Reading image files into Strings

[shadowayex]

For a program I'm working on, it's required to read in data from an image file, store it in as a string in it's own file, and then be able to read the data and put it into a BufferedImage. I wrote some testing code to try to get a hang of what the process might need.

If I read in the data as a byte array and use a byte stream to create the BufferedImage, it displays fine in a Swing application:

import java.awt.Dimension;
import java.awt.image.BufferedImage;
import java.io.ByteArrayInputStream;
import java.io.File;
import java.io.IOException;

import javax.imageio.ImageIO;
import javax.imageio.stream.FileImageInputStream;
import javax.swing.ImageIcon;
import javax.swing.JFrame;
import javax.swing.JLabel;

public class TestCode
{
	public static void main(String[] args)
	{
		try
		{
			File f = new File("star.jpg");
			FileImageInputStream im = new FileImageInputStream(f);

			byte[] data = new byte[(int) f.length()];
			int read = 0;
			int bytesRead = 0;

			while (read < data.length
					&& (bytesRead = im.read(data, read, data.length - read)) > -1)
			{
				read += bytesRead;
			}

			BufferedImage img = ImageIO.read(new ByteArrayInputStream(data));

			JFrame frame = new JFrame();
			frame.setDefaultCloseOperation(JFrame.EXIT_ON_CLOSE);
			frame.setPreferredSize(new Dimension(800, 600));
			frame.getContentPane().add(new JLabel(new ImageIcon(img)));
			frame.pack();
			frame.setVisible(true);
		}
		catch (IOException e)
		{
			e.printStackTrace();
		}
	}
}

The image displays just as it should. But if I convert the byte array into a String, use getBytes on that String, and supply that array to the byte stream, the image comes out wrong.

import java.awt.Dimension;
import java.awt.image.BufferedImage;
import java.io.ByteArrayInputStream;
import java.io.File;
import java.io.IOException;

import javax.imageio.ImageIO;
import javax.imageio.stream.FileImageInputStream;
import javax.swing.ImageIcon;
import javax.swing.JFrame;
import javax.swing.JLabel;

public class TestCode
{
	public static void main(String[] args)
	{
		try
		{
			File f = new File("star.jpg");
			FileImageInputStream im = new FileImageInputStream(f);

			byte[] data = new byte[(int) f.length()];
			int read = 0;
			int bytesRead = 0;

			while (read < data.length
					&& (bytesRead = im.read(data, read, data.length - read)) > -1)
			{
				read += bytesRead;
			}

			String dataStr = new String(data);
			BufferedImage img = ImageIO.read(new ByteArrayInputStream(dataStr.getBytes()));

			JFrame frame = new JFrame();
			frame.setDefaultCloseOperation(JFrame.EXIT_ON_CLOSE);
			frame.setPreferredSize(new Dimension(800, 600));
			frame.getContentPane().add(new JLabel(new ImageIcon(img)));
			frame.pack();
			frame.setVisible(true);
		}
		catch (IOException e)
		{
			e.printStackTrace();
		}
	}
}

Is there a particular process I have to do when converting between the byte array and String that will make this work? Also, as soon as I find a solid method I can use in a closed-source proprietary program, I plan on encoding the data in Base64. I don't know if that changes anything, but I figured I should mention it.

[Fubarable]

You may be running into a Charset issue where bytes that aren't part of the default Charset are being transformed into the byte equivalent of '?'. What if you read data into a String with a different Charset, and read it out with the same?

For example:

         while (read < data.length
               && (bytesRead = im.read(data, read, data.length - read)) > -1) {
            read += bytesRead;
         }

         System.out.println("charset: "
               + Charset.defaultCharset().displayName());

         Charset charset = Charset.forName("ISO-8859-1");
         String dataStr = new String(data, charset);

         byte[] data2 = dataStr.getBytes(charset);

         BufferedImage img = ImageIO.read(new ByteArrayInputStream(data2));

[shadowayex]

ISO-8859-1 does work on this image. Is there any way to figure out how images are encoded? Or is there a standard that says all images have to be encoded under a certain charset?

[Fubarable]

The images aren't "encoded" since they aren't Strings. You do need to use a Charset that allows maximum flexibility and accepts all bytes. Probably better would be to manually write and read the byte Strings in and out.

[shadowayex]

Ah ok. Well the only one that doesn't give me an error in my code is ISO-8859-1. Would that be a safe one? I'll know nothing about the incoming files. They are user supplied.

[Fubarable]

I don't know, but the whole thing sounds iffy to me. Again if I were to do this and I absolutely had to use a String, I'd go the safe route and write each byte out explicitly, possibly to a StringBuilder. Then I'd have no need to worry about Charset.

[shadowayex]

Ok. Well it's being saved into an XML file with other data. The current design is to generate the XML and save it in a String and write that all to a file using an output stream. That's the reason for converting it to a String. I do happen to use a StringBuilder for that functionality, so it wouldn't be any extra work to use a StringBuilder.

Thanks for all the help by the way. I've (obviously) not learned much about encoding, working with bytes, and other things of that nature.

[Fubarable]

And I'm not a pro regarding Charsets either, so please take all advice I give on this subject with a grain of salt. I'm learning as I go! Best of luck!