day 23 py (#75)

* day 23 py

* plotting script for day 23 py

* improve speed

src/python/23_rk.py

@@ -0,0 +1,321 @@
+#!/usr/bin/env python3
+import sys, math, util
+import collections
+
+def parse():
+    m = util.matrix()
+    with open(sys.argv[1], 'r') as infile:
+        for i, line in enumerate(infile):
+            m[i] = [i for i in line.strip()]
+    return m
+
+
+def find_start_end(m):
+    s = None
+    e = None
+    for i, c in enumerate(m[0]):
+        if c == ".":
+            s =  (0, i)
+    for i, c in enumerate(m[-1]):
+        if c == ".":
+            e =  (m.dim()[0] - 1, i)
+    #  print(m.dim())
+    #  print(s, e)
+    return (s, e)
+
+def walk(pos, m, l_visited, sol2):
+    """
+    yield one by one all possible direction to go
+    """
+    slope = [">", "v", "<", "^"]
+    if not sol2:
+        if m[pos] in slope:
+            di = slope.index(m[pos])
+            new_pos = util.tuple_add(pos, util.direction_map[util.diDecode[di]])
+            # ASSUMPTION: no check of whether slope is legal
+            yield (di, new_pos)
+            return
+
+    for di in range(4):
+        new_pos = util.tuple_add(pos, util.direction_map[util.diDecode[di]])
+        if new_pos in l_visited :
+            continue
+        if not m.inbound(new_pos):
+            continue
+        if m[new_pos] == "#":
+            continue
+        if not sol2:
+            # you must follow the arrow
+            # ASSUMPTION: assume you will never access a slope from the side of it
+            if m[new_pos] in slope and di != slope.index(m[new_pos]):
+                continue
+        yield (di, new_pos)
+    pass
+
+def sol_bf(m, sol2):
+    """
+    brute force first
+    disadvantage: there's a lot of list copy and addition, memory use intensive. And slow due to all possibility explored.
+        10s for part 1
+        part 2 > 10min
+
+    cannot use due to data format change
+    """
+    (start, end) = find_start_end(m)
+    q = collections.deque([ [start] ])
+    max_length = 0
+    while len(q):
+        l_visited = q.popleft()
+        #  print(l_visited)
+        if l_visited[-1] == end:
+            max_length = max(max_length, len(l_visited))
+            continue
+        for _, new_pos in walk(l_visited[-1], m, l_visited, sol2):
+            q.append( l_visited + [new_pos] ) 
+
+    #  excluding start
+    return max_length -1 
+
+def check_surrounding(pos, prevpos, m):
+    """
+    check possible direction to go
+    if it's <=2, keep going forward
+    """
+    l_possible_sol1 = []
+    l_possible_sol2 = []
+    slope = [">", "v", "<", "^"]
+    for di in range(4):
+        new_pos = util.tuple_add(pos, util.direction_map[util.diDecode[di]])
+        if prevpos == new_pos:
+            continue
+        if m[new_pos] == "#":
+            continue
+        if not m.inbound(new_pos):
+            continue
+        # you must follow the arrow
+        # ASSUMPTION: assume you will never access a slope from the side of it
+        l_possible_sol2.append(new_pos)
+        if m[new_pos] in slope and di != slope.index(m[new_pos]):
+            continue
+        l_possible_sol1.append(new_pos)
+    return l_possible_sol1, l_possible_sol2
+
+
+def conversion(c, d_graph, pos):
+    if c == "#":
+        return 0
+    elif pos in d_graph:
+        return 2
+    else:
+        return 1
+
+def plot(d_graph, m, sol2, solution):
+    from matplotlib import pyplot as plt
+    import numpy as np
+
+    num_m = [[conversion(c, d_graph, (irow, icol)) for icol, c in enumerate(row)] for irow, row in enumerate(m)]
+    #  print(num_m)
+
+    map = np.array(num_m)
+    #  print(map)
+
+    fig, ax = plt.subplots()
+    im = ax.imshow(map)
+
+    #  ax.set_xticks(np.arange(len(farmers)), labels=farmers)
+    #  ax.set_yticks(np.arange(len(vegetables)), labels=vegetables)
+
+    fig.savefig("23_full_trace.png")
+
+    import networkx as nx
+    G = None
+    G = nx.DiGraph()
+
+
+
+    for i in range(len(solution)-1):
+        start = solution[i]
+        end = solution[i+1]
+        for end2, step in d_graph[start]:
+            if end != end2: continue
+
+            G.add_edge(f"{(start[0], start[1])}", f"{(end[0], end[1])}", weight = 1, length = step, color = "tab:red")
+            break
+        pass
+
+    esolution = [(u, v) for (u, v, d) in G.edges(data=True) if d["color"] == "tab:red"]
+
+    for start, l in d_graph.items():
+        for end, step in l:
+            if (end, start) in esolution or (start, end) in esolution:
+                continue
+            G.add_edge(f"{(start[0], start[1])}", f"{(end[0], end[1])}", weight = 1, length = step, color = "k")
+
+
+    eother    = [(u, v) for (u, v, d) in G.edges(data=True) if d["color"] != "tab:red"]
+
+
+    options = {
+		#  'node_color': 'black',
+		'node_size': 1000,
+		'font_size': 7,
+        'width': 6,
+        #  'horizontalalignment' : 'left',
+        #  'verticalalignment' : '',
+        }
+
+    #  fig, ax = plt.subplots(figsize=(30, 30))
+    fig, ax = plt.subplots(figsize=(30, 30))
+    pos = nx.spectral_layout(G)
+    #  pos = nx.spring_layout(G)
+    #  pos = nx.spring_layout(G, seed = 0)
