pythonanswer-set-programmingclingo

How to parse program parts in Clingo Python?


I'm trying to run the following Blockworld Planning Program in python:

#include <incmode>.

#program base.
% Define
location(table).
location(X) :- block(X).
holds(F,0) :- init(F).

#program step(t).
% Generate
{ move(X,Y,t) : block(X), location(Y), X != Y } = 1.
% Test
:- move(X,Y,t), holds(on(A,X),t-1).
:- move(X,Y,t), holds(on(B,Y),t-1), B != X, Y != table.
% Define
moved(X,t) :- move(X,Y,t).
holds(on(X,Y),t) :- move(X,Y,t).
holds(on(X,Z),t) :- holds(on(X,Z),t-1), not moved(X,t).

#program check(t).
% Test
:- query(t), goal(F), not holds(F,t).

% Display
#show move/3.

#program base.
% Sussman Anomaly
%
block(b0).
block(b1).
block(b2).
%
% initial state:
%
%  2
%  0 1
% -------
%
init(on(b1,table)).
init(on(b2,b0)).
init(on(b0,table)).
%
% goal state:
%
%  2
%  1
%  0
% -------
%
goal(on(b1,b0)).
goal(on(b2,b1)).
goal(on(b0,table)).

I do this by making use of the following simple function

def print_answer_sets(program):
    control = clingo.Control()
    control.add("base", [], program)
    control.ground([("base", [])])
    control.configuration.solve.models = 0
    for model in control.solve(yield_=True):
        sorted_model = [str(atom) for atom in model.symbols(shown=True)]
        sorted_model.sort()
        print("Answer set: {{{}}}".format(", ".join(sorted_model)))

To which I pass a long triple-quote string of the program above. I.e.

print_answer_sets("""
#include <incmode>.

#program base.
% Define
location(table).
location(X) :- bl

...etc.
""")

This has generally worked with other clingo programs.

However, I can't seem to produce an answer set, and I get an error

no atoms over signature occur in program:
  move/3

Note that running this same program directly with clingo in the terminal or on their web interface works fine.

I believe this is because this string makes use of multiple "#program" statements and an "#include" statement, which I might not be parsing correctly (or at all) in print_answer_sets. Is there a way to properly handle program parts in python? I've tried looking at the docs and the user manual but its a bit dense.

Thanks!


Solution

  • I got help from my professor who managed to find a solution. I added some comments for myself and others and am posting below:

    import clingo
    
    asp_code_base = """
        #include <incmode>.
        % Define
        location(table).
        location(X) :- block(X).
        holds(F,0) :- init(F).
    
        block(b0).
        block(b1).
        block(b2).
    
        init(on(b1,table)).
        init(on(b2,b0)).
        init(on(b0,table)).
    
        goal(on(b1,b0)).
        goal(on(b2,b1)).
        goal(on(b0,table)).
    
        #show move/3.
    """
    asp_code_step = """
        % Generate
        { move(X,Y,t) : block(X), location(Y), X != Y } = 1.
    
        % Test
        :- move(X,Y,t), holds(on(A,X),t-1).
        :- move(X,Y,t), holds(on(B,Y),t-1), B != X, Y != table.
        % Define
        moved(X,t) :- move(X,Y,t).
        holds(on(X,Y),t) :- move(X,Y,t).
        holds(on(X,Z),t) :- holds(on(X,Z),t-1), not moved(X,t).
    """
    asp_code_check = """
        #external query(t).
    
        % Test
        :- query(t), goal(F), not holds(F,t).
    """
    
    
    def on_model(model):
        print("Found solution:", model)
    
    
    # this is an arbitrary upper limit set to ensure process terminates
    max_step = 5
    
    control = clingo.Control()
    control.configuration.solve.models = 1  # find one answer only
    
    # add each #program
    control.add("base", [], asp_code_base)
    control.add("step", ["t"], asp_code_step)
    control.add("check", ["t"], asp_code_check)
    
    # for grounding, we make use of a parts array
    parts = []
    parts.append(("base", []))
    control.ground(parts)
    ret, step = None, 1
    # try solving until max number of steps or until solved
    while step <= max_step and (step == 1 or not ret.satisfiable):
        parts = []
        # handle #external call
        control.release_external(clingo.Function("query", [clingo.Number(step - 1)]))
        parts.append(("step", [clingo.Number(step)]))
        parts.append(("check", [clingo.Number(step)]))
        control.cleanup()  # cleanup previous grounding call, so we can ground again
        control.ground(parts)
        # finish handling #external call
        control.assign_external(clingo.Function("query", [clingo.Number(step)]), True)
        print(f"Solving step: t={step}")
        ret = control.solve(on_model=on_model)
        print(f"Returned: {ret}")
        step += 1
    

    This outputs, as expected (omitting warning)

    Solving step: t=1
    Returned: UNSAT
    Solving step: t=2
    Returned: UNSAT
    Solving step: t=3
    Found solution: move(b2,table,1) move(b1,b0,2) move(b2,b1,3)
    Returned: SAT