mirror of
https://github.com/zhenxun-org/zhenxun_bot.git
synced 2026-10-17 18:20:01 +08:00
♻️ refactor(llm): 重构 LLM 服务架构,引入中间件与组件化适配器 (#2073)
检查bot是否运行正常 / bot check (push) Has been cancelled
CodeQL Code Security Analysis / Analyze (${{ matrix.language }}) (none, python) (push) Has been cancelled
Sequential Lint and Type Check / ruff-call (push) Has been cancelled
Release Drafter / Update Release Draft (push) Has been cancelled
Force Sync to Aliyun / sync (push) Has been cancelled
Update Version / update-version (push) Has been cancelled
Sequential Lint and Type Check / pyright-call (push) Has been cancelled
检查bot是否运行正常 / bot check (push) Has been cancelled
CodeQL Code Security Analysis / Analyze (${{ matrix.language }}) (none, python) (push) Has been cancelled
Sequential Lint and Type Check / ruff-call (push) Has been cancelled
Release Drafter / Update Release Draft (push) Has been cancelled
Force Sync to Aliyun / sync (push) Has been cancelled
Update Version / update-version (push) Has been cancelled
Sequential Lint and Type Check / pyright-call (push) Has been cancelled
* ♻️ refactor(llm): 重构 LLM 服务架构,引入中间件与组件化适配器 - 【重构】LLM 服务核心架构: - 引入中间件管道,统一处理请求生命周期(重试、密钥选择、日志、网络请求)。 - 适配器重构为组件化设计,分离配置映射、消息转换、响应解析和工具序列化逻辑。 - 移除 `with_smart_retry` 装饰器,其功能由中间件接管。 - 移除 `LLMToolExecutor`,工具执行逻辑集成到 `ToolInvoker`。 - 【功能】增强配置系统: - `LLMGenerationConfig` 采用组件化结构(Core, Reasoning, Visual, Output, Safety, ToolConfig)。 - 新增 `GenConfigBuilder` 提供语义化配置构建方式。 - 新增 `LLMEmbeddingConfig` 用于嵌入专用配置。 - `CommonOverrides` 迁移并更新至新配置结构。 - 【功能】强化工具系统: - 引入 `ToolInvoker` 实现更灵活的工具执行,支持回调与结构化错误。 - `function_tool` 装饰器支持动态 Pydantic 模型创建和依赖注入 (`ToolParam`, `RunContext`)。 - 平台原生工具支持 (`GeminiCodeExecution`, `GeminiGoogleSearch`, `GeminiUrlContext`)。 - 【功能】高级生成与嵌入: - `generate_structured` 方法支持 In-Context Validation and Repair (IVR) 循环和 AutoCoT (思维链) 包装。 - 新增 `embed_query` 和 `embed_documents` 便捷嵌入 API。 - `OpenAIImageAdapter` 支持 OpenAI 兼容的图像生成。 - `SmartAdapter` 实现模型名称智能路由。 - 【重构】消息与类型系统: - `LLMContentPart` 扩展支持更多模态和代码执行相关内容。 - `LLMMessage` 和 `LLMResponse` 结构更新,支持 `content_parts` 和思维链签名。 - 统一 `LLMErrorCode` 和用户友好错误消息,提供更详细的网络/代理错误提示。 - `pyproject.toml` 移除 `bilireq`,新增 `json_repair`。 - 【优化】日志与调试: - 引入 `DebugLogOptions`,提供细粒度日志脱敏控制。 - 增强日志净化器,处理更多敏感数据和长字符串。 - 【清理】删除废弃模块: - `zhenxun/services/llm/memory.py` - `zhenxun/services/llm/executor.py` - `zhenxun/services/llm/config/presets.py` - `zhenxun/services/llm/types/content.py` - `zhenxun/services/llm/types/enums.py` - `zhenxun/services/llm/tools/__init__.py` - `zhenxun/services/llm/tools/manager.py` * 📦️ build(deps): 移除 bilireq 并添加 json_repair 依赖 * 🐛 (llm): 移除图片生成模型能力预检查 * ♻️ refactor(llm.session): 重构记忆系统以分离存储和策略 * 🐛 fix(reload_setting): 重载配置时清除LLM缓存 * ✨ feat(llm): 支持结构化生成函数接收 UniMessage * ✨ feat(search): 为搜索功能默认启用 Gemini Google Search 工具 * 🚨 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>
This commit is contained in:
co-authored by
webjoin111
pre-commit-ci[bot]
parent
e5b2a872d3
commit
c9f0a8b9d9
@@ -2,10 +2,17 @@
|
||||
LLM 适配器工厂类
|
||||
"""
|
||||
|
||||
from typing import ClassVar
|
||||
import fnmatch
|
||||
from typing import TYPE_CHECKING, Any, ClassVar
|
||||
|
||||
from ..types.exceptions import LLMErrorCode, LLMException
|
||||
from .base import BaseAdapter
|
||||
from ..types.models import ToolChoice
|
||||
from .base import BaseAdapter, RequestData, ResponseData
|
||||
|
||||
if TYPE_CHECKING:
|
||||
from ..config.generation import LLMEmbeddingConfig, LLMGenerationConfig
|
||||
from ..service import LLMModel
|
||||
from ..types import LLMMessage
|
||||
|
||||
|
||||
class LLMAdapterFactory:
|
||||
@@ -21,10 +28,13 @@ class LLMAdapterFactory:
|
||||
return
|
||||
|
||||
from .gemini import GeminiAdapter
|
||||
from .openai import OpenAIAdapter
|
||||
