public boolean nextKeyValue()

in core/src/main/java/com/twitter/elephantbird/mapreduce/input/LzoBinaryBlockRecordReader.java [126:171]

• 33 lines of code
• 7 McCabe index (conditional complexity)

  public boolean nextKeyValue() throws IOException, InterruptedException {
    // If we are past the end of the file split, tell the reader not to read any more new blocks.
    // Then continue reading until the last of the reader's already-parsed values are used up.
    while (true) { // loop to skip over bad records
      if (pos_ > end_) {
        reader_.markNoMoreNewBlocks();
        // Why not pos_ >= end_, stop when we just reach the end?
        // we don't know if we have read all the bytes uncompressed in the current lzo block,
        // only way to make sure that we have read all of the split is to read till the
        // first record that has at least one byte in the next split.
        // As a consequence of this, next split reader skips at least one byte
        // (i.e. skips either partial or full record at the beginning).
      }
      byte[] byteArray = reader_.readNextProtoBytes();

      if(byteArray == null) {
        return false;
      }

      errorTracker.incRecords();
      M decoded = null;
      try {
        decoded = deserializer_.fromBytes(byteArray);
      } catch (DecodeException e) {
        errorTracker.incErrors(e);
        HadoopCompat.incrementCounter(recordErrorsCounter, 1);
        continue;
      }

      if (updatePosition) {
        pos_ = getLzoFilePos();
        updatePosition = false;
      }

      if (decoded != null) {
        key_.set(pos_);
        value_.set(decoded);
        pos_ = getLzoFilePos();

        HadoopCompat.incrementCounter(recordsReadCounter, 1);
        return true;
      } else {
        HadoopCompat.incrementCounter(recordsSkippedCounter, 1);
      }
    }
  }