从平面表构建节点关系

Question

我有一个平台形式的传输网络。内容可以描述如下：

Start_Node,Next_Node,End_Node
10,100,1000
100,200,1000
200,1000,1000
20,1000,1000

我想找到start_end节点对之间的所有节点。所以对于上面的例子，它看起来像：

{10_1000: [100,200,1000]
20_1000: [1000]
100_1000: [200,1000]
200_1000: [1000]}

以下是我的尝试（我正在尝试学习python，因此会欣赏该语言中的任何方向）

import pandas as pd

reg = pd.read_csv('nw.csv')

start_node = reg.start.unique()
end_node = reg.end.unique()
hop_dic = {}

for i in start_node:
    for j in end_node:
        key = str(i) + " - " + str(j)
        node = []
        iter = 0
        for row in reg.itertuples():
            while(row[1]==i and row[3]==j and row[2]!=j and iter < 10):
                hop_dic[key] = node.append(row[2])
                i = row[2]
                iter = iter + 1
                print "i =",i,"j= ",j,"iter= ",iter

首先这很慢，因为我显然在运行循环并迭代所有节点对的所有行，其次node.append（）不起作用，并在hop_dic中将所有值都赋予None。 Iter只是为了避免无限循环，我找不到连接图。

更新

import pandas as pd

reg = pd.read_csv('regularbagfwd.csv')

start_node = reg.start.unique()
end_node = reg.end.unique()
hop_dic = {}
for i in start_node:
    for j in end_node:
        key = str(i) + " - " + str(j)
        iter = 0
        reg_sub = reg.loc[reg.end == j]
        node = [0]
        while(node[-1] != j): #Checking if the last element is the end node
            for row in reg_sub.itertuples():
                if row[1]==i and row[2]==j: #last element 
                    node.append(row[2])
                    hop_dic[key] = node
                    print "i =",i,"j= ",j,"iter= ",iter,"next= ",row[2]
                    print hop_dic
                    break
                elif row[1]==i and row[2]!=j: #Not last element
                    node.append(row[2])
                    hop_dic[key] = node
                    i = row[2]
                    print "i =",i,"j= ",j,"iter= ",iter,"next= ",row[2]
                    print hop_dic
                iter = iter + 1
            if iter > 500: #I will break the while loop, untidy I know
                break

我得到它的工作，我需要在我的列表中添加一个虚拟0，我可以在以后删除，但必须有一个更好的方法。