wip: 完善了worker_individual和workflow_running_engine的逻辑

docs/problem.md

@@ -3,7 +3,7 @@
 ## 问题栏
 #### 🔴 核心缺陷与修复 (Bug Fixes & Stability)
 - [x] /pretor/core/individual每个template进行优化
-- [ ] /pretor/worker_individual待完善复合子个体和基础子个体
+- [x] /pretor/worker_individual待完善复合子个体和基础子个体
 
 #### 🛡️ 安全与合规 (Security & Auth)
 - [ ] 优化安全架构防止模型注入
@@ -15,36 +15,10 @@
 - [ ] 优化import
 
 #### 🏗️ 架构演进 (Architecture & Refactoring)
-- 】~~使用fastapi-users完善用户系统~~(2026/4/19 fastapi-users会严重摧毁代码的优雅性)
-- [ ] 升级auth功能
-- [x] /pretor/api的接口函数进行重构
-- [x] /dockerfile待完善
-- [ ] 完善沙箱功能
-- [ ] 完善爬虫功能
-- [ ] 对接更多的provider
-
----
-## 日志
-#### 2026/4/12
-- [x] /pretor/api的接口函数进行重构
-- [x] /pretor/core/individual每个template进行优化
-- [ ] /pretor/worker_individual待完善复合子个体和基础子个体
-- [ ] /pretor/api待完善
-- [x] /dockerfile待完善
-
-#### 2026/4/16
-- [ ] 发布v0.1.0正式版
-- [ ] 增加对应全workflow的情况追踪，使得在任务运行中人机交互更加自然方便
-- ~~[ ] 使用fastapi-users完善用户系统~~
-
-#### 2026/4/19
-- [ ] 完善沙箱功能
-- [ ] 完善爬虫功能
-- [ ] 对接更多的provider
-- [ ] 优化import
+- [x] ~~使用fastapi-users完善用户系统~~(2026/4/19 fastapi-users会严重摧毁代码的优雅性)
 - [x] 升级auth功能
-
-#### 2026/4/20
-- [ ] 优化安全架构防止模型注入
-- [ ] 设计workflowEngine的自动扩缩容设计
-- [ ] 完善错误捕获和日志系统
+- [x] /pretor/api的接口函数进行重构
+- [x] /dockerfile待完善
+- [ ] 完善沙箱功能
+- [ ] 完善爬虫功能
+- [ ] 对接更多的provider

pretor/core/database/module/individual.py

@@ -74,3 +74,10 @@ class IndividualDatabase:
             session.delete(individual)
             await session.commit()
             return True
+
+    @database_exception
+    async def get_all_worker_individual(self) -> List[WorkerIndividual]:
+        async with self.async_session_maker() as session:
+            statement = select(WorkerIndividual)
+            results = await session.execute(statement)
+            return list(results.scalars().all())

pretor/core/global_state_machine/global_state_machine.py

@@ -22,6 +22,8 @@ import asyncio
 from pretor.core.workflow.workflow import PretorWorkflow
 from pretor.core.workflow.workflow_template_manager import WorkflowManager
 from pretor.core.global_state_machine.skill_manager import GlobalSkillManager
+from pretor.core.global_state_machine.individual_manager import GlobalIndividualManager
+
 
 @ray.remote
 class GlobalStateMachine:
@@ -33,11 +35,13 @@ class GlobalStateMachine:
         self._global_tool_manager = GlobalToolManager()
         self._global_workflow_template_manager = WorkflowManager()
         self._global_skill_manager = GlobalSkillManager()
+        self._global_individual_manager = GlobalIndividualManager()
 
         self.postgres_database = postgres_database
 
     async def init_state_machine(self):
         await self._global_provider_manager.init_provider_register(self.postgres_database)
+        await self._global_individual_manager.init_individual_register(self.postgres_database)
 
     async def add_provider_wrap(self, provider_type, provider_title, provider_url, provider_apikey, provider_owner):
         return await self._global_provider_manager.add_provider(
@@ -73,6 +77,12 @@ class GlobalStateMachine:
             return await method(*args, **kwargs)
         return method(*args, **kwargs)
 
