Search in sources :

Example 1 with CarbonDictionaryWriter

use of org.apache.carbondata.core.writer.CarbonDictionaryWriter in project carbondata by apache.

the class IncrementalColumnDictionaryGenerator method writeDictionaryData.

@Override
public void writeDictionaryData(String tableUniqueName) throws IOException {
    // initialize params
    CarbonMetadata metadata = CarbonMetadata.getInstance();
    CarbonTable carbonTable = metadata.getCarbonTable(tableUniqueName);
    CarbonTableIdentifier tableIdentifier = carbonTable.getCarbonTableIdentifier();
    ColumnIdentifier columnIdentifier = dimension.getColumnIdentifier();
    String storePath = carbonTable.getStorePath();
    DictionaryService dictionaryService = CarbonCommonFactory.getDictionaryService();
    // create dictionary cache from dictionary File
    DictionaryColumnUniqueIdentifier identifier = new DictionaryColumnUniqueIdentifier(tableIdentifier, columnIdentifier, columnIdentifier.getDataType());
    Boolean isDictExists = CarbonUtil.isFileExistsForGivenColumn(storePath, identifier);
    Dictionary dictionary = null;
    long t1 = System.currentTimeMillis();
    if (isDictExists) {
        Cache<DictionaryColumnUniqueIdentifier, Dictionary> dictCache = CacheProvider.getInstance().createCache(CacheType.REVERSE_DICTIONARY, storePath);
        dictionary = dictCache.get(identifier);
    }
    long dictCacheTime = System.currentTimeMillis() - t1;
    long t2 = System.currentTimeMillis();
    // write dictionary
    CarbonDictionaryWriter dictionaryWriter = null;
    dictionaryWriter = dictionaryService.getDictionaryWriter(tableIdentifier, columnIdentifier, storePath);
    List<String> distinctValues = writeDictionary(dictionaryWriter, isDictExists);
    long dictWriteTime = System.currentTimeMillis() - t2;
    long t3 = System.currentTimeMillis();
    // write sort index
    if (distinctValues.size() > 0) {
        writeSortIndex(distinctValues, dictionary, dictionaryService, tableIdentifier, columnIdentifier, storePath);
    }
    long sortIndexWriteTime = System.currentTimeMillis() - t3;
    // update Meta Data
    updateMetaData(dictionaryWriter);
    LOGGER.audit("\n columnName: " + dimension.getColName() + "\n columnId: " + dimension.getColumnId() + "\n new distinct values count: " + distinctValues.size() + "\n create dictionary cache: " + dictCacheTime + "\n sort list, distinct and write: " + dictWriteTime + "\n write sort info: " + sortIndexWriteTime);
}
Also used : Dictionary(org.apache.carbondata.core.cache.dictionary.Dictionary) BiDictionary(org.apache.carbondata.core.devapi.BiDictionary) CarbonTable(org.apache.carbondata.core.metadata.schema.table.CarbonTable) DictionaryService(org.apache.carbondata.core.service.DictionaryService) DictionaryColumnUniqueIdentifier(org.apache.carbondata.core.cache.dictionary.DictionaryColumnUniqueIdentifier) CarbonTableIdentifier(org.apache.carbondata.core.metadata.CarbonTableIdentifier) ColumnIdentifier(org.apache.carbondata.core.metadata.ColumnIdentifier) CarbonMetadata(org.apache.carbondata.core.metadata.CarbonMetadata) CarbonDictionaryWriter(org.apache.carbondata.core.writer.CarbonDictionaryWriter)

Example 2 with CarbonDictionaryWriter

use of org.apache.carbondata.core.writer.CarbonDictionaryWriter in project carbondata by apache.

the class CarbonDictionarySortIndexReaderImplTest method read.

