edge_subgraph fails for MultiDiGraph

[ARF1]

Current Behavior

Attempting to create a subgraph induced by edges using edge_subgraph() of a MultiDiGraph fails with error:

File 'networkx\classes\function.py", line 452, in edge_subgraph
    edges = set(edges)
TypeError: unhashable type: 'dict

Expected Behavior

Should create a subgraph view

Steps to Reproduce

Example of filtering a graph by edge data:

import networkx as nx

G = nx.MultiDiGraph()
G.add_nodes_from([1, 2])
G.add_edge(1, 2, None, kind='kind1', more_data='foo')
G.add_edge(1, 2, None, kind='kind2', even_more_data='bar')

sub_graph = G.edge_subgraph(
    (u, v, d)
    for u, v, d
    in G.edges(data=True)
    if d['kind'] == 'kind1'
)

Environment

Python version: 3.13.7 NetworkX version: 3.5

[amcandio]

You should pass keys=True to edges as opposed to data=True. These subgraphs are constructed as views that "filter" the underlaying graph. So you don't need to pass the data, it is inferred from the underlying graph.

[dschult]

Just what @amcandio said -- and I'll add two other minor slightly subtle points:

• G.edges without any parentheses returns the (u, v, ekey) triples for multigraphs. So you could iterate over that if you weren't filtering on the data.
• G.edges(keys=True, data=True) iterates over the 4-tuples: (u, v, ekey, data). This is probably what you will need to be able to transmit the 3-tuples needed to identify the edge while using the edge data to filter the edge.

G = nx.MultiDiGraph()
G.add_nodes_from([1, 2])
G.add_edge(1, 2, None, kind='kind1', more_data='foo')
G.add_edge(1, 2, None, kind='kind2', even_more_data='bar')

sub_graph = G.edge_subgraph(
    (u, v, k)
    for u, v, k, d
    in G.edges(keys=True, data=True)
    if d['kind'] == 'kind1'
)

[Jiya873]

I have updated the edge_subgraph docstring to clarify the required edge formats as suggested by @amcandio and @dschult. Please have a look at the PR.