/**
* writes a single page
*
* @param valueCount count of values
* @param uncompressedPageSize the size of the data once uncompressed
* @param bytes the compressed data for the page without header
* @param rlEncoding encoding of the repetition level
* @param dlEncoding encoding of the definition level
* @param valuesEncoding encoding of values
*/
public void writeDataPage(
int valueCount,
int uncompressedPageSize,
BytesInput bytes,
Statistics statistics,
parquet.column.Encoding rlEncoding,
parquet.column.Encoding dlEncoding,
parquet.column.Encoding valuesEncoding)
throws IOException {
state = state.write();
long beforeHeader = out.getPos();
if (DEBUG) LOG.debug(beforeHeader + ": write data page: " + valueCount + " values");
int compressedPageSize = (int) bytes.size();
metadataConverter.writeDataPageHeader(
uncompressedPageSize,
compressedPageSize,
valueCount,
statistics,
rlEncoding,
dlEncoding,
valuesEncoding,
out);
long headerSize = out.getPos() - beforeHeader;
this.uncompressedLength += uncompressedPageSize + headerSize;
this.compressedLength += compressedPageSize + headerSize;
if (DEBUG) LOG.debug(out.getPos() + ": write data page content " + compressedPageSize);
bytes.writeAllTo(out);
currentStatistics.mergeStatistics(statistics);
currentEncodings.add(rlEncoding);
currentEncodings.add(dlEncoding);
currentEncodings.add(valuesEncoding);
}
/**
* start a column inside a block
*
* @param descriptor the column descriptor
* @param valueCount the value count in this column
* @param statistics the statistics in this column
* @param compressionCodecName
* @throws IOException
*/
public void startColumn(
ColumnDescriptor descriptor, long valueCount, CompressionCodecName compressionCodecName)
throws IOException {
state = state.startColumn();
if (DEBUG) LOG.debug(out.getPos() + ": start column: " + descriptor + " count=" + valueCount);
currentEncodings = new HashSet<parquet.column.Encoding>();
currentChunkPath = ColumnPath.get(descriptor.getPath());
currentChunkType = descriptor.getType();
currentChunkCodec = compressionCodecName;
currentChunkValueCount = valueCount;
currentChunkFirstDataPage = out.getPos();
compressedLength = 0;
uncompressedLength = 0;
// need to know what type of stats to initialize to
// better way to do this?
currentStatistics = Statistics.getStatsBasedOnType(currentChunkType);
}
