agent.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from langchain.tools import tool\n",
    "from langchain import hub\n",
    "\n",
    "from langchain_openai import ChatOpenAI\n",
    "from langchain_core.prompts.prompt import PromptTemplate\n",
    "from langchain.chains import LLMChain"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "@tool\n",
    "def get_columns(file_name: str) -> list:\n",
    "    \"\"\"Get columns from xlsx file. Output list of column name\"\"\"\n",
    "    import pandas as pd\n",
    "\n",
    "    dataframe = pd.read_excel(file_name)\n",
    "    return dataframe.columns.tolist()\n",
    "\n",
    "@tool\n",
    "def generate_cleaning_code(actions:str, cols:list) -> str:\n",
    "    \"\"\"Generate cleaning code using open ai. Input is the actions to perform and the available list of columns in the excel file. Output is a string of code.\"\"\"\n",
    "        \n",
    "    openai = ChatOpenAI( model=\"gpt-4-0125-preview\",api_key=\"sk-\", temperature = 0)\n",
    "    template = \"\"\"The excel has the following columns {cols}.\n",
    "\n",
    "    Write a react code to do the following cleaning operations:\n",
    "\n",
    "    {question}\n",
    "\n",
    "    Use the following code template:\n",
    "\n",
    "    ```typescript\n",
    "    {code}\n",
    "        ```\n",
    "\n",
    "    make sure to convert the types of item to the appropriate type before performing any operations on them.\n",
    "    Also handle any null values with ? operator.\n",
    "        \n",
    "    Provide only the code without any explanations.     \n",
    "    \"\"\"\n",
    "\n",
    "\n",
    "    prompt = PromptTemplate.from_template(template)\n",
    "    prompt = prompt.partial(code = \"\"\" <NuvoImporter\n",
    "    licenseKey=\"Your License Key\"\n",
    "    settings={\\{\n",
    "        developerMode: true,\n",
    "        identifier: \"customer_data\",\n",
    "        columns: [\n",
    "        {\n",
    "            label: \"Customer Code\",\n",
    "            key: \"customer_code\",\n",
    "            columnType: \"string\",\n",
    "        },\n",
    "        {\n",
    "            label: \"Phone Number\",\n",
    "            key: \"phone_number\",\n",
    "            columnType: \"string\",\n",
    "        },\n",
    "        ],\n",
    "    }}\n",
    "    columnHooks={{\n",
    "        phoneNumber: (values) => {\n",
    "        values.map(([item, index]) => {\n",
    "            let phoneNumber = item;\n",
    "            if (/^[0]{{2}}/.test(phoneNumber)) {\n",
    "            phoneNumber = `+${item.slice(2, item.length)}`;\n",
    "            }\n",
    "            return [\n",
    "            {\n",
    "                value: phoneNumber,\n",
    "                info: [\n",
    "                {\n",
    "                    message:\n",
    "                    'We have automatically transformed your input into the correct format by converting \"00\" to \"+\".',\n",
    "                    level: \"info\",\n",
    "                },\n",
    "                ],\n",
    "            },\n",
    "            index,\n",
    "            ];\n",
    "        });\n",
    "        },\n",
    "    }}\n",
    "    >\n",
    "    Import data\n",
    "    </NuvoImporter>\"\"\")\n",
    "    prompt = prompt.partial(cols= cols)\n",
    "    llm_chain = LLMChain(llm=openai, prompt=prompt)\n",
    "    output = llm_chain.invoke(actions)\n",
    "    return output[\"text\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "prompt = hub.pull(\"hwchase17/react\")\n",
