word splitting regex fails with underdot characters

[chchch]

Hi,

I'm using hypher.js with transliterated Sanskrit, and it doesn't play well with characters such as ṇ, ṣ, ḍ, ṭ, etc. The problem seems to be the long regex used to split a string into words (line 107 of hypher.js). I guess your character class doesn't include the unicode ranges for underdot characters. I've replaced it with a simpler expression:
var words = str.split(/([\s\n\r\t.,:;'"!?-])/g);
which matches word boundary characters instead of word characters. It works for me but it's not totally comprehensive... you would have to add a few more boundary characters to it to make it work for more languages...