Java Code Examples for org.apache.kylin.dict.DictionaryGenerator#newDictionaryBuilder()

The following examples show how to use org.apache.kylin.dict.DictionaryGenerator#newDictionaryBuilder() . You can vote up the ones you like or vote down the ones you don't like, and go to the original project or source file by following the links above each example. You may check out the related API usage on the sidebar.
Example 1
Source File: UHCDictionaryReducer.java    From kylin-on-parquet-v2 with Apache License 2.0 5 votes vote down vote up 
@Override
protected void doSetup(Context context) throws IOException {
    super.bindCurrentConfiguration(context.getConfiguration());
    Configuration conf = context.getConfiguration();
    mos = new MultipleOutputs(context);

    KylinConfig config = AbstractHadoopJob.loadKylinPropsAndMetadata();
    String cubeName = conf.get(BatchConstants.CFG_CUBE_NAME);
    CubeInstance cube = CubeManager.getInstance(config).getCube(cubeName);
    CubeDesc cubeDesc = cube.getDescriptor();
    List<TblColRef> uhcColumns = cubeDesc.getAllUHCColumns();

    int taskId = context.getTaskAttemptID().getTaskID().getId();
    col = uhcColumns.get(taskId);
    logger.info("column name: " + col.getIdentity());

    if (cube.getDescriptor().getShardByColumns().contains(col)) {
        //for ShardByColumns
        builder = DictionaryGenerator.newDictionaryBuilder(col.getType());
        builder.init(null, 0, null);
    } else {
        //for GlobalDictionaryColumns
        String hdfsDir = conf.get(BatchConstants.CFG_GLOBAL_DICT_BASE_DIR);
        DictionaryInfo dictionaryInfo = new DictionaryInfo(col.getColumnDesc(), col.getDatatype());
        String builderClass = cubeDesc.getDictionaryBuilderClass(col);
        builder = (IDictionaryBuilder) ClassUtil.newInstance(builderClass);
        builder.init(dictionaryInfo, 0, hdfsDir);
    }
}
Example 2
Source File: SparkUHCDictionary.java    From kylin-on-parquet-v2 with Apache License 2.0 5 votes vote down vote up 
@Override
public Tuple2<String, Tuple3<Writable, Writable, String>> call(Tuple2<Integer, List<String>> columnValues) throws Exception {
    if (initialized == false) {
        synchronized (SparkFactDistinct.class) {
            if (initialized == false) {
                init();
            }
        }
    }

    try (KylinConfig.SetAndUnsetThreadLocalConfig autoUnset = KylinConfig.setAndUnsetThreadLocalConfig(config);
         ByteArrayOutputStream baos = new ByteArrayOutputStream();
         DataOutputStream outputStream = new DataOutputStream(baos)) {
        TblColRef col = uhcColumns.get(columnValues._1);
        logger.info("Processing column " + col.getName());
        if (cube.getDescriptor().getShardByColumns().contains(col)) {
            //for ShardByColumns
            builder = DictionaryGenerator.newDictionaryBuilder(col.getType());
            builder.init(null, 0, null);
        } else {
            //for GlobalDictionaryColumns
            DictionaryInfo dictionaryInfo = new DictionaryInfo(col.getColumnDesc(), col.getDatatype());
            String builderClass = cubeDesc.getDictionaryBuilderClass(col);
            builder = (IDictionaryBuilder) ClassUtil.newInstance(builderClass);
            builder.init(dictionaryInfo, 0, hdfsDir);
        }
        Iterator<String> values = columnValues._2.iterator();
        while (values.hasNext()) {
            builder.addValue(values.next());
        }
        Dictionary<String> dict = builder.build();
        String dictFileName = col.getIdentity() + "/" + col.getName() + DICT_FILE_POSTFIX;
        logger.info("Dictionary file name is " + dictFileName);

        outputStream.writeUTF(dict.getClass().getName());
        dict.write(outputStream);
        Tuple3 tuple3 = new Tuple3(NullWritable.get(), new ArrayPrimitiveWritable(baos.toByteArray()), dictFileName);
        return new Tuple2<>(BatchConstants.CFG_OUTPUT_DICT, tuple3);
    }
}
Example 3
Source File: UHCDictionaryReducer.java    From kylin with Apache License 2.0 5 votes vote down vote up 
@Override
protected void doSetup(Context context) throws IOException {
    super.bindCurrentConfiguration(context.getConfiguration());
    Configuration conf = context.getConfiguration();
    mos = new MultipleOutputs(context);

