Merge branch 'main' into sitemapoppty

CHANGELOG.md

@@ -1,3 +1,24 @@
+## [1.32.12](https://github.com/adobe/spacecat-audit-worker/compare/v1.32.11...v1.32.12) (2024-11-13)
+
+
+### Bug Fixes
+
+* broken backlinks no longer because of migration to jobs-dispatcher ([#464](https://github.com/adobe/spacecat-audit-worker/issues/464)) ([cc38be2](https://github.com/adobe/spacecat-audit-worker/commit/cc38be247b1cea86977f3268e7c7bfe5ec6ceee5))
+
+## [1.32.11](https://github.com/adobe/spacecat-audit-worker/compare/v1.32.10...v1.32.11) (2024-11-13)
+
+
+### Bug Fixes
+
+* **deps:** update dependency @adobe/spacecat-shared-data-access to v1.52.0 ([#463](https://github.com/adobe/spacecat-audit-worker/issues/463)) ([e8eb5a3](https://github.com/adobe/spacecat-audit-worker/commit/e8eb5a393b3e18799076c984d9c8b43608524e1f))
+
+## [1.32.10](https://github.com/adobe/spacecat-audit-worker/compare/v1.32.9...v1.32.10) (2024-11-13)
+
+
+### Bug Fixes
+
+* broken internal links result schema update ([#460](https://github.com/adobe/spacecat-audit-worker/issues/460)) ([58d8171](https://github.com/adobe/spacecat-audit-worker/commit/58d81717fcbeedf4aa20d358a6fe7ec3a3b8f919))
+
 ## [1.32.9](https://github.com/adobe/spacecat-audit-worker/compare/v1.32.8...v1.32.9) (2024-11-12)
 
 

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@adobe/spacecat-audit-worker",
-  "version": "1.32.9",
+  "version": "1.32.12",
   "description": "SpaceCat Audit Worker",
   "main": "src/index.js",
   "type": "module",
@@ -57,7 +57,7 @@
     "@adobe/helix-universal": "5.0.6",
     "@adobe/helix-universal-logger": "3.0.21",
     "@adobe/spacecat-shared-ahrefs-client": "1.5.11",
-    "@adobe/spacecat-shared-data-access": "1.51.0",
+    "@adobe/spacecat-shared-data-access": "1.52.0",
     "@adobe/spacecat-shared-google-client": "1.2.21",
     "@adobe/spacecat-shared-http-utils": "1.6.18",
     "@adobe/spacecat-shared-rum-api-client": "2.9.10",

src/backlinks/handler.js

@@ -61,12 +61,12 @@ export async function filterOutValidBacklinks(backlinks, log) {
 }
 
 export default async function auditBrokenBacklinks(message, context) {
-  const { type, url: siteId, auditContext = {} } = message;
+  const { type, auditContext = {} } = message;
   const { dataAccess, log, sqs } = context;
   const {
     AUDIT_RESULTS_QUEUE_URL: queueUrl,
   } = context.env;
-
+  const siteId = message.url || message.siteId;
   try {
     log.info(`Received ${type} audit request for siteId: ${siteId}`);
     const site = await retrieveSiteBySiteId(dataAccess, siteId, log);

src/internal-links/handler.js

@@ -49,11 +49,12 @@ function transform404LinksData(responseData, hostUrl, auditUrl, log) {
       const sameDomainSources = allSources.filter(
         (source) => source && hasSameHost(source, hostUrl),
       );
-      if (sameDomainSources.length) {
+
+      for (const source of sameDomainSources) {
         result.push({
-          url,
-          views,
-          sources: sameDomainSources,
+          url_to: url,
+          url_from: source,
+          traffic_domain: views,
         });
       }
     } catch {
@@ -87,9 +88,13 @@ export async function internalLinksAuditRunner(auditUrl, context, site) {
     granularity: 'hourly',
   };
 
+  log.info('broken-internal-links: Options for RUM call: ', JSON.stringify(options));
+
   const all404Links = await rumAPIClient.query('404', options);
   const auditResult = {
-    internalLinks: transform404LinksData(all404Links, finalUrl, auditUrl, log),
+    brokenInternalLinks: transform404LinksData(all404Links, finalUrl, auditUrl, log),
+    fullAuditRef: auditUrl,
+    finalUrl,
     auditContext: {
       interval: INTERVAL,
     },

test/audits/backlinks.test.js

@@ -326,6 +326,46 @@ describe('Backlinks Tests', function () {
     expect(context.log.info).to.have.been.calledWith('Successfully audited site1 for broken-backlinks type audit');
   });
 
