@adobe/spacecat-shared-utils 1.50.0 → 1.50.2

package/CHANGELOG.md

@@ -1,3 +1,17 @@
+# [@adobe/spacecat-shared-utils-v1.50.2](https://github.com/adobe/spacecat-shared/compare/@adobe/spacecat-shared-utils-v1.50.1...@adobe/spacecat-shared-utils-v1.50.2) (2025-09-06)
+
+
+### Bug Fixes
+
+* get pageId from content-page-ref if content-page-id not available ([#935](https://github.com/adobe/spacecat-shared/issues/935)) ([df585ce](https://github.com/adobe/spacecat-shared/commit/df585ce09d01a1e1c3b282d3a967b9ab138e3fa4))
+
+# [@adobe/spacecat-shared-utils-v1.50.1](https://github.com/adobe/spacecat-shared/compare/@adobe/spacecat-shared-utils-v1.50.0...@adobe/spacecat-shared-utils-v1.50.1) (2025-09-06)
+
+
+### Bug Fixes
+
+* **deps:** update external fixes ([#920](https://github.com/adobe/spacecat-shared/issues/920)) ([1a6b1e1](https://github.com/adobe/spacecat-shared/commit/1a6b1e1ac9531a41c86406ada4bd4ab903307fdc))
+
 # [@adobe/spacecat-shared-utils-v1.50.0](https://github.com/adobe/spacecat-shared/compare/@adobe/spacecat-shared-utils-v1.49.0...@adobe/spacecat-shared-utils-v1.50.0) (2025-08-28)
 
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@adobe/spacecat-shared-utils",
-  "version": "1.50.0",
+  "version": "1.50.2",
   "description": "Shared modules of the Spacecat Services - utils",
   "type": "module",
   "engines": {
@@ -36,21 +36,21 @@
   "devDependencies": {
     "@adobe/helix-shared-wrap": "2.0.2",
     "@types/validator": "^13.15.2",
-    "chai": "5.2.1",
-    "chai-as-promised": "8.0.1",
-    "esmock": "2.7.1",
+    "chai": "5.3.3",
+    "chai-as-promised": "8.0.2",
+    "esmock": "2.7.2",
     "husky": "9.1.7",
-    "nock": "14.0.9",
+    "nock": "14.0.10",
     "sinon": "20.0.0",
-    "sinon-chai": "4.0.0"
+    "sinon-chai": "4.0.1"
   },
   "dependencies": {
     "@adobe/fetch": "4.2.2",
     "@adobe/spacecat-shared-data-access": "2.45.0",
     "@adobe/spacecat-shared-ims-client": "1.8.3",
-    "@aws-sdk/client-s3": "3.864.0",
-    "@aws-sdk/client-secrets-manager": "3.864.0",
-    "@aws-sdk/client-sqs": "3.864.0",
+    "@aws-sdk/client-s3": "3.883.0",
+    "@aws-sdk/client-secrets-manager": "3.883.0",
+    "@aws-sdk/client-sqs": "3.883.0",
     "@json2csv/plainjs": "7.0.6",
     "aws-xray-sdk": "3.10.3",
     "date-fns": "2.30.0",

package/src/aem.js

@@ -201,22 +201,89 @@ export function detectAEMVersion(htmlSource, headers = {}) {
 }
 
 /**
- * Determines the AEM CS page ID for Content API, from the page URL
+ * Determines the AEM CS/AMS page ID for Content API, from the page URL
  * @param {string} pageURL - The URL of the page
- * @return {string|null} - The AEM CS page ID
+ * @param {string} authorURL - The URL of the author instance
+ * @param {string} bearerToken - The access token for the author instance
+ * @param {boolean} preferContentApi - Whether to prefer the Content API over the PSS, default is
+ * false
+ * @param {Object} log - The logger object, default is console
+ * @return {string|null} - The AEM page ID
  */
-export async function determineAEMCSPageId(pageURL) {
+export async function determineAEMCSPageId(
+  pageURL,
+  authorURL,
+  bearerToken,
+  preferContentApi = false,
+  log = console,
+) {
+  log.info(`Fetching HTML from ${pageURL} to retrieve content-page-id for AEM CS Content API mode.`);
   const htmlResponse = await fetch(pageURL);
+
   if (!htmlResponse.ok) {
     return null;
   }
+
   const html = await htmlResponse.text();
-  const metaTagRegex = /<meta\s+name=['"]content-page-id['"]\s+content=['"]([^'"]*)['"]\s*\/?>/i;
-  const match = html.match(metaTagRegex);
+
+  // First try to find a content-page-ref meta tag
+  const contentPageRefRegex = /<meta\s+name=['"]content-page-ref['"]\s+content=['"]([^'"]*)['"]\s*\/?>/i;
+  const refMatch = html.match(contentPageRefRegex);
+
+  if (refMatch?.[1]?.trim()) {
+    if (!authorURL || !bearerToken) {
+      // If ref was present but resolution failed, return null per spec
+      log.warn('Content-page-ref found but authorURL or bearerToken is missing, skipping resolution.');
+      return null;
+    }
+    const contentPageRef = refMatch[1].trim();
+    try {
+      const base = preferContentApi
+        ? `${authorURL}/adobe/experimental/aspm-expires-20251231`
+        : `${authorURL}/adobe/experimental/pss`;
+      const resolveUrl = `${base}/pages/resolve?pageRef=${contentPageRef}`;
+      log.info(`Resolving content-page-ref via ${resolveUrl} (preferContentApi=${preferContentApi})`);
+      const resp = await fetch(resolveUrl, {
+        method: 'GET',
+        headers: { Authorization: bearerToken },
+        redirect: 'follow',
+      });
+      if (resp.status === 200) {
+        let pageId = null;
+        if (preferContentApi) {
+          const data = await resp.json();
+          pageId = data?.id || null;
+        } else {
+          const data = await resp.text();
+          pageId = data || null;
+        }
+
+        if (pageId) {
+          log.info(`Resolved pageId: "${pageId}" from JSON directly for ref "${contentPageRef}"`);
+          return pageId;
+        }
+        log.error('resolve response did not contain an "id" property.');
+        return null;
+      } else {
+        log.warn(`Unexpected status ${resp.status} when resolving content-page-ref.`);
+      }
+    } catch (e) {
+      log.error(`Error while resolving content-page-ref: ${e.message}`);
+    }
+    // If ref was present but resolution failed, return null per spec
+    return null;
+  }
+
+  // Fallback to content-page-id meta tag
+  const contentPageIdRegex = /<meta\s+name=['"]content-page-id['"]\s+content=['"]([^'"]*)['"]\s*\/?>/i;
+  const idMatch = html.match(contentPageIdRegex);
 
   let pageId = null;
-  if (match && match[1] && match[1].trim() !== '') {
-    pageId = match[1].trim();
+  if (idMatch?.[1]?.trim()) {
+    pageId = idMatch[1].trim();
+    if (pageId) {
+      log.info(`Extracted pageId: "${pageId}" from "content-page-id" meta tag at ${pageURL}`);
+    }
   }
   return pageId;
 }