sudoku.py

import numpy as np

problem = [[6 , 7 , 0 , 0 , 0 , 0 , 0 , 0 , 2],
           [0 , 0 , 1 , 7 , 9 , 0 , 0 , 3 , 0],
           [0 , 5 , 0 , 0 , 6 , 2 , 0 , 0 , 0],
           [0 , 0 , 0 , 0 , 0 , 0 , 0 , 2 , 5],
           [0 , 0 , 0 , 3 , 0 , 8 , 0 , 0 , 0],
           [2 , 4 , 0 , 0 , 0 , 0 , 0 , 0 , 0],
           [0 , 0 , 0 , 1 , 3 , 0 , 0 , 5 , 0],
           [0 , 8 , 0 , 0 , 5 , 7 , 1 , 0 , 0],
           [5 , 0 , 0 , 0 , 0 , 0 , 0 , 9 , 6]]

np_problem = np.array(problem)

fixed_coordinates = [] # first getting the coordinates where fixed numbers are present
empty_coordinates = []
for i , sub_array in enumerate(problem) : 
    temp =  [[i , c] for c , sub_element  in enumerate(sub_array) if sub_element > 0]
    temp2 = [[i , j] for j , sub_element2 in enumerate(sub_array) if sub_element2 == 0]
    for z in temp : fixed_coordinates.append(z)
    for w in temp2 : empty_coordinates.append(w)

l , m , r  = [0 , 3 , 6] , [1 , 4 , 7] ,  [2 , 5 , 8]

avoid_dict = {idx : [] for idx in list(range(0 , len(empty_coordinates)))}

def generate_bounds(r , c) -> list:

            lower_bound_c = c if c in l else c - 1 if c in m else c - 2
            upper_bound_c = c + 3 if c in l else c + 2 if c in m else c + 1 

            lower_bound_r = r if r in l else r - 1 if r in m else r - 2
            upper_bound_r = r + 3 if r in l else r + 2 if r in m else r + 1
            
            return [lower_bound_c , upper_bound_c , lower_bound_r , upper_bound_r]


def backtrack(return_coordinates) : 
   
    n_r , n_c = empty_coordinates[empty_coordinates.index(return_coordinates) - 1]  # getting back element coordinates
     
    while [n_r , n_c] != empty_coordinates[empty_coordinates.index(return_coordinates) + 1]: 
            
        if np_problem[n_r , n_c] != 0 : 
              avoid_dict[empty_coordinates.index([n_r , n_c])].append(np_problem[n_r , n_c])

        fix_flag = False
        r , c = n_r , n_c
        for num in range(1 , 10) : 
                
                l_b_c , u_b_c , l_b_r , u_b_r = generate_bounds(r , c)
                
                if all([num not in np_problem[l_b_r : u_b_r , l_b_c : u_b_c] , num not in np_problem[r , :] , num not in np_problem[: , c]]) : 
                   if num not in avoid_dict.get(empty_coordinates.index([n_r , n_c])) :         
                           np_problem[n_r , n_c] , fix_flag = num , True
                           break
                           
        if fix_flag : n_r , n_c = empty_coordinates[empty_coordinates.index([n_r , n_c]) + 1] 
                
        if not fix_flag : 
              np_problem[n_r , n_c] = 0
              avoid_dict[empty_coordinates.index([n_r , n_c])].clear()
              n_r , n_c = empty_coordinates[empty_coordinates.index([n_r , n_c]) - 1]
              
        
for r in range(9) : 
    for c in range(9) : 
        
      if [r , c] not in fixed_coordinates : 

        fix_flag = False

        for num in range(1 , 10) : 
           
            l_b_c , u_b_c , l_b_r , u_b_r = generate_bounds(r , c)

            if all([num not in np_problem[l_b_r : u_b_r , l_b_c : u_b_c] , num not in np_problem[r , :] , num not in np_problem[: , c]]) : 
                       
                np_problem[r , c] , fix_flag = num , True
                break

        if not fix_flag : backtrack([r , c])

           
print(np_problem)