Version: 0.9.2.dev.260406
后端:
1.Chat 四路由升级(二分类 chat/task → 四路由 direct_reply/execute/deep_answer/plan)
- 新建model/chat_contract.go:路由决策模型,含 NeedsRoughBuild 粗排标记
- 更新node/chat.go:四路由分流;新增 deep_answer 深度回答路径(二次 LLM 开 thinking)
- 更新prompt/chat.go:意图分类 prompt 升级为四路由 prompt;新增 deep_answer prompt
2.粗排节点(RoughBuild)全链路
- 新建node/rough_build.go:粗排节点,调用注入的算法函数,结果写入 ScheduleState 后进 Execute 微调
- 更新graph/common_graph.go:注册 RoughBuild 节点;Chat/Confirm 后可路由至粗排
- 更新model/graph_run_state.go:新增 RoughBuildPlacement/RoughBuildFunc 类型;Deps 注入入口
- 更新model/plan_contract.go:PlanDecision 新增 NeedsRoughBuild/TaskClassIDs 字段
- 更新node/plan.go:plan_done 时写入粗排标记和 TaskClassIDs
3.任务类约束元数据(TaskClassMeta)贯穿 prompt → tools → 持久化
- 更新tools/state.go:新增 TaskClassMeta;ScheduleState.TaskClasses;ScheduleTask.TaskClassID;Clone 深拷贝
- 更新conv/schedule_state.go:加载时构建 TaskClassMeta;Diff 支持 HostEventID 嵌入关系
- 更新conv/schedule_provider.go:新增 LoadTaskClassMetas 按需加载
- 更新model/state_store.go:ScheduleStateProvider 接口新增 LoadTaskClassMetas
- 更新prompt/base.go:renderStateSummary 渲染任务类约束
- 更新prompt/plan.go:注入任务类 ID 上下文和粗排识别规则
- 更新tools/read_tools.go:GetOverview 展示任务类约束
- 更新model/common_state.go:CommonState 新增 TaskClassIDs/TaskClasses/NeedsRoughBuild
4.Execute 健壮性增强(correction 重试 + 纯 ReAct 模式)
- 更新node/execute.go:未知工具名/空文本走 correction 重试而非 fatal;maxConsecutiveCorrections 提升为包级常量;新增无 plan 纯ReAct 模式;工具结果截断;speak 排除 ask_user/confirm
- 更新prompt/execute.go:新增 ReAct 模式 system prompt 和 contract
5.写入持久化完善(task_item source + 嵌入水课)
- 更新conv/schedule_persist.go:place/move/unplace 支持 task_item source,含嵌入水课和普通 task event 两条路径
- 新建conv/schedule_preview.go:ScheduleState → 排程预览缓存,复用旧格式,前端无需改动
6.状态持久化体系(Redis → MySQL outbox 异步)
- 更新dao/cache.go:Redis 快照 TTL 从 24h 改为 2h,配合 MySQL outbox
- 新建model/agent_state_snapshot_record.go:快照 MySQL 记录模型
- 新建service/events/agent_state_persist.go:outbox 异步持久化处理器
- 更新cmd/start.go + inits/mysql.go:注册快照事件处理器 + AutoMigrate
- 更新service/agentsvc/agent_newagent.go:注入 RoughBuildFunc;outbox 异步写快照;排程结果写 Redis 预览缓存
7.基础设施与稳定性
- 更新stream/sse_adapter.go:outChan 满时静默丢弃,保证持久化不被 SSE 阻断
- 更新service/agentsvc/agent.go:新增 readAgentExtraIntSlice;outChan 容量 8→256
- 更新node/agent_nodes.go:Chat 注入工具 schema;Deliver 改 saveAgentState 替代 deleteAgentState
前端:无
仓库:无
This commit is contained in:
@@ -3,6 +3,7 @@ package newagentnode
|
||||
import (
|
||||
"context"
|
||||
"fmt"
|
||||
"log"
|
||||
"strings"
|
||||
"time"
|
||||
|
||||
@@ -36,89 +37,222 @@ type ChatNodeInput struct {
|
||||
ChunkEmitter *newagentstream.ChunkEmitter
|
||||
}
|
||||
|
||||
// chatIntentDecision 是意图分类的结构化输出。
|
||||
type chatIntentDecision struct {
|
||||
Intent string `json:"intent"`
|
||||
Reply string `json:"reply,omitempty"`
|
||||
Reason string `json:"reason,omitempty"`
|
||||
}
|
||||
|
||||
// Normalize 清洗意图分类结果中的字符串字段。
|
||||
func (d *chatIntentDecision) Normalize() {
|
||||
if d == nil {
|
||||
return
|
||||
}
|
||||
d.Intent = strings.TrimSpace(d.Intent)
|
||||
d.Reply = strings.TrimSpace(d.Reply)
|
||||
d.Reason = strings.TrimSpace(d.Reason)
|
||||
}
|
||||
|
||||
// Validate 校验意图分类结果的最小合法性。
|
||||
func (d *chatIntentDecision) Validate() error {
|
||||
if d == nil {
|
||||
return fmt.Errorf("chat intent decision 不能为空")
|
||||
}
|
||||
d.Normalize()
|
||||
switch d.Intent {
