merge tenders which have the same procedure id - worker performance fix

internal/tender/repository.go

@@ -20,6 +20,7 @@ type TenderRepository interface {
 	GetByID(ctx context.Context, id string) (*Tender, error)
 	GetByContractNoticeID(ctx context.Context, contractNoticeID string) (*Tender, error)
 	GetByNoticePublicationID(ctx context.Context, noticePublicationID string) (*Tender, error)
+	GetByContractFolderID(ctx context.Context, contractFolderID string) (*Tender, error)
 	FindTendersWithContentXML(ctx context.Context, limit, skip int) ([]Tender, error)
 	GetTenderCountByCountry(ctx context.Context) (map[string]int64, error)
 	GetTenderCountByType(ctx context.Context) (map[string]int64, error)
@@ -52,6 +53,7 @@ func NewRepository(mongoManager *orm.ConnectionManager, logger logger.Logger) Te
 		*orm.CreateUniqueIndex("tender_id_idx", bson.D{{Key: "tender_id", Value: 1}}),
 		*orm.NewIndex("project_name_idx", bson.D{{Key: "project_name", Value: 1}}),
 		*orm.NewIndex("contract_notice_id_idx", bson.D{{Key: "contract_notice_id", Value: 1}}),
+		*orm.NewIndex("contract_folder_id_idx", bson.D{{Key: "contract_folder_id", Value: 1}}),
 		*orm.NewIndex("status_idx", bson.D{{Key: "status", Value: 1}}),
 		*orm.NewIndex("source_idx", bson.D{{Key: "source", Value: 1}}),
 		*orm.NewIndex("country_code_idx", bson.D{{Key: "country_code", Value: 1}}),
@@ -158,9 +160,16 @@ func (r *tenderRepository) GetByContractNoticeID(ctx context.Context, contractNo
 	return &result.Items[0], nil
 }
 
-// GetByNoticePublicationID retrieves a tender by notice publication ID
+// GetByNoticePublicationID retrieves a tender by notice publication ID.
+// Matches both the canonical NoticePublicationID and any id retained in RelatedNoticePublicationIDs
+// so historical TED detail links keep resolving after a tender update from a follow-up notice.
 func (r *tenderRepository) GetByNoticePublicationID(ctx context.Context, noticePublicationID string) (*Tender, error) {
-	filter := bson.M{"notice_publication_id": noticePublicationID}
+	filter := bson.M{
+		"$or": []bson.M{
+			{"notice_publication_id": noticePublicationID},
+			{"related_notice_publication_ids": noticePublicationID},
+		},
+	}
 	pagination := orm.Pagination{Limit: 1}
 
 	result, err := r.ormRepo.FindAll(ctx, filter, pagination)
@@ -179,6 +188,34 @@ func (r *tenderRepository) GetByNoticePublicationID(ctx context.Context, noticeP
 	return &result.Items[0], nil
 }
 
+// GetByContractFolderID retrieves a tender by its procedure-level identifier (ContractFolderID).
+// TED groups multiple notice publications (e.g. competition notice + result notice) under the same
+// ContractFolderID, so this lookup is the authoritative way to find the existing tender for a procedure
+// when a follow-up notice arrives. Returns the most recently updated match if there are several.
+func (r *tenderRepository) GetByContractFolderID(ctx context.Context, contractFolderID string) (*Tender, error) {
+	filter := bson.M{"contract_folder_id": contractFolderID}
+	pagination := orm.Pagination{
+		Limit:     1,
+		SortField: "updated_at",
+		SortOrder: -1,
+	}
+
+	result, err := r.ormRepo.FindAll(ctx, filter, pagination)
+	if err != nil {
+		r.logger.Error("Failed to get tender by contract folder ID", map[string]interface{}{
+			"contract_folder_id": contractFolderID,
+			"error":              err.Error(),
+		})
+		return nil, err
+	}
+
+	if len(result.Items) == 0 {
+		return nil, orm.ErrDocumentNotFound
+	}
+
+	return &result.Items[0], nil
+}
+
 // FindTendersWithContentXML returns tenders that still have TED XML (for backfill after notices were deleted).
 func (r *tenderRepository) FindTendersWithContentXML(ctx context.Context, limit, skip int) ([]Tender, error) {
 	filter := bson.M{"content_xml": bson.M{"$exists": true, "$ne": ""}}