get node data?

[Tim Howard]

I think I am missing something extremely basic. Given this graph:

>>> import networkx as nx
>>> G = nx.Graph()
>>> G.add_path(['a','b','c','d'])
>>> G.add_node('d', time='5pm')

How do I call a *single* node, plus its data, by name? I essentially
am looking for the equivalent of G.get_edge_data, but for nodes.

I can get the information by calling the item in the list:

>>> G.nodes(data=True)[3]
('d', {'time': '5pm'})

But I don't know how to get it by calling it by name (e.g. 'd'). I
can get the nodes connected to it by calling:
>>> G['d']
{'c': {}}

But that doesn't seem to provide any node data. I can only think of
iterating over the entire graph using G.nodes_iter(), but that seems
inefficient.

I've also seen code that uses dictionaries as follows:

>>> G.time = {}
>>> G.time['d'] = '6pm'
>>> G.time['d']
'6pm'

Is this still a preferred way to track node (and edge) information? I
was getting the feeling networkX was migrating away from this
approach. Is this correct?

I apologize in advance if there is an obvious and easy solution to
this that I am just missing.

cheers,
Tim

[Tim Howard]

It looks like I can turn the nodes into a dictionary and get node data
that way:

>>> nodeDict = dict(G.nodes(data=True))
>>> nodeDict['d']
{'time': '5pm'}

I'll use this approach for now...
Tim

[Richard Careaga]

I don't think that quite satisfies the condition of your requirement, which was to return and object that contains both the node and its data--you're only returning the key:item pair for the data, not the node identifier, 'd'

If that's what you want, you need to go back to

>>> G.nodes(data=True)
[('a', {}), ('c', {}), ('b', {}), ('d', {'time': '5pm'})]

and extract the tuple

('d', {'time': '5pm'})

>>> called_node = ('d,', nodeDict['d'])
>>> called_node
('d,', {'time': '5pm'})

[Aric Hagberg]

Actually it's simpler. You can access the node data dictionary directly:

In [1]: import networkx

In [2]: G=networkx.Graph()

In [3]: G.add_node(1,time='5pm')

In [4]: G.node[1]
Out[4]: {'time': '5pm'}

This works the same for edges with G.edge[u][v] which
is the same as G[u][v]

n [5]: G.add_edge(1,2,amount=7)

In [6]: G.edge[1][2]
Out[6]: {'amount': 7}

In [7]: G[1][2]
Out[7]: {'amount': 7}

Aric

[Tim Howard]

Excellent! I had found the edge usage, but had not (and still can't)
find the G.node usage in your online documentation. I had kept
trying some permutation of G.nodes() and was getting nowhere.

Thanks so much, everyone!
Tim

[Richard Careaga]

Yes, Aric's way is simpler, but you still don't get the node key to go along with the key:item pair in 'data,' just the data. I don't know that it's necessary to provide for that in the package, but thought I should mention it, again.

[Aric Hagberg]

On Thu, Apr 15, 2010 at 11:56 AM, Richard Careaga <leu...@gmail.com> wrote:
> Yes, Aric's way is simpler, but you still don't get the node key to go along
> with the key:item pair in 'data,' just the data. I don't know that it's
> necessary to provide for that in the package, but thought I should mention
> it, again.

Richard is right - but you can access the node data dictionary directly too.
(Note that add_nodes_from() below with node,attr pairs is new with
networkx-1.1).

In [1]: import networkx

In [2]: G=networkx.Graph()

In [3]: G.add_nodes_from([('a', {}), ('c', {}), ('b', {}), ('d',
{'time': '5pm'})])

In [4]: G.node
Out[4]: {'a': {}, 'b': {}, 'c': {}, 'd': {'time': '5pm'}}

In [5]: G.nodes(data=True)
Out[5]: [('a', {}), ('c', {}), ('b', {}), ('d', {'time': '5pm'})]

Aric

[Dan Schult]

If you want the pairs (node, attrdict) you can use:
(n,G.node[n]) # for a single node n
G.node.iteritems() # for all nodes, or
[(n,G.node[n]) for n in nbunch] # for a subset of the nodes
Dan

> --
> You received this message because you are subscribed to the Google
> Groups "networkx-discuss" group.
> To post to this group, send email to networkx-
> dis...@googlegroups.com.
> To unsubscribe from this group, send email to networkx-discuss
> +unsub...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/
> group/networkx-discuss?hl=en.