|
||||
case "chat", "task":
|
||||
return nil
|
||||
default:
|
||||
return fmt.Errorf("未知 intent: %s", d.Intent)
|
||||
}
|
||||
}
|
||||
|
||||
// RunChatNode 执行一轮聊天节点逻辑。
|
||||
//
|
||||
// 核心职责:
|
||||
// 1. 恢复判定:有 pending interaction 则处理恢复,不生成 speak;
|
||||
// 2. 意图分流:无 pending 时,调 LLM 分类 chat / task;
|
||||
// 3. 闲聊回复:纯 chat 场景直接生成回复并流式推送,phase → chatting → END;
|
||||
// 4. 任务路由:task 场景 phase → planning,交给后续 Plan 节点处理。
|
||||
//
|
||||
// 保守原则:分类失败或意图不明时,一律走 task,不丢失用户意图。
|
||||
// 1. 恢复判定:有 pending interaction 则处理恢复;
|
||||
// 2. 路由分流:无 pending 时,调 LLM 判断复杂度并路由;
|
||||
// 3. direct_reply:简单任务,直接输出回复 → END;
|
||||
// 4. execute:中等任务,推 Execute ReAct;
|
||||
// 5. deep_answer:复杂问答,原地开 thinking 深度回答 → END;
|
||||
// 6. plan:复杂规划,推 Plan 节点。
|
||||
func RunChatNode(ctx context.Context, input ChatNodeInput) error {
|
||||
runtimeState, conversationContext, emitter, err := prepareChatNodeInput(input)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
|
||||
// 1. 有 pending interaction → 纯状态传递,不生成 speak。
|
||||
// 1. 有 pending interaction → 纯状态传递,处理恢复。
|
||||
if runtimeState.HasPendingInteraction() {
|
||||
return handleChatResume(input, runtimeState, conversationContext, emitter)
|
||||
}
|
||||
|
||||
// 2. 无 pending → 调 LLM 做意图分类。
|
||||
messages := newagentprompt.BuildChatIntentMessages(conversationContext, input.UserInput)
|
||||
decision, _, err := newagentllm.GenerateJSON[chatIntentDecision](
|
||||
// 2. 无 pending → 路由决策(一次快速 LLM 调用,不开 thinking)。
|
||||
flowState := runtimeState.EnsureCommonState()
|
||||
messages := newagentprompt.BuildChatRoutingMessages(conversationContext, input.UserInput, flowState)
|
||||
|
||||
decision, rawResult, err := newagentllm.GenerateJSON[newagentmodel.ChatRoutingDecision](
|
||||
ctx,
|
||||
input.Client,
|
||||
messages,
|
||||
newagentllm.GenerateOptions{
|
||||
Temperature: 0.1,
|
||||
MaxTokens: 300,
|
||||
MaxTokens: 500,
|
||||
Thinking: newagentllm.ThinkingModeDisabled,
|
||||
Metadata: map[string]any{
|
||||
"stage": chatStageName,
|
||||
"phase": "routing",
|
||||
},
|
||||
},
|
||||
)
|
||||
if err != nil || decision.Validate() != nil {
|
||||
// 分类失败 → 保守:走 task。
|
||||
runtimeState.EnsureCommonState().Phase = newagentmodel.PhasePlanning
|
||||
|
||||
rawText := ""
|
||||
if rawResult != nil {
|
||||
rawText = strings.TrimSpace(rawResult.Text)
|
||||
}
|
||||
|
||||
if err != nil {
|
||||
// 路由失败 → 保守:走 plan。
|
||||
log.Printf("[WARN] chat routing LLM failed chat=%s raw=%s err=%v",
|
||||
flowState.ConversationID, rawText, err)
|
||||
flowState.Phase = newagentmodel.PhasePlanning
|
||||
return nil
|
||||
}
|
||||
|
||||
// 3. 按意图分流。
|
||||
flowState := runtimeState.EnsureCommonState()
|
||||
switch decision.Intent {
|
||||
case "task":
|
||||
if validateErr := decision.Validate(); validateErr != nil {
|
||||
log.Printf("[WARN] chat routing decision invalid chat=%s raw=%s err=%v",
|
||||
flowState.ConversationID, rawText, validateErr)
|
||||
flowState.Phase = newagentmodel.PhasePlanning
|
||||
return nil
|
||||
case "chat":
|
||||
return handleChatReply(ctx, decision, conversationContext, emitter, flowState)
|
||||
}
|
||||
|
||||
log.Printf("[DEBUG] chat routing chat=%s route=%s reason=%s",
|
||||
flowState.ConversationID, decision.Route, decision.Reason)
