add syn

Author: xixingde

internal/handler/completed_tasks.go

@@ -0,0 +1,42 @@
+package handler
+
+import (
+	"fmt"
+	"net/http"
+
+	"github.com/zgsm-ai/codebase-indexer/internal/logic"
+	"github.com/zgsm-ai/codebase-indexer/internal/response"
+	"github.com/zgsm-ai/codebase-indexer/internal/svc"
+)
+
+// completedTasksHandler 已完成任务处理器
+type completedTasksHandler struct {
+	svcCtx *svc.ServiceContext
+}
+
+// CompletedTasksHandler 创建已完成任务处理器
+func CompletedTasksHandler(svcCtx *svc.ServiceContext) http.HandlerFunc {
+	return func(w http.ResponseWriter, r *http.Request) {
+		handler := &completedTasksHandler{svcCtx: svcCtx}
+		handler.ServeHTTP(w, r)
+	}
+}
+
+// ServeHTTP 处理HTTP请求
+func (h *completedTasksHandler) ServeHTTP(w http.ResponseWriter, r *http.Request) {
+	// 验证请求方法
+	if r.Method != http.MethodGet {
+		response.Error(w, fmt.Errorf("method not allowed"))
+		return
+	}
+
+	// 查询已完成任务
+	completedTasksLogic := logic.NewCompletedTasksLogic(r.Context(), h.svcCtx)
+	resp, err := completedTasksLogic.GetCompletedTasks()
+	if err != nil {
+		response.Error(w, err)
+		return
+	}
+
+	response.Json(w, resp)
+}

internal/handler/routes.go

@@ -0,0 +1,61 @@
+package logic
+
+import (
+	"context"
+	"fmt"
+	"time"
+
+	"github.com/zgsm-ai/codebase-indexer/internal/svc"
+	"github.com/zgsm-ai/codebase-indexer/internal/types"
+)
+
+// CompletedTasksLogic 已完成任务查询逻辑
+type CompletedTasksLogic struct {
+	ctx    context.Context
+	svcCtx *svc.ServiceContext
+}
+
+// NewCompletedTasksLogic 创建已完成任务查询逻辑
+func NewCompletedTasksLogic(ctx context.Context, svcCtx *svc.ServiceContext) *CompletedTasksLogic {
+	return &CompletedTasksLogic{
+		ctx:    ctx,
+		svcCtx: svcCtx,
+	}
+}
+
+// GetCompletedTasks 获取已完成任务
+func (l *CompletedTasksLogic) GetCompletedTasks() (*types.CompletedTasksResponse, error) {
+	// 检查Redis连接
+	if err := l.checkRedisConnection(); err != nil {
+		return nil, fmt.Errorf("Redis服务不可用，请稍后再试")
+	}
+
+	// 扫描已完成任务
+	tasks, err := l.scanCompletedTasks()
+	if err != nil {
+		return nil, fmt.Errorf("查询已完成任务时发生内部错误: %w", err)
+	}
+
+	return &types.CompletedTasksResponse{
+		Code:    0,
+		Message: "ok",
+		Success: true,
+		Data: &types.CompletedTasksData{
+			TotalTasks: len(tasks),
+			Tasks:      tasks,
+		},
+	}, nil
+}
+
+// checkRedisConnection 检查Redis连接
+func (l *CompletedTasksLogic) checkRedisConnection() error {
+	ctx, cancel := context.WithTimeout(l.ctx, 5*time.Second)
+	defer cancel()
+
+	return l.svcCtx.StatusManager.CheckConnection(ctx)
+}
+
+// scanCompletedTasks 扫描已完成的任务
+func (l *CompletedTasksLogic) scanCompletedTasks() ([]types.CompletedTaskInfo, error) {
+	return l.svcCtx.StatusManager.ScanCompletedTasks(l.ctx)
+}

internal/logic/completed_tasks.go

@@ -431,7 +431,7 @@ func (l *TaskLogic) extractFileListFromShenmaSync(content []byte, fileName strin
 		CodebasePath  string                 `json:"codebasePath"`
 		CodebaseName  string                 `json:"codebaseName"`
 		ExtraMetadata map[string]interface{} `json:"extraMetadata"`
-		FileList      map[string]string      `json:"fileList"`
+		FileList      interface{}            `json:"fileList"` // 使用interface{}来兼容两种格式
 		Timestamp     int64                  `json:"timestamp"`
 	}
 
