fix: avoid duplicate document names when linking files to datasets (#17181)

internal/service/document/document_crud.go

@@ -306,6 +306,22 @@ func (s *DocumentService) InsertDocument(doc *entity.Document) error {
 	})
 }
 
+// UniqueDocumentName returns a non-colliding document name within the dataset,
+// mirroring Python duplicate_name(DocumentService.query, name=..., kb_id=...).
+// Returns an error when the existing-name lookup fails so callers never write
+// blind and risk duplicated document names.
+func (s *DocumentService) UniqueDocumentName(kbID, name string) (string, error) {
+	names, err := s.documentDAO.ListNamesByKbID(kbID)
+	if err != nil {
+		return "", err
+	}
+	taken := make(map[string]bool, len(names))
+	for _, n := range names {
+		taken[n] = true
+	}
+	return uniqueUploadName(name, taken), nil
+}
+
 // resolveDocAndKB loads the document and its knowledgebase, returning both or
 // an error.
 func (s *DocumentService) resolveDocAndKB(docID string) (*entity.Document, *entity.Knowledgebase, error) {

internal/service/document/file2document.go

@@ -224,8 +224,18 @@ func (s *File2DocumentService) convertFiles(fileIDs, kbIDs []string, userID, mod
 				continue
 			}
 
+			// Mirror Python duplicate_name: generate a non-colliding document
+			// name within the dataset so that linking does not silently create
+			// duplicate-name documents in the same KB.
+			docName, err := s.documentSvc.UniqueDocumentName(kbID, file.Name)
+			if err != nil {
+				common.Warn("convertFiles: UniqueDocumentName failed",
+					zap.String("kbID", kbID), zap.String("fileID", fileID), zap.Error(err))
+				continue
+			}
+
 			parserID := kb.ParserID
-			suffix := strings.TrimPrefix(filepath.Ext(file.Name), ".")
+			suffix := strings.TrimPrefix(filepath.Ext(docName), ".")
 			doc := &entity.Document{
 				ID:           utility.GenerateUUID(),
 				KbID:         kb.ID,
@@ -233,7 +243,7 @@ func (s *File2DocumentService) convertFiles(fileIDs, kbIDs []string, userID, mod
 				ParserConfig: kb.ParserConfig,
 				CreatedBy:    userID,
 				Type:         file.Type,
-				Name:         &file.Name,
+				Name:         &docName,
 				Suffix:       suffix,
 				Size:         file.Size,
 			}