Examples with FileHolder org.apache.carbondata.core.datastore.FileHolder used on opensource projects

Example 1 with FileHolder

use of org.apache.carbondata.core.datastore.FileHolder in project carbondata by apache.

the class CarbonUtil method calculateMetaSize.

/**
   * The method calculate the B-Tree metadata size.
   *
   * @param tableBlockInfo
   * @return
   */
public static long calculateMetaSize(TableBlockInfo tableBlockInfo) throws IOException {
    FileHolder fileReader = null;
    try {
        long completeBlockLength = tableBlockInfo.getBlockLength();
        long footerPointer = completeBlockLength - 8;
        String filePath = tableBlockInfo.getFilePath();
        fileReader = FileFactory.getFileHolder(FileFactory.getFileType(filePath));
        long actualFooterOffset = fileReader.readLong(filePath, footerPointer);
        return footerPointer - actualFooterOffset;
    } finally {
        if (null != fileReader) {
            try {
                fileReader.finish();
            } catch (IOException e) {
            // ignore the exception as nothing we can do about it
            }
        }
    }
}

Example 2 with FileHolder

use of org.apache.carbondata.core.datastore.FileHolder in project carbondata by apache.

the class DataFileFooterConverter method readDataFileFooter.

/**
   * Below method will be used to convert thrift file meta to wrapper file meta
   */
@Override
public DataFileFooter readDataFileFooter(TableBlockInfo tableBlockInfo) throws IOException {
    DataFileFooter dataFileFooter = new DataFileFooter();
    FileHolder fileReader = null;
    try {
        long completeBlockLength = tableBlockInfo.getBlockLength();
        long footerPointer = completeBlockLength - 8;
        fileReader = FileFactory.getFileHolder(FileFactory.getFileType(tableBlockInfo.getFilePath()));
        long actualFooterOffset = fileReader.readLong(tableBlockInfo.getFilePath(), footerPointer);
        CarbonFooterReader reader = new CarbonFooterReader(tableBlockInfo.getFilePath(), actualFooterOffset);
        FileFooter footer = reader.readFooter();
        dataFileFooter.setVersionId(ColumnarFormatVersion.valueOf((short) footer.getVersion()));
        dataFileFooter.setNumberOfRows(footer.getNum_rows());
        dataFileFooter.setSegmentInfo(getSegmentInfo(footer.getSegment_info()));
        List<ColumnSchema> columnSchemaList = new ArrayList<ColumnSchema>();
        List<org.apache.carbondata.format.ColumnSchema> table_columns = footer.getTable_columns();
        for (int i = 0; i < table_columns.size(); i++) {
            columnSchemaList.add(thriftColumnSchmeaToWrapperColumnSchema(table_columns.get(i)));
        }
        dataFileFooter.setColumnInTable(columnSchemaList);
        List<org.apache.carbondata.format.BlockletIndex> leaf_node_indices_Thrift = footer.getBlocklet_index_list();
        List<BlockletIndex> blockletIndexList = new ArrayList<BlockletIndex>();
        for (int i = 0; i < leaf_node_indices_Thrift.size(); i++) {
            BlockletIndex blockletIndex = getBlockletIndex(leaf_node_indices_Thrift.get(i));
            blockletIndexList.add(blockletIndex);
        }
        List<org.apache.carbondata.format.BlockletInfo> leaf_node_infos_Thrift = footer.getBlocklet_info_list();
        List<BlockletInfo> blockletInfoList = new ArrayList<BlockletInfo>();
        for (int i = 0; i < leaf_node_infos_Thrift.size(); i++) {
            BlockletInfo blockletInfo = getBlockletInfo(leaf_node_infos_Thrift.get(i));
            blockletInfo.setBlockletIndex(blockletIndexList.get(i));
            blockletInfoList.add(blockletInfo);
        }
        dataFileFooter.setBlockletList(blockletInfoList);
        dataFileFooter.setBlockletIndex(getBlockletIndexForDataFileFooter(blockletIndexList));
    } finally {
        if (null != fileReader) {
            fileReader.finish();
        }
    }
    return dataFileFooter;
}

Example 3 with FileHolder

use of org.apache.carbondata.core.datastore.FileHolder in project carbondata by apache.

the class FilterScanner method fillScannedResult.