|
||||
|
||||
// 3. 按路由决策推进。
|
||||
switch decision.Route {
|
||||
case newagentmodel.ChatRouteDirectReply:
|
||||
return handleDirectReply(ctx, decision, conversationContext, emitter, flowState)
|
||||
|
||||
case newagentmodel.ChatRouteExecute:
|
||||
return handleRouteExecute(decision, emitter, flowState)
|
||||
|
||||
case newagentmodel.ChatRouteDeepAnswer:
|
||||
return handleDeepAnswer(ctx, input, decision, conversationContext, emitter, flowState)
|
||||
|
||||
case newagentmodel.ChatRoutePlan:
|
||||
return handleRoutePlan(decision, emitter, flowState)
|
||||
|
||||
default:
|
||||
flowState.Phase = newagentmodel.PhasePlanning
|
||||
return nil
|
||||
}
|
||||
}
|
||||
|
||||
// handleDirectReply 处理简单任务:直接输出回复。
|
||||
func handleDirectReply(
|
||||
ctx context.Context,
|
||||
decision *newagentmodel.ChatRoutingDecision,
|
||||
conversationContext *newagentmodel.ConversationContext,
|
||||
emitter *newagentstream.ChunkEmitter,
|
||||
flowState *newagentmodel.CommonState,
|
||||
) error {
|
||||
if strings.TrimSpace(decision.Speak) != "" {
|
||||
if err := emitter.EmitPseudoAssistantText(
|
||||
ctx, chatSpeakBlockID, chatStageName,
|
||||
decision.Speak,
|
||||
newagentstream.DefaultPseudoStreamOptions(),
|
||||
); err != nil {
|
||||
return fmt.Errorf("闲聊回复推送失败: %w", err)
|
||||
}
|
||||
conversationContext.AppendHistory(schema.AssistantMessage(decision.Speak, nil))
|
||||
}
|
||||
|
||||
flowState.Phase = newagentmodel.PhaseChatting
|
||||
return nil
|
||||
}
|
||||
|
||||
// handleRouteExecute 处理中等任务:推送简短确认,设 PhaseExecuting。
|
||||
//
|
||||
// 不把 speak 写入 history,因为真正的回复由 Execute 节点产出。
|
||||
func handleRouteExecute(
|
||||
decision *newagentmodel.ChatRoutingDecision,
|
||||
emitter *newagentstream.ChunkEmitter,
|
||||
flowState *newagentmodel.CommonState,
|
||||
) error {
|
||||
speak := strings.TrimSpace(decision.Speak)
|
||||
if speak == "" {
|
||||
speak = "好的,我来处理。"
|
||||
}
|
||||
|
||||
// 推送轻量状态通知,让前端知道请求已接收。
|
||||
_ = emitter.EmitStatus(chatStatusBlockID, chatStageName, "accepted", speak, false)
|
||||
|
||||
flowState.Phase = newagentmodel.PhaseExecuting
|
||||
|
||||
// 安全兜底:只有真正持有 task_class_ids 时才开粗排。
|
||||
if decision.NeedsRoughBuild && len(flowState.TaskClassIDs) > 0 {
|
||||
flowState.NeedsRoughBuild = true
|
||||
}
|
||||
|
||||
return nil
|
||||
}
|
||||
|
||||
// handleDeepAnswer 处理复杂问答:推送过渡语 → 原地开 thinking 再调一次 LLM → 输出深度回答。
|
||||
func handleDeepAnswer(
|
||||
ctx context.Context,
|
||||
input ChatNodeInput,
|
||||
decision *newagentmodel.ChatRoutingDecision,
|
||||
conversationContext *newagentmodel.ConversationContext,
|
||||
emitter *newagentstream.ChunkEmitter,
|
||||
flowState *newagentmodel.CommonState,
|
||||
) error {
|
||||
// 1. 推送过渡语。
|
||||
briefSpeak := strings.TrimSpace(decision.Speak)
|
||||
if briefSpeak == "" {
|
||||
briefSpeak = "让我想想。"
|
||||
}
|
||||
if err := emitter.EmitPseudoAssistantText(
|
||||
ctx, chatSpeakBlockID, chatStageName,
|
||||
briefSpeak,
|
||||
newagentstream.DefaultPseudoStreamOptions(),
|
||||
); err != nil {
|
||||
return fmt.Errorf("过渡文案推送失败: %w", err)
|
||||
}
|
||||
|
||||
// 2. 第二次 LLM 调用:开 thinking,深度回答。
|
||||
deepMessages := newagentprompt.BuildDeepAnswerMessages(conversationContext, input.UserInput)
