http://wiki.postgresql.org/wiki/Sudoku_puzzle

it is PostgreSQL select query and is quite fast – 120 seconds for Easter Monster compared to 4 seconds of my JavaScript solver and indeterminate amount of time of yours .

For comparison: my old JavaScript solution yields result after 4 seconds on Chrome 4. Look could at it here: http://wildwezyr-sudoku-solver.blogspot.com/

To facilitate population of sudoku table before execution of your code, I use this function:

CREATE OR REPLACE FUNCTION func_full_solve_sudoku(input_data text)
RETURNS varchar(81) as
$BODY$DECLARE
x int;
y int;
inp varchar(81);
d int;
c varchar(1);
BEGIN
inp := regexp_replace(regexp_replace(input_data, \’[\\n\\r]\’, \’\', \’g\’), \’[^1-9]\’, \’ \’, \’g\’);

truncate table sudoku;

for y in 0..8 loop
for x in 0..8 loop
d := null;
c := substring(inp, y * 9 + x + 1, 1);
if c is not null and length(c) = 1 and c <> \’ \’ then d := cast(c as int); end if;
if d is not null then
insert into sudoku (x_col, y_col, val, is_permanent) values (y + 1, x + 1, d, true);
end if;
end loop;
end loop;

perform func_solve_sudoku();

inp := \’\';

for y in 0..8 loop
for x in 0..8 loop
select cast(val as varchar(1)) into c from sudoku where x_col = y + 1 and y_col = x + 1;
inp := inp || coalesce(c, \’?\’);
end loop;
end loop;

return inp;
END;$BODY$
LANGUAGE \’plpgsql\’ VOLATILE;

Then for Easter Monster I just run:

select func_full_solve_sudoku(\’1…….2.9.4…5…6…7…5.9.3…….7…….85..4.7…..6…3…9.8…2…..1\’)

If you want to see how is my solution performing for this puzzle, use this matrix

1…….2
.9.4…5.
..6…7..
.5.9.3…
….7….
…85..4.
7…..6..
.3…9.8.
..2…..1

in textarea on the left and click [import->] and then [solve].