|
| 1 | +# FileAgent Module Documentation |
| 2 | + |
| 3 | +This document provides an overview of the `FileAgent` module, which is a part of the `patchwork` project. It consists of three Python files: `typed.py`, `FileAgent.py`, and `__init__.py`. This module is focused on processing file-related tasks using various tools and strategies, particularly with tabular data such as CSV files. |
| 4 | + |
| 5 | +## Overview |
| 6 | + |
| 7 | +The `FileAgent` is designed to handle file processing tasks, leveraging large language models and various tools to assist with file management, data extraction, and conversion tasks. It operates on tasks described with templates, using data inputs, and utilizes a strategy pattern for task execution. |
| 8 | + |
| 9 | +## Files |
| 10 | + |
| 11 | +### 1. typed.py |
| 12 | + |
| 13 | +This file defines the data types for inputs and outputs used by classes in the `FileAgent` module. It utilizes Python's typing extensions to ensure type safety. |
| 14 | + |
| 15 | +#### Inputs |
| 16 | + |
| 17 | +- **FileAgentInputs**: |
| 18 | + - `task`: A string defining the task to perform. |
| 19 | + - `base_path`: Optional string indicating the base directory for file operations. |
| 20 | + - `prompt_value`: A dictionary containing data for template rendering. |
| 21 | + - `max_llm_calls`: An integer specifying the maximum number of LLM calls. |
| 22 | + - `anthropic_api_key`: A string for API authentication. |
| 23 | + |
| 24 | +#### Outputs |
| 25 | + |
| 26 | +- **FileAgentOutputs**: |
| 27 | + - `request_tokens`: An integer representing the number of tokens in the request. |
| 28 | + - `response_tokens`: An integer for the number of tokens in the response. |
| 29 | + |
| 30 | +### 2. FileAgent.py |
| 31 | + |
| 32 | +This file contains the implementation of the `FileAgent` class, which extends the `Step` class. It utilizes an agentic strategy pattern to manage file-related tasks and uses various tools for file manipulation and data extraction. |
| 33 | + |
| 34 | +#### Inputs |
| 35 | + |
| 36 | +- **FileAgentInputs**: As described above. |
| 37 | + |
| 38 | +#### Outputs |
| 39 | + |
| 40 | +- **FileAgentOutputs**: As described above. |
| 41 | + |
| 42 | +#### Usage |
| 43 | + |
| 44 | +The `FileAgent` class is initialized with a set of inputs. It processes tasks using a predefined strategy and various tools such as: |
| 45 | +- `FindTextTool`: Searches for text within files. |
| 46 | +- `FileViewTool`: Views file contents. |
| 47 | +- `In2CSVTool`: Converts files to CSV format. |
| 48 | +- `CSVSQLTool`: Performs SQL operations on CSV files. |
| 49 | + |
| 50 | +The `run` method initiates the task execution within a temporary directory, returning results and usage statistics. |
| 51 | + |
| 52 | +### 3. __init__.py |
| 53 | + |
| 54 | +This file serves as the package initializer and is currently empty, indicating no specific initialization code is needed for the module. |
| 55 | + |
| 56 | +## Intended Usage |
| 57 | + |
| 58 | +Developers can utilize the `FileAgent` module to automate file processing tasks, particularly when dealing with structured data files. By configuring the inputs, files can be processed, converted, and summarized following custom tasks defined by users. It is particularly useful for tasks involving large volumes of tabular data and when integrated with advanced language models for context-aware operations. |
0 commit comments