@@ -473,9 +473,49 @@ func (l *TaskLogic) extractFileListFromShenmaSync(content []byte, fileName strin
 	metadata.ClientId = tempMetadata.ClientId
 	metadata.CodebasePath = tempMetadata.CodebasePath
 	metadata.CodebaseName = tempMetadata.CodebaseName
-	metadata.FileList = tempMetadata.FileList
 	metadata.Timestamp = tempMetadata.Timestamp
 
+	// 处理FileList的两种格式
+	switch fileList := tempMetadata.FileList.(type) {
+	case map[string]interface{}:
+		// 格式1: "fileList":{"pkg/codegraph/analyzer/package_classifier/cpp_classifier.go":"add"}
+		l.Logger.Infof("检测到对象格式的fileList")
+		for filePath, operation := range fileList {
+			if opStr, ok := operation.(string); ok {
+				metadata.FileList[filePath] = opStr
+			} else {
+				l.Logger.Errorf("fileList中文件 %s 的操作类型不是字符串: %v", filePath, operation)
+			}
+		}
+	case []interface{}:
+		// 格式2: "fileList":[{"path":"pkg/codegraph/proto/codegraphpb/types.pb.go","targetPath":"","hash":"1755050845505","status":"modify","requestId":""}]
+		l.Logger.Infof("检测到数组格式的fileList")
+		for _, item := range fileList {
+			if itemMap, ok := item.(map[string]interface{}); ok {
+				if path, ok := itemMap["path"].(string); ok {
+					// 优先使用status字段，如果没有则使用其他字段作为操作类型
+					var operation string
+					if status, ok := itemMap["status"].(string); ok {
+						operation = status
+					} else if operate, ok := itemMap["operate"].(string); ok {
+						operation = operate
+					} else {
+						operation = "unknown"
+						l.Logger.Errorf("fileList数组项中未找到status或operate字段: %v", itemMap)
+					}
+					metadata.FileList[path] = operation
+				} else {
+					l.Logger.Errorf("fileList数组项中缺少path字段: %v", itemMap)
+				}
+			} else {
+				l.Logger.Errorf("fileList数组项不是map类型: %v", item)
+			}
+		}
+	default:
+		l.Logger.Errorf("不支持的fileList格式: %T", tempMetadata.FileList)
+		return
+	}
+
 	l.Logger.Infof("从 %s 中提取到 %d 个文件:", fileName, len(metadata.FileList))
 
