litellm-pdf-renamer

Modular, LLM-powered PDF renamer: batch-renames PDFs and updates their internal metadata (author, title, pubdate) using AI-inferred bibliographic details. Ensures filenames and metadata are always in sync for robust archival search and retrieval.

---

Table of Contents

1. About
2. Features
3. Installation
4. Configuration
5. Usage
6. License

---

About

litellm-pdf-renamer uses large language models (LLMs) to:

• Extract bibliographic metadata (author, title, pubdate) from the first pages of each PDF
• Rename files and update their internal metadata accordingly
• Guarantee that filenames and PDF metadata are always synchronized, even after repeated or batch renames

Designed for CLI batch processing, prompt customization, and easy extension to new LLMs/providers.

---

Features

• ✅ AI-powered renaming: Uses LLMs to infer author, title, and publication date for each PDF
• ✅ Metadata synchronization: After renaming, PDF metadata is always updated to match the filename (author/title/pubdate)
• ✅ Batch processing: Rename and tag all PDFs in a directory in one command
• ✅ Prompt-editable: Change extraction logic by editing prompt.txt
• ✅ Provider-agnostic: Works with Anthropic, OpenAI, Gemini, Perplexity, Llama, and any litellm-supported model
• ✅ Extensible: Modular CLI, ready for web UI or new LLMs

---

Installation

1. Clone the repo:

   git clone https://github.com/yourusername/litellm-pdf-renamer.git
   cd litellm-pdf-renamer

2. Install dependencies:

   pip install -r requirements.txt

---

Configuration

• Prompt: Edit prompt.txt to tune LLM extraction instructions
• Model/Provider: Any model supported by litellm; set API keys in .env (e.g., OPENAI_API_KEY, ANTHROPIC_API_KEY)

---

Usage

Basic usage:

python pdf_renamer.py <directory> --model <model-name> [--prompt prompt.txt]

Examples:

• OpenAI GPT-4 Turbo:

  python pdf_renamer.py ./sandbox --model gpt-4-turbo --prompt prompt.txt

• Anthropic Claude:

  python pdf_renamer.py ./sandbox --model claude-3-haiku-20240307 --prompt prompt.txt

After running, all PDFs in the target directory will be renamed and their internal metadata updated to match the inferred author, title, and publication date.

---

License

Distributed under the MIT License. See LICENSE for more information.