def solve(assignment, domains):
"""Solve a CSP problem. Starts the backtrack process"""
global backtracks, splits, constraints
backtracks = splits = 0
# Fast deepcopy the initial state
new_domains = {k: [v for v in values] for k,values in domains.iteritems()}
new_assignment = {k: v for k,v in assignment.iteritems()}
# Use initial assignment to prune domains
assigned_vars = assignment.keys()
while assigned_vars != False and len(assigned_vars) > 0:
assigned_vars = propagate(new_domains, new_assignment, constraints, assigned_vars)
# Start the solver
return bt(new_assignment, new_domains, [])
def bt(assignment, domains, last_var):
"""Performs a CSP solving with the Backtrack algorithm"""
global backtracks, splits, constraints
# Fast deepcopy the CSP state
new_domains = {k: [v for v in values] for k,values in domains.iteritems()}
new_assignment = {k: v for k,v in assignment.iteritems()}
# Prune the domains connected with the assigned variables
assigned_vars = [last_var]
while assigned_vars != False and len(assigned_vars) > 0:
assigned_vars = propagate(new_domains, new_assignment, constraints, assigned_vars)
# Check for inconsistency
if (assigned_vars == False):
return False
# Pick the next variable
var = hr.pick_variable(new_domains, constraints, method=variable_heuristic)
#sudoku.print_board_domains(domains)
# Check if a solution is found done
if var == None:
return new_assignment
# Backtracking for each value
values = hr.pick_values(var, domains, constraints, method=value_heuristic)
for value in values:
new_assignment[var] = value
res = bt(new_assignment, new_domains, var)
if res != False:
return res
backtracks += 1
splits += 1
return False