|
||||
deepResult, err := input.Client.GenerateText(ctx, deepMessages, newagentllm.GenerateOptions{
|
||||
Temperature: 0.5,
|
||||
MaxTokens: 2000,
|
||||
Thinking: newagentllm.ThinkingModeEnabled,
|
||||
Metadata: map[string]any{
|
||||
"stage": chatStageName,
|
||||
"phase": "deep_answer",
|
||||
},
|
||||
})
|
||||
|
||||
if err != nil || deepResult == nil {
|
||||
// 深度回答失败 → 降级,只保留过渡语。
|
||||
log.Printf("[WARN] deep answer LLM failed chat=%s err=%v", flowState.ConversationID, err)
|
||||
conversationContext.AppendHistory(schema.AssistantMessage(briefSpeak, nil))
|
||||
flowState.Phase = newagentmodel.PhaseChatting
|
||||
return nil
|
||||
}
|
||||
|
||||
// 3. 输出深度回答。
|
||||
deepText := strings.TrimSpace(deepResult.Text)
|
||||
if deepText == "" {
|
||||
conversationContext.AppendHistory(schema.AssistantMessage(briefSpeak, nil))
|
||||
flowState.Phase = newagentmodel.PhaseChatting
|
||||
return nil
|
||||
}
|
||||
|
||||
if err := emitter.EmitPseudoAssistantText(
|
||||
ctx, chatSpeakBlockID, chatStageName,
|
||||
deepText,
|
||||
newagentstream.DefaultPseudoStreamOptions(),
|
||||
); err != nil {
|
||||
return fmt.Errorf("深度回答推送失败: %w", err)
|
||||
}
|
||||
|
||||
// 将完整回复(过渡语 + 深度回答)写入 history。
|
||||
fullReply := briefSpeak + "\n\n" + deepText
|
||||
conversationContext.AppendHistory(schema.AssistantMessage(fullReply, nil))
|
||||
|
||||
flowState.Phase = newagentmodel.PhaseChatting
|
||||
return nil
|
||||
}
|
||||
|
||||
// handleRoutePlan 处理复杂规划:推送确认语,设 PhasePlanning。
|
||||
func handleRoutePlan(
|
||||
decision *newagentmodel.ChatRoutingDecision,
|
||||
emitter *newagentstream.ChunkEmitter,
|
||||
flowState *newagentmodel.CommonState,
|
||||
) error {
|
||||
speak := strings.TrimSpace(decision.Speak)
|
||||
if speak == "" {
|
||||
speak = "好的,让我来规划一下。"
|
||||
}
|
||||
|
||||
_ = emitter.EmitStatus(chatStatusBlockID, chatStageName, "planning", speak, false)
|
||||
|
||||
flowState.Phase = newagentmodel.PhasePlanning
|
||||
return nil
|
||||
}
|
||||
|
||||
// ─── 恢复处理(保持原有逻辑不变)───
|
||||
|
||||
// handleChatResume 处理 pending interaction 恢复。
|
||||
//
|
||||
// 职责边界:
|
||||
@@ -216,31 +350,6 @@ func handleConfirmResume(
|
||||
return nil
|
||||
}
|
||||
|
||||
// handleChatReply 处理纯闲聊意图 — 把分类时产出的 reply 流式推给前端。
|
||||
func handleChatReply(
|
||||
ctx context.Context,
|
||||
decision *chatIntentDecision,
|
||||
conversationContext *newagentmodel.ConversationContext,
|
||||
emitter *newagentstream.ChunkEmitter,
|
||||
flowState *newagentmodel.CommonState,
|
||||
) error {
|
||||
reply := strings.TrimSpace(decision.Reply)
|
||||
|
||||
if reply != "" {
|
||||
if err := emitter.EmitPseudoAssistantText(
|
||||
ctx, chatSpeakBlockID, chatStageName,
|
||||
reply,
|
||||
newagentstream.DefaultPseudoStreamOptions(),
|
||||
); err != nil {
|
||||
return fmt.Errorf("闲聊回复推送失败: %w", err)
|
||||
}
|
||||
conversationContext.AppendHistory(schema.AssistantMessage(reply, nil))
|
||||
}
|
||||
|
||||
flowState.Phase = newagentmodel.PhaseChatting
|
||||
return nil
|
||||
}
|
||||
|
||||
// prepareChatNodeInput 校验并准备聊天节点的运行态依赖。
|
||||
func prepareChatNodeInput(input ChatNodeInput) (
|
||||
*newagentmodel.AgentRuntimeState,
|
||||
|
||||
Reference in New Issue
Block a user