@adobe/mysticat-shared-seo-client 1.5.0 → 1.6.0

package/CHANGELOG.md

@@ -1,3 +1,16 @@
+## [@adobe/mysticat-shared-seo-client-v1.6.0](https://github.com/adobe/spacecat-shared/compare/@adobe/mysticat-shared-seo-client-v1.5.1...@adobe/mysticat-shared-seo-client-v1.6.0) (2026-06-08)
+
+### Features
+
+* **seo-client:** add quality filters to broken-backlinks Semrush call (404+410, follow, text, no lostlinks) ([#1652](https://github.com/adobe/spacecat-shared/issues/1652)) ([875d92c](https://github.com/adobe/spacecat-shared/commit/875d92c4ceb306b5010118e29287bc81906e980e))
+
+## [@adobe/mysticat-shared-seo-client-v1.5.1](https://github.com/adobe/spacecat-shared/compare/@adobe/mysticat-shared-seo-client-v1.5.0...@adobe/mysticat-shared-seo-client-v1.5.1) (2026-05-29)
+
+### Bug Fixes
+
+* **deps:** update adobe fixes ([#1609](https://github.com/adobe/spacecat-shared/issues/1609)) ([ccdcc2b](https://github.com/adobe/spacecat-shared/commit/ccdcc2b88c1b03e5c58ed682daaaadb382638395))
+* **deps:** update external major (major) ([#1611](https://github.com/adobe/spacecat-shared/issues/1611)) ([5a80797](https://github.com/adobe/spacecat-shared/commit/5a807979187f1ebcc32b088faee9165aa38cc461))
+
 ## [@adobe/mysticat-shared-seo-client-v1.5.0](https://github.com/adobe/spacecat-shared/compare/@adobe/mysticat-shared-seo-client-v1.4.0...@adobe/mysticat-shared-seo-client-v1.5.0) (2026-05-27)
 
 ### Features

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@adobe/mysticat-shared-seo-client",
-  "version": "1.5.0",
+  "version": "1.6.0",
   "description": "Shared modules of the SpaceCat Services - SEO Client",
   "type": "module",
   "engines": {
@@ -35,7 +35,7 @@
   },
   "dependencies": {
     "@adobe/fetch": "4.3.0",
-    "@adobe/helix-universal": "5.4.1",
+    "@adobe/helix-universal": "5.4.2",
     "@adobe/spacecat-shared-utils": "1.81.1",
     "urijs": "1.19.11"
   },
@@ -43,7 +43,7 @@
     "chai": "6.2.2",
     "chai-as-promised": "8.0.2",
     "nock": "14.0.15",
-    "sinon": "21.1.2",
+    "sinon": "22.0.0",
     "sinon-chai": "4.0.1",
     "typescript": "6.0.3"
   }

package/src/client.js

@@ -744,25 +744,49 @@ export default class SeoClient {
     const epLinks = ENDPOINTS.brokenBacklinks;
     const effectiveLimit = getLimit(limit, 100);
 
-    // Step 1: Find broken (404) target pages, sorted by referring domains
-    const { body: pagesBody, fullAuditRef } = await this.sendRawRequest({
+    // Step 1: Find broken target pages — 404 and 410 in parallel (Semrush does not support OR
+    // within a single display_filter, so we issue two requests and merge the results).
+    const pageParams = (code) => ({
       type: epPages.type,
       target: url,
       target_type: 'domain',
       export_columns: epPages.columns,
       display_limit: effectiveLimit,
       display_filter: buildFilter([{
-        sign: '+', field: 'responsecode', op: 'Eq', value: '404',
+        sign: '+', field: 'responsecode', op: 'Eq', value: code,
       }]),
       ...epPages.defaultParams,
-    }, epPages.path);
-    const brokenPages = parseCsvResponse(pagesBody);
+    });
+
+    const [{ body: pages404Body, fullAuditRef }, { body: pages410Body }] = await Promise.all([
+      this.sendRawRequest(pageParams('404'), epPages.path),
+      this.sendRawRequest(pageParams('410'), epPages.path),
+    ]);
+
+    // Merge, deduplicate by source_url, re-sort by domains_num descending, cap at limit.
+    const seenUrls = new Set();
+    const brokenPages = [
+      ...parseCsvResponse(pages404Body),
+      ...parseCsvResponse(pages410Body),
+    ]
+      .filter((p) => {
+        if (seenUrls.has(p.source_url)) {
+          return false;
+        }
+        seenUrls.add(p.source_url);
+        return true;
+      })
+      .sort((a, b) => Number(b.domains_num) - Number(a.domains_num))
+      .slice(0, effectiveLimit);
 
     if (brokenPages.length === 0) {
       return { result: { backlinks: [] }, fullAuditRef };
     }
 
-    // Step 2: For each broken page, fetch the top backlink by authority score.
+    // Step 2: For each broken page, fetch the top backlink by authority score with quality filters:
+    //   - follow links only (nofollow links carry no SEO value)
+    //   - text links only (more actionable than image/frame links)
+    //   - exclude lostlinks (only active broken backlinks, not already-removed ones)
     const brokenUrls = brokenPages.map((p) => p.source_url);
     const linkResults = await this.fanOut(brokenUrls, (brokenUrl) => this.sendRawRequest({
       type: epLinks.type,
@@ -770,6 +794,17 @@ export default class SeoClient {
       target_type: 'url',
       export_columns: epLinks.columns,
       display_limit: 1,
+      display_filter: buildFilter([
+        {
+          sign: '+', field: 'type', op: '', value: 'follow',
+        },
+        {
+          sign: '+', field: 'type', op: '', value: 'text',
+        },
+        {
+          sign: '-', field: 'type', op: '', value: 'lostlink',
+        },
+      ]),
       ...epLinks.defaultParams,
     }, epLinks.path), 'getBrokenBacklinks');