Parse: adding token start and end position in input string

[dimashi]

I would like to remove some parts from the address string keeping original casing. For example I'd like to remove unit number from the address. I parse full address, get all lower cased parts, then I want to construct address without unit number. Since parsed output skips all the stop words (spaces, commas) it would help to add to each token start and end position in the original string so I can link token to exact postion in the string.
Another option is to return original input in original casing, and stop words.