    KylinConfig config = AbstractHadoopJob.loadKylinPropsAndMetadata();
    String cubeName = conf.get(BatchConstants.CFG_CUBE_NAME);
    CubeInstance cube = CubeManager.getInstance(config).getCube(cubeName);
    CubeDesc cubeDesc = cube.getDescriptor();
    List<TblColRef> uhcColumns = cubeDesc.getAllUHCColumns();

    int taskId = context.getTaskAttemptID().getTaskID().getId();
    col = uhcColumns.get(taskId);
    logger.info("column name: " + col.getIdentity());

    if (cube.getDescriptor().getShardByColumns().contains(col)) {
        //for ShardByColumns
        builder = DictionaryGenerator.newDictionaryBuilder(col.getType());
        builder.init(null, 0, null);
    } else {
        //for GlobalDictionaryColumns
        String hdfsDir = conf.get(BatchConstants.CFG_GLOBAL_DICT_BASE_DIR);
        DictionaryInfo dictionaryInfo = new DictionaryInfo(col.getColumnDesc(), col.getDatatype());
        String builderClass = cubeDesc.getDictionaryBuilderClass(col);
        builder = (IDictionaryBuilder) ClassUtil.newInstance(builderClass);
        builder.init(dictionaryInfo, 0, hdfsDir);
    }
}
Example 4
Source File: FactDistinctColumnsBase.java    From kylin with Apache License 2.0 5 votes vote down vote up 
public void setupReduce(int taskId) throws IOException {
    this.taskId = taskId;
    try (KylinConfig.SetAndUnsetThreadLocalConfig autoUnset = KylinConfig
            .setAndUnsetThreadLocalConfig(envConfig)) {
        cube = CubeManager.getInstance(envConfig).getCube(cubeName);
        cubeDesc = cube.getDescriptor();
        reducerMapping = new FactDistinctColumnsReducerMapping(cube);
        logger.info("reducer no " + taskId + ", role play " + reducerMapping.getRolePlayOfReducer(taskId));

        if (reducerMapping.isCuboidRowCounterReducer(taskId)) {
            // hll
            isStatistics = true;
            baseCuboidId = cube.getCuboidScheduler().getBaseCuboidId();
            baseCuboidRowCountInMappers = Lists.newArrayList();
            cuboidHLLMap = Maps.newHashMap();
            logger.info("Reducer " + taskId + " handling stats");
        } else {
            // normal col
            col = reducerMapping.getColForReducer(taskId);
            Preconditions.checkNotNull(col);

            // local build dict
            buildDictInReducer = envConfig.isBuildDictInReducerEnabled();
            if (cubeDesc.getDictionaryBuilderClass(col) != null) { // only works with default dictionary builder
                buildDictInReducer = false;
            }
            if (reducerMapping.getReducerNumForDimCol(col) > 1) {
                buildDictInReducer = false; // only works if this is the only reducer of a dictionary column
            }
            if (buildDictInReducer) {
                builder = DictionaryGenerator.newDictionaryBuilder(col.getType());
                builder.init(null, 0, null);
            }
            logger.info("Reducer " + taskId + " handling column " + col + ", buildDictInReducer=" + buildDictInReducer);
        }
    }
}
Example 5
Source File: SparkUHCDictionary.java    From kylin with Apache License 2.0 5 votes vote down vote up 
@Override
public Tuple2<String, Tuple3<Writable, Writable, String>> call(Tuple2<Integer, List<String>> columnValues) throws Exception {
    if (initialized == false) {
        synchronized (SparkFactDistinct.class) {
            if (initialized == false) {
                init();
            }
        }
    }

    try (KylinConfig.SetAndUnsetThreadLocalConfig autoUnset = KylinConfig.setAndUnsetThreadLocalConfig(config);
         ByteArrayOutputStream baos = new ByteArrayOutputStream();
         DataOutputStream outputStream = new DataOutputStream(baos)) {
        TblColRef col = uhcColumns.get(columnValues._1);
        logger.info("Processing column " + col.getName());
        if (cube.getDescriptor().getShardByColumns().contains(col)) {
            //for ShardByColumns
            builder = DictionaryGenerator.newDictionaryBuilder(col.getType());
            builder.init(null, 0, null);
        } else {
            //for GlobalDictionaryColumns
            DictionaryInfo dictionaryInfo = new DictionaryInfo(col.getColumnDesc(), col.getDatatype());
            String builderClass = cubeDesc.getDictionaryBuilderClass(col);
            builder = (IDictionaryBuilder) ClassUtil.newInstance(builderClass);
            builder.init(dictionaryInfo, 0, hdfsDir);
        }
        Iterator<String> values = columnValues._2.iterator();
        while (values.hasNext()) {
            builder.addValue(values.next());
        }
        Dictionary<String> dict = builder.build();
        String dictFileName = col.getIdentity() + "/" + col.getName() + DICT_FILE_POSTFIX;
        logger.info("Dictionary file name is " + dictFileName);

