Enhance Scraper with One-Time Date Range Functionality

- Added command line flags for specifying a date range and a one-time scraping mode in the scraper's main function.
- Implemented validation for the date inputs to ensure both dates are provided and correctly formatted.
- Introduced a new RunOneTimeScraping function to handle the one-time scraping logic, including context management and error handling.
- Updated the TED scraper initialization to support both scheduled and one-time scraping modes, improving flexibility in data retrieval.
- Enhanced logging to provide clear feedback on the scraping process and any errors encountered.

cmd/scraper/bootstrap/bootstrap.go

@@ -104,7 +104,7 @@ func InitTEDScraper(config Config, mongoManager *mongo.ConnectionManager, appLog
 	schedule.NewCronScheduler(appLogger, mongoManager, noticeRepo).AddJob(schedule.Job{
 		Name: "TED Scraper Job",
 		Func: func() {
-			_ = tedScraper.Run(context.Background())
+			_ = tedScraper.Run(context.Background(), nil, nil)
 		},
 		Expr: config.TED.ScrapingInterval,
 	})
@@ -123,7 +123,7 @@ func InitTEDScraper(config Config, mongoManager *mongo.ConnectionManager, appLog
 		appLogger.Info("Starting TED scraper", map[string]interface{}{
 			"interval": config.TED.ScrapingInterval,
 		})
-		scraperDone <- tedScraper.Run(ctx)
+		scraperDone <- tedScraper.Run(ctx, nil, nil)
 	}()
 
 	// Wait for shutdown signal or scraper completion
@@ -183,3 +183,63 @@ func InitNotificationService(conf config.NotificationConfig, log logger.Logger)
 
 	return *notificationSDK
 }
+
+// RunOneTimeScraping runs the TED scraper once for a specific date range
+func RunOneTimeScraping(
+	config Config,
+	mongoManager *mongo.ConnectionManager,
+	appLogger logger.Logger,
+	notificationService notification.SDK,
+	from, to *time.Time,
+) {
+	// Initialize notice repository
+	noticeRepo := notice.NewRepository(mongoManager, appLogger)
+
+	// Convert scraper config to TED config
+	tedConfig := &ted.Config{
+		BaseURL:          config.TED.BaseURL,
+		UserAgent:        config.TED.UserAgent,
+		DownloadDir:      config.TED.DownloadDir,
+		MaxRetries:       config.TED.MaxRetries,
+		MaxConcurrency:   config.TED.MaxConcurrency,
+		Timeout:          config.TED.Timeout,
+		RetryDelay:       config.TED.RetryDelay,
+		CleanupAfter:     config.TED.CleanupAfter,
+		ScrapingInterval: config.TED.ScrapingInterval,
+		AlertMail:        config.AlertMail,
+	}
+
+	// Initialize TED scraper
+	tedScraper := ted.NewTEDScraper(
+		tedConfig,
+		appLogger,
+		mongoManager,
+		noticeRepo,
+		notificationService,
+	)
+
+	appLogger.Info("Starting one-time TED scraping", map[string]interface{}{
+		"from_date": from.Format("02/01/2006"),
+		"to_date":   to.Format("02/01/2006"),
+	})
+
+	// Create context with timeout for the entire operation
+	ctx, cancel := context.WithTimeout(context.Background(), 24*time.Hour) // 24 hour timeout
+	defer cancel()
+
+	// Run the scraper with date range
+	err := tedScraper.Run(ctx, from, to)
+	if err != nil {
+		appLogger.Error("One-time scraping failed", map[string]interface{}{
+			"error":     err.Error(),
+			"from_date": from.Format("02/01/2006"),
+			"to_date":   to.Format("02/01/2006"),
+		})
+		os.Exit(1)
+	}
+
+	appLogger.Info("One-time TED scraping completed successfully", map[string]interface{}{
+		"from_date": from.Format("02/01/2006"),
+		"to_date":   to.Format("02/01/2006"),
+	})
+}