    "llm = ChatOpenAI(model = \"gpt-4-0125-preview\", api_key=\"sk-c\", temperature = 0)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "PromptTemplate(input_variables=['agent_scratchpad', 'input', 'tool_names', 'tools'], metadata={'lc_hub_owner': 'hwchase17', 'lc_hub_repo': 'react', 'lc_hub_commit_hash': 'd15fe3c426f1c4b3f37c9198853e4a86e20c425ca7f4752ec0c9b0e97ca7ea4d'}, template='Answer the following questions as best you can. You have access to the following tools:\\n\\n{tools}\\n\\nUse the following format:\\n\\nQuestion: the input question you must answer\\nThought: you should always think about what to do\\nAction: the action to take, should be one of [{tool_names}]\\nAction Input: the input to the action\\nObservation: the result of the action\\n... (this Thought/Action/Action Input/Observation can repeat N times)\\nThought: I now know the final answer\\nFinal Answer: the final answer to the original input question\\n\\nBegin!\\n\\nQuestion: {input}\\nThought:{agent_scratchpad}')"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "prompt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "d:\\Programs\\minconda3\\envs\\langchain\\Lib\\site-packages\\langchain_core\\_api\\deprecation.py:117: LangChainDeprecationWarning: The function `initialize_agent` was deprecated in LangChain 0.1.0 and will be removed in 0.2.0. Use Use new agent constructor methods like create_react_agent, create_json_agent, create_structured_chat_agent, etc. instead.\n",
      "  warn_deprecated(\n"
     ]
    }
   ],
   "source": [
    "# agent_executor = AgentExecutor(agent = agent, tools=[get_columns, generate_cleaning_code], handle_parsing_errors=True, verbose=True, max_iterations=3)\n",
    "# columns = agent_executor.invoke({\"input\": \"\"\"get the list of columns from crm_demo.xlsx and then pass all the columns to the cleaning_function along with the following actions to perform:\n",
    "#                                 1. City names to capital\\n2. Names to capital \n",
    "#                              Output just the code without any explanations\"\"\"})['output']\n",
    "\n",
    "from langchain.agents import initialize_agent\n",
    "from langchain.chains.conversation.memory import ConversationBufferWindowMemory\n",
    "from langchain.agents import AgentType\n",
    "\n",
    "conversational_memory = ConversationBufferWindowMemory(\n",
    "    memory_key=\"chat_history\", k=5, return_messages=True\n",
    ")\n",
    "\n",
    "agent = initialize_agent(\n",
    "    agent=AgentType.STRUCTURED_CHAT_ZERO_SHOT_REACT_DESCRIPTION,\n",
    "    tools=[get_columns, generate_cleaning_code],\n",
    "    llm=llm,\n",
    "    verbose=True,\n",
    "    max_iterations=3,\n",
    "    early_stopping_method=\"generate\",\n",
    "    memory=conversational_memory,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "d:\\Programs\\minconda3\\envs\\langchain\\Lib\\site-packages\\langchain_core\\_api\\deprecation.py:117: LangChainDeprecationWarning: The function `__call__` was deprecated in LangChain 0.1.0 and will be removed in 0.2.0. Use invoke instead.\n",
      "  warn_deprecated(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "\u001b[1m> Entering new AgentExecutor chain...\u001b[0m\n",
      "\u001b[32;1m\u001b[1;3mAction:\n",
      "```\n",
      "{\n",
      "  \"action\": \"get_columns\",\n",
      "  \"action_input\": {\n",
      "    \"file_name\": \"crm_demo.xlsx\"\n",
      "  }\n",
      "}\n",
      "```\u001b[0m\n",
      "Observation: \u001b[36;1m\u001b[1;3m['Company Identification Number ', 'Organisation', 'Mail', 'Number', 'Website', 'Street', 'City', 'Country', 'Amount', 'Department', 'Expenditures', 'Closing Date', 'First Name', 'Last Name']\u001b[0m\n",
      "Thought:\u001b[32;1m\u001b[1;3mAction:\n",
      "```\n",
      "{\n",
      "  \"action\": \"generate_cleaning_code\",\n",
      "  \"action_input\": {\n",
      "    \"actions\": \"City names to capital, Names to capital\",\n",
      "    \"cols\": [\"Company Identification Number \", \"Organisation\", \"Mail\", \"Number\", \"Website\", \"Street\", \"City\", \"Country\", \"Amount\", \"Department\", \"Expenditures\", \"Closing Date\", \"First Name\", \"Last Name\"]\n",
      "  }\n",
      "}\n",
      "```\u001b[0m\n",
      "Observation: \u001b[33;1m\u001b[1;3m```typescript\n",
      "<NuvoImporter\n",
      "  licenseKey=\"Your License Key\"\n",
