fix: reindex ignore empty rag_id (#248)

Author: dhsifss

backend/repo/kb_document.go

@@ -161,7 +161,7 @@ func (d *KBDocument) UpsertSpaceFolderTree(ctx context.Context, folder *model.KB
 	})
 }
 
-func (d *KBDocument) UpdateSpaceFolderAllDoc(ctx context.Context, folderID uint, statusFilter []model.DocStatus, docStatus model.DocStatus, msg string) error {
+func (d *KBDocument) UpdateSpaceFolderAllDoc(ctx context.Context, folderID uint, statusFilter []model.DocStatus, docStatus model.DocStatus, msg string, ignoreEmptyRag bool) error {
 	sql := `WITH RECURSIVE all_folder_doc AS (
 	SELECT id, parent_id FROM kb_documents WHERE parent_id = ?
 	UNION
@@ -172,6 +172,10 @@ UPDATE kb_documents SET status = ?, message = ?, updated_at = ? FROM all_folder_
 		folderID, docStatus, msg, time.Now(), model.DocTypeSpace, model.FileTypeFolder,
 	}
 
+	if ignoreEmptyRag {
+		sql += " AND kb_documents.rag_id != ''"
+	}
+
 	if len(statusFilter) > 0 {
 		sql += " AND kb_documents.status IN (?)"
 		params = append(params, statusFilter)

backend/sub/kb_space.go

@@ -351,7 +351,7 @@ func (k *kbSpace) handleUpdate(ctx context.Context, logger *glog.Logger, msg top
 		if err != nil {
 			logger.WithErr(err).Warn("list doc failed")
 
-			e := k.repoDoc.UpdateSpaceFolderAllDoc(ctx, exportFolderID, []model.DocStatus{model.DocStatusPendingExec}, model.DocStatusExportFailed, err.Error())
+			e := k.repoDoc.UpdateSpaceFolderAllDoc(ctx, exportFolderID, []model.DocStatus{model.DocStatusPendingExec}, model.DocStatusExportFailed, err.Error(), false)
 			if e != nil {
 				logger.WithErr(e).Warn("set doc export failed error")
 			}

backend/svc/kb_docment.go

@@ -1091,12 +1091,16 @@ func (d *KBDocument) ReindexSpaceFolder(ctx context.Context, folderID uint) erro
 		return err
 	}
 
-	err = d.repoDoc.UpdateSpaceFolderAllDoc(ctx, folderID, nil, model.DocStatusPendingApply, "")
+	err = d.repoDoc.UpdateSpaceFolderAllDoc(ctx, folderID, nil, model.DocStatusPendingApply, "", true)
 	if err != nil {
 		return err
 	}
 
 	for _, doc := range docs {
+		if doc.RagID == "" {
+			continue
+		}
+
 		err = d.rag.ReindexDocument(ctx, d.repoDataset.GetBackendID(ctx), doc.RagID)
 		if err != nil {
 			e := d.repoDoc.Update(ctx, map[string]any{
@@ -1408,7 +1412,8 @@ func (d *KBDocument) DocReindex(ctx context.Context, req DocReindexReq) error {
 	err := d.repoDoc.Update(ctx, map[string]any{
 		"status": model.DocStatusPendingApply,
 	}, repo.QueryWithEqual("id", req.IDs, repo.EqualOPEqAny),
-		repo.QueryWithEqual("file_type", model.FileTypeFolder, repo.EqualOPNE))
+		repo.QueryWithEqual("file_type", model.FileTypeFolder, repo.EqualOPNE),
+		repo.QueryWithEqual("rag_id", "", repo.EqualOPNE))
 	if err != nil {
 		return err
 	}
@@ -1432,6 +1437,7 @@ func (d *KBDocument) DocReindex(ctx context.Context, req DocReindexReq) error {
 	},
 		repo.QueryWithEqual("id", req.IDs, repo.EqualOPEqAny),
 		repo.QueryWithEqual("file_type", model.FileTypeFolder, repo.EqualOPNE),
+		repo.QueryWithEqual("rag_id", "", repo.EqualOPNE),
 		repo.QueryWithSelectColumn("id, rag_id"),
 	)
 }