This is a fork of phoenix_live_view_sudoku. The essential difference is that here clingo is used to do the solving (via python via erlport). read-my-gif shows the app in action.
Because solution is more-or-less immediate (less than three hundredths of a second — see clingo test runs below), I have changed the UI a little bit:
- Instead of “Start Solving” and “Interval”, just “Solve”;
- Validity feedback after every entered digit;
- Number of possible solutions shown.
An introduction to clingo and answer set programming is beyond the scope of this README, but Solving “Greater Than” Sudoku with Answer Set Programming (a blog post I wrote a few years ago) covers most of the basics.
As well as the dependencies that mix deps.get will pull in, this project requires clingo, python 3.6+, and clyngor. I haven’t tried clyngor-with-clingo, but that might simplify things a little.
One caveat: for maybe_valid_grid, clyngor needs to access clingo directly, not via clingo’s own python API. If clingo’s python module is squatting in the way, maybe_valid_grid will fail. The python tests (see below) will catch this.
$ mix deps.get $ cd apps/sudoku_web/assets/ $ npm install $ cd ../../../ # back to top level
$ python -m unittest apps/composer_cli/priv/python/test.py $ mix test
The test grids are also available in the asp/ directory for direct testing of the clingo rule file, e.g.:
$ cd apps/composer_cli/priv/asp/ $ clingo -n 100 sudoku9.lp test_gridbad.lp clingo version 5.4.0 Reading from sudoku9.lp ... Solving... UNSATISFIABLE Models : 0 Calls : 1 Time : 0.011s (Solving: 0.00s 1st Model: 0.00s Unsat: 0.00s) CPU Time : 0.011s $ clingo -n 100 sudoku9.lp test_gridok1.lp ... SATISFIABLE Models : 100+ # There are at least 100 possible solutions. Calls : 1 Time : 0.025s (Solving: 0.01s 1st Model: 0.00s Unsat: 0.00s) CPU Time : 0.024s $ clingo -n 100 sudoku9.lp test_gridok2.lp clingo version 5.4.0 Reading from sudoku9.lp ... Solving... Answer: 1 cellHasDigit(1,4,7) cellHasDigit(1,8,1) cellHasDigit(1,9,6) cellHasDigit(2,2,9) cellHasDigit(2,3,5) cellHasDigit(2,4,3) cellHasDigit(2,7,7) cellHasDigit(2,8,8) cellHasDigit(3,7,3) cellHasDigit(4,1,5) cellHasDigit(4,3,2) cellHasDigit(4,5,3) cellHasDigit(5,4,5) cellHasDigit(5,5,7) cellHasDigit(6,1,1) cellHasDigit(6,3,6) cellHasDigit(6,5,9) cellHasDigit(7,7,5) cellHasDigit(8,2,3) cellHasDigit(8,3,7) cellHasDigit(8,4,2) cellHasDigit(8,7,8) cellHasDigit(8,8,6) cellHasDigit(9,4,4) cellHasDigit(9,8,3) cellHasDigit(9,9,1) cellHasDigit(1,1,3) cellHasDigit(2,1,6) cellHasDigit(3,1,7) cellHasDigit(5,1,9) cellHasDigit(7,1,2) cellHasDigit(8,1,4) cellHasDigit(9,1,8) cellHasDigit(1,2,2) cellHasDigit(3,2,1) cellHasDigit(4,2,8) cellHasDigit(5,2,4) cellHasDigit(6,2,7) cellHasDigit(7,2,6) cellHasDigit(9,2,5) cellHasDigit(1,3,4) cellHasDigit(3,3,8) cellHasDigit(5,3,3) cellHasDigit(7,3,1) cellHasDigit(9,3,9) cellHasDigit(3,4,6) cellHasDigit(4,4,1) cellHasDigit(6,4,8) cellHasDigit(7,4,9) cellHasDigit(1,5,5) cellHasDigit(2,5,4) cellHasDigit(3,5,2) cellHasDigit(7,5,8) cellHasDigit(8,5,1) cellHasDigit(9,5,6) cellHasDigit(1,6,8) cellHasDigit(2,6,1) cellHasDigit(3,6,9) cellHasDigit(4,6,4) cellHasDigit(5,6,6) cellHasDigit(6,6,2) cellHasDigit(7,6,3) cellHasDigit(8,6,5) cellHasDigit(9,6,7) cellHasDigit(1,7,9) cellHasDigit(4,7,6) cellHasDigit(5,7,1) cellHasDigit(6,7,4) cellHasDigit(9,7,2) cellHasDigit(3,8,4) cellHasDigit(4,8,9) cellHasDigit(5,8,2) cellHasDigit(6,8,5) cellHasDigit(7,8,7) cellHasDigit(2,9,2) cellHasDigit(3,9,5) cellHasDigit(4,9,7) cellHasDigit(5,9,8) cellHasDigit(6,9,3) cellHasDigit(7,9,4) cellHasDigit(8,9,9) SATISFIABLE Models : 1 # There is exactly one possible solution. Calls : 1 Time : 0.019s (Solving: 0.00s 1st Model: 0.00s Unsat: 0.00s) CPU Time : 0.019s
$ mix phx.server
Navigating to: http://localhost:4000/demo
Some directions I might take this:
- MOAR puzzles!!!
- Richer feedback for the composer
- visualisation and UI for other ASP applications (scheduling, traffic control, music generation, protein design, etc.)