ci: Bump version to 0.9.52 [skip ci]

package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "flyer-crawler",
-  "version": "0.9.50",
+  "version": "0.9.52",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "flyer-crawler",
-      "version": "0.9.50",
+      "version": "0.9.52",
       "dependencies": {
         "@bull-board/api": "^6.14.2",
         "@bull-board/express": "^6.14.2",

package.json

@@ -1,7 +1,7 @@
 {
   "name": "flyer-crawler",
   "private": true,
-  "version": "0.9.50",
+  "version": "0.9.52",
   "type": "module",
   "scripts": {
     "dev": "concurrently \"npm:start:dev\" \"vite\"",

src/services/aiService.server.ts

@@ -543,6 +543,20 @@ export class AIService {
     logger.info(
       `[extractCoreDataFromFlyerImage] Entering method with ${imagePaths.length} image(s).`,
     );
+
+    // [TEST HOOK] Simulate an AI failure if the filename contains specific text.
+    // This allows integration tests to verify error handling.
+    if (imagePaths.some((f) => f.path.includes('ai-fail-test'))) {
+      logger.warn('[TEST HOOK] Simulating AI failure for test file.');
+      throw new Error('AI model failed to extract data.');
+    }
+
+    // [TEST HOOK] Simulate a specific failure for the cleanup test
+    if (imagePaths.some((f) => f.path.includes('cleanup-fail-test'))) {
+      logger.warn('[TEST HOOK] Simulating AI failure for cleanup test.');
+      throw new Error('Simulated AI failure for cleanup test.');
+    }
+
     const prompt = this._buildFlyerExtractionPrompt(masterItems, submitterIp, userProfileAddress);
 
     const imageParts = await Promise.all(

src/services/db/flyer.db.ts

@@ -64,7 +64,17 @@ export class FlyerRepository {
    */
   async insertFlyer(flyerData: FlyerDbInsert, logger: Logger): Promise<Flyer> {
     console.error('[DEBUG] FlyerRepository.insertFlyer called with:', JSON.stringify(flyerData, null, 2));
+    // [TEST HOOK] Simulate a database failure if the filename contains specific text.
+    // This allows integration tests to verify error handling without mocking the entire DB connection.
+    if (flyerData.file_name.includes('db-fail-test')) {
+      logger.warn('[TEST HOOK] Simulating DB transaction failure for test file.');
+      throw new Error('DB transaction failed for test.');
+    }
+
     try {
+      // Sanitize icon_url: Ensure empty strings become NULL to avoid regex constraint violations
+      const iconUrl = flyerData.icon_url && flyerData.icon_url.trim() !== '' ? flyerData.icon_url : null;
+
       const query = `
         INSERT INTO flyers (
           file_name, image_url, icon_url, checksum, store_id, valid_from, valid_to, store_address,
@@ -76,7 +86,7 @@ export class FlyerRepository {
       const values = [
         flyerData.file_name, // $1
         flyerData.image_url, // $2
-        flyerData.icon_url, // $3
+        iconUrl, // $3
         flyerData.checksum, // $4
         flyerData.store_id, // $5
         flyerData.valid_from, // $6

src/services/flyerAiProcessor.server.ts

@@ -103,6 +103,7 @@ export class FlyerAiProcessor {
     jobData: FlyerJobData,
     logger: Logger,
   ): Promise<AiProcessorResult> {
+    console.error(`[WORKER DEBUG] FlyerAiProcessor: extractAndValidateData called with ${imagePaths.length} images`);
     logger.info(`Starting AI data extraction for ${imagePaths.length} pages.`);
     const { submitterIp, userProfileAddress } = jobData;
     const masterItems = await this.personalizationRepo.getAllMasterItems(logger);
@@ -159,6 +160,7 @@ export class FlyerAiProcessor {
     }
 
     logger.info(`Batch processing complete. Total items extracted: ${mergedData.items.length}`);
+    console.error(`[WORKER DEBUG] FlyerAiProcessor: Merged AI Data:`, JSON.stringify(mergedData, null, 2));
 
