {
+              cy.log(`✅ Anchor target exists: #${anchorId}`);
+            } else {
+              // Just warn about the missing anchor
+              cy.log(`⚠️ WARNING: Missing anchor target: #${anchorId}`);
+              missingAnchors.push(anchorId);
+            }
+          });
+        }
+      }).then(() => {
+        // After checking all anchors, log a summary
+        if (missingAnchors.length > 0) {
+          cy.log(`⚠️ Found ${missingAnchors.length} missing anchor targets: ${missingAnchors.join(', ')}`);
+        } else {
+          cy.log('✅ All anchor targets are valid');
+        }
       });
-    it('contains valid external links', function () {
+    });
+
+    it(`contains valid external links on ${subject}`, function () {
       cy.visit(`${subject}`);
-      cy.get('article a[href^="http"]')
-        .each(($a) => {
-          // cy.request doesn't show in your browser's Developer Tools
-          cy.log(`** Testing external link ${$a.attr('href')} **`);
-          // because the request comes from Node, not from the browser.
-          cy.request($a.attr('href')).its('status').should('eq', 200);
-        });
+
+      // Test external links
+      cy.get('article a[href^="http"]').each(($a) => {
+        const href = $a.attr('href');
+        testLink(href);
+      });
     });
   });
-});
+});

cypress/support/map-files-to-urls.mjs

@@ -0,0 +1,79 @@
+#!/usr/bin/env node
+
+import { execSync } from 'child_process';
+import process from 'process';
+
+// Get file paths from command line arguments
+const filePaths = process.argv.slice(2);
+
+// Parse options
+const debugMode = process.argv.includes('--debug');
+
+// Filter for content files
+const contentFiles = filePaths.filter(file => 
+  file.startsWith('content/') && (file.endsWith('.md') || file.endsWith('.html'))
+);
+
+if (contentFiles.length === 0) {
+  console.log('No content files to check.');
+  process.exit(0);
+}
+
+// Map file paths to URL paths
+function mapFilePathToUrl(filePath) {
+  // Remove content/ prefix
+  let url = filePath.replace(/^content/, '');
+
+  // Handle _index files (both .html and .md)
+  url = url.replace(/\/_index\.(html|md)$/, '/');
+
+  // Handle regular .md files
+  url = url.replace(/\.md$/, '/');
+
+  // Handle regular .html files
+  url = url.replace(/\.html$/, '/');
+
+  // Ensure URL starts with a slash
+  if (!url.startsWith('/')) {
+    url = '/' + url;
+  }
+
+  return url;
+}
+
+const urls = contentFiles.map(mapFilePathToUrl);
+const urlList = urls.join(',');
+
+console.log(`Testing links in URLs: ${urlList}`);
+
+// Create environment object with the cypress_test_subjects variable
+const envVars = {
+  ...process.env,
+  cypress_test_subjects: urlList,
+  NODE_OPTIONS: '--max-http-header-size=80000 --max-old-space-size=4096'
+};
+
+// Run Cypress tests with the mapped URLs
+try {
+  // Choose run mode based on debug flag
+  if (debugMode) {
+    // For debug mode, set the environment variable and open Cypress
+    // The user will need to manually select the test file
+    console.log('Opening Cypress in debug mode.');
+    console.log('Please select the "article-links.cy.js" test file when Cypress opens.');
+
+    execSync('npx cypress open --e2e', {
+      stdio: 'inherit',
+      env: envVars
+    });
+  } else {
+    // For normal mode, run the test automatically
+    execSync(`npx cypress run --spec "cypress/e2e/content/article-links.cy.js"`, {
+      stdio: 'inherit',
+      env: envVars
+    });
+  }
+} catch (error) {
+  console.error('Link check failed');
+  process.exit(1);
+}

layouts/partials/article/alpha.html

@@ -19,7 +19,7 @@ <h4>{{ $displayName }} is in Public Alpha</h4>
       invite you to <strong>join our public channels</strong> for updates and to share feedback.
     </p>
     <div class="expand-wrapper">
-      <div class="expand" id="#alpha-expecations">
+      <div class="expand" id="alpha-expecations">
         <p class="expand-label">
           <span class="expand-toggle"></span><span>Alpha expectations and recommendations</span>
         </p>
@@ -39,7 +39,7 @@ <h4>{{ $displayName }} is in Public Alpha</h4>
           </ul>
         </div>
       </div>
-      <div class="expand" id="#alpha-feedback-channels">
+      <div class="expand" id="alpha-feedback-channels">
         <p class="expand-label">
           <span class="expand-toggle"></span><span>Join our public channels</span>
         </p>

