Excel → Calculations → Results → Visuals

A small, production-ready template to:

1. Upload an Excel
2. Run custom formulas with pandas
3. Download a results Excel
4. Ask ChatGPT for chart ideas and render charts with matplotlib (no model code execution)

Quickstart

1) Backend

cd backend
python -m venv .venv && source .venv/bin/activate  # Windows: .venv\Scripts\activate
pip install -r requirements.txt
cp ../.env.example .env  # and set your OPENAI_API_KEY
uvicorn main:app --reload --port 8000

2) Web (static)

Open web/index.html directly in your browser, or serve it with any static server.

By default CORS allows http://localhost:5173. Adjust ALLOWED_ORIGINS in .env or edit the list in main.py.

Customize formulas

Edit backend/processors/formulas.py → apply_all(). Replace the sample logic with your actual business rules, calculations, and summaries. Return a dict of sheet_name -> DataFrame, which will be written to the output Excel.

How visuals work

• We detect a simple schema (numeric, categorical, time columns).
• We ask OpenAI to propose ~2-4 chart specs as JSON.
• We validate those specs and render with matplotlib.
• We never execute code from the model—only our own renderer.

API

POST /process

Form-data:

• file: Excel file (.xlsx / .xls)
• sheet (optional): Sheet name

Returns: an Excel file with your result sheets.

POST /visualize

Form-data:

• file: Excel file (.xlsx / .xls)
• sheet (optional): Sheet name

Returns: a ZIP containing chart_*.png and chart_plan.json.

Docker (optional)

# Simple dev Dockerfile
FROM python:3.11-slim
WORKDIR /app
COPY backend/requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt
COPY backend .
ENV PORT=8000
EXPOSE 8000
CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8000"]

Build & run:

docker build -t excel-viz .
docker run -it --rm -p 8000:8000 -e OPENAI_API_KEY=your_key excel-viz

Notes

• For Excel engine we use openpyxl.
• If your files are very large, consider streaming chunked processing or parquet conversion.
• For security, this app does not execute code returned by the LLM.

Integrate your own calculations

1. Open backend/processors/user_main.py and paste your complete function analizar_inventarios(archivo_entrada, output_path).

2. Save and run the backend. When you POST to /process, the server will call your function and return its Excel.

3. If the module or function is missing, the server falls back to the default processors/formulas.py pipeline.