        outputStream.writeUTF(dict.getClass().getName());
        dict.write(outputStream);
        Tuple3 tuple3 = new Tuple3(NullWritable.get(), new ArrayPrimitiveWritable(baos.toByteArray()), dictFileName);
        return new Tuple2<>(BatchConstants.CFG_OUTPUT_DICT, tuple3);
    }
}
Example 6
Source File: FactDistinctColumnsReducer.java    From kylin-on-parquet-v2 with Apache License 2.0 4 votes vote down vote up 
@Override
protected void doSetup(Context context) throws IOException {
    super.bindCurrentConfiguration(context.getConfiguration());
    Configuration conf = context.getConfiguration();
    mos = new MultipleOutputs(context);

    KylinConfig config = AbstractHadoopJob.loadKylinPropsAndMetadata();
    String cubeName = conf.get(BatchConstants.CFG_CUBE_NAME);
    CubeInstance cube = CubeManager.getInstance(config).getCube(cubeName);
    cubeConfig = cube.getConfig();
    cubeDesc = cube.getDescriptor();

    taskId = context.getTaskAttemptID().getTaskID().getId();

    reducerMapping = new FactDistinctColumnsReducerMapping(cube);

    logger.info("reducer no " + taskId + ", role play " + reducerMapping.getRolePlayOfReducer(taskId));

    if (reducerMapping.isCuboidRowCounterReducer(taskId)) {
        // hll
        isStatistics = true;
        baseCuboidId = cube.getCuboidScheduler().getBaseCuboidId();
        baseCuboidRowCountInMappers = Lists.newArrayList();
        cuboidHLLMap = Maps.newHashMap();
        samplingPercentage = Integer
                .parseInt(context.getConfiguration().get(BatchConstants.CFG_STATISTICS_SAMPLING_PERCENT));
        logger.info("Reducer " + taskId + " handling stats");
    } else {
        // normal col
        col = reducerMapping.getColForReducer(taskId);
        Preconditions.checkNotNull(col);

        // local build dict
        buildDictInReducer = config.isBuildDictInReducerEnabled();
        if (cubeDesc.getDictionaryBuilderClass(col) != null) { // only works with default dictionary builder
            buildDictInReducer = false;
        }
        if (reducerMapping.getReducerNumForDimCol(col) > 1) {
            buildDictInReducer = false; // only works if this is the only reducer of a dictionary column
        }
        if (buildDictInReducer) {
            builder = DictionaryGenerator.newDictionaryBuilder(col.getType());
            builder.init(null, 0, null);
        }
        logger.info("Reducer " + taskId + " handling column " + col + ", buildDictInReducer=" + buildDictInReducer);
    }
}
Example 7
Source File: SparkFactDistinct.java    From kylin-on-parquet-v2 with Apache License 2.0 4 votes vote down vote up 
private void init() throws IOException {
    taskId = TaskContext.getPartitionId();
    kConfig = AbstractHadoopJob.loadKylinConfigFromHdfs(conf, metaUrl);
    try (KylinConfig.SetAndUnsetThreadLocalConfig autoUnset = KylinConfig
            .setAndUnsetThreadLocalConfig(kConfig)) {
        CubeInstance cubeInstance = CubeManager.getInstance(kConfig).getCube(cubeName);
        cubeDesc = cubeInstance.getDescriptor();
        cubeConfig = cubeInstance.getConfig();
        reducerMapping = new FactDistinctColumnsReducerMapping(cubeInstance);

        result = Lists.newArrayList();

        if (reducerMapping.isCuboidRowCounterReducer(taskId)) {
            // hll
            isStatistics = true;
            baseCuboidId = cubeInstance.getCuboidScheduler().getBaseCuboidId();
            baseCuboidRowCountInMappers = Lists.newArrayList();
            cuboidHLLMap = Maps.newHashMap();

            logger.info("Partition {} handling stats", taskId);
        } else {
            // normal col
            col = reducerMapping.getColForReducer(taskId);
            Preconditions.checkNotNull(col);

            isDimensionCol = cubeDesc.listDimensionColumnsExcludingDerived(true).contains(col) && col.getType().needCompare();
            isDictCol = cubeDesc.getAllColumnsNeedDictionaryBuilt().contains(col);