      "  settings={{\n",
      "    developerMode: true,\n",
      "    identifier: \"customer_data\",\n",
      "    columns: [\n",
      "      {\n",
      "        label: \"Company Identification Number\",\n",
      "        key: \"company_identification_number\",\n",
      "        columnType: \"string\",\n",
      "      },\n",
      "      {\n",
      "        label: \"Organisation\",\n",
      "        key: \"organisation\",\n",
      "        columnType: \"string\",\n",
      "      },\n",
      "      {\n",
      "        label: \"Mail\",\n",
      "        key: \"mail\",\n",
      "        columnType: \"string\",\n",
      "      },\n",
      "      {\n",
      "        label: \"Number\",\n",
      "        key: \"number\",\n",
      "        columnType: \"string\",\n",
      "      },\n",
      "      {\n",
      "        label: \"Website\",\n",
      "        key: \"website\",\n",
      "        columnType: \"string\",\n",
      "      },\n",
      "      {\n",
      "        label: \"Street\",\n",
      "        key: \"street\",\n",
      "        columnType: \"string\",\n",
      "      },\n",
      "      {\n",
      "        label: \"City\",\n",
      "        key: \"city\",\n",
      "        columnType: \"string\",\n",
      "      },\n",
      "      {\n",
      "        label: \"Country\",\n",
      "        key: \"country\",\n",
      "        columnType: \"string\",\n",
      "      },\n",
      "      {\n",
      "        label: \"Amount\",\n",
      "        key: \"amount\",\n",
      "        columnType: \"number\",\n",
      "      },\n",
      "      {\n",
      "        label: \"Department\",\n",
      "        key: \"department\",\n",
      "        columnType: \"string\",\n",
      "      },\n",
      "      {\n",
      "        label: \"Expenditures\",\n",
      "        key: \"expenditures\",\n",
      "        columnType: \"number\",\n",
      "      },\n",
      "      {\n",
      "        label: \"Closing Date\",\n",
      "        key: \"closing_date\",\n",
      "        columnType: \"date\",\n",
      "      },\n",
      "      {\n",
      "        label: \"First Name\",\n",
      "        key: \"first_name\",\n",
      "        columnType: \"string\",\n",
      "      },\n",
      "      {\n",
      "        label: \"Last Name\",\n",
      "        key: \"last_name\",\n",
      "        columnType: \"string\",\n",
      "      },\n",
      "    ],\n",
      "  }}\n",
      "  columnHooks={{\n",
      "    city: (values) => {\n",
      "      return values.map(([item, index]) => {\n",
      "        const city = item?.toUpperCase() ?? \"\";\n",
      "        return [\n",
      "          {\n",
      "            value: city,\n",
      "            info: [\n",
      "              {\n",
      "                message: \"City name has been converted to uppercase.\",\n",
      "                level: \"info\",\n",
      "              },\n",
      "            ],\n",
      "          },\n",
      "          index,\n",
      "        ];\n",
      "      });\n",
      "    },\n",
      "    first_name: (values) => {\n",
      "      return values.map(([item, index]) => {\n",
      "        const firstName = item?.toUpperCase() ?? \"\";\n",
      "        return [\n",
      "          {\n",
      "            value: firstName,\n",
      "            info: [\n",
      "              {\n",
      "                message: \"First name has been converted to uppercase.\",\n",
      "                level: \"info\",\n",
      "              },\n",
      "            ],\n",
      "          },\n",
      "          index,\n",
      "        ];\n",
      "      });\n",
      "    },\n",
      "    last_name: (values) => {\n",
      "      return values.map(([item, index]) => {\n",
      "        const lastName = item?.toUpperCase() ?? \"\";\n",
      "        return [\n",
      "          {\n",
      "            value: lastName,\n",
      "            info: [\n",
      "              {\n",
