The New Sudoku Players' Forum

by **1to9only** » Sun Nov 18, 2018 4:24 pm

[Introduction: A C/C++ SudokuExplainer replacement was discussed many many years ago here: http://forum.enjoysudoku.com/team-project-c-or-c-explainer-like-rating-program-t30083.html.]

This is from one of my attempts to write a SudokuExplainer clone, the project was to explore how to implement a SudokuExplainer clone in C. This was done, and further developement has now stopped. I'm busy with other stuff.

I've added a simple(!) sudoku generator, tidied the code a bit, and released all under version 3 of the GNU General Public License (GPLv3):

"This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version."

"This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details."

If you do not agree with the terms of the GPLv3 license: https://www.gnu.org/licenses/, please do not download this program.

The program (seclone.exe) generates, solves and rates (a single) sudoku that needs singles only to solve. The singles solving code is converted from SudokuExplainer. Sudokus requiring other (not implemented) solving techniques are given a rating of 0.0.

The code has bits I've borrowed from elsewhere, but it's mostly my own. It's coded for speed(!), using global variables (no parameters in function calls) and similar code are repeated (no function calls). There are numerous areas where the code can be improved. As with free software, there is no support!

Some usage information is provided below (this is all the documentation!):

To generate a sudoku [-g generate option]:

Code: Select all: seclone.exe -g

To generate a sudoku, redirecting the output to a file (puzzle.txt):

Code: Select all: seclone.exe -g > puzzle.txt

Example puzzle.txt:

Code: Select all: 7..1.2...3..6.74....9.......85........6.1..7........8..5....2...3.5........84.61. ED=2.3/1.2/1.2

To solve/rate a sudoku [no option specified]:

Code: Select all: seclone.exe 7..1.2...3..6.74....9.......85........6.1..7........8..5....2...3.5........84.61.

Code: Select all: seclone.exe puzzle.txt

Example output:

Code: Select all: 748152369312697458569384721285739146496218573173465982851976234634521897927843615 ED=2.3/1.2/1.2

To solve a sudoku (with hints output) [-s solve option]:

Code: Select all: seclone.exe -s 7..1.2...3..6.74....9.......85........6.1..7........8..5....2...3.5........84.61.

Code: Select all: seclone.exe -s puzzle.txt

Example output:

Hidden Text: Show

The solution is 'exactly' as SudokuExplainer's solution, as verifed using the hints produced by SukakuExplainer from here: http://forum.enjoysudoku.com/help-with-sudoku-explainer-t6677-60.html#p271963.

To solve a sudoku (with pencilmarks output) [-v verbose option]:

Code: Select all: seclone.exe -v 7..1.2...3..6.74....9.......85........6.1..7........8..5....2...3.5........84.61.

Code: Select all: seclone.exe -v puzzle.txt

The output is best redirected to a file for viewing!

The code compiles with TCC (Tiny C Compiler: https://bellard.org/tcc/), using tcc version 0.92.27 downloaded from github: https://github.com/TinyCC/tinycc/releases.

To build seclone.exe:

Code: Select all: tcc seclone.c

ATTACHMENTS
seclone.zip (7,173 bytes)
seclone_src.zip (26,438 bytes)

by **1to9only** » Thu Jan 03, 2019 10:44 am

I see the source code has been downloaded a number of times, hope it's been useful. Here are some possible modifications:

To generate symmetric (180 rotational) puzzles, change the '// add a clue' section with the '// add 2 clues' code below:

Hidden Text: Show

For puzzles with diagonal symmetry, change the '// 180 rotational' line with the '// diagonal' line below:

Hidden Text: Show

For puzzles with antidiagonal symmetry, change the '// 180 rotational' line with the '// antidiagonal' line below:

Hidden Text: Show

by **1to9only** » Fri Jan 11, 2019 10:55 am

It's an interesting exercise to modify the source code from post#1 for 6x6 sudoku puzzles - after having recently modified SudokuExplainer to solve/rate 6x6 sudoku puzzles, see here.

For usage, see post#1 - replacing with s6clone.exe. I've made a few cosmetic changes to this program! In testing, I generated a batch of just under 6000 puzzles producing some (241) 9-clue puzzles, less (23) ED=2.3/1.2/1.2 rated puzzles, and even less (7) 8-clue puzzles. More testing, another batch of just over 5000 puzzles producing a lone (1) 12-clue ED=2.3/1.5/1.5 rated puzzle! All puzzles can be solved with singles only, e.g.

Code: Select all: 2...36....2..14...5....4.5...1...65. #12 ED=2.3/1.5/1.5

Hidden Text: Show

No source! As mentioned, it's an interesting project...