            // local build dict
            buildDictInReducer = kConfig.isBuildDictInReducerEnabled();
            if (cubeDesc.getDictionaryBuilderClass(col) != null) { // only works with default dictionary builder
                buildDictInReducer = false;
            }

            if (reducerMapping.getReducerNumForDimCol(col) > 1) {
                buildDictInReducer = false; // only works if this is the only reducer of a dictionary column
            }

            if (buildDictInReducer) {
                builder = DictionaryGenerator.newDictionaryBuilder(col.getType());
                builder.init(null, 0, null);
            }
            logger.info("Partition {} handling column {}, buildDictInReducer={}", taskId, col, buildDictInReducer);
        }

        initialized = true;
    }
}
Example 8
Source File: FactDistinctColumnsReducer.java    From kylin with Apache License 2.0 4 votes vote down vote up 
@Override
protected void doSetup(Context context) throws IOException {
    super.bindCurrentConfiguration(context.getConfiguration());
    Configuration conf = context.getConfiguration();
    mos = new MultipleOutputs(context);

    KylinConfig config = AbstractHadoopJob.loadKylinPropsAndMetadata();
    String cubeName = conf.get(BatchConstants.CFG_CUBE_NAME);
    CubeInstance cube = CubeManager.getInstance(config).getCube(cubeName);
    cubeConfig = cube.getConfig();
    cubeDesc = cube.getDescriptor();

    taskId = context.getTaskAttemptID().getTaskID().getId();

    reducerMapping = new FactDistinctColumnsReducerMapping(cube);

    logger.info("reducer no " + taskId + ", role play " + reducerMapping.getRolePlayOfReducer(taskId));

    if (reducerMapping.isCuboidRowCounterReducer(taskId)) {
        // hll
        isStatistics = true;
        baseCuboidId = cube.getCuboidScheduler().getBaseCuboidId();
        baseCuboidRowCountInMappers = Lists.newArrayList();
        cuboidHLLMap = Maps.newHashMap();
        samplingPercentage = Integer
                .parseInt(context.getConfiguration().get(BatchConstants.CFG_STATISTICS_SAMPLING_PERCENT));
        logger.info("Reducer " + taskId + " handling stats");
    } else {
        // normal col
        col = reducerMapping.getColForReducer(taskId);
        Preconditions.checkNotNull(col);

        // local build dict
        buildDictInReducer = config.isBuildDictInReducerEnabled();
        if (cubeDesc.getDictionaryBuilderClass(col) != null) { // only works with default dictionary builder
            buildDictInReducer = false;
        }
        if (reducerMapping.getReducerNumForDimCol(col) > 1) {
            buildDictInReducer = false; // only works if this is the only reducer of a dictionary column
        }
        if (buildDictInReducer) {
            builder = DictionaryGenerator.newDictionaryBuilder(col.getType());
            builder.init(null, 0, null);
        }
        logger.info("Reducer " + taskId + " handling column " + col + ", buildDictInReducer=" + buildDictInReducer);
    }
}
Example 9
Source File: SparkFactDistinct.java    From kylin with Apache License 2.0 4 votes vote down vote up 
private void init() throws IOException {
    taskId = TaskContext.getPartitionId();
    kConfig = AbstractHadoopJob.loadKylinConfigFromHdfs(conf, metaUrl);
    try (KylinConfig.SetAndUnsetThreadLocalConfig autoUnset = KylinConfig
            .setAndUnsetThreadLocalConfig(kConfig)) {
        CubeInstance cubeInstance = CubeManager.getInstance(kConfig).getCube(cubeName);
        cubeDesc = cubeInstance.getDescriptor();
        cubeConfig = cubeInstance.getConfig();
        reducerMapping = new FactDistinctColumnsReducerMapping(cubeInstance);

        result = Lists.newArrayList();

        if (reducerMapping.isCuboidRowCounterReducer(taskId)) {
            // hll
            isStatistics = true;
            baseCuboidId = cubeInstance.getCuboidScheduler().getBaseCuboidId();
            baseCuboidRowCountInMappers = Lists.newArrayList();
            cuboidHLLMap = Maps.newHashMap();

            logger.info("Partition {} handling stats", taskId);
        } else {
            // normal col
            col = reducerMapping.getColForReducer(taskId);
            Preconditions.checkNotNull(col);

            isDimensionCol = cubeDesc.listDimensionColumnsExcludingDerived(true).contains(col) && col.getType().needCompare();
            isDictCol = cubeDesc.getAllColumnsNeedDictionaryBuilt().contains(col);

            // local build dict
            buildDictInReducer = kConfig.isBuildDictInReducerEnabled();
            if (cubeDesc.getDictionaryBuilderClass(col) != null) { // only works with default dictionary builder
                buildDictInReducer = false;
            }

            if (reducerMapping.getReducerNumForDimCol(col) > 1) {
                buildDictInReducer = false; // only works if this is the only reducer of a dictionary column
            }

            if (buildDictInReducer) {
                builder = DictionaryGenerator.newDictionaryBuilder(col.getType());
                builder.init(null, 0, null);
            }
            logger.info("Partition {} handling column {}, buildDictInReducer={}", taskId, col, buildDictInReducer);
        }

        initialized = true;
    }
}