♻️ refactor(core): 重构 AI 编排框架与记忆及 RAG 子系统 (#2149)

* ♻️ refactor(core): 重构 AI 编排框架与记忆及 RAG 子系统

- 【重构】重构 `BaseRunnable` 并引入统一的 `RunIntent` 意图载体，规范 Agent、Team 和 Workflow 的执行流
- 【解耦】将中期记忆槽和长期向量记忆从 `MemoryConfig` 中解耦，转为独立的能力组件与工具箱进行管理
- 【记忆】移除 `MemoryReader` 和 `MemoryWriter`，统一封装为 `SessionMemoryContext` 会话记忆门面
- 【RAG】重构检索器与存储后端接口，统一采用 `QueryRequest` 进行多维度联合检索，并引入 `InMemoryScorer` 提升打分性能
- 【事件】优化 `EventBus` 异步事件分发机制，引入队列机制确保事件按序处理，避免并发竞态问题
- 【依赖注入】移除 `memory` 注入项，优化 `DependencyInjector` 的签名解析缓存以提升性能

* 🚨 auto fix by pre-commit hooks

---------

Co-authored-by: webjoin111 <455457521@qq.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

zhenxun/services/ai/llm/api.py

@@ -2,11 +2,13 @@
 LLM 服务的高级 API 接口 - 便捷函数入口 (无状态)
 """
 
+import json
 from pathlib import Path
 from typing import Any, Literal, TypeVar, overload
 
 from pydantic import BaseModel
 
+from zhenxun.services.ai.config import get_llm_config
 from zhenxun.services.ai.core.exceptions import (
     ControlFlowExit,
     LLMException,
@@ -27,14 +29,18 @@ from zhenxun.services.ai.core.messages import (
     RerankRequest,
     RerankResult,
     SpeechRequest,
+    UsageInfo,
 )
 from zhenxun.services.ai.core.models import ModelName
 from zhenxun.services.ai.core.options import (
     GenerationConfig,
     LLMEmbeddingConfig,
+    OutputFormatConfig,
+    ResponseFormat,
+    StructuredOutputStrategy,
     TTSConfig,
 )
-from zhenxun.services.ai.guardrails import GuardrailSource
+from zhenxun.services.ai.guardrails import GuardrailSource, parse_guardrails
 from zhenxun.services.ai.utils.logger import log_llm as logger
 
 from .builder import IntentBuilder
@@ -109,7 +115,7 @@ async def embed(
 
 @overload
 async def embed(
-    input_batch: list[Any],
+    input_batch: list[PromptInput],
     *,
     model: ModelName = None,
     task: Literal[
@@ -122,7 +128,7 @@ async def embed(
 
 
 async def embed(
-    input_batch: PromptInput | list[Any],
+    input_batch: PromptInput | list[PromptInput],
     *,
     model: ModelName = None,
     task: Literal[
@@ -158,8 +164,6 @@ async def embed(
     )
 
     if not batch.payloads:
-        from zhenxun.services.ai.core.messages import UsageInfo
-
         return EmbeddingResponse(
             embeddings=[], usage=UsageInfo(), model_name=str(model)
         )
@@ -256,21 +260,13 @@ async def generate_structured(
         T: 解析验证通过后的 Pydantic 模型实例。
     """  # noqa: E501
     try:
-        from zhenxun.services.ai.config import get_llm_config
         from zhenxun.services.ai.core.engine.structured_parser import (
             BaseOutputProcessor,
         )
-        from zhenxun.services.ai.core.options import (
-            OutputFormatConfig,
-            ResponseFormat,
-            StructuredOutputStrategy,
-        )
 
         if max_retries is None:
             max_retries = get_llm_config().client_settings.structured_retries
 
-        from zhenxun.services.ai.guardrails import parse_guardrails
-
         parsed_guardrails = parse_guardrails(guardrails)
 
         output_processor = BaseOutputProcessor(
@@ -291,8 +287,6 @@ async def generate_structured(
         if instruction:
             prompt_parts.append(instruction)
 
-        import json
-
         schema_str = json.dumps(json_schema, ensure_ascii=False, indent=2)
         prompt_parts.append(
             "### ⚠️ [结构化输出要求]\n"
@@ -393,7 +387,9 @@ async def generate(
             llm_context = LLMContext(request=request)
             combined_cap = CombinedCapability(sys_caps)
 
