mx-cloud 0.0.25 → 0.0.27

package/build/interpret.d.ts

@@ -38,12 +38,13 @@ interface InterpreterOptions {
     serializableCallback: (output: any) => (void | Promise<void>);
     binaryCallback: (output: any, mimeType: string) => (void | Promise<void>);
     debug: boolean;
-    robotType?: 'extract' | 'scrape' | 'deep-extract';
+    robotType?: 'extract' | 'scrape' | 'crawl' | 'search' | 'deep-extract';
     debugChannel: Partial<{
         activeId: (id: number) => void;
         debugMessage: (msg: string) => void;
         setActionType: (type: string) => void;
         incrementScrapeListIndex: () => void;
+        progressUpdate: (current: number, total: number, percentage: number) => void;
     }>;
 }
 /**
@@ -66,28 +67,14 @@ export default class Interpreter extends EventEmitter {
     private serializableDataByType;
     private pendingDeepExtraction;
     private isInDeepExtractionPhase;
+    private deepExtractionStats;
+    private totalActions;
+    private executedActions;
     constructor(workflow: WorkflowFile, options?: Partial<InterpreterOptions>);
     trackAutohealFailure(error: string): void;
     private applyAdBlocker;
     private disableAdBlocker;
-    private getSelectors;
-    /**
-      * Returns the context object from given Page and the current workflow.\
-      * \
-      * `workflow` is used for selector extraction - function searches for used selectors to
-      * look for later in the page's context.
-      * @param page Playwright Page object
-      * @param workflow Current **initialized** workflow (array of where-what pairs).
-      * @returns {PageState} State of the current page.
-      */
-    private getState;
-    /**
-     * Tests if the given action is applicable with the given context.
-     * @param where Tested *where* condition
-     * @param context Current browser context.
-     * @returns True if `where` is applicable in the given context, false otherwise
-     */
-    private applicable;
+    private callWithTimeout;
     /**
      * Sets the abort flag to immediately stop all operations
      */
@@ -107,12 +94,7 @@ export default class Interpreter extends EventEmitter {
    */
     private carryOutSteps;
     private handlePagination;
-    private getMatchingActionId;
-    private removeShadowSelectors;
     private removeSpecialSelectors;
-    private generatePageNodeInformation;
-    private detectElementChanges;
-    private validateWorkflowAction;
     /**
      * Test if a selector is working on the current page
      * @param {Page} page - Playwright page object
@@ -189,6 +171,16 @@ export default class Interpreter extends EventEmitter {
     /**
      * Helper function to check if a URL matches a goto pattern.
      */
+    /**
+     * Generic pattern matching for deep extraction URLs.
+     * Works across any website by analyzing URL structure rather than relying on keywords.
+     *
+     * Strategy:
+     * 1. Match URLs with same origin and path length
+     * 2. Identify "structural" segments (numbers, short words, etc.) that should match exactly
+     * 3. Allow other segments to vary (dynamic content like IDs, slugs, names)
+     * 4. Skip exact matches to avoid duplicates
+     */
     private matchesGotoPattern;
     /**
      * Executes hierarchical deep extraction by processing each level recursively.