@adobe/spacecat-shared-data-access 2.28.0 → 2.29.0

package/CHANGELOG.md

@@ -1,3 +1,10 @@
+# [@adobe/spacecat-shared-data-access-v2.29.0](https://github.com/adobe/spacecat-shared/compare/@adobe/spacecat-shared-data-access-v2.28.0...@adobe/spacecat-shared-data-access-v2.29.0) (2025-06-20)
+
+
+### Features
+
+* introduce findByPreviewURL in site collection ([#773](https://github.com/adobe/spacecat-shared/issues/773)) ([3b7c092](https://github.com/adobe/spacecat-shared/commit/3b7c0925ce66bcb8d38db3e05cc04fefd0cac229))
+
 # [@adobe/spacecat-shared-data-access-v2.28.0](https://github.com/adobe/spacecat-shared/compare/@adobe/spacecat-shared-data-access-v2.27.0...@adobe/spacecat-shared-data-access-v2.28.0) (2025-06-20)
 
 

package/README.md

@@ -81,6 +81,8 @@ The module provides the following DAOs:
 - `addSite`
 - `updateSite`
 - `removeSite`
+- `findByPreviewURL`
+- `findByExternalOwnerIdAndExternalSiteId`
 
 ### Site Candidate Functions
 - `getSiteCandidateByBaseURL`

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@adobe/spacecat-shared-data-access",
-  "version": "2.28.0",
+  "version": "2.29.0",
   "description": "Shared modules of the Spacecat Services - Data Access",
   "type": "module",
   "engines": {
@@ -35,7 +35,7 @@
     "access": "public"
   },
   "dependencies": {
-    "@adobe/spacecat-shared-utils": "1.38.4",
+    "@adobe/spacecat-shared-utils": "1.39.1",
     "@aws-sdk/client-dynamodb": "3.828.0",
     "@aws-sdk/lib-dynamodb": "3.828.0",
     "@types/joi": "17.2.3",

package/src/models/site/index.d.ts

@@ -176,4 +176,8 @@ export interface SiteCollection extends BaseCollection<Organization> {
   findByBaseURL(baseURL: string): Promise<Site | null>;
   findByDeliveryType(deliveryType: string): Promise<Site | null>;
   findByOrganizationId(organizationId: string): Promise<Site | null>;
+  findByPreviewURL(previewURL: string): Promise<Site | null>;
+  findByExternalOwnerIdAndExternalSiteId(
+    externalOwnerId: string, externalSiteId: string
+  ): Promise<Site | null>;
 }

package/src/models/site/site.collection.js

@@ -10,12 +10,12 @@
  * governing permissions and limitations under the License.
  */
 
-import { hasText } from '@adobe/spacecat-shared-utils';
+import { hasText, isValidHelixPreviewUrl, isValidUrl } from '@adobe/spacecat-shared-utils';
 
 import DataAccessError from '../../errors/data-access.error.js';
 import BaseCollection from '../base/base.collection.js';
 
-import Site from './site.model.js';
+import Site, { AEM_CS_HOST, getPreviewType } from './site.model.js';
 
 /**
  * SiteCollection - A collection class responsible for managing Site entities.
@@ -71,6 +71,35 @@ class SiteCollection extends BaseCollection {
 
     return orderedSites;
   }
+
+  async findByPreviewURL(previewURL) {
+    if (!isValidUrl(previewURL)) {
+      throw new DataAccessError(`Invalid preview URL: ${previewURL}`, this);
+    }
+
+    const { hostname } = new URL(previewURL);
+    const previewType = getPreviewType(hostname, Site.DELIVERY_TYPES);
+
+    switch (previewType) {
+      case Site.DELIVERY_TYPES.AEM_EDGE: {
+        if (!isValidHelixPreviewUrl(previewURL)) {
+          throw new DataAccessError(`Invalid Helix preview URL: ${previewURL}`, this);
+        }
+        const [host] = hostname.split('.');
+        const [ref, site, owner] = host.split('--');
+        const externalOwnerId = `${ref}#${owner}`;
+        return this.findByExternalOwnerIdAndExternalSiteId(externalOwnerId, site);
+      }
+      case Site.DELIVERY_TYPES.AEM_CS: {
+        const [, programId, envId] = AEM_CS_HOST.exec(hostname);
+        const externalOwnerId = `p${programId}`;
+        const externalSiteId = `e${envId}`;
+        return this.findByExternalOwnerIdAndExternalSiteId(externalOwnerId, externalSiteId);
+      }
+      default:
+        throw new DataAccessError(`Unsupported preview URL: ${previewURL}`, this);
+    }
+  }
 }
 
