mirror of
				https://github.com/hwchase17/langchain.git
				synced 2025-10-30 23:29:54 +00:00 
			
		
		
		
	Add other File Utilities, include - List Directory - Search for file - Move - Copy - Remove file Bundle as toolkit Add a notebook that connects to the Chat Agent, which somewhat supports multi-arg input tools Update original read/write files to return the original dir paths and better handle unsupported file paths. Add unit tests
		
			
				
	
	
		
			191 lines
		
	
	
		
			7.8 KiB
		
	
	
	
		
			Plaintext
		
	
	
	
	
	
			
		
		
	
	
			191 lines
		
	
	
		
			7.8 KiB
		
	
	
	
		
			Plaintext
		
	
	
	
	
	
| {
 | |
|  "cells": [
 | |
|   {
 | |
|    "cell_type": "markdown",
 | |
|    "metadata": {},
 | |
|    "source": [
 | |
|     "# File System Tools\n",
 | |
|     "\n",
 | |
|     "LangChain provides tools for interacting with a local file system out of the box. This notebook walks through some of them.\n",
 | |
|     "\n",
 | |
|     "Note: these tools are not recommended for use outside a sandboxed environment! "
 | |
|    ]
 | |
|   },
 | |
|   {
 | |
|    "cell_type": "markdown",
 | |
|    "metadata": {},
 | |
|    "source": [
 | |
|     "First, we'll import the tools."
 | |
|    ]
 | |
|   },
 | |
|   {
 | |
|    "cell_type": "code",
 | |
|    "execution_count": 1,
 | |
|    "metadata": {
 | |
|     "tags": []
 | |
|    },
 | |
|    "outputs": [],
 | |
|    "source": [
 | |
|     "from langchain.tools.file_management import (\n",
 | |
|     "    ReadFileTool,\n",
 | |
|     "    CopyFileTool,\n",
 | |
|     "    DeleteFileTool,\n",
 | |
|     "    MoveFileTool,\n",
 | |
|     "    WriteFileTool,\n",
 | |
|     "    ListDirectoryTool,\n",
 | |
|     ")\n",
 | |
|     "from langchain.agents.agent_toolkits import FileManagementToolkit\n",
 | |
|     "from tempfile import TemporaryDirectory\n",
 | |
|     "\n",
 | |
|     "# We'll make a temporary directory to avoid clutter\n",
 | |
|     "working_directory = TemporaryDirectory()"
 | |
|    ]
 | |
|   },
 | |
|   {
 | |
|    "cell_type": "markdown",
 | |
|    "metadata": {},
 | |
|    "source": [
 | |
|     "## The FileManagementToolkit\n",
 | |
|     "\n",
 | |
|     "If you want to provide all the file tooling to your agent, it's easy to do so with the toolkit. We'll pass the temporary directory in as a root directory as a workspace for the LLM.\n",
 | |
|     "\n",
 | |
|     "It's recommended to always pass in a root directory, since without one, it's easy for the LLM to pollute the working directory, and without one, there isn't any validation against\n",
 | |
|     "straightforward prompt injection."
 | |
|    ]
 | |
|   },
 | |
|   {
 | |
|    "cell_type": "code",
 | |
|    "execution_count": 2,
 | |
|    "metadata": {
 | |
|     "tags": []
 | |
|    },
 | |
|    "outputs": [
 | |
|     {
 | |
|      "data": {
 | |
|       "text/plain": [
 | |
|        "[CopyFileTool(name='copy_file', description='Create a copy of a file in a specified location', args_schema=<class 'langchain.tools.file_management.copy.FileCopyInput'>, return_direct=False, verbose=False, callback_manager=<langchain.callbacks.shared.SharedCallbackManager object at 0x1156f4350>, root_dir='/var/folders/gf/6rnp_mbx5914kx7qmmh7xzmw0000gn/T/tmpxb8c3aug'),\n",
 | |
|        " DeleteFileTool(name='file_delete', description='Delete a file', args_schema=<class 'langchain.tools.file_management.delete.FileDeleteInput'>, return_direct=False, verbose=False, callback_manager=<langchain.callbacks.shared.SharedCallbackManager object at 0x1156f4350>, root_dir='/var/folders/gf/6rnp_mbx5914kx7qmmh7xzmw0000gn/T/tmpxb8c3aug'),\n",
 | |
|        " FileSearchTool(name='file_search', description='Recursively search for files in a subdirectory that match the regex pattern', args_schema=<class 'langchain.tools.file_management.file_search.FileSearchInput'>, return_direct=False, verbose=False, callback_manager=<langchain.callbacks.shared.SharedCallbackManager object at 0x1156f4350>, root_dir='/var/folders/gf/6rnp_mbx5914kx7qmmh7xzmw0000gn/T/tmpxb8c3aug'),\n",
 | |
|        " MoveFileTool(name='move_file', description='Move or rename a file from one location to another', args_schema=<class 'langchain.tools.file_management.move.FileMoveInput'>, return_direct=False, verbose=False, callback_manager=<langchain.callbacks.shared.SharedCallbackManager object at 0x1156f4350>, root_dir='/var/folders/gf/6rnp_mbx5914kx7qmmh7xzmw0000gn/T/tmpxb8c3aug'),\n",
 | |
|        " ReadFileTool(name='read_file', description='Read file from disk', args_schema=<class 'langchain.tools.file_management.read.ReadFileInput'>, return_direct=False, verbose=False, callback_manager=<langchain.callbacks.shared.SharedCallbackManager object at 0x1156f4350>, root_dir='/var/folders/gf/6rnp_mbx5914kx7qmmh7xzmw0000gn/T/tmpxb8c3aug'),\n",
 | |
