Remove ContentXML field from Tender entity and related repository method

- Eliminated the `ContentXML` field from the `Tender` struct in the entity definition, streamlining the data model.
- Removed the `FindTendersWithContentXML` method from the `TenderRepository` interface, as it is no longer necessary.
- Updated the `ToTender` method in the `NoticeWorker` to reflect these changes, ensuring consistency across the codebase.

This update simplifies the tender data structure and repository interface, improving maintainability and clarity in the tender management system.

cmd/worker/workers/notice.go

@@ -565,7 +565,6 @@ func (w *NoticeWorker) ToTender(n *notice.Notice) (*tender.Tender, error) {
 	t.ProcurementProjectID = n.ProcurementProjectID
 	t.SourceFileURL = n.SourceFileURL
 	t.SourceFileName = n.SourceFileName
-	t.ContentXML = n.ContentXML
 	// Preserve document-scraping/summarization flags on the tender (managed by
 	// downstream workers), but refresh the notice-side processing metadata.
 	t.ProcessingMetadata.ScrapedAt = n.ProcessingMetadata.ScrapedAt

internal/tender/entity.go

@@ -88,7 +88,6 @@ type Tender struct {
 	Source                        TenderSource         `bson:"source" json:"source"`
 	SourceFileURL                 string               `bson:"source_file_url" json:"source_file_url"`
 	SourceFileName                string               `bson:"source_file_name" json:"source_file_name"`
-	ContentXML                    string               `bson:"content_xml,omitempty" json:"content_xml,omitempty"`
 	ProcessingMetadata            ProcessingMetadata   `bson:"processing_metadata" json:"processing_metadata"`
 	// AIOverallSummary optional denormalized overall summary for search/list (e.g. synced from AI pipeline storage).
 	AIOverallSummary string `bson:"ai_overall_summary,omitempty" json:"ai_overall_summary,omitempty"`

internal/tender/repository.go

@@ -27,7 +27,6 @@ type TenderRepository interface {
 	// GetLatestByContractFolderIDs returns the most recently updated tender per distinct contract_folder_id.
 	GetLatestByContractFolderIDs(ctx context.Context, contractFolderIDs []string) (map[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)
 	GetTenderCountByClassification(ctx context.Context) (map[string]int64, error)
@@ -667,27 +666,6 @@ func (r *tenderRepository) GetByProcurementProjectID(ctx context.Context, procur
 	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": ""}}
-	pagination := orm.Pagination{
-		Limit:     limit,
-		Skip:      skip,
-		SortField: "_id",
-		SortOrder: 1,
-	}
-
-	result, err := r.ormRepo.FindAll(ctx, filter, pagination)
-	if err != nil {
-		r.logger.Error("Failed to list tenders with content_xml", map[string]interface{}{
-			"error": err.Error(),
-		})
-		return nil, err
-	}
-
-	return result.Items, nil
-}
-
 // Update updates an existing tender
 func (r *tenderRepository) Update(ctx context.Context, tender *Tender) error {
 	tender.UpdatedAt = time.Now().Unix()