 export default SiteCollection;

package/src/models/site/site.model.js

@@ -13,6 +13,50 @@
 import { composeAuditURL, hasText, isValidUrl } from '@adobe/spacecat-shared-utils';
 import BaseModel from '../base/base.model.js';
 
+const HLX_HOST = /\.(?:aem|hlx)\.(?:page|live)$/i;
+export const AEM_CS_HOST = /^author-p(\d+)-e(\d+)/i;
+
+/**
+ * Computes external IDs based on delivery type and configuration
+ */
+export const computeExternalIds = (attrs) => {
+  const { hlxConfig, deliveryConfig } = attrs;
+
+  if (hlxConfig) {
+    const rso = hlxConfig.rso ?? {};
+    const { ref, owner, site } = rso;
+
+    return {
+      externalOwnerId: ref && owner ? `${ref}#${owner}` : undefined,
+      externalSiteId: site || undefined,
+    };
+  }
+
+  if (deliveryConfig) {
+    const { programId, environmentId } = deliveryConfig;
+
+    return {
+      externalOwnerId: programId ? `p${programId}` : undefined,
+      externalSiteId: environmentId ? `e${environmentId}` : undefined,
+    };
+  }
+
+  return { externalOwnerId: undefined, externalSiteId: undefined };
+};
+
+/**
+ * Determines the preview type based on hostname
+ */
+export const getPreviewType = (hostname, deliveryTypes) => {
+  if (HLX_HOST.test(hostname)) {
+    return deliveryTypes.AEM_EDGE;
+  }
+  if (AEM_CS_HOST.test(hostname)) {
+    return deliveryTypes.AEM_CS;
+  }
+  return null;
+};
+
 /**
  * A class representing a Site entity. Provides methods to access and manipulate Site-specific data.
  * @class Site
@@ -35,7 +79,7 @@ class Site extends BaseModel {
   }
 
   /**
-   * Resolves the site's base URL to a  final URL by fetching the URL,
+   * Resolves the site's base URL to a final URL by fetching the URL,
    * following the redirects and returning the final URL.
    *
    * If the site has a configured overrideBaseURL, that one will be returned.

package/src/models/site/site.schema.js

@@ -22,7 +22,7 @@ import {
 import { Config, DEFAULT_CONFIG, validateConfiguration } from './config.js';
 import SchemaBuilder from '../base/schema.builder.js';
 
-import Site from './site.model.js';
+import Site, { computeExternalIds } from './site.model.js';
 import SiteCollection from './site.collection.js';
 
 /*
@@ -94,10 +94,28 @@ const schema = new SchemaBuilder(Site, SiteCollection)
     set: () => new Date().toISOString(),
     validate: (value) => !value || isIsoDate(value),
   })
+  .addAttribute('externalOwnerId', {
+    type: 'string',
+    hidden: true,
+    readOnly: true,
+    watch: ['hlxConfig', 'deliveryConfig'],
+    set: (_, attrs) => computeExternalIds(attrs).externalOwnerId,
+  })
+  .addAttribute('externalSiteId', {
+    type: 'string',
+    hidden: true,
+    readOnly: true,
+    watch: ['hlxConfig', 'deliveryConfig'],
+    set: (_, attrs) => computeExternalIds(attrs).externalSiteId,
+  })
   .addAllIndex(['baseURL'])
   .addIndex(
     { composite: ['deliveryType'] },
     { composite: ['updatedAt'] },
+  )
+  .addIndex(
+    { composite: ['externalOwnerId'] },
+    { composite: ['externalSiteId'] },
   );
 
 export default schema.build();