DocsReader

Not the 1000th PDF or document chatbot.
DocsReader is a chatbot designed to be your learning partner for up to 13 Python libraries, using official documentation as its source.

• Can it build your whole project with one prompt? No
• Can it help you learn new tools with accurate syntax and information? Yes

---

How It Works

Official documentations of Python libraries are scraped as vector embeddings to be used in Retrieval-Augmented Generation (RAG) for the AI chatbot. Meaning, the chatbot receives accurate information to prevent hallucination.

As latest Python libraries are updated frequently, mainstream chatbots like ChatGPT are prone to give outdated and deprecated code suggestions due to their slow updates. This chatbot solves this problems by directly scraping official documentation, without having to rely on inconsistent Web Search feature by other coding assistants.

---

Usage

You may run the app via Docker or local deployment.

via Docker

(Fixing)

via Local Deployment

1. Clone the repo

git clone https://github.com/JYNgithub/Docs-Reader-AI-Chatbot

2. Create and activate a venv in the project folder

python -m venv .venv
call .venv/Scripts/activate

or if you are on Linux/macOS:

python3 -m venv .venv
source .venv/bin/activate

3. Install dependencies

pip install -r requirements.txt

4. Run the app

streamlit run app.py

5. Access the web app at http://localhost:8501/

---

Supported Libraries

1. Chroma
2. Pillow
3. PySpark
4. Transformers
5. Tokenizers
6. Gradio
7. FastAPI
8. Psycopg2
9. Psycopg3
10. Redis
11. SQLAlchemy
12. DataBricks-SDK
13. NiceGUI

---

Limitations

As this is a hobby project, it has some limitations:

• Limited number of libraries supported
• Limited scalability
• Inconsistent updates of scraped documentation until a CI/CD process is established