lefthook.yml

@@ -50,6 +50,37 @@ pre-commit:
       run: '.ci/vale/vale.sh
         --config=content/influxdb/v2/.vale.ini
         --minAlertLevel=error {staged_files}'
+
+    # Link checking for InfluxDB v2
+    v2-links:
+      tags: test,links,v2
+      glob: "content/influxdb/v2/**/*.{md,html}"
+      run: 'node cypress/support/map-files-to-urls.mjs {staged_files}'
+
+    # Link checking for InfluxDB v3 core
+    v3-core-links:
+      tags: test,links,v3
+      glob: "content/influxdb3/core/**/*.{md,html}"
+      run: 'node cypress/support/map-files-to-urls.mjs {staged_files}'
+
+    # Link checking for InfluxDB v3 enterprise
+    v3-enterprise-links:
+      tags: test,links,v3
+      glob: "content/influxdb3/enterprise/**/*.{md,html}"
+      run: 'node cypress/support/map-files-to-urls.mjs {staged_files}'
+
+    # Link checking for Cloud products
+    cloud-links:
+      tags: test,links,cloud
+      glob: "content/influxdb/{cloud,cloud-dedicated,cloud-serverless}/**/*.{md,html}"
+      run: 'node cypress/support/map-files-to-urls.mjs {staged_files}'
+
+    # Link checking for Telegraf
+    telegraf-links:
+      tags: test,links
+      glob: "content/telegraf/**/*.{md,html}"
+      run: 'node cypress/support/map-files-to-urls.mjs {staged_files}'
+
     cloud-pytest:
       glob: content/influxdb/cloud/**/*.md
       tags: test,codeblocks,v2

package.json

@@ -35,9 +35,8 @@
   "scripts": {
     "e2e:chrome": "npx cypress run --browser chrome",
     "e2e:o": "npx cypress open",
-    "e2e:o:links": "export cypress_test_subjects=\"http://localhost:1313/influxdb3/core/,http://localhost:1313/influxdb3/enterprise/\"; npx cypress open cypress/e2e/content/article-links.cy.js",
-    "e2e:links": "export cypress_test_subjects=\"http://localhost:1313/influxdb3/core/,http://localhost:1313/influxdb3/enterprise/\"; npx cypress run --spec cypress/e2e/content/article-links.cy.js",
-    "e2e:api-docs": "export cypress_test_subjects=\"http://localhost:1313/influxdb3/core/api/,http://localhost:1313/influxdb3/enterprise/api/,http://localhost:1313/influxdb3/cloud-dedicated/api/,http://localhost:1313/influxdb3/cloud-dedicated/api/v1/,http://localhost:1313/influxdb/cloud-dedicated/api/v1/,http://localhost:1313/influxdb/cloud-dedicated/api/management/,http://localhost:1313/influxdb3/cloud-dedicated/api/management/\"; npx cypress run --spec cypress/e2e/content/article-links.cy.js",
+    "e2e:o:links": "node cypress/support/map-files-to-urls.mjs content/influxdb3/core/get-started/_index.md --debug",
+    "e2e:api-docs": "export cypress_test_subjects=\"http://localhost:1313/influxdb3/core/api/,http://localhost:1313/influxdb3/enterprise/api/,http://localhost:1313/influxdb3/cloud-dedicated/api/,http://localhost:1313/influxdb3/cloud-dedicated/api/v1/,http://localhost:1313/influxdb/cloud-dedicated/api/v1/,http://localhost:1313/influxdb/cloud-dedicated/api/management/,http://localhost:1313/influxdb3/cloud-dedicated/api/management/\"; npx cypress run --spec cypress/e2e/article-links.cy.js",
     "lint": "LEFTHOOK_EXCLUDE=test lefthook run pre-commit && lefthook run pre-push",
     "pre-commit": "lefthook run pre-commit",
     "test-content": "docker compose --profile test up"