@adobe/spacecat-shared-data-access 2.53.0 → 2.55.0

package/CHANGELOG.md

@@ -1,3 +1,17 @@
+# [@adobe/spacecat-shared-data-access-v2.55.0](https://github.com/adobe/spacecat-shared/compare/@adobe/spacecat-shared-data-access-v2.54.0...@adobe/spacecat-shared-data-access-v2.55.0) (2025-08-20)
+
+
+### Features
+
+* add scrape-client destination  ([#913](https://github.com/adobe/spacecat-shared/issues/913)) ([e208a87](https://github.com/adobe/spacecat-shared/commit/e208a87214874a2708ac2d7614fcfd4c0770fe17))
+
+# [@adobe/spacecat-shared-data-access-v2.54.0](https://github.com/adobe/spacecat-shared/compare/@adobe/spacecat-shared-data-access-v2.53.0...@adobe/spacecat-shared-data-access-v2.54.0) (2025-08-13)
+
+
+### Features
+
+* add isOriginal field to scrapeUrl ([#908](https://github.com/adobe/spacecat-shared/issues/908)) ([03dfd59](https://github.com/adobe/spacecat-shared/commit/03dfd59c94f91204ac995aeeac53d4e128317dd5))
+
 # [@adobe/spacecat-shared-data-access-v2.53.0](https://github.com/adobe/spacecat-shared/compare/@adobe/spacecat-shared-data-access-v2.52.0...@adobe/spacecat-shared-data-access-v2.53.0) (2025-08-12)
 
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@adobe/spacecat-shared-data-access",
-  "version": "2.53.0",
+  "version": "2.55.0",
   "description": "Shared modules of the Spacecat Services - Data Access",
   "type": "module",
   "engines": {

package/src/models/audit/audit.model.js

@@ -73,6 +73,7 @@ class Audit extends BaseModel {
   static AUDIT_STEP_DESTINATIONS = {
     CONTENT_SCRAPER: 'content-scraper',
     IMPORT_WORKER: 'import-worker',
+    SCRAPE_CLIENT: 'scrape-client',
   };
 
   /**
@@ -86,7 +87,9 @@ class Audit extends BaseModel {
    *   [Audit.AUDIT_STEP_DESTINATIONS.IMPORT_WORKER]: {
    *     getQueueUrl: function,
    *     formatPayload: function
-   *   }
+   *   },
+   *   [Audit.AUDIT_STEP_DESTINATIONS.SCRAPE_CLIENT]: {
+   *   formatPayload: function
    * }}
    */
   static AUDIT_STEP_DESTINATION_CONFIGS = {
@@ -153,6 +156,26 @@ class Audit extends BaseModel {
         auditContext,
       }),
     },
+    [Audit.AUDIT_STEP_DESTINATIONS.SCRAPE_CLIENT]: {
+      /**
+       *
+       * @param stepResult - The result of the audit step.
+       * @param auditContext - The audit context.
+       * @param context - The context object.
+       * @returns {object} - The formatted payload for the scrape client.
+       */
+      formatPayload: (stepResult, auditContext, context) => ({
+        urls: stepResult.urls.map((urlObj) => urlObj.url),
+        processingType: stepResult.processingType || 'default',
+        options: stepResult.options || {},
+        maxScrapeAge: stepResult.maxScrapeAge || 24,
+        auditData: {
+          siteId: stepResult.siteId,
+          completionQueueUrl: stepResult.completionQueueUrl || context.env?.AUDIT_JOBS_QUEUE_URL,
+          auditContext,
+        },
+      }),
+    },
   };
 
   /**

package/src/models/scrape-url/index.d.ts

@@ -20,12 +20,14 @@ export interface ScrapeUrl extends BaseModel {
   getReason(): string,
   getStatus(): string,
   getUrl(): string,
+  getIsOriginal(): boolean,
   setFile(file: string): void,
   setScrapeJobId(ScrapeJobId: string): void,
   setPath(path: string): void,
   setReason(reason: string): void,
   setStatus(status: string): void,
   setUrl(url: string): void,
+  setIsOriginal(isOriginal: boolean): void,
 }
 
 export interface ScrapeUrlCollection extends BaseCollection<ScrapeUrl> {

package/src/models/scrape-url/scrape-url.schema.js

@@ -45,6 +45,10 @@ const schema = new SchemaBuilder(ScrapeUrl, ScrapeUrlCollection)
     type: 'string',
     required: true,
     validate: (value) => isValidUrl(value),
+  })
+  .addAttribute('isOriginal', {
+    type: 'boolean',
+    default: true,
   });
 
 export default schema.build();