Merge pull request #36 from graille/Thibault

Update

Author: graille

Glouton.py

@@ -0,0 +1,363 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+import sys
+import os
+import numpy as np
+import heapq
+
+# Set the syspath
+f_name = "main.py"
+a_path = str(os.path.abspath(__file__))
+new_sys_entry = a_path[0:len(a_path) - len(f_name)]
+
+print("Add " + new_sys_entry + "to sys path")
+sys.path.insert(0, new_sys_entry)
+
+# Initialize vars
+TEAM_NAME = "JoJo Le Glouton"
+maze = None
+
+class PriorityQueue(object):
+    def __init__(self, heap = []):
+        heapq.heapify(heap)
+        self.heap = heap
+
+    def insert(self, node, priority = 0):
+        heapq.heappush(self.heap, (priority, node))
+
+    def pop(self):
+        return heapq.heappop(self.heap)[1]
+
+
+class Dijkstra:
+    def __init__(self, maze, origin = None, goal = None):
+        self.graph = maze
+        self.setOrigin(origin)
+        self.setGoal(goal)
+
+    def setOrigin(self, n):
+        if n:
+            self.origin = n
+
+    def setGoal(self, n):
+        if isinstance(n, list):
+            self.goal = n.copy()
+        else:
+            self.goal = n
+
+    def clear(self):
+        self.pathArray = {}
+        self.dist = {}
+        self.Q = PriorityQueue()
+
+        self.dist[self.origin] = 0
+
+        for node in self.graph.nodes:
+            if node != self.origin:
+                self.pathArray[node] = None
+                self.dist[node] = np.inf
+
+        self.Q.insert(self.origin, self.dist[self.origin])
+
+    def process(self):
+        self.algorithm()
+
+    def algorithm(self):
+        if self.origin != self.goal:
+            self.clear()
+
+            while self.Q.heap:
+                u = self.Q.pop()
+                for v in self.graph.getNeighbors(u):
+                    alt = self.dist[u] + self.graph.getDistance(u, v)
+                    if alt < self.dist[v]:
+                        self.dist[v] = alt
+                        self.pathArray[v] = u
+                        self.Q.insert(v, alt)
+        else:
+            pass
+
+    def reconstructPath(self, node):
+        #litteral_path = ""
+        total_distance = 0
+        current = node
+        real_path = []
+        while current != self.origin:
+            new = self.pathArray[current]
+
+            #litteral_path += self.graph.getMove(new, current)
+            total_distance += self.graph.getDistance(current, new)
+            real_path.append(current)
+
+            current = new
+
+        real_path.append(current)
+        return (total_distance, real_path[::-1])
+
+    def getResult(self, node = None):
+        if self.goal != self.origin:
+            if (not node) and (self.goal):
+                return self.reconstructPath(self.goal)
+            else:
+                return self.reconstructPath(node)
+        else:
+            return (0, [])
+
+
+class Maze:
+    def __init__(self, mazeMap, mazeWidth, mazeHeight):
+        self.mazeMap = mazeMap
+        self.mazeWidth = mazeWidth
+        self.mazeHeight = mazeHeight
+
+        self.NB_NODES = self.mazeWidth * self.mazeHeight
+        self.nodes = list(self.mazeMap.keys())
+
+        # Init matrixMap
+        self.matrixMap = {}
+
+        # Init metagraph
+        self.distanceMetagraph = {} # matrice des distances du métagraph sous forme de dictionnaire
+        self.pathMetagraph = {} # matrice des chemins du métagraph sous forme de dictionnaire
+
+    def convertToMatrix(self):
+        """
+        Prend en argument
+        Renvoie une un dictionnaire où les clefs sont les clefs sont des cases et où
+        np.inf signifie qu'il n'y a pas de passage direct entre les deux cases,
+        n>0 signifie qu'il y a passage en n coups et 0 que l'on reste sur la même case
+        """
+        for n in self.nodes:
+            self.matrixMap[n] = {}
+
+        for n1 in self.nodes:
+            for n2 in self.nodes:
+                if n1 == n2:
+                    self.matrixMap[n1][n1] = 0
+                elif n2 in self.mazeMap[n1]:
+                    self.matrixMap[n1][n2] = self.mazeMap[n1][n2]
+                    self.matrixMap[n2][n1] = self.mazeMap[n2][n1]
+                else:
+                    self.matrixMap[n1][n2] = np.inf
+                    self.matrixMap[n2][n1] = np.inf
+
+    def reversePath(self, path):
+        """
+        Inverse un chemin symétriquement par rapport a la diagonale
+        :param path:
+        :return:
+        """
+        r = ""
+        for l in path:
+            if l == 'D':
+                r += 'U'
+            if l == 'U':
+                r += 'D'
+            if l == 'R':
+                r += 'L'
+            if l == 'L':
+                r += 'R'
+
+        return r
+
+    def getDistance(self, from_location, to_location):
+        try:
+            return self.mazeMap[from_location][to_location]
+        except KeyError:
+            if to_location == from_location:
+                return 0
+            else:
+                return np.inf
+
+    def getMove(self, origin, goal):
+        """
+        Retourne le mouvement à effectuer pour aller de origin à goal,
+        qui doivent être adjacentes
+        :param origin:
+        :param goal:
+        :return: bool
+        """
+        if origin != goal:
+            i1, j1 = origin
+            i2, j2 = goal
+
+            if i1 - i2 == -1:
+                return 'D'
+            elif i1 - i2 == 1:
+                return 'U'