 	// 打印fileList中的文件

internal/logic/embedding_task.go

@@ -284,3 +284,102 @@ func (sm *StatusManager) ResetPendingAndProcessingTasksToFailed(ctx context.Cont
 	logx.Infof("Successfully reset %d pending/processing tasks to failed", updatedCount)
 	return nil
 }
+
+// ScanCompletedTasks 扫描已完成的任务
+func (sm *StatusManager) ScanCompletedTasks(ctx context.Context) ([]types.CompletedTaskInfo, error) {
+	var completedTasks []types.CompletedTaskInfo
+
+	// 使用SCAN命令避免阻塞
+	iter := sm.client.Scan(ctx, 0, "request:id:*", 0).Iterator()
+	for iter.Next(ctx) {
+		key := iter.Val()
+
+		// 获取任务状态数据
+		data, err := sm.client.Get(ctx, key).Result()
+		if err != nil {
+			if err == redis.Nil {
+				continue // 键可能已过期
+			}
+			return nil, fmt.Errorf("failed to get task data for key %s: %w", key, err)
+		}
+
+		// 解析任务状态
+		var status types.FileStatusResponseData
+		if err := json.Unmarshal([]byte(data), &status); err != nil {
+			continue // 跳过格式错误的数据
+		}
+
+		// 过滤已完成的任务状态
+		if sm.isCompletedStatus(status.Process) {
+			taskInfo, err := sm.parseCompletedTaskInfo(key, status)
+			if err != nil {
+				continue
+			}
+			completedTasks = append(completedTasks, taskInfo)
+		}
+	}
+
+	if err := iter.Err(); err != nil {
+		return nil, fmt.Errorf("redis scan error: %w", err)
+	}
+
+	// 按完成时间排序（最新的在前）
+	sort.Slice(completedTasks, func(i, j int) bool {
+		return completedTasks[i].CompletedTime.After(completedTasks[j].CompletedTime)
+	})
+
+	return completedTasks, nil
+}
+
+// isCompletedStatus 检查是否为已完成的状态
+func (sm *StatusManager) isCompletedStatus(status string) bool {
+	return status == "complete" || status == "success" || status == "failed" || status == "cancelled" || status == "timeout"
+}
+
+// parseCompletedTaskInfo 解析已完成任务信息
+func (sm *StatusManager) parseCompletedTaskInfo(key string, status types.FileStatusResponseData) (types.CompletedTaskInfo, error) {
+	// 从key中提取任务ID
+	taskId := strings.TrimPrefix(key, "request:id:")
+
+	// 解析任务状态数据中的时间信息
+	var completedTime time.Time
+
+	// 设置当前时间为完成时间
+	completedTime = time.Now()
+
+	// 尝试从文件列表中提取客户端ID
+	var clientId string
+	var fileCount int
+	var successCount, failedCount int
+
+	if len(status.FileList) > 0 {
+		// 统计文件状态
+		for _, file := range status.FileList {
+			fileCount++
+			if file.Status == "complete" || file.Status == "success" {
+				successCount++
+			} else if file.Status == "failed" {
+				failedCount++
+			}
+		}
+	}
+
+	// 计算成功率
+	var successRate float64
+	if fileCount > 0 {
+		successRate = float64(successCount) / float64(fileCount) * 100
+	}
+
+	return types.CompletedTaskInfo{
+		TaskId:        taskId,
+		ClientId:      clientId,
+		Status:        status.Process,
+		Process:       status.Process,
+		TotalProgress: status.TotalProgress,
+		CompletedTime: completedTime,
+		FileCount:     fileCount,
+		SuccessCount:  successCount,
+		FailedCount:   failedCount,
+		SuccessRate:   successRate,
+	}, nil
+}

internal/store/redis/status_manager.go

@@ -0,0 +1,31 @@
+package types
+
+import "time"
+
+// CompletedTaskInfo 已完成任务信息
+type CompletedTaskInfo struct {
+	TaskId        string    `json:"taskId"`        // 任务ID
+	ClientId      string    `json:"clientId"`      // 客户端ID
+	Status        string    `json:"status"`        // 任务状态
+	Process       string    `json:"process"`       // 处理进程
+	TotalProgress int       `json:"totalProgress"` // 总进度
+	CompletedTime time.Time `json:"completedTime"` // 完成时间
+	FileCount     int       `json:"fileCount"`     // 文件总数
+	SuccessCount  int       `json:"successCount"`  // 成功文件数
+	FailedCount   int       `json:"failedCount"`   // 失败文件数
+	SuccessRate   float64   `json:"successRate"`   // 成功率
+}
+
+// CompletedTasksResponse 已完成任务查询响应
+type CompletedTasksResponse struct {
+	Code    int                 `json:"code"`    // 响应码
+	Message string              `json:"message"` // 响应消息
+	Success bool                `json:"success"` // 是否成功
+	Data    *CompletedTasksData `json:"data"`    // 任务数据
+}
+
+// CompletedTasksData 已完成任务数据
+type CompletedTasksData struct {
+	TotalTasks int                 `json:"totalTasks"` // 任务总数
+	Tasks      []CompletedTaskInfo `json:"tasks"`      // 任务列表
+}