/**
 * Test to read the data from dictionary sort index file
 *
 * @throws Exception
 */
@Test
public void read() throws Exception {
    deleteStorePath();
    CarbonTableIdentifier carbonTableIdentifier = new CarbonTableIdentifier("testSchema", "carbon", UUID.randomUUID().toString());
    AbsoluteTableIdentifier absoluteTableIdentifier = AbsoluteTableIdentifier.from(storePath + "/testSchema/carbon", carbonTableIdentifier);
    ColumnIdentifier columnIdentifier = new ColumnIdentifier("Name", null, null);
    DictionaryColumnUniqueIdentifier dictionaryColumnUniqueIdentifier = new DictionaryColumnUniqueIdentifier(absoluteTableIdentifier, columnIdentifier, columnIdentifier.getDataType());
    CarbonDictionaryWriter dictionaryWriter = new CarbonDictionaryWriterImpl(dictionaryColumnUniqueIdentifier);
    String metaFolderPath = storePath + File.separator + carbonTableIdentifier.getDatabaseName() + File.separator + carbonTableIdentifier.getTableName() + File.separator + "Metadata";
    CarbonUtil.checkAndCreateFolder(metaFolderPath);
    CarbonDictionarySortIndexWriter dictionarySortIndexWriter = new CarbonDictionarySortIndexWriterImpl(dictionaryColumnUniqueIdentifier);
    List<int[]> expectedData = prepareExpectedData();
    int[] data = expectedData.get(0);
    for (int i = 0; i < data.length; i++) {
        dictionaryWriter.write(String.valueOf(data[i]));
    }
    dictionaryWriter.close();
    dictionaryWriter.commit();
    List<Integer> sortIndex = Arrays.asList(ArrayUtils.toObject(expectedData.get(0)));
    List<Integer> invertedSortIndex = Arrays.asList(ArrayUtils.toObject(expectedData.get(1)));
    dictionarySortIndexWriter.writeSortIndex(sortIndex);
    dictionarySortIndexWriter.writeInvertedSortIndex(invertedSortIndex);
    dictionarySortIndexWriter.close();
    CarbonDictionarySortIndexReader dictionarySortIndexReader = new CarbonDictionarySortIndexReaderImpl(dictionaryColumnUniqueIdentifier);
    List<Integer> actualSortIndex = dictionarySortIndexReader.readSortIndex();
    List<Integer> actualInvertedSortIndex = dictionarySortIndexReader.readInvertedSortIndex();
    for (int i = 0; i < actualSortIndex.size(); i++) {
        Assert.assertEquals(sortIndex.get(i), actualSortIndex.get(i));
        Assert.assertEquals(invertedSortIndex.get(i), actualInvertedSortIndex.get(i));
    }
}
Also used : CarbonDictionaryWriterImpl(org.apache.carbondata.core.writer.CarbonDictionaryWriterImpl) CarbonDictionarySortIndexWriter(org.apache.carbondata.core.writer.sortindex.CarbonDictionarySortIndexWriter) DictionaryColumnUniqueIdentifier(org.apache.carbondata.core.cache.dictionary.DictionaryColumnUniqueIdentifier) CarbonTableIdentifier(org.apache.carbondata.core.metadata.CarbonTableIdentifier) AbsoluteTableIdentifier(org.apache.carbondata.core.metadata.AbsoluteTableIdentifier) CarbonDictionarySortIndexWriterImpl(org.apache.carbondata.core.writer.sortindex.CarbonDictionarySortIndexWriterImpl) ColumnIdentifier(org.apache.carbondata.core.metadata.ColumnIdentifier) CarbonDictionaryWriter(org.apache.carbondata.core.writer.CarbonDictionaryWriter) Test(org.junit.Test)

Example 3 with CarbonDictionaryWriter

use of org.apache.carbondata.core.writer.CarbonDictionaryWriter in project carbondata by apache.

the class AbstractDictionaryCacheTest method prepareWriterAndWriteData.

