fangfufu/LLM-book-generator：基于 llama.cpp 的整本书自动生成工具

LLM Book Generator is a powerful Python-based command-line tool that automates the creation of entire books, from initial concept to a fully formatted DOCX manuscript. It leverages Large Language Models (LLMs) via llama.cpp to generate diverse content, including titles, chapter outlines, detailed section text, front and back matter, and even marketing materials.

Features

llama.cpp Support: Seamlessly integrates with llama.cpp server (llama-server).
Comprehensive Content Generation: Creates everything from titles and chapter outlines to front and back matter.
LaTeX Math Rendering: Converts LaTeX math expressions into high-quality images.
Advanced DOCX Styling: Produces professionally formatted DOCX files with customizable styles.
Intelligent Caching: Caches LLM responses to speed up subsequent runs.
Flexible Configuration: Uses a config.yaml file for easy customization.

Getting Started

Prerequisites

Python 3.8+
A LaTeX distribution (e.g., MiKTeX, MacTeX, TeX Live)
A running llama.cpp server instance (llama-server).

Installation

Clone the repository:

git clone <repository-url>
cd <repository-directory>

Install dependencies:
```
pip install -r requirements.txt
```
Set up environment (Optional): If using Hugging Face tokenizers requiring authentication, create a .env file in the project root:
```
HF_TOKEN="YOUR_HUGGING_FACE_TOKEN"
```

How to Run

Configure your book: Edit the config.yaml file to define your book's topic, author, style, and other parameters. For a detailed explanation of all the available options, see CONFIG.md.
Run the script:
```
python main.py
```
The generated book and marketing materials will be saved in the directory specified in config.yaml (default is "output").

Troubleshooting

LaTeX Math Rendering Issues: Ensure your LaTeX distribution is installed correctly and accessible from your system's PATH.
API Errors: Ensure that your llama.cpp server is running and accessible (default: http://localhost:8080).
File Permissions: Make sure the script has write access to the output and cache directories.

Contributing

Contributions are welcome! Please feel free to submit a pull request or open an issue.

License

This project is licensed under GPLv3. See the LICENSE file for details.