在使用Python Networkx 中的relabel_nodes函数时，出现：
AttributeError: 'str' object has no attribute 'copy'
找了半天也没发现错误出现在哪里，通过比较发现：
如果在定义图的类型时候使用G=nx.MultiDiGraph()就会出现这个错误，改为G=nx.Graph()错误消失。
目测是relabel_nodes函数不支持MultiDiGraph。
附：
1 def relabel_nodes(G, mapping, copy=True): 2 """Relabel the nodes of the graph G. 3 4 Parameters 5 ---------- 6 G : graph 7 A NetworkX graph 8 9 mapping : dictionary 10 A dictionary with the old labels as keys and new labels as values. 11 A partial mapping is allowed. 12 13 copy : bool (optional, default=True) 14 If True return a copy, or if False relabel the nodes in place. 15 16 Examples 17 -------- 18 >>> G=nx.path_graph(3) # nodes 0-1-2 19 >>> mapping={0:'a',1:'b',2:'c'} 20 >>> H=nx.relabel_nodes(G,mapping) 21 >>> print(sorted(H.nodes())) 22 ['a', 'b', 'c'] 23 24 >>> G=nx.path_graph(26) # nodes 0..25 25 >>> mapping=dict(zip(G.nodes(),"abcdefghijklmnopqrstuvwxyz")) 26 >>> H=nx.relabel_nodes(G,mapping) # nodes a..z 27 >>> mapping=dict(zip(G.nodes(),range(1,27))) 28 >>> G1=nx.relabel_nodes(G,mapping) # nodes 1..26 29 30 Partial in-place mapping: 31 32 >>> G=nx.path_graph(3) # nodes 0-1-2 33 >>> mapping={0:'a',1:'b'} # 0->'a' and 1->'b' 34 >>> G=nx.relabel_nodes(G,mapping, copy=False) 35 36 print(G.nodes()) 37 [2, 'b', 'a'] 38 39 Mapping as function: 40 41 >>> G=nx.path_graph(3) 42 >>> def mapping(x): 43 ... return x**2 44 >>> H=nx.relabel_nodes(G,mapping) 45 >>> print(H.nodes()) 46 [0, 1, 4] 47 48 Notes 49 ----- 50 Only the nodes specified in the mapping will be relabeled. 51 52 The keyword setting copy=False modifies the graph in place. 53 This is not always possible if the mapping is circular. 54 In that case use copy=True. 55 56 See Also 57 -------- 58 convert_node_labels_to_integers 59 """ 60 # you can pass a function f(old_label)->new_label 61 # but we'll just make a dictionary here regardless 62 if not hasattr(mapping,"__getitem__"): 63 m = dict((n, mapping(n)) for n in G) 64 else: 65 m = mapping 66 if copy: 67 return _relabel_copy(G, m) 68 else: 69 return _relabel_inplace(G, m)