+    async def individual_manager(self, method_name: str, *args, **kwargs):
+        method = getattr(self._global_individual_manager, method_name)
+        if asyncio.iscoroutinefunction(method):
+            return await method(*args, **kwargs)
+        return method(*args, **kwargs)
+
     ###以下方法为event_dict方法
     def add_event(self, event: PretorEvent) -> None:
         event.pending_queue = asyncio.Queue()

pretor/core/global_state_machine/individual_manager.py

@@ -0,0 +1,61 @@
+#  Copyright 2026 zhaoxi826
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#       http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+
+from typing import Dict, Any
+from loguru import logger
+
+class GlobalIndividualManager:
+    def __init__(self):
+        self._individuals: Dict[str, Dict[str, Any]] = {}
+
+    async def init_individual_register(self, postgres) -> None:
+        try:
+            try:
+                individuals = await postgres.individual_database.remote("get_all_worker_individual")
+                for ind in individuals:
+                    agent_id = getattr(ind, 'agent_id', None)
+                    if agent_id:
+                        self._individuals[agent_id] = ind.model_dump() if hasattr(ind, 'model_dump') else dict(ind)
+                logger.info(f"成功从数据库拉取了 {len(self._individuals)} 个 Worker Individual 配置。")
+            except AttributeError:
+                logger.warning("数据库中 get_all_worker_individual 方法未实现，跳过全量加载。可以在将来完善该接口。")
+            except Exception as e:
+                # 捕获因 Ray 调用目标方法不存在引发的异常
+                if "has no attribute 'get_all_worker_individual'" in str(e):
+                    logger.warning("数据库 individual_database 中缺少 get_all_worker_individual 方法，无法全量拉取。")
+                else:
+                    raise e
+        except Exception as e:
+            logger.error(f"从数据库拉取 Worker Individual 配置失败: {e}")
+
+    def add_individual(self, agent_id: str, config: Dict[str, Any]) -> None:
+        """
+        注册一个 worker individual
+        config 可以包含 type, prompt, provider_title, model_id 等
+        """
+        config["agent_id"] = agent_id
+        self._individuals[agent_id] = config
+
+    def get_individual(self, agent_id: str) -> Dict[str, Any]:
+        """
+        获取一个 worker individual 的配置
+        """
+        return self._individuals.get(agent_id, None)
+
+    def remove_individual(self, agent_id: str) -> None:
+        if agent_id in self._individuals:
+            del self._individuals[agent_id]
+
+    def list_individuals(self) -> Dict[str, Dict[str, Any]]:
+        return self._individuals

pretor/core/workflow/workflow_runner.py

@@ -187,7 +187,6 @@ class WorkflowEngine:
             elif step.node == "consciousness_node":
                 if not self.consciousness_node:
                     raise WorkflowError("未提供 consciousness_node 句柄！")
-                # 这里将 command 作为 original_command，可根据业务调整
                 original_cmd = self.workflow.command or ""
                 payload = ConsciousnessForWorkflowInput(
                     workflow_step=step,
@@ -199,11 +198,31 @@ class WorkflowEngine:
                 return result_obj, True
 
             elif step.node in ["primary_individual", "composite_individual"]:
-                logger.warning(f"当前节点 {step.node} 暂未实现完整调度支持，这里将模拟执行。")
-                await asyncio.sleep(1)
-                simulated_result = f"这是 {step.node} 执行 {step.action} 产生的模拟结果 (输入: {input_data})"
-                return simulated_result, True
+                logger.info(f"正在通过 WorkerCluster 调度 {step.node} 的 {step.action} 动作。")
+                try:
+                    from pretor.utils.ray_hook import ray_actor_hook
+                    worker_cluster = ray_actor_hook("worker_cluster").worker_cluster
+                    task_id = f"{self.workflow.trace_id}_step_{step.step}"
+                    agent_id = getattr(step, 'agent_id', f"default_{step.node}")
+                    if isinstance(input_data, dict) and "agent_id" in input_data:
+                        agent_id = input_data.get("agent_id")
+                    task_event = {
+                        "action": step.action,
+                        "description": step.description,
+                        "input_data": input_data,
+                        "context_memory": self.workflow.context_memory
+                    }
+                    result_response = await worker_cluster.submit_task.remote(task_id, agent_id, task_event)
 
