SnipSum — OCR Snipping Tool for Instant Totals

A tiny, pure-Python desktop tool that lets you drag a rectangle on your screen (like Snipping Tool), runs OCR on the selection, extracts all numbers, treats values in parentheses as negative (accounting style), and pops up a clean result window with the total and the list of parsed numbers.

Why? Manually adding amounts from tables, invoices, dashboards, or web apps is slow and error-prone. SnipSum makes it one drag-and-copy.

---

Table of Contents

• Demo
• Features
• Tech Stack
• Folder Structure
• Quick Start
  • 1) Install Python packages
  • 2) Install Tesseract OCR (system)
  • 3) Run
• Usage
• How It Works
• Configuration & Customization
• Troubleshooting
• Tests
• Packaging (Optional)
• Roadmap
• Security & Privacy
• Contributing
• License
• Credits
• Why I built this

---

Demo

---

Features

• 🖼️ Snip like a pro: full-screen overlay; drag to select any region.
• 🔎 OCR on device: uses Tesseract OCR via pytesseract (no internet required).
• ➕ Smart parsing: handles 1,234.56, 1.234,56, $120.00, (45.67), and 120- (trailing minus).
• 🧮 Accounting aware: parentheses → negative values.
• 📋 Copy friendly: copy the total; inspect parsed numbers + their raw tokens.
• 🔐 Private by design: processing is 100% local.
• 🪟 Beginner-friendly code: small, modular files you can read in one sitting.

---

Tech Stack

• GUI: Tkinter (Python standard library)
• Screen capture: mss
• Images: Pillow (PIL)
• OCR: Tesseract via pytesseract
  (Optional swap: EasyOCR for a pure-pip OCR with no system EXE)

---

Folder Structure

snipsum_py/
├─ app.py                 # Entry point & glue
├─ ui.py                  # Overlay snipper + results window (Tkinter + MSS + PIL)
├─ ocr.py                 # OCR helpers (Tesseract; lightweight preprocessing)
├─ parse.py               # Number extraction & summing (Decimal-based)
├─ requirements.txt
└─ tests/
   └─ test_parse.py       # A few sanity tests for the parser

---

Quick Start

1) Install Python packages

pip install -r requirements.txt

requirements.txt:

mss
pillow
pytesseract

2) Install Tesseract OCR (system)

• Windows:

  • Install the 64-bit build (commonly “UB Mannheim” build).
  • Default path: C:\Program Files\Tesseract-OCR\tesseract.exe
  • You can skip editing PATH; SnipSum tries common paths and also allows hard-coding.

• macOS (Homebrew):

  brew install tesseract

• Ubuntu/Debian:

  sudo apt-get update
  sudo apt-get install tesseract-ocr

Verify (optional):

tesseract -v

or in Python:

import pytesseract; print(pytesseract.get_tesseract_version())

3) Run

python app.py

• Your screen will dim; drag to select a region.
• Release the mouse → a small window appears with the total and parsed numbers.
• Buttons: Copy Total, Save Snapshot, Re-Snip.

---

Usage

• Cancel snip: Press Esc while the overlay is active.
• Re-Snip: Use the “Re-Snip” button on the result window.
• Save snapshot: Saves your cropped selection as a PNG file.

---

How It Works

1. Overlay & Capture (ui.py)
   Full-screen Tkinter window shows a screenshot via mss. Drag a rectangle → crop to a PIL image.

2. OCR (ocr.py)
   Light preprocessing (grayscale + autocontrast). pytesseract runs with a character whitelist optimized for numeric/currency characters.

3. Parsing & Sum (parse.py)

   • Regex finds numeric tokens (supports thousands/decimal separators, currencies, parentheses, trailing minus).
   • Detects decimal separator via the rightmost-separator heuristic (EU formats handled).
   • Converts using decimal.Decimal (avoids float rounding errors).
   • Returns a total and the list of (Decimal value, raw_token).

---

Configuration & Customization

• Tesseract path (Windows):
  Auto-detection tries common install paths. If needed, set explicitly in ocr.py:

  import pytesseract
  pytesseract.pytesseract.tesseract_cmd = r"C:\Program Files\Tesseract-OCR\tesseract.exe"

• Character whitelist: tweak for your data:

  # ocr.py
  WHITELIST = "0123456789.,-()$€£₹% RsPKR"

• Ignore percentages:
  The parser skips tokens ending with %. To include them, adjust the PCT regex logic in parse.py.

• Parentheses as negative:
  Implemented in _to_decimal() in parse.py. Change the rule if your workflow differs.

---

Troubleshooting

pytesseract.pytesseract.TesseractNotFoundError

• Ensure Tesseract is installed (see above).
• On Windows, verify the path:

  "C:\Program Files\Tesseract-OCR\tesseract.exe" -v
  

• Set pytesseract.pytesseract.tesseract_cmd to that path in ocr.py.

Blurry OCR / Low accuracy

• Zoom your source (e.g., browser zoom 125–150%) before snipping.
• Increase font size if possible.
• Keep the selection tight around the numbers to reduce noise.

EU formats not summed correctly

• The parser infers decimal separator from tokens. If sources are mixed (, and . used inconsistently), normalize the source or adjust _decide_decimal_sep().

Multiple monitors / High DPI

• mss captures the virtual screen (all monitors). If coordinates seem off with unusual DPI scaling, try 100–150% OS scaling.

---

Tests

pip install pytest
pytest -q

tests/test_parse.py includes:

• Parentheses → negative
• EU decimal comma support
• Date-like strings ignored

Add more tests as you encounter new formats.

---

Packaging (Optional)

Create a standalone Windows EXE:

pip install pyinstaller
pyinstaller --noconsole --onefile --name SnipSum app.py

Note: Tesseract remains an external dependency; your EXE still requires Tesseract installed on the target machine.

---

Roadmap

• v1.1 (Python-only)

  • Global hotkey to launch/re-launch snipper.
  • Include/exclude checkboxes for individual values.
  • CSV export (numbers + raw token).
  • Session history with thumbnails.
  • Optional EasyOCR backend (pure-pip; no system Tesseract).

• v2 (ideas)

  • Column-mode (sum a visually aligned column).
  • Currency-only filter & per-currency totals.
  • Regex presets (include/exclude patterns).
  • PDF/image file import (no snip required).

---

Security & Privacy

• All processing happens locally on your machine.
• No data is sent anywhere by default.
• A snapshot is saved only if you click Save Snapshot.

---

Contributing

Pull requests welcome! Good first issues:

• Add a global hotkey launcher.
• Add a “Copy CSV” button with value + raw token.
• Toggle “ignore percentages” in the UI.
• Improve EU format detection heuristics.
• Add column-mode.

---

License

This project is licensed under the MIT License.
Create a LICENSE file with:

MIT License

Copyright (c) 2025 Muhammad Ibrahim

Permission is hereby granted, free of charge, to any person obtaining a copy...

---

Credits

• Tesseract OCR
• pytesseract
• Pillow
• python-mss
• Tkinter (Python standard library)

---

Why I built this

At a previous job, I had to total many order amounts by hand from screenshots and web pages. It was slow and error-prone. SnipSum scratches that itch: select an area on screen, and it instantly totals all the amounts—treating accounting parentheses as negatives. It’s a beginner-friendly, pure-Python desktop app you can clone, read, and improve in an afternoon.