/**
   * This method will process the data in below order
   * 1. first apply min max on the filter tree and check whether any of the filter
   * is fall on the range of min max, if not then return empty result
   * 2. If filter falls on min max range then apply filter on actual
   * data and get the filtered row index
   * 3. if row index is empty then return the empty result
   * 4. if row indexes is not empty then read only those blocks(measure or dimension)
   * which was present in the query but not present in the filter, as while applying filter
   * some of the blocks where already read and present in chunk holder so not need to
   * read those blocks again, this is to avoid reading of same blocks which was already read
   * 5. Set the blocks and filter indexes to result
   *
   * @param blocksChunkHolder
   * @throws FilterUnsupportedException
   */
private AbstractScannedResult fillScannedResult(BlocksChunkHolder blocksChunkHolder) throws FilterUnsupportedException, IOException {
    long startTime = System.currentTimeMillis();
    QueryStatistic totalBlockletStatistic = queryStatisticsModel.getStatisticsTypeAndObjMap().get(QueryStatisticsConstants.TOTAL_BLOCKLET_NUM);
    totalBlockletStatistic.addCountStatistic(QueryStatisticsConstants.TOTAL_BLOCKLET_NUM, totalBlockletStatistic.getCount() + 1);
    // apply filter on actual data
    BitSetGroup bitSetGroup = this.filterExecuter.applyFilter(blocksChunkHolder);
    // if indexes is empty then return with empty result
    if (bitSetGroup.isEmpty()) {
        CarbonUtil.freeMemory(blocksChunkHolder.getDimensionRawDataChunk(), blocksChunkHolder.getMeasureRawDataChunk());
        QueryStatistic scanTime = queryStatisticsModel.getStatisticsTypeAndObjMap().get(QueryStatisticsConstants.SCAN_BLOCKlET_TIME);
        scanTime.addCountStatistic(QueryStatisticsConstants.SCAN_BLOCKlET_TIME, scanTime.getCount() + (System.currentTimeMillis() - startTime));
        QueryStatistic scannedPages = queryStatisticsModel.getStatisticsTypeAndObjMap().get(QueryStatisticsConstants.PAGE_SCANNED);
        scannedPages.addCountStatistic(QueryStatisticsConstants.PAGE_SCANNED, scannedPages.getCount() + bitSetGroup.getScannedPages());
        return createEmptyResult();
    }
    AbstractScannedResult scannedResult = new FilterQueryScannedResult(blockExecutionInfo);
    scannedResult.setBlockletId(blockExecutionInfo.getBlockId() + CarbonCommonConstants.FILE_SEPARATOR + blocksChunkHolder.getDataBlock().nodeNumber());
    // valid scanned blocklet
    QueryStatistic validScannedBlockletStatistic = queryStatisticsModel.getStatisticsTypeAndObjMap().get(QueryStatisticsConstants.VALID_SCAN_BLOCKLET_NUM);
    validScannedBlockletStatistic.addCountStatistic(QueryStatisticsConstants.VALID_SCAN_BLOCKLET_NUM, validScannedBlockletStatistic.getCount() + 1);
    // adding statistics for valid number of pages
    QueryStatistic validPages = queryStatisticsModel.getStatisticsTypeAndObjMap().get(QueryStatisticsConstants.VALID_PAGE_SCANNED);
    validPages.addCountStatistic(QueryStatisticsConstants.VALID_PAGE_SCANNED, validPages.getCount() + bitSetGroup.getValidPages());
    QueryStatistic scannedPages = queryStatisticsModel.getStatisticsTypeAndObjMap().get(QueryStatisticsConstants.PAGE_SCANNED);
    scannedPages.addCountStatistic(QueryStatisticsConstants.PAGE_SCANNED, scannedPages.getCount() + bitSetGroup.getScannedPages());
    int[] rowCount = new int[bitSetGroup.getNumberOfPages()];
    // get the row indexes from bot set
    int[][] indexesGroup = new int[bitSetGroup.getNumberOfPages()][];
    for (int k = 0; k < indexesGroup.length; k++) {
        BitSet bitSet = bitSetGroup.getBitSet(k);
        if (bitSet != null && !bitSet.isEmpty()) {
            int[] indexes = new int[bitSet.cardinality()];
            int index = 0;
            for (int i = bitSet.nextSetBit(0); i >= 0; i = bitSet.nextSetBit(i + 1)) {
                indexes[index++] = i;
            }
            rowCount[k] = indexes.length;
            indexesGroup[k] = indexes;
        }
    }
    // loading delete data cache in blockexecutioninfo instance
    DeleteDeltaCacheLoaderIntf deleteCacheLoader = new BlockletDeleteDeltaCacheLoader(scannedResult.getBlockletId(), blocksChunkHolder.getDataBlock(), blockExecutionInfo.getAbsoluteTableIdentifier());
    deleteCacheLoader.loadDeleteDeltaFileDataToCache();
    scannedResult.setBlockletDeleteDeltaCache(blocksChunkHolder.getDataBlock().getDeleteDeltaDataCache());
    FileHolder fileReader = blocksChunkHolder.getFileReader();
    int[][] allSelectedDimensionBlocksIndexes = blockExecutionInfo.getAllSelectedDimensionBlocksIndexes();
    long dimensionReadTime = System.currentTimeMillis();
    DimensionRawColumnChunk[] projectionListDimensionChunk = blocksChunkHolder.getDataBlock().getDimensionChunks(fileReader, allSelectedDimensionBlocksIndexes);
    dimensionReadTime = System.currentTimeMillis() - dimensionReadTime;
    DimensionRawColumnChunk[] dimensionRawColumnChunks = new DimensionRawColumnChunk[blockExecutionInfo.getTotalNumberDimensionBlock()];
