A pygame-based ubongo puzzle solver that visualizes a backtracking algorithm in real time.
The solver runs in a background thread and automatically captures and stores every valid solution as an image for later analysis.
So I can keep my braincells 😁
Ubongo puzzles look simple… until you actually try solving them programmatically.
This project is my way of exploring:
- Backtracking algorithms 🧠
- Real-time visualization 🎮
- Grid-based puzzle systems 🧩
-
Left Mouse Click – Fill a grid cell (toggle between clear and blocked depending on mode)
-
Mouse Drag – Paint multiple grid cells while dragging 🖱️
-
S Key – Switch fill mode:
- ⬛ Blocked (draw obstacles)
- ⬜ Clear (erase cells)
-
R Key – Restart the current board (reset everything)
-
Spacebar – Start the solver 🤖
-
Q Key – Quit the application 🚪
Every solved puzzle is automatically captured and saved as an image for later review.
When a solution is found, the current board state is exported as a screenshot and stored in a structured folder system based on the puzzle configuration.
The system encodes both:
- 🧩 The puzzle layout (grid state)
- 🧱 The block combination used
This ensures every solution is uniquely identifiable and reproducible.
📁 All outputs are saved under SCREENSHOT_SAVE_PATH/, organized by a generated puzzle hash derived from the board state.
📝 Each filename includes:
- A sequential solution index (e.g.
01.jpg) - The IDs of the blocks used in that specific run
Which also helps trace:
- Which block set produced which solution 🔍
- How different combinations behave over time 📊