/**
 * write dictionary data
 *
 * @param data
 * @throws IOException
 */
protected void prepareWriterAndWriteData(List<String> data, String columnId) throws IOException {
    ColumnIdentifier columnIdentifier = new ColumnIdentifier(columnId, null, null);
    DictionaryColumnUniqueIdentifier dictionaryColumnUniqueIdentifier = new DictionaryColumnUniqueIdentifier(identifier, columnIdentifier, columnIdentifier.getDataType());
    CarbonDictionaryWriter carbonDictionaryWriter = new CarbonDictionaryWriterImpl(dictionaryColumnUniqueIdentifier);
    CarbonUtil.checkAndCreateFolder(CarbonTablePath.getMetadataPath(identifier.getTablePath()));
    List<byte[]> valueList = convertStringListToByteArray(data);
    try {
        carbonDictionaryWriter.write(valueList);
    } finally {
        carbonDictionaryWriter.close();
        carbonDictionaryWriter.commit();
    }
}
Also used : CarbonDictionaryWriterImpl(org.apache.carbondata.core.writer.CarbonDictionaryWriterImpl) ColumnIdentifier(org.apache.carbondata.core.metadata.ColumnIdentifier) CarbonDictionaryWriter(org.apache.carbondata.core.writer.CarbonDictionaryWriter)

Example 4 with CarbonDictionaryWriter

use of org.apache.carbondata.core.writer.CarbonDictionaryWriter in project carbondata by apache.

the class StoreCreator method writeDictionary.

private static void writeDictionary(String factFilePath, CarbonTable table) throws Exception {
    BufferedReader reader = new BufferedReader(new FileReader(factFilePath));
    String header = reader.readLine();
    String[] split = header.split(",");
    List<CarbonColumn> allCols = new ArrayList<CarbonColumn>();
    List<CarbonDimension> dims = table.getDimensionByTableName(table.getFactTableName());
    allCols.addAll(dims);
    List<CarbonMeasure> msrs = table.getMeasureByTableName(table.getFactTableName());
    allCols.addAll(msrs);
    Set<String>[] set = new HashSet[dims.size()];
    for (int i = 0; i < set.length; i++) {
        set[i] = new HashSet<String>();
    }
    String line = reader.readLine();
    while (line != null) {
        String[] data = line.split(",");
        for (int i = 0; i < set.length; i++) {
            set[i].add(data[i]);
        }
        line = reader.readLine();
    }
    Cache dictCache = CacheProvider.getInstance().createCache(CacheType.REVERSE_DICTIONARY, absoluteTableIdentifier.getStorePath());
    for (int i = 0; i < set.length; i++) {
        ColumnIdentifier columnIdentifier = new ColumnIdentifier(dims.get(i).getColumnId(), null, null);
        CarbonDictionaryWriter writer = new CarbonDictionaryWriterImpl(absoluteTableIdentifier.getStorePath(), absoluteTableIdentifier.getCarbonTableIdentifier(), columnIdentifier);
        for (String value : set[i]) {
            writer.write(value);
        }
        writer.close();
        writer.commit();
        Dictionary dict = (Dictionary) dictCache.get(new DictionaryColumnUniqueIdentifier(absoluteTableIdentifier.getCarbonTableIdentifier(), columnIdentifier, dims.get(i).getDataType()));
        CarbonDictionarySortInfoPreparator preparator = new CarbonDictionarySortInfoPreparator();
        List<String> newDistinctValues = new ArrayList<String>();
        CarbonDictionarySortInfo dictionarySortInfo = preparator.getDictionarySortInfo(newDistinctValues, dict, dims.get(i).getDataType());
        CarbonDictionarySortIndexWriter carbonDictionaryWriter = new CarbonDictionarySortIndexWriterImpl(absoluteTableIdentifier.getCarbonTableIdentifier(), columnIdentifier, absoluteTableIdentifier.getStorePath());
        try {
            carbonDictionaryWriter.writeSortIndex(dictionarySortInfo.getSortIndex());
            carbonDictionaryWriter.writeInvertedSortIndex(dictionarySortInfo.getSortIndexInverted());
        } finally {
            carbonDictionaryWriter.close();
        }
    }
    reader.close();
}
Also used : Dictionary(org.apache.carbondata.core.cache.dictionary.Dictionary) CarbonColumn(org.apache.carbondata.core.metadata.schema.table.column.CarbonColumn) Set(java.util.Set) HashSet(java.util.HashSet) CarbonDictionaryWriterImpl(org.apache.carbondata.core.writer.CarbonDictionaryWriterImpl) ArrayList(java.util.ArrayList) CarbonDictionarySortIndexWriterImpl(org.apache.carbondata.core.writer.sortindex.CarbonDictionarySortIndexWriterImpl) FileReader(java.io.FileReader) HashSet(java.util.HashSet) CarbonDictionarySortInfoPreparator(org.apache.carbondata.core.writer.sortindex.CarbonDictionarySortInfoPreparator) CarbonDictionarySortInfo(org.apache.carbondata.core.writer.sortindex.CarbonDictionarySortInfo) CarbonDimension(org.apache.carbondata.core.metadata.schema.table.column.CarbonDimension) CarbonDictionarySortIndexWriter(org.apache.carbondata.core.writer.sortindex.CarbonDictionarySortIndexWriter) DictionaryColumnUniqueIdentifier(org.apache.carbondata.core.cache.dictionary.DictionaryColumnUniqueIdentifier) CarbonMeasure(org.apache.carbondata.core.metadata.schema.table.column.CarbonMeasure) BufferedReader(java.io.BufferedReader) ColumnIdentifier(org.apache.carbondata.core.metadata.ColumnIdentifier) CarbonDictionaryWriter(org.apache.carbondata.core.writer.CarbonDictionaryWriter) Cache(org.apache.carbondata.core.cache.Cache)