+    nx.draw_networkx_nodes(G, pos = pos, node_size = options['node_size'])
+    nx.draw_networkx_labels(G, pos = pos, font_size = options['font_size'])
+    #  colors = nx.get_edge_attributes(G,'color')
+    #  [print(i) for i in colors.values()]
+    #  nx.draw_networkx_edges(G, pos, edge_color = colors)
+    if sol2:
+        nx.draw_networkx_edges(G, pos, edgelist=eother,    edge_color = "k", arrows = True, width = 6, arrowsize = 15, arrowstyle = "-")
+    else:
+        nx.draw_networkx_edges(G, pos, edgelist=eother,    edge_color = "k", arrows = True, width = 6, arrowsize = 15)
+    nx.draw_networkx_edges(G, pos, edgelist=esolution, edge_color = "tab:red", arrows = True, width = 6, arrowsize = 15)
+    labels = nx.get_edge_attributes(G,'length')
+    nx.draw_networkx_edge_labels(G, pos = pos, edge_labels=labels, font_size=options["font_size"])
+    plt.savefig(f"23_graph_part{int(sol2) + 1}.png", dpi = 400)
+
+    pass
+
+def sol(m, sol2):
+    """
+    find the graph feature and then loop inside
+    """
+    (start, end) = find_start_end(m)
+    q = [(start, (-1, -1))]
+
+    #  here study every junction/start
+    d_graph = {}
+    while len(q):
+        # DFS find all nodes, same as WFS
+        startpos, prevpos = q.pop()
+        if startpos in d_graph:
+            continue
+        d_graph[startpos] = []
+        pos = startpos
+        for pos in check_surrounding(pos, prevpos, m)[sol2]:
+            # this is the actual possible step after the current pos
+            prevpos = startpos
+            step = 1
+
+            while True:
+                # in fact just check junction here.
+                # this is l_possible_sol2, which is effectively any junction(considering direction going in)
+                l_possible = check_surrounding(pos, prevpos, m)[1]
+                if len(l_possible) != 1:
+                    break
+
+                prevpos = pos
+                pos = l_possible[0]
+                step += 1
+
+
+            # outside, then pos is the junction/end
+            d_graph[startpos].append((pos, step))
+
+            # if you have already visited this junction
+            if pos in d_graph:
+                continue
+
+            # this is a node, and consider it next
+            # consider all possible direction(even back to this node itself due to symmetry in part 2, it might take slightly longer but code is simpler)
+            if pos != end:
+                q.append((pos, (-1, -1)))
+
+
+    if len(sys.argv) <= 2:
+        # prepare for dfs
+        max_s = [-999]
+
+        l_mapped = list(d_graph.keys()) + [end]
+        istart = l_mapped.index(start)
+        iend = l_mapped.index(end)
+        graph_I = [[] for i in l_mapped]
+        visited = [False for i in l_mapped]
+        for this, l in d_graph.items():
+            ithis = l_mapped.index(this)
+            for next, step in l:
+                inext = l_mapped.index(next)
+                graph_I[ithis].append((inext, step))
+            pass
+
+        def dfs(graph_I, visited, this, step, max_s):
+            if this == iend:
+                max_s[0] = max(max_s[0], step)
+                return
+
+            for next, added_step in graph_I[this]:
+                if visited[next]:
+                    continue
+                visited[next] = True
+                dfs(graph_I, visited, next, step+added_step, max_s)
+                visited[next] = False
+
+        visited[istart] = True
+        dfs(graph_I, visited, istart, 0, max_s)
+        return max_s[0]
+
+
+        # 7 seconds
+        visited = set(start)
+        def dfs(d_graph, visited, this, step, max_s):
+            if this == end:
+                max_s[0] = max(max_s[0], step)
+                return
+
+            for next, added_step in d_graph[this]:
+                if next in visited:
+                    continue
+                visited.add(next)
+                dfs(d_graph, visited, next, step+added_step, max_s)
+                visited.remove(next)
+
+        dfs(d_graph, visited, start, 0, max_s)
+        return max_s[0]
+
+
+
+    #  40 second to map out the path for plotting
+    if len(sys.argv) > 2:
+        q = [([start], 0)]
+        solution = []
+        max_s = 0
+        while len(q):
+            visited_nodes, step = q.pop()
+            this = visited_nodes[-1]
+            #  print(visited_nodes)
+            if this == end:
+                #  print(visited_nodes, step)
+                max_s = max(max_s, step)
+                if max_s == step:
+                    solution = visited_nodes
+                continue
+
+            for next, added_step in d_graph[this]:
+                #  print("next", next)
+                if next in visited_nodes:
+                    continue
+                #  visited_nodes.append(next)
+                q.append((visited_nodes + [next], step+added_step))
+
+        plot(d_graph, m, sol2, solution)
+
+    return max_s
+
+def main():
+    m = parse()
+    #  print(m)
+    #  print(sol_bf(m, False))
+    #  print(sol_bf(m, True))
+    print(sol(m, False))
+    print(sol(m, True))
+
+main()