     // Validate the final merged dataset
     return this._validateAiData(mergedData, logger);

src/services/flyerFileHandler.server.ts

@@ -253,7 +253,9 @@ export class FlyerFileHandler {
     job: Job<FlyerJobData>,
     logger: Logger,
   ): Promise<{ imagePaths: { path: string; mimetype: string }[]; createdImagePaths: string[] }> {
+    console.error(`[WORKER DEBUG] FlyerFileHandler: prepareImageInputs called for ${filePath}`);
     const fileExt = path.extname(filePath).toLowerCase();
+    console.error(`[WORKER DEBUG] FlyerFileHandler: Detected extension: ${fileExt}`);
 
     if (fileExt === '.pdf') {
       return this._handlePdfInput(filePath, job, logger);

src/services/flyerProcessingService.server.ts

@@ -69,6 +69,7 @@ export class FlyerProcessingService {
       // Stage 1: Prepare Inputs (e.g., convert PDF to images)
       stages[0].status = 'in-progress';
       await job.updateProgress({ stages });
+      console.error(`[WORKER DEBUG] ProcessingService: Calling fileHandler.prepareImageInputs for ${job.data.filePath}`);
 
       const { imagePaths, createdImagePaths } = await this.fileHandler.prepareImageInputs(
         job.data.filePath,
@@ -76,6 +77,7 @@ export class FlyerProcessingService {
         logger,
       );
       allFilePaths.push(...createdImagePaths);
+      console.error(`[WORKER DEBUG] ProcessingService: fileHandler returned ${imagePaths.length} images.`);
       stages[0].status = 'completed';
       stages[0].detail = `${imagePaths.length} page(s) ready for AI.`;
       await job.updateProgress({ stages });
@@ -84,7 +86,9 @@ export class FlyerProcessingService {
       stages[1].status = 'in-progress';
       await job.updateProgress({ stages });
 
+      console.error(`[WORKER DEBUG] ProcessingService: Calling aiProcessor.extractAndValidateData`);
       const aiResult = await this.aiProcessor.extractAndValidateData(imagePaths, job.data, logger);
+      console.error(`[WORKER DEBUG] ProcessingService: aiProcessor returned data for store: ${aiResult.data.store_name}`);
       stages[1].status = 'completed';
       await job.updateProgress({ stages });
 
@@ -97,7 +101,9 @@ export class FlyerProcessingService {
       const primaryImagePath = imagePaths[0].path;
       const imageFileName = path.basename(primaryImagePath);
       const iconsDir = path.join(path.dirname(primaryImagePath), 'icons');
+      console.error(`[WORKER DEBUG] ProcessingService: Generating icon from ${primaryImagePath} to ${iconsDir}`);
       const iconFileName = await generateFlyerIcon(primaryImagePath, iconsDir, logger);
+      console.error(`[WORKER DEBUG] ProcessingService: Icon generated: ${iconFileName}`);
 
       // Add the newly generated icon to the list of files to be cleaned up.
       // The main processed image path is already in `allFilePaths` via `createdImagePaths`.
@@ -120,6 +126,7 @@ export class FlyerProcessingService {
         baseUrl,
       );
       console.error('[DEBUG] FlyerProcessingService transformer output URLs:', { imageUrl: flyerData.image_url, iconUrl: flyerData.icon_url });
+      console.error('[DEBUG] Full Flyer Data to be saved:', JSON.stringify(flyerData, null, 2));
       stages[2].status = 'completed';
       await job.updateProgress({ stages });
 
@@ -150,6 +157,12 @@ export class FlyerProcessingService {
         });
         flyerId = flyer.flyer_id;
       } catch (error) {
+        // Capture specific validation errors and append context for debugging
+        if (error instanceof Error && error.message.includes('Invalid URL')) {
+             const msg = `DB Validation Failed: ${error.message}. ImageURL: '${flyerData.image_url}', IconURL: '${flyerData.icon_url}'`;
+             console.error('[ERROR] ' + msg);
+             throw new Error(msg);
+        }
         if (error instanceof FlyerProcessingError) throw error;
         throw new DatabaseError(error instanceof Error ? error.message : String(error));
       }

src/tests/integration/flyer-processing.integration.test.ts

@@ -175,6 +175,7 @@ describe('Flyer Processing Background Job Integration Test', () => {
     if (jobStatus?.state === 'failed') {
       console.error('[DEBUG] Job failed with reason:', jobStatus.failedReason);
       console.error('[DEBUG] Job stack trace:', jobStatus.stacktrace);
+      console.error('[DEBUG] Job return value:', JSON.stringify(jobStatus.returnValue, null, 2));
       console.error('[DEBUG] Full Job Status:', JSON.stringify(jobStatus, null, 2));
     }
     expect(jobStatus?.state).toBe('completed');