How exactly do numberOfPaths1Visit and numberOfPaths2Visit work?

How exactly do they work? 
It seems they're deterministic recursive algorithms, so how can they generate paths (aka currentPaths aka covering subgraphs) between recursive calls and don't get duplicates?
How could it be that efficient?