Example 5 with CarbonDictionaryWriter

use of org.apache.carbondata.core.writer.CarbonDictionaryWriter in project carbondata by apache.

the class StoreCreator method writeDictionary.

private static void writeDictionary(String factFilePath, CarbonTable table) throws Exception {
    BufferedReader reader = new BufferedReader(new FileReader(factFilePath));
    String header = reader.readLine();
    String[] split = header.split(",");
    List<CarbonColumn> allCols = new ArrayList<CarbonColumn>();
    List<CarbonDimension> dims = table.getDimensionByTableName(table.getTableName());
    allCols.addAll(dims);
    List<CarbonMeasure> msrs = table.getMeasureByTableName(table.getTableName());
    allCols.addAll(msrs);
    Set<String>[] set = new HashSet[dims.size()];
    for (int i = 0; i < set.length; i++) {
        set[i] = new HashSet<String>();
    }
    String line = reader.readLine();
    while (line != null) {
        String[] data = line.split(",");
        for (int i = 0; i < set.length; i++) {
            set[i].add(data[i]);
        }
        line = reader.readLine();
    }
    Cache dictCache = CacheProvider.getInstance().createCache(CacheType.REVERSE_DICTIONARY);
    for (int i = 0; i < set.length; i++) {
        ColumnIdentifier columnIdentifier = new ColumnIdentifier(dims.get(i).getColumnId(), null, null);
        DictionaryColumnUniqueIdentifier dictionaryColumnUniqueIdentifier = new DictionaryColumnUniqueIdentifier(table.getAbsoluteTableIdentifier(), columnIdentifier, columnIdentifier.getDataType());
        CarbonDictionaryWriter writer = new CarbonDictionaryWriterImpl(dictionaryColumnUniqueIdentifier);
        for (String value : set[i]) {
            writer.write(value);
        }
        writer.close();
        writer.commit();
        Dictionary dict = (Dictionary) dictCache.get(new DictionaryColumnUniqueIdentifier(identifier, columnIdentifier, dims.get(i).getDataType()));
        CarbonDictionarySortInfoPreparator preparator = new CarbonDictionarySortInfoPreparator();
        List<String> newDistinctValues = new ArrayList<String>();
        CarbonDictionarySortInfo dictionarySortInfo = preparator.getDictionarySortInfo(newDistinctValues, dict, dims.get(i).getDataType());
        CarbonDictionarySortIndexWriter carbonDictionaryWriter = new CarbonDictionarySortIndexWriterImpl(dictionaryColumnUniqueIdentifier);
        try {
            carbonDictionaryWriter.writeSortIndex(dictionarySortInfo.getSortIndex());
            carbonDictionaryWriter.writeInvertedSortIndex(dictionarySortInfo.getSortIndexInverted());
        } finally {
            carbonDictionaryWriter.close();
        }
    }
    reader.close();
}
Also used : Dictionary(org.apache.carbondata.core.cache.dictionary.Dictionary) CarbonColumn(org.apache.carbondata.core.metadata.schema.table.column.CarbonColumn) Set(java.util.Set) HashSet(java.util.HashSet) CarbonDictionaryWriterImpl(org.apache.carbondata.core.writer.CarbonDictionaryWriterImpl) ArrayList(java.util.ArrayList) CarbonDictionarySortIndexWriterImpl(org.apache.carbondata.core.writer.sortindex.CarbonDictionarySortIndexWriterImpl) FileReader(java.io.FileReader) HashSet(java.util.HashSet) CarbonDictionarySortInfoPreparator(org.apache.carbondata.core.writer.sortindex.CarbonDictionarySortInfoPreparator) CarbonDictionarySortInfo(org.apache.carbondata.core.writer.sortindex.CarbonDictionarySortInfo) CarbonDimension(org.apache.carbondata.core.metadata.schema.table.column.CarbonDimension) CarbonDictionarySortIndexWriter(org.apache.carbondata.core.writer.sortindex.CarbonDictionarySortIndexWriter) DictionaryColumnUniqueIdentifier(org.apache.carbondata.core.cache.dictionary.DictionaryColumnUniqueIdentifier) CarbonMeasure(org.apache.carbondata.core.metadata.schema.table.column.CarbonMeasure) BufferedReader(java.io.BufferedReader) ColumnIdentifier(org.apache.carbondata.core.metadata.ColumnIdentifier) CarbonDictionaryWriter(org.apache.carbondata.core.writer.CarbonDictionaryWriter) Cache(org.apache.carbondata.core.cache.Cache)

