mirror of
https://github.com/zhenxun-org/zhenxun_bot.git
synced 2026-10-05 11:49:59 +08:00
♻️ refactor!: 重构LLM服务架构并统一Pydantic兼容性处理 (#2002)
检查bot是否运行正常 / bot check (push) Waiting to run
Sequential Lint and Type Check / ruff-call (push) Waiting to run
Sequential Lint and Type Check / pyright-call (push) Blocked by required conditions
Release Drafter / Update Release Draft (push) Waiting to run
Force Sync to Aliyun / sync (push) Waiting to run
Update Version / update-version (push) Waiting to run
CodeQL Code Security Analysis / Analyze (${{ matrix.language }}) (none, javascript-typescript) (push) Has been cancelled
CodeQL Code Security Analysis / Analyze (${{ matrix.language }}) (none, python) (push) Has been cancelled
检查bot是否运行正常 / bot check (push) Waiting to run
Sequential Lint and Type Check / ruff-call (push) Waiting to run
Sequential Lint and Type Check / pyright-call (push) Blocked by required conditions
Release Drafter / Update Release Draft (push) Waiting to run
Force Sync to Aliyun / sync (push) Waiting to run
Update Version / update-version (push) Waiting to run
CodeQL Code Security Analysis / Analyze (${{ matrix.language }}) (none, javascript-typescript) (push) Has been cancelled
CodeQL Code Security Analysis / Analyze (${{ matrix.language }}) (none, python) (push) Has been cancelled
* ♻️ refactor(pydantic): 提取 Pydantic 兼容函数到独立模块 * ♻️ refactor!(llm): 重构LLM服务,引入现代化工具和执行器架构 🏗️ **架构变更** - 引入ToolProvider/ToolExecutable协议,取代ToolRegistry - 新增LLMToolExecutor,分离工具调用逻辑 - 新增BaseMemory抽象,解耦会话状态管理 🔄 **API重构** - 移除:analyze, analyze_multimodal, pipeline_chat - 新增:generate_structured, run_with_tools - 重构:chat, search, code变为无状态调用 🛠️ **工具系统** - 新增@function_tool装饰器 - 统一工具定义到ToolExecutable协议 - 移除MCP工具系统和mcp_tools.json --------- Co-authored-by: webjoin111 <455457521@qq.com>
This commit is contained in:
@@ -12,6 +12,7 @@ from zhenxun.services.llm.core import KeyStatus
|
||||
from zhenxun.services.llm.manager import (
|
||||
reset_key_status,
|
||||
)
|
||||
from zhenxun.services.llm.types import LLMMessage
|
||||
|
||||
|
||||
class DataSource:
|
||||
@@ -58,7 +59,7 @@ class DataSource:
|
||||
start_time = time.monotonic()
|
||||
try:
|
||||
async with await get_model_instance(model_name_str) as model:
|
||||
await model.generate_text("你好")
|
||||
await model.generate_response([LLMMessage.user("你好")])
|
||||
end_time = time.monotonic()
|
||||
latency = (end_time - start_time) * 1000
|
||||
return (
|
||||
|
||||
Reference in New Issue
Block a user