But hashes are ONE WAY FUNCTIONS!
Trivial example - the hash function is strlen().
That means the hash of "SEND ONE DOLLAR" and "SEND 1M DOLLARS" are the same. But there is zero information in the hash result to tell you which one it was.
The hash of an entire book (say war and peace) would still only be 40 characters. Are you seriously thinking you could recover the entire book? Who needs ZIP, when we've got your reversible hash.
Now SHA1 is more secure, because there is a lot less chance of getting the same result from any given two messages.
It is also more secure, because it is hard to arrange for a particular message to have a particular SHA1 result.
You can't get back the original text just by "reversing" the process, even if you could.
Read Bruce Schneier's post from 5 YEARS AGO.
If you've only managed a couple of rounds, are you doing anything which isn't just brute force (or something slightly less expensive).
Sha512. That's a great algorithm to crack. But sha1 first.
As for about how bytes are lost and removed, I have created algorithms that generates those bytes from existing data. That is a very simple and effective concept. Got me very far until I came across this one variable which I will need to write a big algorithm for. So when you think about it I am like Macgyver and Neo combined. I see everything as a matrix and use existing items to solve the hardest problems. With that concept in mind I should be able to crack this layer of sha1 and possibly the entire algorithm.