update dictionary

Author: xixingde

internal/logic/update_embedding.go

@@ -2,17 +2,11 @@ package logic
 
 import (
 	"context"
-	"errors"
 	"fmt"
-	"strings"
 
 	"github.com/zeromicro/go-zero/core/logx"
-	"github.com/zgsm-ai/codebase-indexer/internal/dao/model"
-	"github.com/zgsm-ai/codebase-indexer/internal/errs"
-	"github.com/zgsm-ai/codebase-indexer/internal/store/vector"
 	"github.com/zgsm-ai/codebase-indexer/internal/svc"
 	"github.com/zgsm-ai/codebase-indexer/internal/types"
-	"gorm.io/gorm"
 )
 
 type UpdateEmbeddingLogic struct {
@@ -30,125 +24,30 @@ func NewUpdateEmbeddingLogic(ctx context.Context, svcCtx *svc.ServiceContext) *U
 }
 
 func (l *UpdateEmbeddingLogic) UpdateEmbeddingPath(req *types.UpdateEmbeddingPathRequest) (resp *types.UpdateEmbeddingPathResponseData, err error) {
-	clientId := req.ClientId
 	codebasePath := req.CodebasePath
 	oldPath := req.OldPath
 	newPath := req.NewPath
 
-	// 查找代码库记录
-	codebase, err := l.svcCtx.Querier.Codebase.FindByClientIdAndPath(l.ctx, clientId, codebasePath)
-	if errors.Is(err, gorm.ErrRecordNotFound) {
-		return nil, errs.NewRecordNotFoundErr(types.NameCodeBase, fmt.Sprintf("client_id: %s, codebasePath: %s", clientId, codebasePath))
-	}
-	if err != nil {
-		return nil, err
-	}
-
-	// 检查是否是目录
-	// fullOldPath := filepath.Join(codebasePath, oldPath)
-	// info, err := os.Stat(fullOldPath)
-	// if err != nil {
-	// 	return nil, fmt.Errorf("failed to stat path %s: %w", fullOldPath, err)
-	// }
-
 	var modifiedFiles []string
 
-	// if info.IsDir() {
-	// 	// 处理目录情况
-	// 	modifiedFiles, err = l.updateDirectoryPaths(codebase, oldPath, newPath)
-	// 	if err != nil {
-	// 		return nil, fmt.Errorf("failed to update directory paths: %w", err)
-	// 	}
-	// } else {
-	// 	// 处理文件情况
-	modifiedFiles, err = l.updateFilePath(codebase, oldPath, newPath)
+	// 处理目录情况，使用 UpdateCodeChunksDictionary 接口
+	err = l.svcCtx.VectorStore.UpdateCodeChunksDictionary(l.ctx, codebasePath, oldPath, newPath)
 	if err != nil {
-		return nil, fmt.Errorf("failed to update file path: %w", err)
-	}
-	// }
-
-	return &types.UpdateEmbeddingPathResponseData{
-		ModifiedFiles: modifiedFiles,
-		TotalFiles:    len(modifiedFiles),
-	}, nil
-}
-
-func (l *UpdateEmbeddingLogic) updateDirectoryPaths(codebase *model.Codebase, oldDirPath, newDirPath string) ([]string, error) {
-	// 获取该目录下所有的文件路径
-	records, err := l.svcCtx.VectorStore.GetCodebaseRecords(l.ctx, codebase.ID, codebase.Path)
-	if err != nil {
-		return nil, fmt.Errorf("failed to get codebase records: %w", err)
-	}
-
-	var modifiedFiles []string
-	var pathUpdates []*types.CodeChunkPathUpdate
-
-	for _, record := range records {
-		// 检查文件路径是否以旧目录路径开头
-		if strings.HasPrefix(record.FilePath, oldDirPath) {
-			// 构建新的文件路径
-			newFilePath := strings.Replace(record.FilePath, oldDirPath, newDirPath, 1)
-
-			// 创建路径更新请求
-			pathUpdate := &types.CodeChunkPathUpdate{
-				CodebaseId:  codebase.ID,
-				OldFilePath: record.FilePath,
-				NewFilePath: newFilePath,
-			}
-
-			pathUpdates = append(pathUpdates, pathUpdate)
-			modifiedFiles = append(modifiedFiles, newFilePath)
-		}
+		return nil, fmt.Errorf("failed to update directory paths: %w", err)
 	}
 
