Speed up topsort in parsing

[masinag]

Currently the parser uses the toposort_flatten from the toposort library

py-aiger/aiger/parser.py

Line 367 in 19b3fe3

for lit in toposort_flatten(state.nodes):

However, I noticed that this really slows down the parsing of files with many gates, e.g. this file.

After some investigation, I noticed that the implementation of topsort in the toposort library is very inefficient, as it is O(n^2) instead of O(n+m).
https://gitlab.com/ericvsmith/toposort/-/blob/master/src/toposort.py#L47

I suggest using the graphlib library, it should be enough to replace the above line with:

    for lit in graphlib.TopologicalSorter(state.nodes).static_order():

What do you think?

[mvcisback]

thanks for looking into this. Would you be open to creating a PR?

[masinag]

I found out that graphlib is supported only since python 3.9, I'm looking for an alternative

[masinag]

I can implement a simple topsort function instead of relying on an external library, what do you think?
It shouldn't be longer than 10-15 sloc