Scanlly / README.md
VsinghiAssociates's picture
Upload folder using huggingface_hub
8304ec0 verified
|
Raw History Blame Contribute Delete
3.4 kB
---
title: SCANLY
emoji: 📄
colorFrom: blue
colorTo: green
sdk: docker
app_port: 7860
---
# SCANLY
SCANLY is a Flask web app for:
- PDF OCR conversion with `ocrmypdf`
- Excel to CSV conversion with `pandas` and `openpyxl`
- PDF to Word conversion
- PDF splitting
- PDF page deletion
- PDF text summary
- Fast default OCR mode that skips pages that already contain text
- Streaming `.xlsx` conversion to reduce memory use on large workbooks
- Razorpay payment before paid processing
## Important: GitHub Pages will not run this app
This project uses Python and system OCR tools, so it cannot be hosted directly on GitHub Pages. GitHub Pages only serves static HTML/CSS/JavaScript.
Use GitHub as the source repository, then deploy the app to a Python/container host such as Hugging Face Spaces, Render, Fly.io, Railway, or any Docker-capable server.
## Deploy from GitHub to Hugging Face Spaces
1. Push this folder to a GitHub repository.
2. Create a Hugging Face Space.
3. Select **Docker** as the Space SDK.
4. In your GitHub repository settings, add:
- Repository secret `HF_TOKEN`: a Hugging Face access token with write access
- Repository variable `HF_SPACE`: your Space id, for example `yourname/document-converter-pro`
5. Push to the `main` branch, or run the **Deploy to Hugging Face Space** workflow manually from the GitHub Actions tab.
The included `Dockerfile` installs the required OCR system packages and starts `app.py` on port `7860`.
## Razorpay setup
PDF and paid Excel processing is payment gated. The app counts PDF pages or Excel rows, creates a Razorpay order when the file exceeds the free tier, waits for a successful Checkout payment, verifies the payment signature on the server, and only then starts processing.
Set these environment variables in your deployment host or Hugging Face Space secrets:
```bash
RAZORPAY_KEY_ID=rzp_test_or_live_key_id
RAZORPAY_KEY_SECRET=your_key_secret
```
OCR pricing is configured in `app.py` as `OCR_PRICE_PER_PAGE_PAISE = 500`, which is INR 5.00 per page.
Other PDF tools use `PDF_FREE_PAGES = 10` and `PDF_EXTRA_PAGE_PRICE_PAISE = 200`, which means the first 10 pages are free and additional pages cost INR 2.00 per page.
Excel to CSV uses `EXCEL_FREE_ROWS = 500`, `EXCEL_PRICE_PER_EXTRA_ROW_PAISE = 20`, and `EXCEL_MAX_ROWS = 50000`, which means the first 500 rows are free, additional rows cost INR 0.20 per row, and files above 50,000 rows are rejected.
The owner bypass key defaults to `215836`. To change it without editing code, set this optional secret:
```bash
OCR_BYPASS_KEY=your_private_access_key
```
## Run locally
Install system dependencies first:
- `ocrmypdf`
- `tesseract-ocr`
- `poppler-utils`
- `ghostscript`
- `qpdf`
Then run:
```bash
pip install -r requirements.txt
python app.py
```
Open `http://localhost:7860`.
## Optional AI API keys
The current bank statement parsers are rule-based and do not require API keys. If an AI fallback is added later, put keys in environment variables before starting Flask:
```powershell
$env:GROQ_API_KEY="your_groq_key"
$env:GEMINI_API_KEY="your_gemini_key"
$env:OPENROUTER_API_KEY="your_openrouter_key"
python run_server.py
```
## Files
- `app.py`: deployable Flask app
- `Dockerfile`: container build for deployment
- `requirements.txt`: Python dependencies
- `.github/workflows/deploy-huggingface.yml`: optional GitHub Actions deployment workflow