Multi-Doc Q&A Chatbot

Small Streamlit app that uses Google Gemini (via the google-generativeai SDK) to answer questions about uploaded images (for example, invoices). The main app entrypoint is app.py.

Features

• Upload an image (jpg / jpeg / png) and ask questions about it.
• Choose a Gemini model from the UI (e.g. gemini-2.5-pro, gemini-2.5-flash).
• Simple Streamlit UI for quick testing and iteration.

Files

• app.py — main Streamlit app and functions: get_gemini_response, input_image_setup.
• requirements.txt — Python dependencies used by the project.
• .env — environment variables (not checked in). Add GOOGLE_API_KEY here.

Quickstart

1. Create a virtual environment and activate it (recommended):

   python3 -m venv env
   source env/bin/activate   # macOS / Linux (zsh / bash)

2. Install dependencies (choose one):

   • From requirements.txt:

     pip install -r requirements.txt

   • Or install core packages individually:

     pip install --upgrade google-generativeai streamlit pillow python-dotenv

3. Create a .env file at the project root and add your Google API key:

   GOOGLE_API_KEY=your_api_key_here

   The app loads this via load_dotenv() in app.py.

4. Run the app with Streamlit:

   streamlit run app.py

   Open the URL shown in the terminal (usually http://localhost:8501).

Usage

• Select a Gemini model from the dropdown.
• Enter a question in the text input box.
• Upload an image (jpg / jpeg / png) containing the document you want to ask about.
• Click "Generate Response" to call the Gemini model and display the answer.

Implementation notes

• input_image_setup converts the uploaded file into the MIME+bytes structure expected by the Gemini SDK.
• get_gemini_response creates a genai.GenerativeModel and calls generate_content with the text and image parts.
• The code expects GOOGLE_API_KEY to be available in environment variables.

Troubleshooting

• "No file uploaded": make sure you uploaded an image before clicking "Generate Response".
• API key errors: check that GOOGLE_API_KEY is set in .env and the key is valid for the Google Generative AI / Gemini service.
• Streamlit not launching: ensure dependencies are installed in the active environment and streamlit is available (which streamlit).
• If you see errors from the google-generativeai package, upgrade to the latest version:

  pip install --upgrade google-generativeai

Security & privacy

• Do not commit your .env or API keys to source control.
• Uploaded images are sent to Google Gemini for processing. Ensure you are comfortable sharing the content with the API.

Extending the project

• Add OCR pre-processing to extract text from images before sending to Gemini.
• Add structured extraction to output JSON of invoice fields (date, total, vendor, line items).
• Add more robust error handling and logging.