add line

Author: xixingde

internal/logic/embedding_task.go

@@ -719,28 +719,11 @@ func (l *TaskLogic) deleteFilesFromVectorDB(ctx context.Context, codebase *model
 
 	l.Logger.Infof("准备从向量数据库中删除 %d 个文件，代码库ID: %d", len(filePaths), codebase.ID)
 
-	// 构建需要删除的 CodeChunk 列表
-	var chunks []*types.CodeChunk
 	for _, filePath := range filePaths {
 		// 将文件路径转换为Linux格式（正斜杠）
 		linuxPath := strings.ReplaceAll(filePath, "\\", "/")
-		chunk := &types.CodeChunk{
-			CodebaseId:   codebase.ID,
-			CodebasePath: codebase.Path,
-			FilePath:     linuxPath,
-		}
-		chunks = append(chunks, chunk)
 		l.Logger.Debugf("添加文件到删除列表: %s", linuxPath)
-	}
-
-	// 调用向量数据库的删除方法
-	options := vector.Options{
-		CodebasePath: codebase.Path,
-	}
-	err := l.svcCtx.VectorStore.DeleteCodeChunks(ctx, chunks, options)
-	if err != nil {
-		l.Logger.Errorf("删除向量数据库中的文件失败: %v", err)
-		return fmt.Errorf("failed to delete files from vector database: %w", err)
+		l.svcCtx.VectorStore.DeleteDictionary(ctx, filePath, vector.Options{CodebaseId: codebase.ID, CodebasePath: codebase.Path})
 	}
 
 	l.Logger.Infof("成功从向量数据库中删除了 %d 个文件", len(filePaths))

internal/logic/index.go

@@ -4,9 +4,9 @@ import (
 	"context"
 	"errors"
 	"fmt"
+
 	"github.com/zgsm-ai/codebase-indexer/internal/store/vector"
 	"github.com/zgsm-ai/codebase-indexer/internal/tracer"
-	"strings"
 
 	"github.com/zgsm-ai/codebase-indexer/internal/errs"
 	"gorm.io/gorm"
@@ -55,17 +55,7 @@ func (l *IndexLogic) DeleteIndex(req *types.DeleteIndexRequest) (resp *types.Del
 		return &types.DeleteIndexResponseData{}, nil
 	}
 
-	// 如果指定了filePaths，则只删除指定文件的嵌入数据
-	var deleteChunks []*types.CodeChunk
-	for _, path := range strings.Split(filePaths, ",") {
-		deleteChunks = append(deleteChunks, &types.CodeChunk{
-			CodebaseId:   codebase.ID,
-			CodebasePath: codebase.Path,
-			FilePath:     strings.TrimSpace(path),
-		})
-	}
-
-	if err = l.svcCtx.VectorStore.DeleteCodeChunks(ctx, deleteChunks, vector.Options{CodebaseId: codebase.ID,
+	if err = l.svcCtx.VectorStore.DeleteDictionary(ctx, filePaths, vector.Options{CodebaseId: codebase.ID,
 		CodebasePath: codebase.Path}); err != nil {
 		return nil, fmt.Errorf("failed to delete embedding index, err:%w", err)
 	}

internal/logic/semantic.go

@@ -2,14 +2,11 @@ package logic
 
 import (
 	"context"
-	"errors"
-	"fmt"
 
 	"github.com/zgsm-ai/codebase-indexer/internal/errs"
 	"github.com/zgsm-ai/codebase-indexer/internal/store/vector"
 	"github.com/zgsm-ai/codebase-indexer/internal/tracer"
 	"github.com/zgsm-ai/codebase-indexer/pkg/utils"
-	"gorm.io/gorm"
 
 	"github.com/zgsm-ai/codebase-indexer/internal/svc"
 	"github.com/zgsm-ai/codebase-indexer/internal/types"
@@ -51,21 +48,15 @@ func (l *SemanticLogic) SemanticSearch(req *types.SemanticSearchRequest, authori
 	if err != nil {
 		return nil, err
 	}
-	clientId := req.ClientId
-	clientPath := req.CodebasePath
 
-	codebase, err := l.svcCtx.Querier.Codebase.FindByClientIdAndPath(l.ctx, clientId, clientPath)
-	if errors.Is(err, gorm.ErrRecordNotFound) {
-		return nil, errs.NewRecordNotFoundErr(types.NameCodeBase, fmt.Sprintf("client_id: %s, clientPath: %s", clientId, clientPath))
-	}
-	ctx := context.WithValue(l.ctx, tracer.Key, tracer.RequestTraceId(int(codebase.ID)))
+	ctx := context.WithValue(l.ctx, tracer.Key, req.ClientId)
 
 	documents, err := l.svcCtx.VectorStore.Query(ctx, req.Query, topK,
 		vector.Options{
-			CodebaseId:    codebase.ID,
-			ClientId:      clientId,
-			CodebasePath:  codebase.Path,
-			CodebaseName:  codebase.Name,
+			CodebaseId:    0,
+			ClientId:      req.ClientId,
+			CodebasePath:  req.CodebasePath,
+			CodebaseName:  "",
 			Authorization: authorization,
 		})
 	if err != nil {

