Question

How can I read certain number of elements (characters, specifically) at a time in Java? It’s a little difficult to explain, but my idea is this:

If I have a text file that contains:

This is a text file named text.txt

I want to be able to have a String or a character array of a certain length that iterates through the file. So if I specified the length to be 3, first iteration the char array would contain [T,h,i], and if I iterate through it once, it would become [h,i,s], and then [i,s, ], and so on.

I have tried using the BufferedReader.read(char[], off, len) method which reads certain number of characters at a time from the file, but performance is important for what I’m trying to do.

Is there any method to achieve this in Java? I’ve tried using BufferedReader but I’m not too familiar with it to fully utilize it.

Answer 1

You’ll actually get the best I/O performance by buffering both the input stream and the reader. (Buffering just one gives most of the improvement; double buffering is only a bit better, but it is better.) Here’s sample code to read a file a chunk at a time:

final int CHUNK_SIZE = 3;
final int BUFFER_SIZE = 8192; // explicit buffer size is better
File file = ...
InputStream is = new BufferedInputStream(new FileInputStream(file), BUFFER_SIZE);
Reader rdr = new BufferedReader(new InputStreamReader(is), BUFFER_SIZE);
char buff = new char[CHUNK_SIZE];
int len;
while ((len = rdr.read(buff)) != -1) {
   // buff[0] through buff[len-1] are valid
}
rdr.close();

This, of course, is missing all sorts of error checking, exception handling, etc., but it shows the basic idea of buffering streams and readers. You may also want to specify a character encoding when constructing the InputStreamReader. (You could bypass dealing with input streams by using a FileReader to start with, but then you cannot specify a character set encoding and lose the slight performance boost that comes from double buffering.)