private long getFilePosition() throws IOException {
   long retVal;
   if (isCompressedInput && null != filePosition) {
     retVal = filePosition.getPos();
   } else {
     retVal = pos;
   }
   return retVal;
 }
Exemple #2
0
  /**
   * Creates CompressionInputStream to be used to read off uncompressed data in one of the two
   * reading modes. i.e. Continuous or Blocked reading modes
   *
   * @param seekableIn The InputStream
   * @param start The start offset into the compressed stream
   * @param end The end offset into the compressed stream
   * @param readMode Controls whether progress is reported continuously or only at block boundaries.
   * @return CompressionInputStream for BZip2 aligned at block boundaries
   */
  public SplitCompressionInputStream createInputStream(
      InputStream seekableIn, Decompressor decompressor, long start, long end, READ_MODE readMode)
      throws IOException {

    if (!(seekableIn instanceof Seekable)) {
      throw new IOException("seekableIn must be an instance of " + Seekable.class.getName());
    }

    // find the position of first BZip2 start up marker
    ((Seekable) seekableIn).seek(0);

    // BZip2 start of block markers are of 6 bytes.  But the very first block
    // also has "BZh9", making it 10 bytes.  This is the common case.  But at
    // time stream might start without a leading BZ.
    final long FIRST_BZIP2_BLOCK_MARKER_POSITION =
        CBZip2InputStream.numberOfBytesTillNextMarker(seekableIn);
    long adjStart = Math.max(0L, start - FIRST_BZIP2_BLOCK_MARKER_POSITION);

    ((Seekable) seekableIn).seek(adjStart);
    SplitCompressionInputStream in =
        new BZip2CompressionInputStream(seekableIn, adjStart, end, readMode);

    // The following if clause handles the following case:
    // Assume the following scenario in BZip2 compressed stream where
    // . represent compressed data.
    // .....[48 bit Block].....[48 bit   Block].....[48 bit Block]...
    // ........................[47 bits][1 bit].....[48 bit Block]...
    // ................................^[Assume a Byte alignment here]
    // ........................................^^[current position of stream]
    // .....................^^[We go back 10 Bytes in stream and find a Block marker]
    // ........................................^^[We align at wrong position!]
    // ...........................................................^^[While this pos is correct]

    if (in.getPos() <= start) {
      ((Seekable) seekableIn).seek(start);
      in = new BZip2CompressionInputStream(seekableIn, start, end, readMode);
    }

    return in;
  }