Character reader (like FileReader) with dynamic repositioning based on characters offset

I am looking for something similar to a RandomAccessFile which would allow a positioning inside a big file based on the character (not the byte) offset, and allow reading from there. The FileReader or most implementations I came accross do not have a seek-like method such as the one provided in RandomAccessFile.

Does any such reader exist?

tl;dr No such Reader exists, because it can't easily be made to work with every encoding.

If the encoding you use is a fixed-width encoding (such as ISO-8859-*, a Windows codepage, ASCII, UCS-2, ...) then this could work, since you'd simply have to multiply the character offset by some constant (depending on the encoding, usually 1) to get the byte offset.

In fact you can easily emulate this yourself, by seeking on the underlying InputStream (make sure not to use a BufferedReader, since the buffering could be thrown off by the seeking).

But a couple of very popular encodings (and a few more less popular ones) are variable-width, meaning that each character can be represented by different numbers of bytes. UTF-8 and UTF-16 ones are well-known examples, but others like Shift-JIS have this property as well.

For variable-width encodings creating such a stream without any indexing or prior knowledge is not possible. One could implement the seek() by just reading the desired number of bytes, but that wouldn't actually have the benefit of a real seek, since you'd actually have to read the "skipped" bytes from disk to know how far to go.