Add TED Scraper Configuration and Cron Scheduler Implementation

- Introduced a new configuration file for the TED scraper, defining database connection settings, logging preferences, and scraping parameters.
- Implemented a CronScheduler to manage scheduled tasks for TED operations, utilizing the robfig/cron library for scheduling.
- Added new entities and structures for handling TED XML data, including eForms and tender-related information, enhancing the scraper's functionality.
- Updated the main application to initialize the TED scraper with the new configuration and set up graceful shutdown handling.
- Removed the deprecated scraping implementation to streamline the codebase and focus on the new architecture.

pkg/schedule/cron.go

@@ -0,0 +1,144 @@
+package schedule
+
+import (
+	"time"
+	"tm/internal/tender"
+	"tm/pkg/logger"
+	"tm/pkg/mongo"
+
+	"github.com/robfig/cron/v3"
+)
+
+// CronScheduler handles scheduled tasks for TED operations
+type CronScheduler struct {
+	cron         *cron.Cron
+	mongoManager *mongo.ConnectionManager
+	tenderRepo   tender.TenderRepository
+	logger       logger.Logger
+}
+
+// Logger interface for structured logging
+type Logger interface {
+	Info(msg string, fields map[string]interface{})
+	Error(msg string, fields map[string]interface{})
+	Debug(msg string, fields map[string]interface{})
+}
+
+type Job struct {
+	Name string
+	Func func()
+	Expr string
+}
+
+// NewCronScheduler creates a new cron scheduler instance
+func NewCronScheduler(logger logger.Logger, mongoManager *mongo.ConnectionManager, tenderRepo tender.TenderRepository) *CronScheduler {
+	// Create cron with timezone support and seconds precision
+	c := cron.New(cron.WithLocation(time.UTC), cron.WithSeconds())
+
+	return &CronScheduler{
+		cron:         c,
+		mongoManager: mongoManager,
+		tenderRepo:   tenderRepo,
+		logger:       logger,
+	}
+}
+
+// Start starts the cron scheduler
+func (cs *CronScheduler) Start() {
+	cs.logger.Info("Starting TED cron scheduler", map[string]interface{}{})
+	cs.cron.Start()
+}
+
+// Stop stops the cron scheduler gracefully
+func (cs *CronScheduler) Stop() {
+	cs.logger.Info("Stopping TED cron scheduler", map[string]interface{}{})
+	ctx := cs.cron.Stop()
+	<-ctx.Done()
+}
+
+// ScheduleDailyJob schedules a job to run daily at 10:00 AM UTC
+func (cs *CronScheduler) AddJob(job Job) error {
+	// Cron expression: "0 0 10 * * *" means:
+	// Second: 0
+	// Minute: 0
+	// Hour: 10 (10 AM)
+	// Day of month: * (every day)
+	// Month: * (every month)
+	// Day of week: * (every day of week)
+	_, err := cs.cron.AddFunc(job.Expr, func() {
+		now := time.Now().Local()
+		cs.logger.Info("Starting daily TED job", map[string]interface{}{
+			"job_name":        job.Name,
+			"cron_expression": job.Expr,
+			"time":            time.Now().Format(time.RFC3339),
+		})
+
+		job.Func()
+
+		cs.logger.Info("Completed job successfully", map[string]interface{}{
+			"time":            now.Format(time.RFC3339),
+			"job_name":        job.Name,
+			"cron_expression": job.Expr,
+		})
+	})
+
+	if err != nil {
+		cs.logger.Error("Failed to schedule daily job", map[string]interface{}{
+			"error":           err.Error(),
+			"job_name":        job.Name,
+			"cron_expression": job.Expr,
+			"time":            time.Now().Format(time.RFC3339),
+		})
+		return err
+	}
+
+	cs.logger.Info("Daily job scheduled successfully", map[string]interface{}{
+		"job_name":        job.Name,
+		"cron_expression": job.Expr,
+		"time":            time.Now().Format(time.RFC3339),
+	})
+
+	return nil
+}
+
+// ExampleTEDScrapingJob is an example job function that can be scheduled
+func ExampleTEDScrapingJob() {
+	// now := time.Now().Local()
+	// // Example: Get current date for TED scraping
+	// currentDate := time.Now().Format("02/01/2006")
+	// currentYear := time.Now().Year()
+
+	// // This is just an example - replace with actual TED scraping logic
+	// ojs, err := GetOJS(currentYear, currentDate)
+	// if err != nil {
+	// 	log.Printf("Error getting OJS: %v", err)
+	// 	return
+	// }
+
+	// log.Printf("Successfully retrieved OJS: %s for date: %s", ojs, currentDate)
+
+	// Add your actual TED scraping/processing logic here
+
+	// ojs, err := GetOJS(now.Year(), now.Format("02/01/2006"))
+	// if err != nil {
+	// 	cs.logger.Error("Failed to get OJS", map[string]interface{}{
+	// 		"error": err.Error(),
+	// 	})
+	// 	return
+	// }
+
+	// scraper := NewTEDScraper(cs.config, cs.logger, cs.mongoManager, cs.tenderRepo)
+	// result, err := scraper.DownloadDailyPackage(context.Background(), ojs)
+	// if err != nil {
+	// 	cs.logger.Error("Failed to download daily package", map[string]interface{}{
+	// 		"error": err.Error(),
+	// 	})
+	// 	return
+	// }
+
+	// cs.logger.Info("Downloaded daily package successfully", map[string]interface{}{
+	// 	"result": result,
+	// 	"time":   now.Format(time.RFC3339),
+	// 	"ojs":    ojs,
+	// })
+}