Aggregations

ColumnIdentifier (org.apache.carbondata.core.metadata.ColumnIdentifier)7 CarbonDictionaryWriter (org.apache.carbondata.core.writer.CarbonDictionaryWriter)7 DictionaryColumnUniqueIdentifier (org.apache.carbondata.core.cache.dictionary.DictionaryColumnUniqueIdentifier)6 Dictionary (org.apache.carbondata.core.cache.dictionary.Dictionary)5 CarbonDictionaryWriterImpl (org.apache.carbondata.core.writer.CarbonDictionaryWriterImpl)5 CarbonDictionarySortIndexWriter (org.apache.carbondata.core.writer.sortindex.CarbonDictionarySortIndexWriter)4 CarbonDictionarySortIndexWriterImpl (org.apache.carbondata.core.writer.sortindex.CarbonDictionarySortIndexWriterImpl)4 BufferedReader (java.io.BufferedReader)3 ArrayList (java.util.ArrayList)3 HashSet (java.util.HashSet)3 Set (java.util.Set)3 Cache (org.apache.carbondata.core.cache.Cache)3 CarbonDimension (org.apache.carbondata.core.metadata.schema.table.column.CarbonDimension)3 CarbonDictionarySortInfo (org.apache.carbondata.core.writer.sortindex.CarbonDictionarySortInfo)3 CarbonDictionarySortInfoPreparator (org.apache.carbondata.core.writer.sortindex.CarbonDictionarySortInfoPreparator)3 FileReader (java.io.FileReader)2 BiDictionary (org.apache.carbondata.core.devapi.BiDictionary)2 AbsoluteTableIdentifier (org.apache.carbondata.core.metadata.AbsoluteTableIdentifier)2 CarbonTableIdentifier (org.apache.carbondata.core.metadata.CarbonTableIdentifier)2 CarbonColumn (org.apache.carbondata.core.metadata.schema.table.column.CarbonColumn)2