internal/store/vector/vector_store.go

@@ -20,6 +20,7 @@ type Store interface {
 	InsertCodeChunks(ctx context.Context, docs []*types.CodeChunk, options Options) error
 	UpsertCodeChunks(ctx context.Context, chunks []*types.CodeChunk, options Options) error
 	DeleteCodeChunks(ctx context.Context, chunks []*types.CodeChunk, options Options) error
+	DeleteDictionary(ctx context.Context, dictionary string, 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)

internal/store/vector/weaviate_wrapper.go

@@ -335,18 +335,19 @@ func (r *weaviateWrapper) unmarshalSimilarSearchResponse(res *models.GraphQLResp
 		content := ""
 		filePath := getStringValue(obj, MetadataFilePath)
 
+		// 从MetadataRange中提取startLine和endLine（用于构建映射键）
+		var startLine, endLine int
+		if rangeValue, ok := obj[MetadataRange].([]interface{}); ok && len(rangeValue) >= 2 {
+			if first, ok := rangeValue[0].(float64); ok {
+				startLine = int(first)
+			}
+			if second, ok := rangeValue[2].(float64); ok {
+				endLine = int(second)
+			}
+		}
+
 		// 如果开启获取源码且有批量获取的内容，则使用获取到的内容
 		if r.cfg.FetchSourceCode && filePath != "" && codebasePath != "" {
-			// 从MetadataRange中提取startLine和endLine（用于构建映射键）
-			var startLine, endLine int
-			if rangeValue, ok := obj[MetadataRange].([]interface{}); ok && len(rangeValue) >= 2 {
-				if first, ok := rangeValue[0].(float64); ok {
-					startLine = int(first)
-				}
-				if second, ok := rangeValue[2].(float64); ok {
-					endLine = int(second)
-				}
-			}
 
 			// 构建映射键并查找批量获取的内容
 			fullPath := filepath.Join(codebasePath, filePath)
@@ -358,9 +359,11 @@ func (r *weaviateWrapper) unmarshalSimilarSearchResponse(res *models.GraphQLResp
 
 		// Create SemanticFileItem with proper fields
 		item := &types.SemanticFileItem{
-			Content:  content,
-			FilePath: filePath,
-			Score:    float32(getFloatValue(additional, "certainty")), // Convert float64 to float32
+			Content:   content,
+			FilePath:  filePath,
+			StartLine: startLine,
+			EndLine:   endLine,
+			Score:     float32(getFloatValue(additional, "certainty")), // Convert float64 to float32
 		}
 
 		items = append(items, item)
@@ -1423,6 +1426,79 @@ func (r *weaviateWrapper) getRecordsByPathPrefix(ctx context.Context, pathPrefix
 	return r.unmarshalRecordsResponse(res)
 }
 
+// DeleteDictionary 删除指定目录的记录，通过匹配filePath的前缀
+func (r *weaviateWrapper) DeleteDictionary(ctx context.Context, dictionary string, options Options) error {
+	// 生成租户名称
+	tenantName, err := r.generateTenantName(options.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 += "/"
+	}
+
+	// 获取所有匹配目录前缀的记录
+	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 {
+		// 没有找到需要删除的记录
+		return nil
+	}
+
+	// 将记录转换为CodeChunk格式以便删除
+	chunks := make([]*types.CodeChunk, 0, len(records))
+	for _, record := range records {
+		chunk := &types.CodeChunk{
+			CodebaseId: record.CodebaseId,
+			FilePath:   record.FilePath,
+		}
+		chunks = append(chunks, chunk)
+	}
+
+	if len(chunks) > 0 {
+		chunkFilters := make([]*filters.WhereBuilder, len(chunks))
+		for i, chunk := range chunks {
+			if chunk.FilePath == types.EmptyString {
+				return fmt.Errorf("invalid chunk to delete: and filePath")
+			}
+			chunkFilters[i] = filters.Where().
+				WithOperator(filters.And).
+				WithOperands([]*filters.WhereBuilder{
+					filters.Where().
+						WithPath([]string{MetadataCodebaseId}).
+						WithOperator(filters.Equal).
+						WithValueInt(int64(chunk.CodebaseId)),
+					filters.Where().
+						WithPath([]string{MetadataFilePath}).
+						WithOperator(filters.Equal).
+						WithValueText(chunk.FilePath),
+				})
+		}
+
+		// Combine all chunk filters with OR to support batch deletion of files
+		combinedFilter := filters.Where().
+			WithOperator(filters.Or).
+			WithOperands(chunkFilters)
+
+		do, err := r.client.Batch().ObjectsBatchDeleter().
+			WithTenant(tenantName).WithWhere(
+			combinedFilter,
+		).WithClassName(r.className).Do(ctx)
+		if err != nil {
+			return fmt.Errorf("failed to send delete chunks err:%w", err)
+		}
+		return CheckBatchDeleteErrors(do)
+	}
+
+	// 批量删除记录
+	return nil
+}
+
 // 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