/*

  Railroad switching problem in Picat.

  Problem fromStefan Edelkamp, Stefan Schrödl:
  "Heuristic Search: Theory and Applications, page ff"

  Note: This use a graph representation.

  Model created by Hakan Kjellerstrand, hakank@gmail.com
  See also my Picat page: http://www.hakank.org/picat/

*/

import planner.
import bplan.

main => go.

% planner
go =>
   initial_state(Init),
   time(planner.best_plan(Init,Plan)),
   print_plan(Plan),
   write(len=Plan.length),nl.

%
% planner for optimal path 
% and then bplan for all optimal paths
% 
go2 =>
   % first find the optimal plan
   initial_state(Init),
   time(planner.best_plan(Init,Plan1)),

   % then find all optimal plans
   P = new_list(Plan1.length),
   All=findall(P,bplan.plan(P)),
   foreach(Plan in All) print_plan(Plan) end,
   writeln(len=All.length),
   nl.

print_plan(Plan) => 
   foreach([From,To] in Plan) printf("%w -> %w\n", From,To) end, 
   nl.


index(-)
initial_state([e,a,b]).

index(-)
final([e,b,a]).

% for bplan
index(-)
goal_state([e,b,a]).


action(From,To,Move,Cost) ?=>
   graph2(From,To,_),
   Move = [From,To],
   Cost = 1.


% for bplan
legal_move(From,Move,To) =>
   action(From,To,Move,_).


%
% This is the graph from figure 1.4 (page 13) in "Heuristic Search..."
% (This is an directed graph.)
index(-,-,-)
graph([e,a,b],[b,a,e],eab_bae).
graph([b,a,e],[b,e,a],bae_bea).
graph([b,a,e],[a,e,b],bae_aeb).
graph([b,e,a],[a,b,e],bea_abe).
graph([a,b,e],[a,e,b],abe_aeb).
graph([a,b,e],[e,b,a],abe_eba).

% Make the graph undirected
graph2(From,To,Move) ?=> 
  graph(From,To,Move).

graph2(From,To,Move) ?=> 
  graph(To,From, Move).