The readChar() method displays japanese character

Question

I'm trying to write a code that pick-up a word from a file according to an index entered by the user but the problem is that the method readChar() from the RandomAccessFile class is returning japanese characters, I must admit that it's not the first time that I've seen this on my lenovo laptop , sometimes on some installation wizards I can see mixed stuff with normal characters mixed with japanese characters, do you think it comes from the laptop or rather from the code?

This is the code:

package com.project;

import java.io.*;
import java.util.StringTokenizer;

public class Main {

    public static void main(String[] args) throws IOException {
        int N, i=0;
        char C;
        char[] charArray = new char[100];
        String fileLocation = "file.txt";
        BufferedReader buffer = new BufferedReader(new InputStreamReader(System.in));
        do {
            System.out.println("enter the index of the word");
            N = Integer.parseInt(buffer.readLine());
            if (N!=0) {
                RandomAccessFile word = new RandomAccessFile(new File(fileLocation), "r");
                do {
                    word.seek((2*(N-1))+i);
                    C = word.readChar();
                    charArray[i] = C;
                    i++;
                }while(charArray[i-1] != ' ');
                System.out.println("the word of index " + N + " is: " );
                for (char carTemp : charArray )
                System.out.print(carTemp);
                System.out.print("\n");

            }
        }while(N!=0);
        buffer.close();
    }
}

i get this output :

瑯潕啰灰灥敲牃䍡慳獥攨⠩⤍ഊੴ瑯潌䱯潷睥敲牃䍡慳獥攨⠩⤍ഊ੣捯潭浣捡慴琨⡓却瑲物楮湧朩⤍ഊ੣捨桡慲牁䅴琨⡩楮湴琩⤍ഊੳ獵畢扳獴瑲物楮湧木⠠⁳獴瑡慲牴琠⁩楮湤摥數砬Ⱐ⁥敮湤搠⁩楮湤摥數砩⤍ഊੴ瑲物業洨⠩Exception in thread "main" java.lang.ArrayIndexOutOfBoundsException: Index 100 out of bounds for length 100
    at Main.main(Main.java:21)

Answer 1

char is 16 bits, i.e. 2 bytes.

seek seeks to a byte boundary.

If the file contains chars then they are at even offsets: 0, 2, 4...

The expression (2*(N-1))+i) is even iff i is even; if odd, you are sure to land in the middle of a char, and thus read garbage.

i starts at zero, but you increment by 1, i.e., half a character.

Your seek argument should probably be (2*(N-1+i)).

---

Alternative explanation: your file does not contain chars at all; for example, you created an ASCII file in which a character is a single byte.

In that case, the error is attempting to read ASCII (an obsolete character encoding) with a readChar function.

But if the file contains ASCII, the purpose of multiplying by 2 in the seek argument is obscure. It apparently serves no useful purpose.

Answer 2

There are many things wrong, all of which have to do with fundamental misconceptions.

First off: A file on your disk - never mind the File interface in Java, or any other programming language; the file itself - does not and cannot store text. Ever. It stores bytes. That is, raw data, as (on every machine that's been relevant for decades, but historically there have been other ways to do it) quantified in bits, which are organized into groups of 8 that are called bytes.

Text is an abstraction; an interpretation of some particular sequence of byte values. It depends - fundamentally and unavoidably - on an encoding. Because this isn't a blog, I'll spare you the history lesson here, but suffice to say that Java's char type does not simply store a character of text. It stores an unsigned two-byte value, which may represent a character of text. Because there are more characters of text in Unicode than two bytes can represent, sometimes two adjacent chars in an array are required to represent a character of text. (And, of course, there is probably code out there that abuses the char type simply because someone wanted an unsigned equivalent of short. I may even have written some myself. That era is a blur for me.)

Anyway, the point is: using .readChar() is going to read two bytes from your file, and store them into a char within your char[], and the corresponding numeric value is not going to be anything like the one you wanted - unless your file happens to be encoded using the same encoding that Java uses natively, called UTF-16.

You cannot properly read and interpret the file without knowing the file encoding. Full stop. You can at best delude yourself into believing that you can read it. You also cannot have "random access" to a text file - i.e., indexing according to a number of characters of text - unless the encoding in question is constant width. (Otherwise, of course, you can't just calculate the distance-in-bytes into the file where a given character of text is; it depends on how many bytes the previous characters took up, which depends on which characters they are.) Many text encodings are not constant width. One of the most popular, which frankly is the sane default recommendation for most tasks these days, is not. In which case you are simply out of luck for the problem you describe.

At any rate, once you know the encoding of your file, the expected way to retrieve a character of text from a file in Java is to use one of the Reader classes, such as InputStreamReader:

An InputStreamReader is a bridge from byte streams to character streams: It reads bytes and decodes them into characters using a specified charset. The charset that it uses may be specified by name or may be given explicitly, or the platform's default charset may be accepted.

(Here, charset simply means an instance of the class that Java uses to represent text encodings.)

You may be able to fudge your problem description a little bit: seek to a byte offset, and then grab the text characters starting at that offset. However, there is no guarantee that the "text characters starting at that offset" make any sense, or in fact can be decoded at all. If the offset happens to be in the middle of a multi-byte encoding for a character, the remaining part isn't necessarily valid encoded text.

Answer 3

I changed the encoding of the file to UTF-16 and modified the programe in order to display the right indexes, those that represents the beginning of each word, now it works fine, Thank you guys.

  import java.io.*;

public class Main {
public static void main(String[] args) throws IOException {
    int N, i=0, j=0, k=0;
    char C;
    char[] charArray = new char[100];
    String fileLocation = "file.txt";
    BufferedReader buffer = new BufferedReader(new InputStreamReader(System.in));
    DataInputStream in = new DataInputStream(new FileInputStream(fileLocation));
    boolean EOF=false;
    do {
        try {
            j++;
            C = in.readChar();
           if((C==' ')||(C=='\n')){
                System.out.print(j+1+"\t");
            }

        }catch (IOException e){
            EOF=true;
        }

    }while (EOF!=true);
    System.out.println("\n");
    do {
        System.out.println("enter the index of the word");
        N = Integer.parseInt(buffer.readLine());
        if (N!=0) {
            RandomAccessFile word = new RandomAccessFile(new File(fileLocation), "r");


            do {

                word.seek((2*(N-1+i)));
                C = word.readChar();
                charArray[i] = C;
                i++;
            }while(charArray[i-1] != ' ' && charArray[i-1] != '\n');
            System.out.print("the word of index " + N + " is: " );
            for (char carTemp : charArray )
                System.out.print(carTemp);
            System.out.print("\n");
             i=0;
            charArray = new char[100];
        }
    }while(N!=0);
    buffer.close();


   }
}