Let me get the weird cases we will find down the line to add tests for those and see what happens there.

I can already think of a case that will fail... - seems fine on my machine, but I bet . would do bad things.

However, a simple regex gsub I think would fix it... maybe gsub anything not in a set of acceptable characters? Like:

trie_parents = @name[0...3].gsub(/[^0-9a-zA-Z]/, "_").split("")

hmmm, now that I think of it, the . probably wouldn't really pose a problem... it would essentially get ignored in the File.join, but it wouldn't be added in a trie manner, so I think gsubbing everything outside a certain range might still be the best option

This is the whole data sample of what we can find with current specs:

["_", "-", "0", "1", "2", "3", "4", "5", "6", "7", "8", "9", "a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "k", "l", "m", "n", "o", "p", "q", "r", "s", "t", "u", "v", "w", "x", "y", "z", "."]

So, yeah, maybe replace the . with something else that is not so full of meaning at the filesystem level.

And downcase to prevent the OSX HFS stupidity

👍