      "                message: \"Last name has been converted to uppercase.\",\n",
      "                level: \"info\",\n",
      "              },\n",
      "            ],\n",
      "          },\n",
      "          index,\n",
      "        ];\n",
      "      });\n",
      "    },\n",
      "  }}\n",
      ">\n",
      "  Import data\n",
      "</NuvoImporter>\n",
      "```\u001b[0m\n",
      "Thought:\u001b[32;1m\u001b[1;3mAction:\n",
      "```\n",
      "{\n",
      "  \"action\": \"Final Answer\",\n",
      "  \"action_input\": \"<NuvoImporter\\n  licenseKey=\\\"Your License Key\\\"\\n  settings={{\\n    developerMode: true,\\n    identifier: \\\"customer_data\\\",\\n    columns: [\\n      {\\n        label: \\\"Company Identification Number\\\",\\n        key: \\\"company_identification_number\\\",\\n        columnType: \\\"string\\\",\\n      },\\n      {\\n        label: \\\"Organisation\\\",\\n        key: \\\"organisation\\\",\\n        columnType: \\\"string\\\",\\n      },\\n      {\\n        label: \\\"Mail\\\",\\n        key: \\\"mail\\\",\\n        columnType: \\\"string\\\",\\n      },\\n      {\\n        label: \\\"Number\\\",\\n        key: \\\"number\\\",\\n        columnType: \\\"string\\\",\\n      },\\n      {\\n        label: \\\"Website\\\",\\n        key: \\\"website\\\",\\n        columnType: \\\"string\\\",\\n      },\\n      {\\n        label: \\\"Street\\\",\\n        key: \\\"street\\\",\\n        columnType: \\\"string\\\",\\n      },\\n      {\\n        label: \\\"City\\\",\\n        key: \\\"city\\\",\\n        columnType: \\\"string\\\",\\n      },\\n      {\\n        label: \\\"Country\\\",\\n        key: \\\"country\\\",\\n        columnType: \\\"string\\\",\\n      },\\n      {\\n        label: \\\"Amount\\\",\\n        key: \\\"amount\\\",\\n        columnType: \\\"number\\\",\\n      },\\n      {\\n        label: \\\"Department\\\",\\n        key: \\\"department\\\",\\n        columnType: \\\"string\\\",\\n      },\\n      {\\n        label: \\\"Expenditures\\\",\\n        key: \\\"expenditures\\\",\\n        columnType: \\\"number\\\",\\n      },\\n      {\\n        label: \\\"Closing Date\\\",\\n        key: \\\"closing_date\\\",\\n        columnType: \\\"date\\\",\\n      },\\n      {\\n        label: \\\"First Name\\\",\\n        key: \\\"first_name\\\",\\n        columnType: \\\"string\\\",\\n      },\\n      {\\n        label: \\\"Last Name\\\",\\n        key: \\\"last_name\\\",\\n        columnType: \\\"string\\\",\\n      },\\n    ],\\n  }}\\n  columnHooks={{\\n    city: (values) => {\\n      return values.map(([item, index]) => {\\n        const city = item?.toUpperCase() ?? \\\"\\\";\\n        return [\\n          {\\n            value: city,\\n            info: [\\n              {\\n                message: \\\"City name has been converted to uppercase.\\\",\\n                level: \\\"info\\\",\\n              },\\n            ],\\n          },\\n          index,\\n        ];\\n      });\\n    },\\n    first_name: (values) => {\\n      return values.map(([item, index]) => {\\n        const firstName = item?.toUpperCase() ?? \\\"\\\";\\n        return [\\n          {\\n            value: firstName,\\n            info: [\\n              {\\n                message: \\\"First name has been converted to uppercase.\\\",\\n                level: \\\"info\\\",\\n              },\\n            ],\\n          },\\n          index,\\n        ];\\n      });\\n    },\\n    last_name: (values) => {\\n      return values.map(([item, index]) => {\\n        const lastName = item?.toUpperCase() ?? \\\"\\\";\\n        return [\\n          {\\n            value: lastName,\\n            info: [\\n              {\\n                message: \\\"Last name has been converted to uppercase.\\\",\\n                level: \\\"info\\\",\\n              },\\n            ],\\n          },\\n          index,\\n        ];\\n      });\\n    },\\n  }}\\n>\\n  Import data\\n</NuvoImporter>\"\n",
