Using networkx I created a graph G and added the edge (1,2). Strangely,
(1,2) in G.edges() == False
but
G.has_edge(1,2) == True
I'm baffled. This is a concise summary of the problem. The actual graph data is large and cannot be reproduced here in full.
The issue is that G.edges() only has each edge once. Since it's an undirected graph, it could be as either (1,2) or as (2,1) (you wouldn't want both to appear --- you might be iterating and delete the edge the first time). You can't be sure of what order it will do so because it's coming from a python dict, which does not have predictable order. You could do (1,2) in G.edges() or (2,1) in G.edges(), but you really don't want to do that - it's inefficient to create the edge lists.
So the test should use G.has_edge(1,2) which checks it correctly (and much more efficiently).
Just to show the lack of predictability for different python implementations, this is what I get
In [3]: G=nx.Graph()
In [4]: G.add_edge(1,2)
In [5]: G.edges()
Out[5]: [(1, 2)]
In [6]: (1,2) in G.edges()
Out[6]: True
In [7]: G.add_edge(4,3)
In [8]: G.edges()
Out[8]: [(1, 2), (3, 4)]
In [9]: (4,3) in G.edges()
Out[9]: False
