tender duplication fix

internal/tender/entity.go

@@ -39,6 +39,8 @@ type Tender struct {
 	// (same procedure / contract_folder_id). Ordered oldest-first; the latest is also reflected in NoticePublicationID.
 	RelatedNoticePublicationIDs []string             `bson:"related_notice_publication_ids,omitempty" json:"related_notice_publication_ids,omitempty"`
 	ContractFolderID            string               `bson:"contract_folder_id" json:"contract_folder_id"`
+	// ProcurementProjectID is UBL ProcurementProject/ID — merges per-lot CAN/CN notices into one tender.
+	ProcurementProjectID        string               `bson:"procurement_project_id,omitempty" json:"procurement_project_id,omitempty"`
 	NoticeTypeCode              string               `bson:"notice_type_code" json:"notice_type_code"`
 	FormType                    string               `bson:"form_type,omitempty" json:"form_type,omitempty"`
 	NoticeSubTypeCode           string               `bson:"notice_sub_type_code" json:"notice_sub_type_code"`

internal/tender/repository.go

@@ -21,6 +21,7 @@ type TenderRepository interface {
 	GetByContractNoticeID(ctx context.Context, contractNoticeID string) (*Tender, error)
 	GetByNoticePublicationID(ctx context.Context, noticePublicationID string) (*Tender, error)
 	GetByContractFolderID(ctx context.Context, contractFolderID string) (*Tender, error)
+	GetByProcurementProjectID(ctx context.Context, procurementProjectID 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)
@@ -54,6 +55,14 @@ func NewRepository(mongoManager *orm.ConnectionManager, logger logger.Logger) Te
 		*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}}),
+		// One tender row per TED publication / procedure when ids are present (prevents parallel worker races).
+		*orm.CreateUniqueIndex("notice_publication_id_unique", bson.D{{Key: "notice_publication_id", Value: 1}}).
+			WithPartialFilterExpression(bson.M{"notice_publication_id": bson.M{"$type": "string", "$ne": ""}}),
+		*orm.CreateUniqueIndex("contract_folder_id_unique", bson.D{{Key: "contract_folder_id", Value: 1}}).
+			WithPartialFilterExpression(bson.M{"contract_folder_id": bson.M{"$type": "string", "$ne": ""}}),
+		// One tender per UBL procurement project (merges per-lot notices that share ProcurementProject/ID).
+		*orm.CreateUniqueIndex("procurement_project_id_unique", bson.D{{Key: "procurement_project_id", Value: 1}}).
+			WithPartialFilterExpression(bson.M{"procurement_project_id": bson.M{"$type": "string", "$ne": ""}}),
 		*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}}),
@@ -216,6 +225,31 @@ func (r *tenderRepository) GetByContractFolderID(ctx context.Context, contractFo
 	return &result.Items[0], nil
 }
 
+// GetByProcurementProjectID finds a tender by UBL ProcurementProject ID (shared across per-lot notices).
+func (r *tenderRepository) GetByProcurementProjectID(ctx context.Context, procurementProjectID string) (*Tender, error) {
+	filter := bson.M{"procurement_project_id": procurementProjectID}
+	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 procurement project ID", map[string]interface{}{
+			"procurement_project_id": procurementProjectID,
+			"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": ""}}