基于访问权限的变量类型变化？

Question

I have a directed graph structure consisting of nodes and edges both of which subclass an Event parent class. Depending on external events, edges can either be active or inactive. I then find all the directed paths from a given node to the root node, but I really only care about the nodes along the way, not the edges. For instance, to convert from a set of edges to a set of nodes I use:

>>> paths
[[<Edge F>, <Edge B>]]
>>> lst = [set(map(lambda e: e.tail, path)) for path in paths]

where path is a list of edges. This is what confuses me: when I go to check the contents of lst , it changes depending on how I access it

>>> lst
[set([<Node 2>, <Node 1>])]
>>> [type(n) for n in path for path in lst]
[<class 'libs.network.Edge'>, <class 'libs.network.Edge'>]
>>> [type(n) for n in lst[0]]
[<class 'libs.network.Node'>, <class 'libs.network.Node'>]

Why aren't these two ways of accessing the type information the same?

Answer 1

You have your list comprehension order wrong. Nested loops are listed from left to right .

So the expression

[type(n) for n in path for path in lst]

is executed as

for n in path:
    for path in lst:
        type(n)

so n is taken from some random pre-assigned path variable you had before, and it is that variable that contains Edge instances. Those objects have nothing to do with the contents of lst[0] you loop over in your other expression.

You probably wanted to do this the other way around:

[type(n) for path in lst for n in path]

so that path is actually set from lst before you iterate over it.