Reading a file line-by-line in reverse order is fundamentally tricky.

It's not too bad if you've got a fixed width encoding. It's feasible if you've got a variable width encoding which you can detect the first byte of etc (e.g. UTF-8). It's virtually impossible to do efficiently if the encoding is variable width with no sensible way of determining boundaries (or if it uses "shifting" for example).

I have an implementation in C# in another question, but it would take a fair amount of effort to port that to Java.

Reading a file line-by-line in reverse order is fundamentally tricky.

It's not too bad if you've got a fixed width encoding. It's feasible if you've got a variable width encoding which you can detect the first byte of etc (e.g. UTF-8). It's virtually impossible to do efficiently if the encoding is variable width with no sensible way of determining boundaries (or if it uses "shifting" for example).

I have an implementation in C# in another question, but it would take a fair amount of effort to port that to Java.