-            async def inner_handler(ctx: LLMContext[Any, Any]) -> ChatResponse:
+            async def inner_handler(
+                ctx: LLMContext[ChatRequest, ChatResponse],
+            ) -> ChatResponse:
                 return await LLMOrchestrator.invoke(
                     ctx.request,
                     model_name=model,
@@ -420,7 +416,7 @@ async def generate(
 
 @overload
 async def create_image(
-    prompt: str | Any,
+    prompt: PromptInput,
     *,
     images: None = None,
     model: ModelName = None,
@@ -432,7 +428,7 @@ async def create_image(
 
 @overload
 async def create_image(
-    prompt: str | Any,
+    prompt: PromptInput,
     *,
     images: list[Path | bytes | str] | Path | bytes | str,
     model: ModelName = None,
@@ -443,7 +439,7 @@ async def create_image(
 
 
 async def create_image(
-    prompt: str | Any,
+    prompt: PromptInput,
     *,
     images: list[Path | bytes | str] | Path | bytes | str | None = None,
     model: ModelName = None,

zhenxun/services/ai/llm/builder.py

@@ -2,13 +2,18 @@
 LLM 生成配置相关类和函数
 """
 
-from typing import Any, Literal
+import inspect
+from typing import Any, Literal, cast
 from typing_extensions import Self
 
+from pydantic import BaseModel
+
+from zhenxun.services.ai.config import get_gemini_safety_threshold
 from zhenxun.services.ai.core.exceptions import ConfigurationException
 from zhenxun.services.ai.core.options import (
     GenerationConfig,
     ResponseFormat,
+    StructuredOutputStrategy,
 )
 from zhenxun.services.ai.utils.logger import log_llm as logger
 from zhenxun.utils.pydantic_compat import model_json_schema, model_validate
@@ -94,20 +99,12 @@ class IntentBuilder:
         强制要求结构化输出意图。
         支持自动处理 Pydantic 模型并转换为厂商所需的 JSON Schema。
         """
-        import inspect
-
-        from pydantic import BaseModel
-
-        from zhenxun.services.ai.core.options import StructuredOutputStrategy
-
         self._config.output.response_format = ResponseFormat.JSON
         self._config.output.response_mime_type = "application/json"
         if schema:
             if inspect.isclass(schema) and issubclass(schema, BaseModel):
                 self._config.output.response_schema = model_json_schema(schema)
             else:
-                from typing import cast
-
                 self._config.output.response_schema = cast(dict[str, Any], schema)
         if strict:
             self._config.output.structured_output_strategy = (
@@ -137,8 +134,6 @@ class IntentBuilder:
         安全合规意图。
         level 取值: 'strict' (最严格), 'moderate' (中等), 'none' (完全无限制)。
         """
-        from zhenxun.services.ai.config import get_gemini_safety_threshold
-
         if level == "strict":
             self.gemini.set_safety_threshold("BLOCK_LOW_AND_ABOVE")
         elif level == "none":

zhenxun/services/ai/llm/manager.py

@@ -17,6 +17,7 @@ from zhenxun.services.ai.utils.logger import log_llm as logger
 from zhenxun.utils.manager.priority_manager import PriorityLifecycle
 from zhenxun.utils.pydantic_compat import model_dump
 
+from .system.cache import clear_model_cache, get_or_create_model
 from .system.capabilities import get_model_capabilities
 from .system.network import health_manager
 
@@ -277,8 +278,6 @@ async def get_model_instance(
 
     provider_config_found, model_detail_found = config_tuple_found
 
-    from .system.cache import get_or_create_model
-
     return await get_or_create_model(
         provider_config_found, model_detail_found, override_config
     )
@@ -288,8 +287,6 @@ def clear_all_cache() -> None:
     """
     清空模型实例缓存与路由组解析缓存。
     """
-    from .system.cache import clear_model_cache
-
     clear_model_cache()
     clear_resolved_group_cache()
     logger.debug("已清空全局模型实例与路由组缓存")
@@ -300,11 +297,8 @@ async def _init_llm_config_on_startup():
     """启动时初始化 LLM 配置、密钥状态并预热工具提供者管理器。"""
     logger.info("正在初始化 LLM 配置并加载遥测状态...")
     try:
-        from zhenxun.services.ai.config import get_llm_config
         from zhenxun.services.ai.tools.engine.registry import tool_provider_manager
 
-        from .system.network import health_manager
-
         get_llm_config()
         await health_manager.initialize()
         await tool_provider_manager.initialize()