{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'22222, lizhen'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 测试，自定义tool，一个输入\n",
    "from langchain.agents import load_tools, AgentExecutor\n",
    "from langchain.agents import AgentType\n",
    "from pydantic import BaseModel, Field\n",
    "from langchain.tools import Tool\n",
    "\n",
    "class SingleInput(BaseModel):\n",
    "    msg: str = Field()\n",
    "\n",
    "def singleInputAddLizhen(msg):\n",
    "    return msg+\", lizhen\"\n",
    "\n",
    "singleTool = Tool.from_function(\n",
    "    name = \"SingleInput\",\n",
    "        func=singleInputAddLizhen,\n",
    "        description=\"数字相加\",\n",
    "        args_schema=SingleInput,\n",
    "        return_direct=False\n",
    ")\n",
    "\n",
    "singleTool.run(tool_input=\"22222\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'result=3'"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 测试，自定义tool，多个输入\n",
    "from langchain.tools import StructuredTool\n",
    "\n",
    "class MultiInput(BaseModel):\n",
    "    a: int = Field()\n",
    "    b: int = Field()\n",
    "    \n",
    "def multiInput(a, b) -> str:\n",
    "    return f\"result={a+b}\"\n",
    "\n",
    "multiTool = StructuredTool.from_function(\n",
    "    name = \"MultiInput\",\n",
    "        func=multiInput,\n",
    "        description=\"多个输入，数字相加\",\n",
    "        args_schema=MultiInput,\n",
    "        return_direct=True\n",
    ")\n",
    "\n",
    "multiTool.run(tool_input={\"a\":1, \"b\":2})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "\u001b[1m> Entering new AgentExecutor chain...\u001b[0m\n",
      "\u001b[32;1m\u001b[1;3m\u001b[0m\u001b[36;1m\u001b[1;3mresult=3\u001b[0m\n",
      "\n",
      "\u001b[1m> Finished chain.\u001b[0m\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'result=3'"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#自定义代理\n",
    "from langchain.agents import Tool, AgentExecutor, BaseSingleActionAgent\n",
    "from typing import List, Tuple, Any, Union\n",
    "from langchain.schema import AgentAction, AgentFinish\n",
    "\n",
    "class LizhenAgent(BaseSingleActionAgent):\n",
    "    \"\"\"自定义代理\"\"\"\n",
    "    \n",
    "    @property\n",
    "    def input_keys(self):\n",
    "        return [\"input\"]\n",
    "    \n",
    "    def plan(\n",
    "        self,\n",
    "        intermediate_steps: List[Tuple[AgentAction, str]],\n",
    "        **kwargs: Any,\n",
    "    ) -> Union[AgentAction, AgentFinish]:\n",
    "        return AgentAction(tool=\"MultiInput\", tool_input=kwargs[\"input\"], log=\"\")\n",
    "    \n",
    "    def aplan(\n",
    "        self,\n",
    "        intermediate_steps: List[Tuple[AgentAction, str]],\n",
    "        **kwargs: Any,\n",
    "    ) -> Union[AgentAction, AgentFinish]:\n",
    "        return AgentAction(tool=\"MultiInput\", tool_input=kwargs[\"input\"], log=\"\")\n",
    "    \n",
    "agent = LizhenAgent()\n",
    "tools = [multiTool, singleTool]\n",
    "agent_executor = AgentExecutor.from_agent_and_tools(agent=agent, tools=tools, verbose=True)\n",
    "agent_executor.run(input={\"a\":1, \"b\":2})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'BaseMultiActionAgent , run tool1'"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from langchain.tools import tool\n",
    "\n",
    "@tool\n",
    "def tool1(event) -> str:\n",
    "    \"\"\"这里必须添加描述，不然AssertionError: Function must have a docstring if description not provided.\"\"\"\n",
    "    return f\"BaseMultiActionAgent , {event}\"\n",
    "\n",
    "tool1.run(\"run tool1\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'BaseMultiActionAgent , run tool2'"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from langchain.tools import tool\n",
    "\n",
    "@tool\n",
    "def tool2(event) -> str:\n",
    "    \"\"\"这里必须添加描述，不然AssertionError: Function must have a docstring if description not provided.\"\"\"\n",
    "    return f\"BaseMultiActionAgent , {event}\"\n",
    "\n",
    "tool2.run(\"run tool2\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "\u001b[1m> Entering new AgentExecutor chain...\u001b[0m\n",
      "\u001b[32;1m\u001b[1;3m\u001b[0mMultiInput is not a valid tool, try another one.\u001b[32;1m\u001b[1;3m\u001b[0mSingleInput is not a valid tool, try another one.\u001b[32;1m\u001b[1;3m\u001b[0m\n",
      "\n",
      "\u001b[1m> Finished chain.\u001b[0m\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'bar'"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#自定义代理\n",
    "from langchain.agents import Tool, AgentExecutor, BaseMultiActionAgent\n",
    "from typing import List, Tuple, Any, Union\n",
    "from langchain.schema import AgentAction, AgentFinish\n",
    "\n",
    "class LizhenMultiAgent(BaseMultiActionAgent):\n",
    "    \"\"\"自定义代理\"\"\"\n",
    "    \n",
    "    @property\n",
    "    def input_keys(self):\n",
    "        return [\"input\"]\n",
    "    \n",
    "    def plan(\n",
    "        self,\n",
    "        intermediate_steps: List[Tuple[AgentAction, str]],\n",
    "        **kwargs: Any,\n",
    "    ) -> Union[AgentAction, AgentFinish]:\n",
    "        if len(intermediate_steps) == 0:\n",
    "            return [AgentAction(tool=\"MultiInput\", tool_input=kwargs[\"input\"], log=\"\"), AgentAction(tool=\"SingleInput\", tool_input=kwargs[\"input\"], log=\"\")]\n",
    "        else:\n",
    "            return AgentFinish(return_values={\"output\": \"bar\"}, log=\"\")\n",
    "    def aplan(\n",
    "        self,\n",
    "        intermediate_steps: List[Tuple[AgentAction, str]],\n",
    "        **kwargs: Any,\n",
    "    ) -> Union[AgentAction, AgentFinish]:\n",
    "        if len(intermediate_steps) == 0:\n",
    "            return [AgentAction(tool=\"MultiInput\", tool_input=kwargs[\"input\"], log=\"\"), AgentAction(tool=\"SingleInput\", tool_input=kwargs[\"input\"], log=\"\")]\n",
    "        else:\n",
    "            return AgentFinish(return_values={\"output\": \"bar\"}, log=\"\")\n",
    "    \n",
    "multiAgent = LizhenMultiAgent()\n",
    "multiTools = [tool1, tool2]\n",
    "multi_agent_executor = AgentExecutor.from_agent_and_tools(agent=multiAgent, tools=multiTools, verbose=True)\n",
    "multi_agent_executor.run(input={\"a\":1, \"b\":2})"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "base",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.10"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}