+                    if result_response.get("success"):
+                        return result_response.get("data"), True
+                    else:
+                        logger.error(f"WorkerCluster 执行 {step.node} 失败: {result_response.get('error')}")
+                        return result_response.get("error"), False
+
+                except Exception as e:
+                    logger.exception(f"调度 WorkerCluster 执行 {step.node} 时发生异常: {e}")
+                    raise WorkflowError(f"WorkerCluster 调度异常: {e}")
             else:
                 raise WorkflowError(f"未知的节点类型：{step.node}")
 

pretor/worker_individual/worker_cluster.py

@@ -13,6 +13,14 @@
 #  limitations under the License.
 
 import ray
+import time
+import asyncio
+from collections import OrderedDict
+from loguru import logger
+from ray.util.queue import Queue
+from pretor.utils.ray_hook import ray_actor_hook
+from pretor.worker_individual.worker_individual import BaseIndividual, SkillIndividual, OrdinaryIndividual, \
+    SpecialIndividual
 
 
 @ray.remote
@@ -22,35 +30,31 @@ class WorkerCluster:
     设计理念：按需加载，内存 LRU 淘汰，避免 Actor 爆炸
     """
 
-    def __init__(self, db_actor, max_capacity: int = 200):
-        self.db = db_actor
+    def __init__(self, max_capacity: int = 200, num_runners: int = 10):
         self.max_capacity = max_capacity
-        # 核心：LRU 活跃 Agent 缓存池
-        self._active_workers: OrderedDict[str, BaseWorkerIndividual] = OrderedDict()
+        self._active_workers: OrderedDict[str, BaseIndividual] = OrderedDict()
         self.status = "running"
+        self.task_queue = Queue()
+        self.results_futures = {}
+        self.runners = []
+        self.num_runners = num_runners
 
-    async def _recruit_worker(self, agent_id: str) -> BaseWorkerIndividual:
+    async def start(self):
+        self.runners = [asyncio.create_task(self._runner(i)) for i in range(self.num_runners)]
+        logger.info(f"WorkerCluster 已启动 {self.num_runners} 个 runner 协程。")
+
+    async def _recruit_worker(self, agent_id: str) -> BaseIndividual:
         """内部方法：招聘/唤醒一个具体的 Agent 对象"""
-
-        # 1. 尝试从缓存直接命中
         if agent_id in self._active_workers:
-            self._active_workers.move_to_end(agent_id)  # 标记为最近使用
+            self._active_workers.move_to_end(agent_id)
             return self._active_workers[agent_id]
 
-        # 2. 缓存未命中，去数据库拉取 Agent 档案配置
-        # agent_config = await self.db.get_agent_config.remote(agent_id)
-
-        # 模拟从数据库取出的配置数据
-        agent_config = {
-            "agent_id": agent_id,
-            "type": "skill",  # 取决于数据库里的设定：ordinary, skill, special
-            "prompt": "你是一个资深架构师..."
-        }
+        global_state_machine = ray_actor_hook("global_state_machine").global_state_machine
+        agent_config = await global_state_machine.individual_manager.remote("get_individual", agent_id)
 
         if not agent_config:
             raise ValueError(f"无法唤醒 Agent {agent_id}：数据库中不存在该档案")
 
-        # 3. 工厂模式：根据类型动态装配不同量级的 Individual
         worker_type = agent_config.get("type", "ordinary")
         if worker_type == "skill":
             worker = SkillIndividual(agent_config)
@@ -59,49 +63,76 @@ class WorkerCluster:
         else:
             worker = OrdinaryIndividual(agent_config)
 