+            elif j1 - j2 == -1:
+                return 'R'
+            elif j1 - j2 == 1:
+                return 'L'
+            else:
+                return False
+        else:
+            return False
+
+    def getNeighbors(self, position):
+        return self.mazeMap[position].keys()
+
+    def convertMetaPathToRealPaths(self, pathMeta):
+        path = []
+        for k in range(len(pathMeta) - 1):
+            path.append(self.pathMetagraph[pathMeta[k]][pathMeta[k+1]])
+
+        return path
+
+    def convertToRealPath(self, origin, pathNodes):
+        L = [origin]
+
+        x, y = origin
+        for elt in pathNodes:
+            if elt == 'L':
+                y -= 1
+            if elt == 'U':
+                x -= 1
+            if elt == 'D':
+                x += 1
+            if elt == 'R':
+                y += 1
+            L.append((x, y))
+
+        return L
+
+    def concatPaths(self, paths):
+        r = ""
+        for path in paths:
+            r += path
+
+        return r
+
+    def createMetaGraph(self, cheeses_list):
+        """
+        Create the pattern of the metagraph(distance between cheeses
+        :param nodes_list: List of currents cheeses
+        :return:
+        """
+        cheeses_list = cheeses_list.copy()
+        dij = Dijkstra(self)
+
+        while cheeses_list:
+            n1 = cheeses_list[0]
+            # Calculate path and distance with Dijkstra
+            dij.setOrigin(n1)
+            dij.setGoal(cheeses_list)
+            dij.process()
+
+            for n2 in cheeses_list:
+                d, p = dij.getResult(n2)
+                self.coupleNodesInMetagraph(n1, n2, d, p)
+
+            cheeses_list.remove(n1) # On supprime le node en cours, pour accelerer le programme
+            cheeses_list.remove(self.getOpposite(n1)) if self.getOpposite(n1) != n1 else () # Par symetrie, on supprime l'opposé
+
+        #print(repr(len(self.distanceMetagraph[(12, 13)]))+ repr(self.distanceMetagraph[(12, 13)]))
+
+    def addNodeToMetagraph(self, node, nodes_list):
+        """
+        Ajoute le noeud "node" par rapports aux nodes "nodes_list" qui doivent déja exister dans le metaGraph
+        :param node: noeud a ajouter ou uploader
+        :param nodes_list:
+        :return:
+        """
+
+        # Create the list of unChecked nodes
+        checked_list = []
+
+        if node in self.distanceMetagraph:
+            for n in nodes_list:
+                if n not in self.distanceMetagraph[node]:
+                    checked_list.append(n)
+        else:
+            checked_list = nodes_list
+
+        # Environ 10^-6 sec pour arriver là
+        if checked_list:
+            dij = Dijkstra(self, node, checked_list)
+            dij.process()
+            # Environ 0.02 sec pour un 25x25
+
+            for n in checked_list:
+                d, p = dij.getResult(n)
+                self.coupleNodesInMetagraph(node, n, d, p)
+
+    def coupleNodesInMetagraph(self, n1, n2, d, p, recurse = True):
+        """
+        :param d: distance from n1 to n2
+        :param p: path from n1 to n2
+        """
+
+        if n1 not in self.distanceMetagraph:
+            self.distanceMetagraph[n1] = {}
+            self.pathMetagraph[n1] = {}
+
+        self.distanceMetagraph[n1][n2] = d
+        self.pathMetagraph[n1][n2] = p
+
+        if n2 not in self.distanceMetagraph:
+            self.distanceMetagraph[n2] = {}
+            self.pathMetagraph[n2] = {}
+
+        self.distanceMetagraph[n2][n1] = d
+        self.pathMetagraph[n2][n1] = p[::-1]  # Path from n2 to n1 is the opposite of the path from n1 to n2
+
+        # Add opposite
+        if recurse:
+            self.coupleNodesInMetagraph(self.getOpposite(n1), self.getOpposite(n2), d, list(map(self.getOpposite, p)), False)
+
+    def getOpposite(self, n):
+        x, y = n
+        return (self.mazeHeight - x - 1, self.mazeWidth - y - 1)
+
+    # Algoritms application
+    def getFastestPath(self, origin, goal):
+        try:
+            return (self.distanceMetagraph[origin][goal], self.pathMetagraph[origin][goal])
+        except KeyError:
+            print("## Need to calculate the fastest path from " + repr(origin) + " to " + repr(goal))
+
+            # Calculate path and distance with Astar
+            dij = Astar(self, origin, goal)
+            dij.process()
+
+            d, p = dij.getResult()
+
+            # Addto metagraph for a next time
+            self.coupleNodesInMetagraph(origin, goal, d, p)
+
+            return (d, p)
+
+    def getNearestNode(self, origin, nodes):
+        dij = Dijkstra(self)
+        dij.setOrigin(origin)
+        dij.setGoal(None)
+
+        dij.process()
+
+        n_list = [dij.getResult(n) for n in nodes]
+        n_list.sort()
+        
+        #print(n_list)
+
+        return n_list[0] if len(n_list) > 0 else (0, [])
+
+
+
+def preprocessing(mazeMap, mazeWidth, mazeHeight, playerLocation, opponentLocation, piecesOfCheese, timeAllowed):
+    global maze
+    
+    print("Start preprocessing (" + str(timeAllowed) + ")")
+
+    maze = Maze(mazeMap, mazeWidth, mazeHeight)
+
+    print("")
+
+def turn(mazeMap, mazeWidth, mazeHeight, playerLocation, opponentLocation, playerScore, opponentScore, piecesOfCheese, timeAllowed):
+    global maze
+
+    path = maze.getNearestNode(playerLocation, piecesOfCheese)[1]
+    action = maze.getMove(path[0], path[1])
+    print('[' + repr(action) + ']')
+    
+    return action