Lucene - searching for a specific part of engine for my science work

Question

Hello all.
First of all, a short introduction:
I`m performing science work in my university that are connected with symbol processing using Markov productions. I give you an example of it:
AB -> BC
It`s a production that means replace all ocurence of AB in word with BC (as a word can be count a sentence or whole document or even a binary file , cause it consists with symbols '1' and '0' in raw, and we can use smth like this '1001'->'1100'). In common case it work in such way: scan word for first occurence of AB, change if to BC, scan word for very begining once again for first occurence in changed word and so on. It stops if it scan whole word but didn`t find Ab. So as you can it`s bad that we have to look the whole word from very begining after every change (cause we are afraid that in previous part of work after change we can find AB). I`m developing some ways to change all ocurences of ABs in word for one pass and I also develope a benchmark that allows to model how fast work Markov production in classic and with my improvements. But now I need to show that this improvments can affect on real project. So I need a project that in its logic have a part with a lot of replacement espessialy with replaceAll, so I can change it my function that will replace for one pass all occurences (so it should be open source). I decided that Lucene engine can help me, cause in theory during building indexes files there are should be a lot of transfers of index words. But in practice, looking in source index building and next step - optimization in this engine doesn`t use replace or replaceAll in explicit or implicit way. Am I right? May be some one can give me a piece of advise where I can find what I need in Lucene? Or may be someone show another opensource project where I can make a real benchmark?
P.S. I suppose my funtions cann`t beat Java replaceAll, cause I suppose it don`t work in raw with return to begin of word, but I need real result even if it will negative.

Answer 1

I decided that Lucene engine can help me, cause in theory during building indexes files there are should be a lot of transfers of index words.I don't understand how you got from 'a lot of transfers of index words' to 'a lot of replacement especially with replaceAll()'. Why would building a Lucene index need replaceAll()?
You'd be better off looking into something like 'sed' I would have thought.