将MultiDiGraph作为dicts的dict读取会忽略multigraph_input

Question

在NetworkX 2.5中，我似乎无法使用一组dicts来阅读多重图。下面是一个尝试往返一些数据的示例：

G = nx.MultiDiGraph()
G.add_edge(2, 3, weight=5)
G.add_edge(2, 3, weight=4)  # Create 2 edges
dict_of_dicts = nx.to_dict_of_dicts(G)
G2 = nx.MultiDiGraph(dict_of_dicts, multigraph_input=True)
print("Original", G.edges(data=True))
print("Round tripped", G2.edges(data=True))
print(G2.number_of_edges(), "edge(s) in round tripped graph")

在往返图G2:中只给出一条边

Original [(2, 3, {'weight': 5}), (2, 3, {'weight': 4})]
Round tripped [(2, 3, {0: {'weight': 5}, 1: {'weight': 4}})]
1 edge(s) in round tripped graph

那么，在MultiDiGraph中作为dicts的dict进行阅读的格式是什么呢？

Answer 1

对我来说，同时使用create_using和multigraph_input参数的nx.from_dict_of_dicts有效：

import networkx as nx
G = nx.MultiDiGraph()
G.add_edge(2, 3, weight=5)
G.add_edge(2, 3, weight=4)  # Create 2 edges
dict_of_dicts = nx.to_dict_of_dicts(G)
G2 = nx.from_dict_of_dicts(dict_of_dicts, create_using=nx.MultiDiGraph(), multigraph_input=True)