-        # 4. 放入内存池，如果爆满则淘汰最老的那个
         self._active_workers[agent_id] = worker
         if len(self._active_workers) > self.max_capacity:
             evicted_id, _ = self._active_workers.popitem(last=False)
-            print(f"[WorkerCluster] 内存池满，休眠老化 Agent: {evicted_id}")
+            logger.info(f"[WorkerCluster] 内存池满，休眠老化 Agent: {evicted_id}")
 
         return worker
 
-    async def execute_task(self, agent_id: str, task_event: dict) -> dict:
-        """
-        对外暴露的唯一干活接口。
-        task_event 应该包含所有的上下文（Context、历史记忆、本次指令）
-        """
+    async def _runner(self, runner_id: int):
+        while True:
+            try:
+                task = await self.task_queue.get_async()
+                task_id = task.get("task_id")
+                agent_id = task.get("agent_id")
+                task_event = task.get("task_event")
+
+                logger.debug(f"[WorkerCluster Runner {runner_id}] 开始处理任务 {task_id} 给 Agent {agent_id}")
+                start_time = time.time()
+
+                try:
+                    worker = await self._recruit_worker(agent_id)
+                    result = await worker.run(task_event)
+                    cost_time = time.time() - start_time
+
+                    response = {
+                        "success": True,
+                        "agent_id": agent_id,
+                        "data": result,
+                        "metrics": {"cost_time_sec": round(cost_time, 2)}
+                    }
+                except Exception as e:
+                    logger.exception(f"[WorkerCluster Runner {runner_id}] 执行任务 {task_id} 时发生错误: {e}")
+                    response = {
+                        "success": False,
+                        "agent_id": agent_id,
+                        "error": str(e)
+                    }
+                if task_id in self.results_futures:
+                    future = self.results_futures[task_id]
+                    if not future.done():
+                        future.set_result(response)
+
+            except Exception as e:
+                logger.error(f"[WorkerCluster Runner {runner_id}] 循环发生异常: {e}")
+                await asyncio.sleep(1)
+
+    async def submit_task(self, task_id: str, agent_id: str, task_event: dict):
+        if not self.runners:
+            await self.start()
+
+        future = asyncio.Future()
+        self.results_futures[task_id] = future
+
+        task = {
+            "task_id": task_id,
+            "agent_id": agent_id,
+            "task_event": task_event
+        }
+        await self.task_queue.put_async(task)
+        logger.debug(f"[WorkerCluster] 任务 {task_id} 已加入队列。")
+
         try:
-            # 1. 获取工作实体（秒级热启动或毫秒级缓存命中）
-            worker = await self._recruit_worker(agent_id)
-
-            # 2. 注入上下文并执行
-            # 这里的 run 方法内部不保存状态，所有记忆都从 task_event 传入
-            start_time = time.time()
-            result = await worker.run(task_event)
-            cost_time = time.time() - start_time
-
-            # 3. 封装标准回包
-            return {
-                "success": True,
-                "agent_id": agent_id,
-                "data": result,
-                "metrics": {"cost_time_sec": round(cost_time, 2)}
-            }
-
-        except Exception as e:
-            # 异常隔离：一个 Agent 报错，绝对不能把整个 Cluster 搞崩
-            return {
-                "success": False,
-                "agent_id": agent_id,
-                "error": str(e)
-            }
+            result = await future
+            return result
+        finally:
+            self.results_futures.pop(task_id, None)
 
     def get_cluster_metrics(self):
-        """监控探针：用于查看当前集群负载"""
         return {
             "active_worker_count": len(self._active_workers),
             "max_capacity": self.max_capacity,
-            "cached_agent_ids": list(self._active_workers.keys())
+            "cached_agent_ids": list(self._active_workers.keys()),
+            "queue_size": self.task_queue.size()
         }

