gistfile1.txt

import java.io.*;

//Custom java md5 implementaiton, use whatever you have available, as long as you stream bytes
//into it.
import com.twmacinta.util.MD5;

public class FileHash
{

  /*
   * Computes the start of mp3 audio data, skipping over any ID3v2 tag, if it exists.
   * The original implementation has bug in it where its reading 10 more bytes into the audio
   * then it actually should be if there is an ID3v2 tag (see if you can see the bug).
   * Unfortunately this needs to stay in as all our MD5Hashes are computed this way.
   */

  private static long mp3StartByte( InputStream mp3Stream ) throws IOException
  {
    byte[] id3header = new byte[4];
    int size = -10;
    mp3Stream.read( id3header, 0, 3 );
    // Look for ID3v2
    if( ( id3header[0] == 'I' ) && ( id3header[1] == 'D' ) && ( id3header[2] == '3' ) )
    {
      mp3Stream.read( id3header, 0, 3 );
      mp3Stream.read( id3header, 0, 4 );
      size = ( id3header[0] << 21 ) + ( id3header[1] << 14 ) + ( id3header[2] << 7 )
              + ( id3header[3] );
    }
    return size + 10;
  }

  public static String mp3Hash(File mp3File) {
    return MD5.asHex(FileHash.mp3ByteHash(mp3File));
  }

  
  public static byte[] mp3ByteHash( File mp3File )
  {
    int buf = 8192;
    //This is the size of a ID3v1 tag, which would be at the end of the mp3.
    int id3 = 128;
    FileInputStream mp3Stream = null;
    MD5 md5 = new MD5();

    try
    {
      //Open file and create necessary buffers
      mp3Stream = new FileInputStream( mp3File );

      long fileLength = mp3File.length();
      byte[] buffer = new byte[buf];
      long curByte;

      //Initializes the md5 object
      md5.Init();
      //compute start byte offset of mp3 audio
      long startByte = mp3StartByte( mp3Stream );
      //closes the stream again so we can compute the md5 from the beginning
      mp3Stream.close();
      //open it back up and skip to the audio start position
      mp3Stream = new FileInputStream( mp3File );
      mp3Stream.skip( startByte );
      //start reading the audio and pipe it to the md5 object.  We stop reading right
      //when we reach the last 128 bytes of the file, to check if its an ID3v1 tag.
      int read = 0;
      for( curByte = startByte; curByte + buf < fileLength - id3; curByte += read )
      {
        read = mp3Stream.read( buffer );
        md5.Update( buffer, read );
      }

      //Read in the last 128 Bytes of the file
      Long tempLong = new Long( fileLength - curByte - id3 );
      buffer = new byte[tempLong.intValue()];
      read = mp3Stream.read( buffer );
      md5.Update( buffer, read );
      buffer = new byte[3];

      read = mp3Stream.read( buffer );

      //Read the first 3 bytes of that last 128 and see if its a TAG string.
      //If it is, its a ID3v1 tag and we should use it to compute the md5, if its
      //not then its not an ID3 tag and we should use it in the md5 computation.
      String tag = new String( buffer );
      if( !tag.equals( "TAG" ) )
      {
        md5.Update( buffer, read );
        buffer = new byte[id3 - read];
        while( read != -1 )
        {
          read = mp3Stream.read( buffer );
          if( read != -1 )
            md5.Update( buffer, read );
        }
      }

      //compute the md5 and return it.
      return md5.Final();
    }
    catch( Exception e )
    {
      e.printStackTrace();
      return null;
    }
    finally
    {
      if( mp3Stream != null )
      {
        try
        {
          mp3Stream.close();
        }
        catch( IOException e )
        {
        }
      }
    }
  }
}