    // read dimension chunk blocks from file which is not present
    for (int i = 0; i < dimensionRawColumnChunks.length; i++) {
        if (null != blocksChunkHolder.getDimensionRawDataChunk()[i]) {
            dimensionRawColumnChunks[i] = blocksChunkHolder.getDimensionRawDataChunk()[i];
        }
    }
    for (int i = 0; i < allSelectedDimensionBlocksIndexes.length; i++) {
        for (int j = allSelectedDimensionBlocksIndexes[i][0]; j <= allSelectedDimensionBlocksIndexes[i][1]; j++) {
            dimensionRawColumnChunks[j] = projectionListDimensionChunk[j];
        }
    }
    long dimensionReadTime1 = System.currentTimeMillis();
    /**
     * in case projection if the projected dimension are not loaded in the dimensionColumnDataChunk
     * then loading them
     */
    int[] projectionListDimensionIndexes = blockExecutionInfo.getProjectionListDimensionIndexes();
    int projectionListDimensionIndexesLength = projectionListDimensionIndexes.length;
    for (int i = 0; i < projectionListDimensionIndexesLength; i++) {
        if (null == dimensionRawColumnChunks[projectionListDimensionIndexes[i]]) {
            dimensionRawColumnChunks[projectionListDimensionIndexes[i]] = blocksChunkHolder.getDataBlock().getDimensionChunk(fileReader, projectionListDimensionIndexes[i]);
        }
    }
    dimensionReadTime += (System.currentTimeMillis() - dimensionReadTime1);
    dimensionReadTime1 = System.currentTimeMillis();
    MeasureRawColumnChunk[] measureRawColumnChunks = new MeasureRawColumnChunk[blockExecutionInfo.getTotalNumberOfMeasureBlock()];
    int[][] allSelectedMeasureBlocksIndexes = blockExecutionInfo.getAllSelectedMeasureBlocksIndexes();
    MeasureRawColumnChunk[] projectionListMeasureChunk = blocksChunkHolder.getDataBlock().getMeasureChunks(fileReader, allSelectedMeasureBlocksIndexes);
    dimensionReadTime += System.currentTimeMillis() - dimensionReadTime1;
    // read the measure chunk blocks which is not present
    for (int i = 0; i < measureRawColumnChunks.length; i++) {
        if (null != blocksChunkHolder.getMeasureRawDataChunk()[i]) {
            measureRawColumnChunks[i] = blocksChunkHolder.getMeasureRawDataChunk()[i];
        }
    }
    for (int i = 0; i < allSelectedMeasureBlocksIndexes.length; i++) {
        for (int j = allSelectedMeasureBlocksIndexes[i][0]; j <= allSelectedMeasureBlocksIndexes[i][1]; j++) {
            measureRawColumnChunks[j] = projectionListMeasureChunk[j];
        }
    }
    dimensionReadTime1 = System.currentTimeMillis();
    /**
     * in case projection if the projected measure are not loaded in the measureColumnDataChunk
     * then loading them
     */
    int[] projectionListMeasureIndexes = blockExecutionInfo.getProjectionListMeasureIndexes();
    int projectionListMeasureIndexesLength = projectionListMeasureIndexes.length;
    for (int i = 0; i < projectionListMeasureIndexesLength; i++) {
        if (null == measureRawColumnChunks[projectionListMeasureIndexes[i]]) {
            measureRawColumnChunks[projectionListMeasureIndexes[i]] = blocksChunkHolder.getDataBlock().getMeasureChunk(fileReader, projectionListMeasureIndexes[i]);
        }
    }
    dimensionReadTime += System.currentTimeMillis() - dimensionReadTime1;
    DimensionColumnDataChunk[][] dimensionColumnDataChunks = new DimensionColumnDataChunk[dimensionRawColumnChunks.length][indexesGroup.length];
    MeasureColumnDataChunk[][] measureColumnDataChunks = new MeasureColumnDataChunk[measureRawColumnChunks.length][indexesGroup.length];
    for (int i = 0; i < dimensionRawColumnChunks.length; i++) {
        for (int j = 0; j < indexesGroup.length; j++) {
            if (dimensionRawColumnChunks[i] != null) {
                dimensionColumnDataChunks[i][j] = dimensionRawColumnChunks[i].convertToDimColDataChunk(j);
            }
        }
    }
    for (int i = 0; i < measureRawColumnChunks.length; i++) {
        for (int j = 0; j < indexesGroup.length; j++) {
            if (measureRawColumnChunks[i] != null) {
                measureColumnDataChunks[i][j] = measureRawColumnChunks[i].convertToMeasureColDataChunk(j);
            }
        }
    }
    scannedResult.setDimensionChunks(dimensionColumnDataChunks);
    scannedResult.setIndexes(indexesGroup);
    scannedResult.setMeasureChunks(measureColumnDataChunks);
    scannedResult.setRawColumnChunks(dimensionRawColumnChunks);
    scannedResult.setNumberOfRows(rowCount);
    // adding statistics for carbon scan time
    QueryStatistic scanTime = queryStatisticsModel.getStatisticsTypeAndObjMap().get(QueryStatisticsConstants.SCAN_BLOCKlET_TIME);
    scanTime.addCountStatistic(QueryStatisticsConstants.SCAN_BLOCKlET_TIME, scanTime.getCount() + (System.currentTimeMillis() - startTime - dimensionReadTime));
    QueryStatistic readTime = queryStatisticsModel.getStatisticsTypeAndObjMap().get(QueryStatisticsConstants.READ_BLOCKlET_TIME);
    readTime.addCountStatistic(QueryStatisticsConstants.READ_BLOCKlET_TIME, readTime.getCount() + dimensionReadTime);
    return scannedResult;
}

