File Parsing question

Question

Say I have a fairly big file. Many megs, for example. My file can be formatted kind of the way I want, but it contains data in a key-value pair format, where the key is a long (ordered) and the value is a string of variable length.
So, for example, it could look something like this:
1blabla
2bobob
3gugugu
99910102923tututu
Now, I know I can get the value for a specific key by parsing the lines, reading a long when reading a string. This is fine.
But I'm trying to be somewhat smarter. Since I know my longs are ordered, I'm thinking I can implement a search algorithm that looks in the middle of an interval for the looked value, to eliminate half the remaining options per pass.
I can use randomaccessfile to get to some random point in the file. That's fine. But now, since I can be at any point of the file, I could be within the long, or somewhere within the string. What I'm looking for, and can't seem to figure out, is how to get back to the beginning of the long value, to read it correctly.
Any pointers appreciated!

Answer 1

First reason is I'm trying to build something for unstructured data. Databases are usually good only for structured data. I could put everything in text, I suppose.
Second reason is databases are, in theory, slower than intelligent file-parsing. You have to generate the connection, have the dbms manipulate the data then return the data then have the driver translate it to something your programming language understands. I know, I Know, I'd have to work years to make something as efficient as an existing DB.
Third reason is databases usually require a database engine, so if I wanted to make a standalone application, with, say, a jar file, I could avoid requiring of my userd to install some database application. Yes, there are some embedded database thignies out there.
Fourth reason is curiosity. I want to see if I can do something like this because I've never done it before, and I'm interested in finding out if it's possible or not!