src/python/util.py

@@ -1,10 +1,22 @@
 class matrix:
-    def __init__(self, l):
+    """
+    a list of M lists with size N of the same type of element, default to be int(0)
+    [[],
+     [],
+     ...
+     []]
+    """
+    def __init__(self, l = [], default = 0):
         self._l = [[e for e in il] for il in l]
-        self._nr = len(self._l)
-        self._nc = len(self._l[0])
+        self.dim()
+        self._default = default
 
     def dim(self):
+        self._nr = len(self._l)
+        if len(self._l):
+            self._nc = len(self._l[0])
+        else:
+            self._nc = 0
         return self._nr, self._nc
 
     def __iter__(self):
@@ -23,7 +35,24 @@ def __getitem__(self, tup):
             return self._l[tup]
 
     def __setitem__(self, tup, val):
-        self._l[tup[0]][tup[1]] = val
+        if type(tup) == type((0,0)):
+            if tup[0] >= self._nr:
+                for i in range(self._nr, tup[0]+1):
+                    self._l.append([])
+            self.dim()
+            if tup[1] >= self._nc:
+                for r in range(0, self._nr):
+                    for c in range(self._nc, tup[1]+1):
+                        self._l[r].append(self._default)
+
+            self._l[tup[0]][tup[1]] = val
+            self.dim()
+        else:
+            if tup >= self._nr:
+                for i in range(self._nr, tup+1):
+                    self._l.append([])
+            self._l[tup] = val
+            self.dim()
 
     def __str__(self):
         s = ""