from .openai import DeepSeekAdapter, OpenAIAdapter, OpenAIImageAdapter
|
||||
|
||||
cls.register_adapter(OpenAIAdapter())
|
||||
cls.register_adapter(DeepSeekAdapter())
|
||||
cls.register_adapter(GeminiAdapter())
|
||||
cls.register_adapter(SmartAdapter())
|
||||
cls.register_adapter(OpenAIImageAdapter())
|
||||
|
||||
@classmethod
|
||||
def register_adapter(cls, adapter: BaseAdapter) -> None:
|
||||
@@ -74,3 +84,100 @@ def get_adapter_for_api_type(api_type: str) -> BaseAdapter:
|
||||
def register_adapter(adapter: BaseAdapter) -> None:
|
||||
"""注册新的适配器"""
|
||||
LLMAdapterFactory.register_adapter(adapter)
|
||||
|
||||
|
||||
class SmartAdapter(BaseAdapter):
|
||||
"""
|
||||
智能路由适配器。
|
||||
本身不处理序列化,而是根据规则委托给 OpenAIAdapter 或 GeminiAdapter。
|
||||
"""
|
||||
|
||||
@property
|
||||
def log_sanitization_context(self) -> str:
|
||||
return "openai_request"
|
||||
|
||||
_ROUTING_RULES: ClassVar[list[tuple[str, str]]] = [
|
||||
("*nano-banana*", "gemini"),
|
||||
("*gemini*", "gemini"),
|
||||
]
|
||||
_DEFAULT_API_TYPE: ClassVar[str] = "openai"
|
||||
|
||||
def __init__(self):
|
||||
self._adapter_cache: dict[str, BaseAdapter] = {}
|
||||
|
||||
@property
|
||||
def api_type(self) -> str:
|
||||
return "smart"
|
||||
|
||||
@property
|
||||
def supported_api_types(self) -> list[str]:
|
||||
return ["smart"]
|
||||
|
||||
def _get_delegate_adapter(self, model: "LLMModel") -> BaseAdapter:
|
||||
"""
|
||||
核心路由逻辑:决定使用哪个适配器 (带缓存)
|
||||
"""
|
||||
if model.model_detail.api_type:
|
||||
return get_adapter_for_api_type(model.model_detail.api_type)
|
||||
|
||||
model_name = model.model_name
|
||||
if model_name in self._adapter_cache:
|
||||
return self._adapter_cache[model_name]
|
||||
|
||||
target_api_type = self._DEFAULT_API_TYPE
|
||||
model_name_lower = model_name.lower()
|
||||
|
||||
for pattern, api_type in self._ROUTING_RULES:
|
||||
if fnmatch.fnmatch(model_name_lower, pattern):
|
||||
target_api_type = api_type
|
||||
break
|
||||
|
||||
adapter = get_adapter_for_api_type(target_api_type)
|
||||
self._adapter_cache[model_name] = adapter
|
||||
return adapter
|
||||
|
||||
async def prepare_advanced_request(
|
||||
self,
|
||||
model: "LLMModel",
|
||||
api_key: str,
|
||||
messages: list["LLMMessage"],
|
||||
config: "LLMGenerationConfig | None" = None,
|
||||
tools: list[Any] | None = None,
|
||||
tool_choice: "str | dict[str, Any] | ToolChoice | None" = None,
|
||||
) -> RequestData:
|
||||
adapter = self._get_delegate_adapter(model)
|
||||
return await adapter.prepare_advanced_request(
|
||||
model, api_key, messages, config, tools, tool_choice
|
||||
)
|
||||
|
||||
def parse_response(
|
||||
self,
|
||||
model: "LLMModel",
|
||||
response_json: dict[str, Any],
|
||||
is_advanced: bool = False,
|
||||
) -> ResponseData:
|
||||
adapter = self._get_delegate_adapter(model)
|
||||
return adapter.parse_response(model, response_json, is_advanced)
|
||||
|
||||
def prepare_embedding_request(
|
||||
self,
|
||||
model: "LLMModel",
|
||||
api_key: str,
|
||||
texts: list[str],
|
||||
config: "LLMEmbeddingConfig",
|
||||
) -> RequestData:
|
||||
adapter = self._get_delegate_adapter(model)
|
||||
return adapter.prepare_embedding_request(model, api_key, texts, config)
|
||||
|
||||
def parse_embedding_response(
|
||||
self, response_json: dict[str, Any]
|
||||
) -> list[list[float]]:
|
||||
return get_adapter_for_api_type("openai").parse_embedding_response(
|
||||
response_json
|
||||
)
|
||||
|
||||
def convert_generation_config(
|
||||
self, config: "LLMGenerationConfig", model: "LLMModel"
|
||||
) -> dict[str, Any]:
|
||||
adapter = self._get_delegate_adapter(model)
|
||||
return adapter.convert_generation_config(config, model)
|
||||
|
||||
Reference in New Issue
Block a user