+  it('should successfully perform an audit to detect broken backlinks, save and send the proper audit result for message containing siteId', async () => {
+    mockDataAccess.getSiteByID = sinon.stub().withArgs('site1').resolves(site);
+    mockDataAccess.getTopPagesForSite.resolves([]);
+    mockDataAccess.getConfiguration = sinon.stub().resolves(configuration);
+    message.siteId = message.url;
+    delete message.url;
+
+    nock(site.getBaseURL())
+      .get(/.*/)
+      .reply(200);
+
+    nock('https://ahrefs.com')
+      .get(/.*/)
+      .reply(200, auditResult);
+
+    const expectedMessage = {
+      type: message.type,
+      url: site.getBaseURL(),
+      auditContext: {
+        finalUrl: 'bar.foo.com',
+      },
+      auditResult: {
+        finalUrl: 'bar.foo.com',
+        brokenBacklinks: auditResult.backlinks,
+        fullAuditRef: 'https://ahrefs.com/site-explorer/broken-backlinks?select=title%2Curl_from%2Curl_to%2Ctraffic_domain&limit=50&mode=prefix&order_by=domain_rating_source%3Adesc%2Ctraffic_domain%3Adesc&target=bar.foo.com&output=json&where=%7B%22and%22%3A%5B%7B%22field%22%3A%22domain_rating_source%22%2C%22is%22%3A%5B%22gte%22%2C29.5%5D%7D%2C%7B%22field%22%3A%22traffic_domain%22%2C%22is%22%3A%5B%22gte%22%2C500%5D%7D%2C%7B%22field%22%3A%22links_external%22%2C%22is%22%3A%5B%22lte%22%2C300%5D%7D%5D%7D',
+      },
+    };
+
+    const response = await auditBrokenBacklinks(message, context);
+
+    expect(response.status).to.equal(204);
+    expect(mockDataAccess.addAudit).to.have.been.calledOnce;
+    expect(context.sqs.sendMessage).to.have.been.calledOnce;
+    expect(context.sqs.sendMessage).to.have.been
+      .calledWith(context.env.AUDIT_RESULTS_QUEUE_URL, expectedMessage);
+    expect(context.log.info).to.have.been.calledWith('Successfully audited site1 for broken-backlinks type audit');
+    message.url = message.siteId;
+    delete message.siteId;
+  });
+
   it('should successfully perform an audit to detect broken backlinks and suggest fixes based on keywords from top pages if auto-suggest'
     + ' enabled', async () => {
     mockDataAccess.getSiteByID = sinon.stub().withArgs('site1').resolves(site);

test/audits/internal-links.test.js

@@ -23,18 +23,14 @@ import { MockContextBuilder } from '../shared.js';
 
 const AUDIT_RESULT_DATA = [
   {
-    url: 'https://www.example.com/article/dogs/breeds/choosing-an-irish-setter',
-    views: 100,
-    sources: [
-      'https://www.example.com/article/dogs/just-for-fun/dogs-good-for-men-13-manly-masculine-dog-breeds',
-    ],
+    url_to: 'https://www.example.com/article/dogs/breeds/choosing-an-irish-setter',
+    url_from: 'https://www.example.com/article/dogs/just-for-fun/dogs-good-for-men-13-manly-masculine-dog-breeds',
+    traffic_domain: 100,
   },
   {
-    url: 'https://www.example.com/article/dogs/breeds/choosing-a-miniature-poodle',
-    views: 100,
-    sources: [
-      'https://www.example.com/article/dogs/pet-care/when-is-a-dog-considered-senior',
-    ],
+    url_to: 'https://www.example.com/article/dogs/breeds/choosing-a-miniature-poodle',
+    url_from: 'https://www.example.com/article/dogs/pet-care/when-is-a-dog-considered-senior',
+    traffic_domain: 100,
   },
 ];
 
@@ -88,7 +84,9 @@ describe('Broken internal links audit', () => {
     });
     expect(result).to.deep.equal({
       auditResult: {
-        internalLinks: AUDIT_RESULT_DATA,
+        brokenInternalLinks: AUDIT_RESULT_DATA,
+        fullAuditRef: auditUrl,
+        finalUrl: auditUrl,
         auditContext: {
           interval: 30,
         },