-	if len(pathUpdates) > 0 {
-		// 使用新的直接更新路径的方法，而不是删除再插入
-		err = l.svcCtx.VectorStore.UpdateCodeChunksPaths(l.ctx, pathUpdates, vector.Options{
-			CodebaseId:   codebase.ID,
-			CodebasePath: codebase.Path,
-		})
-		if err != nil {
-			return nil, fmt.Errorf("failed to update chunk paths: %w", err)
-		}
-	}
-
-	return modifiedFiles, nil
-}
-
-func (l *UpdateEmbeddingLogic) updateFilePath(codebase *model.Codebase, oldFilePath, newFilePath string) ([]string, error) {
-	// 获取该文件的记录
-	records, err := l.svcCtx.VectorStore.GetCodebaseRecords(l.ctx, codebase.ID, codebase.Path)
+	// 获取更新后的记录以返回修改的文件列表
+	records, err := l.svcCtx.VectorStore.GetDictionaryRecords(l.ctx, codebasePath, newPath)
 	if err != nil {
-		return nil, fmt.Errorf("failed to get codebase records: %w", err)
+		return nil, fmt.Errorf("failed to get updated dictionary records: %w", err)
 	}
 
-	var modifiedFiles []string
-
-	// 检查是否有需要更新的记录
 	for _, record := range records {
-		if record.FilePath == oldFilePath {
-			modifiedFiles = append(modifiedFiles, newFilePath)
-		}
-	}
-
-	if len(modifiedFiles) > 0 {
-		// 使用直接更新路径的方法
-		pathUpdates := []*types.CodeChunkPathUpdate{
-			{
-				CodebaseId:  codebase.ID,
-				OldFilePath: oldFilePath,
-				NewFilePath: newFilePath,
-			},
-		}
-
-		err = l.svcCtx.VectorStore.UpdateCodeChunksPaths(l.ctx, pathUpdates, vector.Options{
-			CodebaseId:   codebase.ID,
-			CodebasePath: codebase.Path,
-		})
-		if err != nil {
-			return nil, fmt.Errorf("failed to update chunk paths: %w", err)
-		}
+		modifiedFiles = append(modifiedFiles, record.FilePath)
 	}
 
-	return modifiedFiles, nil
+	return &types.UpdateEmbeddingPathResponseData{
+		ModifiedFiles: modifiedFiles,
+		TotalFiles:    len(modifiedFiles),
+	}, nil
 }

internal/store/vector/vector_store.go

@@ -21,6 +21,7 @@ type Store interface {
 	UpsertCodeChunks(ctx context.Context, chunks []*types.CodeChunk, options Options) error
 	DeleteCodeChunks(ctx context.Context, chunks []*types.CodeChunk, options Options) error
 	UpdateCodeChunksPaths(ctx context.Context, updates []*types.CodeChunkPathUpdate, options Options) error
+	UpdateCodeChunksDictionary(ctx context.Context, codebasePath string, dictionary string, newDictionary string) error
 	Query(ctx context.Context, query string, topK int, options Options) ([]*types.SemanticFileItem, error)
 	Close()
 }

internal/store/vector/weaviate_wrapper.go

@@ -1422,3 +1422,53 @@ func (r *weaviateWrapper) getRecordsByPathPrefix(ctx context.Context, pathPrefix
 	// 解析响应获取记录
 	return r.unmarshalRecordsResponse(res)
 }
+
+// UpdateCodeChunksDictionary 更新代码块的目录路径，通过匹配filePath的前缀
+func (r *weaviateWrapper) UpdateCodeChunksDictionary(ctx context.Context, codebasePath string, dictionary string, newDictionary string) error {
+	// 生成租户名称
+	tenantName, err := r.generateTenantName(codebasePath)
+	if err != nil {
+		return fmt.Errorf("failed to generate tenant name: %w", err)
+	}
+
+	// 确保路径前缀以/结尾，以便正确匹配子目录和文件
+	if dictionary != "" && !strings.HasSuffix(dictionary, "/") {
+		dictionary += "/"
+	}
+	if newDictionary != "" && !strings.HasSuffix(newDictionary, "/") {
+		newDictionary += "/"
+	}
+
+	// 获取所有匹配原目录前缀的记录
+	records, err := r.getRecordsByPathPrefix(ctx, dictionary, tenantName)
+	if err != nil {
+		return fmt.Errorf("failed to get records by path prefix: %w", err)
+	}
+
+	if len(records) == 0 {
+		// 没有找到需要更新的记录
+		return nil
+	}
+
+	// 批量更新记录路径
+	for _, record := range records {
+		// 构建新的文件路径：将原目录前缀替换为新目录前缀
+		newFilePath := strings.Replace(record.FilePath, dictionary, newDictionary, 1)
+
+		// 获取对象的ID
+		objectIds, err := r.getObjectIdsByPath(ctx, record.CodebaseId, record.FilePath, tenantName)
+		if err != nil {
+			return fmt.Errorf("failed to get object ids by path: %w", err)
+		}
+
+		// 更新每个对象的路径
+		for _, objectId := range objectIds {
+			err = r.updateObjectPath(ctx, objectId, newFilePath, tenantName, record)
+			if err != nil {
+				return fmt.Errorf("failed to update object path: %w", err)
+			}
+		}
+	}
+
+	return nil
+}