Hi!

Infomap doesn't need to generate Huffman codes in practice, instead we calculate the stationary visit-rates.
Using those, we can calculate the lower bound for the expected modular codelength.

We call the visit-rates "flow", and you can find those values in the second column of the tree file.
Check here for more info about the format

If you want to assign Huffman codes, you can use those values and put it into a Huffman code generator or implement it yourself.

We have a visualization of what a modular Huffman coding scheme would look like, but as I said, we don't need to do that in practice.

Hope this helps!