      "}\n",
      "```\u001b[0m\n",
      "\n",
      "\u001b[1m> Finished chain.\u001b[0m\n"
     ]
    }
   ],
   "source": [
    "actions = \"1. City names to capital\\n2. Names to capital\" \n",
    "response = agent(\n",
    "    f\"get the list of columns from crm_demo.xlsx and then pass all the columns to the cleaning_function along with the following actions to perform: {actions} Output just the code without any explanations\"\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'<NuvoImporter\\n  licenseKey=\"Your License Key\"\\n  settings={{\\n    developerMode: true,\\n    identifier: \"customer_data\",\\n    columns: [\\n      {\\n        label: \"Company Identification Number\",\\n        key: \"company_identification_number\",\\n        columnType: \"string\",\\n      },\\n      {\\n        label: \"Organisation\",\\n        key: \"organisation\",\\n        columnType: \"string\",\\n      },\\n      {\\n        label: \"Mail\",\\n        key: \"mail\",\\n        columnType: \"string\",\\n      },\\n      {\\n        label: \"Number\",\\n        key: \"number\",\\n        columnType: \"string\",\\n      },\\n      {\\n        label: \"Website\",\\n        key: \"website\",\\n        columnType: \"string\",\\n      },\\n      {\\n        label: \"Street\",\\n        key: \"street\",\\n        columnType: \"string\",\\n      },\\n      {\\n        label: \"City\",\\n        key: \"city\",\\n        columnType: \"string\",\\n      },\\n      {\\n        label: \"Country\",\\n        key: \"country\",\\n        columnType: \"string\",\\n      },\\n      {\\n        label: \"Amount\",\\n        key: \"amount\",\\n        columnType: \"number\",\\n      },\\n      {\\n        label: \"Department\",\\n        key: \"department\",\\n        columnType: \"string\",\\n      },\\n      {\\n        label: \"Expenditures\",\\n        key: \"expenditures\",\\n        columnType: \"number\",\\n      },\\n      {\\n        label: \"Closing Date\",\\n        key: \"closing_date\",\\n        columnType: \"date\",\\n      },\\n      {\\n        label: \"First Name\",\\n        key: \"first_name\",\\n        columnType: \"string\",\\n      },\\n      {\\n        label: \"Last Name\",\\n        key: \"last_name\",\\n        columnType: \"string\",\\n      },\\n    ],\\n  }}\\n  columnHooks={{\\n    city: (values) => {\\n      return values.map(([item, index]) => {\\n        const city = item?.toUpperCase() ?? \"\";\\n        return [\\n          {\\n            value: city,\\n            info: [\\n              {\\n                message: \"City name has been converted to uppercase.\",\\n                level: \"info\",\\n              },\\n            ],\\n          },\\n          index,\\n        ];\\n      });\\n    },\\n    first_name: (values) => {\\n      return values.map(([item, index]) => {\\n        const firstName = item?.toUpperCase() ?? \"\";\\n        return [\\n          {\\n            value: firstName,\\n            info: [\\n              {\\n                message: \"First name has been converted to uppercase.\",\\n                level: \"info\",\\n              },\\n            ],\\n          },\\n          index,\\n        ];\\n      });\\n    },\\n    last_name: (values) => {\\n      return values.map(([item, index]) => {\\n        const lastName = item?.toUpperCase() ?? \"\";\\n        return [\\n          {\\n            value: lastName,\\n            info: [\\n              {\\n                message: \"Last name has been converted to uppercase.\",\\n                level: \"info\",\\n              },\\n            ],\\n          },\\n          index,\\n        ];\\n      });\\n    },\\n  }}\\n>\\n  Import data\\n</NuvoImporter>'"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "response['output']"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "langchain",
   "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": "3.11.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}