pdfchat 0.1.1

CLI app to quickly chat with your PDFs locally

pdfchat

CLI app to quickly chat with your PDFs locally

---

Overview

pdfchat is a Python-based CLI app that utilizes open-source LLMs using Ollama to quickly open a chat session with a PDF file. The app emphasizes speed as you can start chatting directly from the command line without needing to open a GUI or web interface and security as it runs locally on your machine and your data is not sent to any external servers.

Features

• Parse and extract content from PDF files.
• Chat with the content of the PDF using a conversational interface.
• Select specific pages of a PDF for focused interaction.
• Supports multiple language models via the Ollama platform.

Use Cases

• Students: Quickly find information in textbooks or lecture notes. Have a tutor teach you content specific to your study material.
• Researchers: Extract data from research papers or articles. Ask questions about specific sections of a paper.
• Educators: Create interactive learning materials. Use the app to generate quizzes and questions based on the content of a PDF.
• Business Professionals: Review contracts, reports, or any other documents. Get clarifications on specific sections.
• General Users: Quickly find information in any PDF document. Reduce hallucinations by grounding the model with the content of the PDF.

Installation

Prerequisites

pdfchat assumes you have Ollama installed. If you haven't installed it yet, follow the instructions on the Ollama website. Make sure you have the required models installed. You can check the available models on the Ollama website.

You also need to have marker-pdf installed. You can install it using pip:

pip install marker-pdf

You can also install marker-pdf using pipx.

Install pdfchat

To install pdfchat using pipx, run:

pipx install pdfchat

Usage

Basic Command

To start a chat session with a PDF file, run:

pdfchat <path_to_pdf>

Note: If it is your first time running the app, it will take a few seconds (depending on your internet speed) to download PDF parsing and OCR models.

Options

• --model or -m: Ollama model to use (must be installed). Defaults to the first model returned by ollama list. ( ex: llama3.1:8b)
• --url or -u: Ollama base URL. Defaults to 'http://localhost:11434'
• --pages or -p: The pdf pages to parse. Defaults to all pages. Ex: 1,2,3 or 1-3 or 1-3,5-7. Defaults to all pages.

Help

To see all available options, run:

pdfchat --help

This will display the help message with all available options and their descriptions.

Example

pdfchat example.pdf --model llama3.1:8b

Example

pdfchat example.pdf --model llama3.1:8b --pages 1-5

License

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

Author

Developed by Ibrahim Habib. You can contact me through LinkedIn or Email.