ATTACHMENT
s6clone.zip (5,213 bytes)

by **tarek** » Thu Jan 17, 2019 3:14 pm

1to9only wrote:It's an interesting exercise to modify the source code from post#1 for 6x6 sudoku puzzles - after having recently modified SudokuExplainer to solve/rate 6x6 sudoku puzzles

Thanks for your effort. You and others have certainly re-energized this forum with very helpful contributions recently

tarek

by **tarek** » Sun Jan 20, 2019 12:33 pm

I will try to contribute to your effort here 1to9only & if I have time to others efforts too

1to9only wrote:To generate symmetric (180 rotational) puzzles, change the '// add a clue' section with the '// add 2 clues' code below

I've modified it slightly;

180 symmetry code: Show

Replace \\random cell & \\180 rotational symmetry with the desired symmetry code

main diagonal symmetry: Show

anti diagonal symmetry: Show

Horizontal symmetry: Show

Vertical symmetry: Show

I removed the need to have different candidates for a symmetrical pairing. This will obviously cause an immediate conflict if they exist in the same Line or Box

I'll add the clunky 4-way symmetries following the same code later today

tarek

by **1to9only** » Sun Jan 20, 2019 3:45 pm

tarek wrote:I removed the need to have different candidates for a symmetrical pairing. This will obviously cause an immediate conflict if they exist in the same Line or Box

A pitfall to be wary of! Ideally, there should be additional code to avoid this conflict. ~~I'll leave this to the person using this code.~~

Tarek, thanks for your contributions.

[Edit] I admit this is not good code! Looking at past code, I did something like this [pseudocode]:

Code: Select all: select cell1 select cell2 if cell1 and cell2 empty select number1 from available cell1 candidates apply if cell1 not cell2 [i.e. not center cell] select number2 from avalible cell2 candidates apply

The issue highlighted by tarek does not arise!

by **1to9only** » Sun Jan 20, 2019 3:56 pm

It's another interesting exercise to modify the source code from post#1 for sudoku-X puzzles.

For usage, see post#1 - replacing with secloneX.exe. This program solves the very easy sudoku-x puzzle posted by m_b_metcalf in Puzzles related to Patterns Game patterns.

Code: Select all: .....5..3.9.....5...5.2.7......7..4...15..9..8....3..1..9.6.....4.8...7.2....7..6 pg 0130, x-sudoku, v. easy ..1..4..7....7..5.7..2..4....31....5.4.....6.5.....7....9..7..6.1..6....3..8..1.. pg 0133, x-sudoku, v. easy ..1..7.9.....4...13..8..7....86..4...2..1....4....3.....45....31......8..8....9.. pg 0141, x-sudoku, very easy ...2..3.8...1.8.5.....93..442....96...3...4...86....236..51.....3.4.6...2.1..7... pg 0142, x-sudoku, very easy .....9..6..2.1..9..6.2..4....4....5..9..5...75....81....1..56...8.9...2.7...3.... pg 0146, x-sudoku, very easy 7.....4...2..8......8..1..2.....45...4..9..6...25....75..4.........1..5...7..9..4 pg 0147, x-sudoku, very easy .....3..1..8....6..3..4.8.....2....6..6.51.3.1...9.2....7..2..9.9..7..4.2..4..1.. pg 0249, x-sudoku, v. easy .6.3...9.1...7...8..8...3..5..9.4.......6.......7.5..6..2...1..3...4...5.1...3.6. pg 0250, x-sudoku, v. easy

Except that the pg 0250 sudoku-X requires more that singles to solve!!

No source! This is another interesting (and tougher) project...

Undocumented option: This program (and the previous sudoku 6x6 program as well) has a -gg option that (after a new grid is generated) also dumps it in 'nice' format! E.g. secloneX.exe -gg

Code: Select all: 6..9...1........24............3.7...2...1..5..4.....671............2....8..7..... #X17 ED=2.3/1.2/1.2 +-------+-------+-------+ | 6 . . | 9 . . | . 1 . | | . . . | . . . | . 2 4 | | . . . | . . . | . . . | +-------+-------+-------+ | . . . | 3 . 7 | . . . | | 2 . . | . 1 . | . 5 . | | . 4 . | . . . | . 6 7 | +-------+-------+-------+ | 1 . . | . . . | . . . | | . . . | . 2 . | . . . | | 8 . . | 7 . . | . . . | +-------+-------+-------+

In the #XNN shown, X denotes a sudoku-X, and NN is the number of givens.

Enjoy!

