我想在树中找到元素的完整路径。元素可以位于几个地方。
我当前的代码:
levels = [
{"L3A": ["L4A"]},
{"L3B": ["L4B"]},
{"L3C": ["L4C"]},
{"L1": ["L2", "L4A"]},
{"L2": ["L3A", "L3B", "L3C"]}
]
def get_level(name):
tree = []
recursive(name, tree)
return tree
def recursive(name, tree):
for level in levels:
for k, v in level.items():
if name in v:
tree.append(k)
recursive(k, tree)
levl = get_level("L4A")
print(levl)
结果
是:['L3A', 'L2', 'L1', 'L1']
想要:[['L3A', 'L2', 'L1'], ['L1']]
最终想要:
L4A in L1 > L2 > L3A
L4A in L1
您能给我一些建议如何更改它吗?
答案 0 :(得分:0)
反转关联图,然后应用标准图搜索,例如。 DFS。
答案 1 :(得分:0)
为什么L1
在您的列表中出现两次?因为您有两条通往L4A
的路径:L1 -> L2 -> L3A -> L4A
和L1 -> L4A
,但是只有一个path
变量可以存储这些路径。由于您使用了一种反向DFS,因此您具有以下级别:L4 -> L3A -> L2 -> L1
,然后是L4 -> L1
。
让我们尝试阐述一种算法。您正在处理图形(如果您添加根,您会得到一棵树),因此我将使用通常的词汇:级别是“节点”,级别之间的路径是“边缘”。这是进行操作的好方法:
N
P
,以使边缘P-N
存在并存储路径。P
,找到所有节点Q
,以使边缘P-Q
存在并存储路径。edge
不再有Q
,即当前path
最大,则返回path
。作为真正的算法,它缺乏精确度。让我们关注:
GIVEN: a node N
let paths_to_explore = [N]
while paths_to_explore is not empty:
for every path_to_explore:
try to add a node at the beginning of path_to_explore
if it fails, yield path_to_explore
在获得代码之前,请注意,图形的表示形式不是两种常用的表示形式。就您而言,您具有边列表,但是字典from_node -> [to_nodes]
更好:
edges = {
"L3A": {"L4A"},
"L3B": {"L4B"},
"L3C": {"L4C"},
"L1": {"L2", "L4A"},
"L2": {"L3A", "L3B", "L3C"},
}
这使边缘上的迭代更容易:
for from_node, to_nodes in edges.items():
# do something with nodes
现在,代码:
def find_reverse_path(name):
paths = []
paths_to_explore = [[name]]
while paths_to_explore:
path = paths_to_explore.pop() # next!
to_node = path[0] # the HEAD of the current path
expanded = False
for from_node, to_nodes in edges.items():
if to_node in to_nodes: # there's an edge to the HEAD
new_path_to_explore = [from_node] + path # new path = from_node + old path
paths_to_explore.append(new_path_to_explore) # add it to the exploration list
expanded = True # this path was expanded
if not expanded: # the path is maximal
paths.append(path) # use yield if you want to create a generator
return paths
print(find_reverse_path("L4A"))
输出:
[['L1', 'L4A'], ['L1', 'L2', 'L3A', 'L4A']]
这是一个迭代的DFS。 (我想我们很难知道是否在递归DFS中扩展了路径。)
看看这两行,它们包含“技巧”:
new_path_to_explore = [from_node] + path # new path = from_node - old path
paths_to_explore.append(new_path_to_explore) # add it to the exploration list
请注意,new_path_to_explore
是path
的副本,这意味着我们不只是向paths[-1]
添加节点(就地)。这是为什么?看一下第一步:
1. paths = [[L4A]]
2. paths = [], path = [L4A]
3. append L1-L4A to paths, then append L3A-L4A to paths
4. paths = [[L1, L4A], [L3A, L4A]]
5. paths = [[L1, L4A]], path = [L3A, L4A]
...
如果不进行复制,并且在当前路径的开头找到多个边,则将在步骤4 paths = [[L3A, L1, L4]]
中执行。那几乎就是您在代码中遇到的相同问题。