为何SCI2脚本可用[:]获取NetworkX节点列表而我无法实现？

Hey there! Let's clear up this confusion—you're totally right that Python/NetworkX version differences are the root cause here, so let's break it down simply:

The Core Issue: g.nodes Changed Its Return Type #

In older versions of NetworkX (common with Python 2), g.nodes directly returned a list of nodes. That's why the SCI2 script could use [:] to slice and create a copy of the list.

But in modern NetworkX (2.0+), which is built for Python 3, g.nodes returns a NodeView object—this is a dynamic, efficient view of the graph's nodes, not a static list. Since views don't support slicing with [:], your code throws an error.

Solutions to Get What You Need #

Here are straightforward fixes depending on your goal:

1. Get a list of all node IDs
   Just convert the NodeView to a list explicitly:

   import networkx as nx
   g = nx.Graph()
   g.add_node(1, size=11)
   g.add_node(2, size=12)
   
   a = list(g.nodes)
   print(a)  # Output: [1, 2]
   

2. Get nodes along with their attributes
   If you need attribute data too, use g.nodes(data=True) and convert to a list—each item will be a tuple of (node_id, attribute_dict):

   nodes_with_attrs = list(g.nodes(data=True))
   print(nodes_with_attrs)  # Output: [(1, {'size': 11}), (2, {'size': 12})]
   

3. Modify node attributes directly
   You don't even need a list copy to modify attributes! Access and update them directly via the NodeView:

   # Update the size of node 1
   g.nodes[1]['size'] = 15
   print(g.nodes[1]['size'])  # Output: 15
   

Why This Change Happened #

NetworkX switched to view objects for efficiency—views avoid unnecessary copies of node/edge data, and they automatically reflect graph changes (e.g., new nodes added later show up in the view without re-fetching). The old list approach was far less efficient for large graphs.

内容的提问来源于stack exchange，提问作者LiuQiang8650