ATTACHMENT
secloneX.zip (8,636 bytes)

by **tarek** » Sun Jan 20, 2019 9:04 pm

1to9only wrote:[Edit] I admit this is not good code! Looking at past code, I did something like this [pseudocode]:
Code: Select all
select cell1 select cell2 if cell1 and cell2 empty select number1 from available cell1 candidates apply if cell1 not cell2 [i.e. not center cell] select number2 from avalible cell2 candidates apply

The issue highlighted by tarek does not arise!

As I haven't tested nor looked at the full code yet, I played it safe with another replacement code:

Code: Select all: while ( (gen_clues[ c2] & mbits[ n2]) == 0) when it could be: while (!(gen_clues[ c2] & mbits[ n2]))

Here is the replacement code again!

180 symmetry code adjusted: Show

Code: Select all: // add 2 clues int addingclue = True; while ( addingclue == True ) { // to change symmetry type replace the following lines with the desired 2-way symmetry type int c1 = rand() % 41; // random cell int c2 = Sudsize - c1 - 1; // 180 rotational symmetry // to change symmetry type replace the above lines with the desired 2-way symmetry type if ( gen_grid[c1] == 0 ) // cell is empty? { int n1 = rand() % 9 + 1; int n2 = rand() % 9 + 1; // random number, 1-9 if ( c1 != c2 ) // not symmetrize to same cell { while ( (gen_clues[c2] & mbits[n2]) == 0) { n2 = rand() % 9 + 1; // making sure number is candidate } gen_grid[c2] = n2; // set clue gen_clues[c2] = 0; // clear clue for (int k = 1; k < 21; k++ ) // clear peers { gen_clues[peer[c2][k]] &= ~mbits[n2]; } numclues++; } while ( (gen_clues[c1] & mbits[n1]) == 0) { n1 = rand() % 9 + 1; // making sure number is candidate } gen_grid[c1] = n1; // set clue gen_clues[c1] = 0; // clear clue for (int k = 1; k < 21; k++ ) // clear peers { gen_clues[peer[c1][k]] &= ~mbits[n1]; } numclues++; addingclue = False; } }

And here is a 4-way symmetry (I lost interest after this and didn't continue

). Again trying to remain as faithful to the original code

Double orthogonal symmetry: Show

Code: Select all: // add 4 clues int addingclue = True; while ( addingclue == True ) { // to change symmetry type replace the following lines with the desired 4-way symmetry type int c3 = rand() % 5; // random cell (double orthogonal pool) 1st step int c1 = 9 * c3 + rand() % 5; // random cell (double orthogonal pool) 2nd step c3 = Sudsize - c1 - 1; // double orthogonal plane symmetry int c2 = 9 * (8 - (c1 / 9)) + c1 % 9; // double orthogonal plane symmetry int c4 = 9 * (c1 / 9) + (8 - c1 % 9); // double orthogonal plane symmetry c3 = Sudsize - c1 - 1; // double orthogonal plane symmetry // to change symmetry type replace the above lines with the desired 4-way symmetry type if ( gen_grid[c1] == 0 ) // cell is empty? { int n1 = rand() % 9 + 1; // random number, 1-9 int n2 = rand() % 9 + 1; // random number, 1-9 int n3 = rand() % 9 + 1; // random number, 1-9 int n4 = rand() % 9 + 1; // random number, 1-9 if ( c1 != c3 ) // no 4-way symmetry to same cell { if ( c1 != c2 && // no 2-way symmetry to same cell c1 != c4 ) // no 2-way symmetry to same cell { while ( (gen_clues[c2] & mbits[n2]) == 0) { n2 = rand() % 9 + 1; // making sure number is candidate } gen_grid[c2] = n2; // set clue gen_clues[c2] = 0; // clear clue for (int k = 1; k < 21; k++ ) // clear peers { gen_clues[peer[c2][k]] &= ~mbits[n2]; } numclues++; while ( (gen_clues[c4] & mbits[n4]) == 0) { n4 = rand() % 9 + 1; // making sure number is candidate } gen_grid[c4] = n4; // set clue gen_clues[c4] = 0; // clear clue for (int k = 1; k <21; k++ ) // clear peers { gen_clues[peer[c4][k]] &= ~mbits[n4]; } numclues++; } while ( (gen_clues[c3] & mbits[n3]) == 0) { n3 = rand() % 9 + 1; // making sure number is candidate } gen_grid[c3] = n3; // set clue gen_clues[c3] = 0; // clear clue for (int k = 1; k < 21; k++ ) // clear peers { gen_clues[peer[c3][k]] &= ~mbits[n3]; } numclues++; } while ( (gen_clues[c1] & mbits[n1]) == 0) { n1 = rand() % 9 + 1; // making sure number is candidate } gen_grid[c1] = n1; // set clue gen_clues[c1] = 0; // clear clue for (int k = 1; k < 21; k++ ) // clear peers { gen_clues[peer[c1][k]] &= ~mbits[n1]; } numclues++; addingclue = False; } }

by **1to9only** » Tue Jan 22, 2019 9:45 am

Tarek, thanks for your input. At some point in the future I may want to reissue the code with options to generate symmetric puzzles.

