使用邻接表表示有向图,并且使用回溯法查找有向图中的路径
对于有向图的邻接表表示形式,可以使用字典数据结构来表示
import sys
class Solution:
def __init__(self):
# self.graph = {'A':['B','C'],
# 'B':['C','D'],
# 'C':['D'],
# # 'D':['C'],
# 'E':['F'],
# 'F':['C']
# }
self.graph = {
'1':['3','4'],
'2':['5','4'],
'3':['6'],
'4':['3','7','6'],
'5':['7','4'],
'7':['6']
}
def find_path(self,start,end,path=[]):
path = path+[start]
if start==end:
return path
if not self.graph.has_key(start):
return None
for node in self.graph[start]:
if node not in path:
newpath = self.find_path(node,end,path)
if newpath:
return newpath
return None
def find_paths(self,start,end,path=[]):
path = path+[start]
# path.append(start)
paths=[]
if start==end:
return [path]
if not self.graph.has_key(start):
return None
paths = []
for node in self.graph[start]:
if node not in path:
newpath = self.find_paths(node,end,path)
if newpath:
paths=paths + newpath
return paths
def find_shortest_path(self,start,end,path=[]):
path = path+[start]
# path.append(start)
paths=[]
if start==end:
return path
if not self.graph.has_key(start):
return None
shortest = None
for node in self.graph[start]:
if node not in path:
newpath = self.find_shortest_path(node,end,path)
if newpath:
if not shortest or len(shortest)>len(newpath):
shortest = newpath
return shortest
def dfs(self):
stack = []
visited = set()
for key in self.graph:
# sys.stdout.write(key+'\n')
if key not in visited:
sys.stdout.write(key+" ")
stack.append(key)
visited.add(key)
while len(stack)>0:
tmp = stack[len(stack)-1]
if not self.graph.has_key(tmp):
if len(stack)>0:
stack.pop()
continue
for value in self.graph[tmp]:
if value not in visited:
sys.stdout.write(value+" ")
stack.append(value)
visited.add(value)
else:
if len(stack)>0:
stack.pop()
continue
def bfs(self):
from collections import deque
queue = deque()
visited=set()
for key in self.graph.keys():
if key not in visited:
queue.append(key)
visited.add(key)
while len(queue)>0:
tmp = queue.popleft()
sys.stdout.write(tmp+'\t')
if not self.graph.has_key(tmp):
break
for value in self.graph[tmp]:
if value not in visited:
queue.append(value)
visited.add(value)
def has_circle(self):
from collections import deque
visited = set()
for key in self.graph.keys():
if key not in visited:
queue = deque(key)
visited.add(key)
while len(queue)>0:
tmp = queue.popleft()
if self.graph.has_key(tmp):
for value in self.graph[tmp]:
if key==value:
print "There is circle"
return
if value not in visited:
visited.add(value)
queue.append(value)
print "There is not circle"
return
if __name__=="__main__":
s=Solution()
sys.stdout.write("Breadth first search:"+'\n')
s.bfs()
sys.stdout.write('\n')
sys.stdout.write("Depth first search:"+'\n')
s.dfs()
sys.stdout.write('\n')
sys.stdout.write("find a path:" + '\n')
print s.find_path('1','6')
sys.stdout.write("find all paths:" + '\n')
print s.find_paths('1','6')
sys.stdout.write("find the shortest path:" + '\n')
print s.find_shortest_path('1','6')
sys.stdout.write("judge circle:" + '\n')
s.has_circle()
网友评论