|        " WriteFileTool(name='write_file', description='Write file to disk', args_schema=<class 'langchain.tools.file_management.write.WriteFileInput'>, return_direct=False, verbose=False, callback_manager=<langchain.callbacks.shared.SharedCallbackManager object at 0x1156f4350>, root_dir='/var/folders/gf/6rnp_mbx5914kx7qmmh7xzmw0000gn/T/tmpxb8c3aug'),\n",
 | |
|        " ListDirectoryTool(name='list_directory', description='List files and directories in a specified folder', args_schema=<class 'langchain.tools.file_management.list_dir.DirectoryListingInput'>, return_direct=False, verbose=False, callback_manager=<langchain.callbacks.shared.SharedCallbackManager object at 0x1156f4350>, root_dir='/var/folders/gf/6rnp_mbx5914kx7qmmh7xzmw0000gn/T/tmpxb8c3aug')]"
 | |
|       ]
 | |
|      },
 | |
|      "execution_count": 2,
 | |
|      "metadata": {},
 | |
|      "output_type": "execute_result"
 | |
|     }
 | |
|    ],
 | |
|    "source": [
 | |
|     "toolkit = FileManagementToolkit(root_dir=str(working_directory.name)) # If you don't provide a root_dir, operations will default to the current working directory\n",
 | |
|     "toolkit.get_tools()"
 | |
|    ]
 | |
|   },
 | |
|   {
 | |
|    "cell_type": "markdown",
 | |
|    "metadata": {},
 | |
|    "source": [
 | |
|     "### Selecting File System Tools\n",
 | |
|     "\n",
 | |
|     "If you only want to select certain tools, you can pass them in as arguments when initializing the toolkit, or you can individually initialize the desired tools."
 | |
|    ]
 | |
|   },
 | |
|   {
 | |
|    "cell_type": "code",
 | |
|    "execution_count": 3,
 | |
|    "metadata": {
 | |
|     "tags": []
 | |
|    },
 | |
|    "outputs": [
 | |
|     {
 | |
|      "data": {
 | |
|       "text/plain": [
 | |
|        "[ReadFileTool(name='read_file', description='Read file from disk', args_schema=<class 'langchain.tools.file_management.read.ReadFileInput'>, return_direct=False, verbose=False, callback_manager=<langchain.callbacks.shared.SharedCallbackManager object at 0x1156f4350>, root_dir='/var/folders/gf/6rnp_mbx5914kx7qmmh7xzmw0000gn/T/tmpxb8c3aug'),\n",
 | |
|        " WriteFileTool(name='write_file', description='Write file to disk', args_schema=<class 'langchain.tools.file_management.write.WriteFileInput'>, return_direct=False, verbose=False, callback_manager=<langchain.callbacks.shared.SharedCallbackManager object at 0x1156f4350>, root_dir='/var/folders/gf/6rnp_mbx5914kx7qmmh7xzmw0000gn/T/tmpxb8c3aug'),\n",
 | |
|        " ListDirectoryTool(name='list_directory', description='List files and directories in a specified folder', args_schema=<class 'langchain.tools.file_management.list_dir.DirectoryListingInput'>, return_direct=False, verbose=False, callback_manager=<langchain.callbacks.shared.SharedCallbackManager object at 0x1156f4350>, root_dir='/var/folders/gf/6rnp_mbx5914kx7qmmh7xzmw0000gn/T/tmpxb8c3aug')]"
 | |
|       ]
 | |
|      },
 | |
|      "execution_count": 3,
 | |
|      "metadata": {},
 | |
|      "output_type": "execute_result"
 | |
|     }
 | |
|    ],
 | |
|    "source": [
 | |
|     "tools = FileManagementToolkit(root_dir=str(working_directory.name), selected_tools=[\"read_file\", \"write_file\", \"list_directory\"]).get_tools()\n",
 | |
|     "tools"
 | |
|    ]
 | |
|   },
 | |
|   {
 | |
|    "cell_type": "code",
 | |
|    "execution_count": 4,
 | |
|    "metadata": {},
 | |
|    "outputs": [
 | |
|     {
 | |
|      "data": {
 | |
|       "text/plain": [
 | |
|        "'File written successfully to example.txt.'"
 | |
|       ]
 | |
|      },
 | |
|      "execution_count": 4,
 | |
|      "metadata": {},
 | |
|      "output_type": "execute_result"
 | |
|     }
 | |
|    ],
 | |
|    "source": [
 | |
|     "read_tool, write_tool, list_tool = tools\n",
 | |
|     "write_tool.run({\"file_path\": \"example.txt\", \"text\": \"Hello World!\"})"
 | |
|    ]
 | |
|   },
 | |
|   {
 | |
|    "cell_type": "code",
 | |
|    "execution_count": 5,
 | |
|    "metadata": {},
 | |
|    "outputs": [
 | |
|     {
 | |
|      "data": {
 | |
|       "text/plain": [
 | |
|        "'example.txt'"
 | |
|       ]
 | |
|      },
 | |
|      "execution_count": 5,
 | |
|      "metadata": {},
 | |
|      "output_type": "execute_result"
 | |
|     }
 | |
|    ],
 | |
|    "source": [
 | |
|     "# List files in the working directory\n",
 | |
|     "list_tool.run({})"
 | |
|    ]
 | |
|   },
 | |
|   {
 | |
|    "cell_type": "code",
 | |
|    "execution_count": null,
 | |
|    "metadata": {},
 | |
|    "outputs": [],
 | |
|    "source": []
 | |
|   }
 | |
|  ],
 | |
|  "metadata": {
 | |
|   "kernelspec": {
 | |
|    "display_name": "Python 3 (ipykernel)",
 | |
|    "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": "3.11.2"
 | |
|   }
 | |
|  },
 | |
|  "nbformat": 4,
 | |
|  "nbformat_minor": 4
 | |
| }
 |