The code to update the grid and clues should be in a separate function, this will help with future code changes!

After the secloneX program, adding other variants is possible, NC would be easiest, the rest will require more coding! I have an s6cloneX program, which I'll post soon!

I've also looked into creating jigsaw puzzles, this requires changes to some of the static tables, here's a jigsaw (singles only) I created earlier:

Code: Select all: ........1..7..9..5...........36...48.5....8....1.3....1....7...2..4.3.....8...5.. #JA19 ED=2.3/1.2/1.2 +-----------+---------------+-------+ | . . . | . . . . | . 1 | | | | | | . . 7 | . . 9 . | . 5 | | +---+---+---+ +---+ | | . . | . . | . | . | . . . | | +---+ | +---+---+ | | . | . 3 6 | . . | . | 4 8 | +---+ +---+ | +-------+ | . 5 . | . . . | 8 . . | +-------+ | +---+ +---+ | . . | 1 | . 3 | . . . | . | | +---+---+ | +---+ | | 1 . . | . | . | 7 . | . . | | +---+ +---+---+---+ | | 2 . | . 4 . 3 | . . . | | | | | | . . | 8 . . . | 5 . . | +-------+---------------+-----------+

Lots of ideas to be kept busy with...

by **tarek** » Tue Jan 22, 2019 12:36 pm

I only modified the stuff that you clearly shown that can bedowloafef and changed by other people. So it was intended for the audience interested in downloading and possibly modifying it. A bit of fun. Doesn’t matter if clunky as long that it does the job.

From looking at the code I can see how the Jigsaw can easily replace the boxes by changing the definitions (tedious but easy). The NC can be done by doing 4 x 2 extra eliminations when you set a clue. It will be easy to do the cyclic NC and even toroidal eliminations too!

Great work!!

tarek

by **Mathimagics** » Tue Jan 22, 2019 2:58 pm

tarek wrote:... bedowloafef and changed by other people.

Someone's getting a little excited! :lol:

Ok looks like you guys are having some fun …

A question for 1to9only. Which do you pick first, the solution grid or the jigsaw pattern?

by **1to9only** » Tue Jan 22, 2019 5:21 pm

Mathimagics wrote:A question for 1to9only. Which do you pick first, the solution grid or the jigsaw pattern?

I read the jigsaw pattern first - from a Sumocue file (SumoCueV1=0J0...), update some table definitions as Tarek noted, and generate the jigsaw puzzle from an empty grid. The program reuses the code in post #1.

I decided to go with the Sumocue file format as there are a few around, and you can create/export your own in SudokuSolver.

by **Mathimagics** » Tue Jan 22, 2019 6:33 pm

.
Ok, that's like most generators that I have seen.

But consider the alternative … why not start with a Sudoku grid, and then pick a Jigsaw pattern that matches?

You then have a truly independent generator, no catalog of jigsaw patterns is required. No trust issues (have these patterns been vetted? etc)

Just a thought! 8-)

The New Sudoku Players' Forum

sudokuexplainer clone, SINGLESONLY, solver,generator, c src

sudokuexplainer clone, SINGLESONLY, solver,generator, c src

Re: sudokuexplainer clone, SINGLESONLY, solver,generator, c

6x6 sudoku puzzles, SINGLESONLY, solver,generator, NO src

Re: sudokuexplainer clone, SINGLESONLY, solver,generator, c

Re: sudokuexplainer clone, SINGLESONLY, solver,generator, c

Re: sudokuexplainer clone, SINGLESONLY, solver,generator, c

Re: secloneX, SINGLESONLY, solver,generator, NO src

Re: sudokuexplainer clone, SINGLESONLY, solver,generator, c

Re: sudokuexplainer clone, SINGLESONLY, solver,generator, c

Re: sudokuexplainer clone, SINGLESONLY, solver,generator, c

Re: sudokuexplainer clone, SINGLESONLY, solver,generator, c

Re: sudokuexplainer clone, SINGLESONLY, solver,generator, c

Re: sudokuexplainer clone, SINGLESONLY, solver,generator, c