pretor/worker_individual/worker_individual.py

@@ -11,3 +11,148 @@
 #  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
+
+
+from loguru import logger
+from pydantic_ai import Agent, RunContext
+from pydantic import Field
+from pretor.adapter.model_adapter.agent_factory import AgentFactory
+from pretor.core.global_state_machine.global_state_machine import GlobalStateMachine
+from pretor.core.global_state_machine.model_provider.base_provider import Provider
+from pretor.utils.agent_model import ResponseModel, InputModel, DepsModel
+from pretor.utils.get_tool import get_tool
+import ray
+from pretor.utils.ray_hook import ray_actor_hook
+
+
+class WorkerIndividualResponse(ResponseModel):
+    output: str = Field(..., description="Worker执行任务的输出结果")
+
+
+class WorkerIndividualDeps(DepsModel):
+    task_event: dict
+
+
+class WorkerIndividualInput(InputModel):
+    task_event: dict
+
+
+class BaseIndividual:
+    """
+    Worker Individual 的基类
+    """
+
+    def __init__(self, agent_config: dict):
+        self.agent_config = agent_config
+        self.agent_id = agent_config.get("agent_id")
+        self.agent: Agent | None = None
+
+    async def _init_agent(self, agent_name: str, system_prompt: str):
+        global_state_machine = ray_actor_hook("global_state_machine").global_state_machine
+        provider_title = self.agent_config.get("provider_title", "openai")  # default fallback
+        model_id = self.agent_config.get("model_id", "gpt-4o")  # default fallback
+
+        provider: Provider = await global_state_machine.provider_manager.remote("get_provider", provider_title)
+        agent_factory = AgentFactory()
+        self.agent = agent_factory.create_agent(
+            provider=provider,
+            model_id=model_id,
+            output_type=WorkerIndividualResponse,
+            system_prompt=system_prompt,
+            deps_type=WorkerIndividualDeps,
+            agent_name=agent_name
+        )
+
+        @self.agent.system_prompt
+        async def dynamic_prompt(ctx: RunContext[WorkerIndividualDeps]):
+            prompt = system_prompt + "\n\n"
+            prompt += (
+                f"=== 当前任务上下文 ===\n"
+                f"{ctx.deps.task_event}\n"
+            )
+            return prompt
+
+    async def run(self, task_event: dict) -> dict:
+        raise NotImplementedError("子类必须实现 run 方法")
+
+
+class SkillIndividual(BaseIndividual):
+    """
+    专家子个体：拥有专业 skill 的 agent。
+    """
+
+    def __init__(self, agent_config: dict):
+        super().__init__(agent_config)
+
+    async def run(self, task_event: dict) -> dict:
+        if self.agent is None:
+            system_prompt = self.agent_config.get("prompt",
+                                                  "你是一个拥有专业技能的专家级AI助手，请利用你的专业知识完成给定的任务。")
+            await self._init_agent("skill_individual", system_prompt)
+
+        deps = WorkerIndividualDeps(task_event=task_event)
+        self.agent.retries = 3
+        # In actual usage, tools could be dynamically loaded here based on agent_config
+        # tool = get_tool("skill_individual")
+        try:
+            result = await self.agent.run(
+                f"请执行以下任务：\n{task_event}",
+                deps=deps
+                # tools=tool
+            )
+            return {"output": result.data.output}
+        except Exception as e:
+            logger.exception(f"SkillIndividual {self.agent_id} 执行失败: {e}")
+            raise
+
+
+class OrdinaryIndividual(BaseIndividual):
+    """
+    普通子个体：普通的 agent。
+    """
+
+    def __init__(self, agent_config: dict):
+        super().__init__(agent_config)
+
+    async def run(self, task_event: dict) -> dict:
+        if self.agent is None:
+            system_prompt = self.agent_config.get("prompt", "你是一个普通的AI助手，请尽力完成给定的任务。")
+            await self._init_agent("ordinary_individual", system_prompt)
+
+        deps = WorkerIndividualDeps(task_event=task_event)
+        self.agent.retries = 3
+        try:
+            result = await self.agent.run(
+                f"请执行以下任务：\n{task_event}",
+                deps=deps
+            )
+            return {"output": result.data.output}
+        except Exception as e:
+            logger.exception(f"OrdinaryIndividual {self.agent_id} 执行失败: {e}")
+            raise
+
+
+class SpecialIndividual(BaseIndividual):
+    """
+    特殊子个体：执行特殊任务的 agent，如生成语音、视频等。
+    """
+
+    def __init__(self, agent_config: dict):
+        super().__init__(agent_config)
+
+    async def run(self, task_event: dict) -> dict:
+        if self.agent is None:
+            system_prompt = self.agent_config.get("prompt", "你是一个特殊的AI助手，负责处理特殊类型的任务。")
+            await self._init_agent("special_individual", system_prompt)
+
+        deps = WorkerIndividualDeps(task_event=task_event)
+        self.agent.retries = 3
+        try:
+            result = await self.agent.run(
+                f"请执行以下任务：\n{task_event}",
+                deps=deps
+            )
+            return {"output": result.data.output}
+        except Exception as e:
+            logger.exception(f"SpecialIndividual {self.agent_id} 执行失败: {e}")
+            raise