Example 4 with FileHolder

use of org.apache.carbondata.core.datastore.FileHolder in project carbondata by apache.

the class CompressedDimensionChunkFileBasedReaderV1 method convertToDimensionChunk.

@Override
public DimensionColumnDataChunk convertToDimensionChunk(DimensionRawColumnChunk dimensionRawColumnChunk, int pageNumber) throws IOException {
    int blockIndex = dimensionRawColumnChunk.getBlockletId();
    byte[] dataPage = null;
    int[] invertedIndexes = null;
    int[] invertedIndexesReverse = null;
    int[] rlePage = null;
    FileHolder fileReader = dimensionRawColumnChunk.getFileReader();
    ByteBuffer rawData = dimensionRawColumnChunk.getRawData();
    dataPage = COMPRESSOR.unCompressByte(rawData.array(), dimensionRawColumnChunk.getOffSet(), dimensionRawColumnChunk.getLength());
    // if row id block is present then read the row id chunk and uncompress it
    DataChunk dataChunk = dimensionColumnChunk.get(blockIndex);
    if (CarbonUtil.hasEncoding(dataChunk.getEncodingList(), Encoding.INVERTED_INDEX)) {
        byte[] columnIndexData;
        synchronized (fileReader) {
            columnIndexData = fileReader.readByteArray(filePath, dataChunk.getRowIdPageOffset(), dataChunk.getRowIdPageLength());
        }
        invertedIndexes = CarbonUtil.getUnCompressColumnIndex(dataChunk.getRowIdPageLength(), columnIndexData, numberComressor, 0);
        // get the reverse index
        invertedIndexesReverse = getInvertedReverseIndex(invertedIndexes);
    }
    //then actual data based on rle block
    if (CarbonUtil.hasEncoding(dataChunk.getEncodingList(), Encoding.RLE)) {
        // read and uncompress the rle block
        byte[] key;
        synchronized (fileReader) {
            key = fileReader.readByteArray(filePath, dataChunk.getRlePageOffset(), dataChunk.getRlePageLength());
        }
        rlePage = numberComressor.unCompress(key, 0, dataChunk.getRlePageLength());
        // uncompress the data with rle indexes
        dataPage = UnBlockIndexer.uncompressData(dataPage, rlePage, eachColumnValueSize[blockIndex]);
        rlePage = null;
    }
    // fill chunk attributes
    DimensionColumnDataChunk columnDataChunk = null;
    if (dataChunk.isRowMajor()) {
        // to store fixed length column chunk values
        columnDataChunk = new ColumnGroupDimensionDataChunk(dataPage, eachColumnValueSize[blockIndex], numberOfRows);
    } else // and set to data chunk instance
    if (!CarbonUtil.hasEncoding(dataChunk.getEncodingList(), Encoding.DICTIONARY)) {
        columnDataChunk = new VariableLengthDimensionDataChunk(dataPage, invertedIndexes, invertedIndexesReverse, numberOfRows);
    } else {
        // to store fixed length column chunk values
        columnDataChunk = new FixedLengthDimensionDataChunk(dataPage, invertedIndexes, invertedIndexesReverse, numberOfRows, eachColumnValueSize[blockIndex]);
    }
    return columnDataChunk;
}