Examples with StringColumnStatsData org.apache.hadoop.hive.metastore.api.StringColumnStatsData used on opensource projects

Example 11 with StringColumnStatsData

use of org.apache.hadoop.hive.metastore.api.StringColumnStatsData in project hive by apache.

the class TestHiveMetaStoreStatsMerge method testStatsMerge.

public void testStatsMerge() throws Exception {
    int listSize = 0;
    List<ListenerEvent> notifyList = DummyListener.notifyList;
    assertEquals(notifyList.size(), listSize);
    msc.createDatabase(db);
    listSize++;
    assertEquals(listSize, notifyList.size());
    CreateDatabaseEvent dbEvent = (CreateDatabaseEvent) (notifyList.get(listSize - 1));
    assert dbEvent.getStatus();
    msc.createTable(table);
    listSize++;
    assertEquals(notifyList.size(), listSize);
    CreateTableEvent tblEvent = (CreateTableEvent) (notifyList.get(listSize - 1));
    assert tblEvent.getStatus();
    table = msc.getTable(dbName, tblName);
    ColumnStatistics cs = new ColumnStatistics();
    ColumnStatisticsDesc desc = new ColumnStatisticsDesc(true, dbName, tblName);
    cs.setStatsDesc(desc);
    ColumnStatisticsObj obj = new ColumnStatisticsObj();
    obj.setColName("a");
    obj.setColType("string");
    ColumnStatisticsData data = new ColumnStatisticsData();
    StringColumnStatsData scsd = new StringColumnStatsData();
    scsd.setAvgColLen(10);
    scsd.setMaxColLen(20);
    scsd.setNumNulls(30);
    scsd.setNumDVs(123);
    scsd.setBitVectors("{0, 4, 5, 7}{0, 1}{0, 1, 2}{0, 1, 4}{0}{0, 2}{0, 3}{0, 2, 3, 4}{0, 1, 4}{0, 1}{0}{0, 1, 3, 8}{0, 2}{0, 2}{0, 9}{0, 1, 4}");
    data.setStringStats(scsd);
    obj.setStatsData(data);
    cs.addToStatsObj(obj);
    List<ColumnStatistics> colStats = new ArrayList<>();
    colStats.add(cs);
    SetPartitionsStatsRequest request = new SetPartitionsStatsRequest(colStats);
    msc.setPartitionColumnStatistics(request);
    List<String> colNames = new ArrayList<>();
    colNames.add("a");
    StringColumnStatsData getScsd = msc.getTableColumnStatistics(dbName, tblName, colNames).get(0).getStatsData().getStringStats();
    assertEquals(getScsd.getNumDVs(), 123);
    cs = new ColumnStatistics();
    scsd = new StringColumnStatsData();
    scsd.setAvgColLen(20);
    scsd.setMaxColLen(5);
    scsd.setNumNulls(70);
    scsd.setNumDVs(456);
    scsd.setBitVectors("{0, 1}{0, 1}{1, 2, 4}{0, 1, 2}{0, 1, 2}{0, 2}{0, 1, 3, 4}{0, 1}{0, 1}{3, 4, 6}{2}{0, 1}{0, 3}{0}{0, 1}{0, 1, 4}");
    data.setStringStats(scsd);
    obj.setStatsData(data);
    cs.addToStatsObj(obj);
    request = new SetPartitionsStatsRequest(colStats);
    request.setNeedMerge(true);
    msc.setPartitionColumnStatistics(request);
    getScsd = msc.getTableColumnStatistics(dbName, tblName, colNames).get(0).getStatsData().getStringStats();
    assertEquals(getScsd.getAvgColLen(), 20.0);
    assertEquals(getScsd.getMaxColLen(), 20);
    assertEquals(getScsd.getNumNulls(), 100);
    // since metastore is ObjectStore, we use the max function to merge.
    assertEquals(getScsd.getNumDVs(), 456);
}

Example 12 with StringColumnStatsData

use of org.apache.hadoop.hive.metastore.api.StringColumnStatsData in project hive by apache.

the class StatObjectConverter method convertToMPartitionColumnStatistics.

public static MPartitionColumnStatistics convertToMPartitionColumnStatistics(MPartition partition, ColumnStatisticsDesc statsDesc, ColumnStatisticsObj statsObj) throws MetaException, NoSuchObjectException {
    if (statsDesc == null || statsObj == null) {
        return null;
    }
    MPartitionColumnStatistics mColStats = new MPartitionColumnStatistics();
    mColStats.setPartition(partition);
    mColStats.setDbName(statsDesc.getDbName());
    mColStats.setTableName(statsDesc.getTableName());
    mColStats.setPartitionName(statsDesc.getPartName());
    mColStats.setLastAnalyzed(statsDesc.getLastAnalyzed());
    mColStats.setColName(statsObj.getColName());
    mColStats.setColType(statsObj.getColType());
    if (statsObj.getStatsData().isSetBooleanStats()) {
        BooleanColumnStatsData boolStats = statsObj.getStatsData().getBooleanStats();
        mColStats.setBooleanStats(boolStats.isSetNumTrues() ? boolStats.getNumTrues() : null, boolStats.isSetNumFalses() ? boolStats.getNumFalses() : null, boolStats.isSetNumNulls() ? boolStats.getNumNulls() : null);
    } else if (statsObj.getStatsData().isSetLongStats()) {
        LongColumnStatsData longStats = statsObj.getStatsData().getLongStats();
        mColStats.setLongStats(longStats.isSetNumNulls() ? longStats.getNumNulls() : null, longStats.isSetNumDVs() ? longStats.getNumDVs() : null, longStats.isSetLowValue() ? longStats.getLowValue() : null, longStats.isSetHighValue() ? longStats.getHighValue() : null);
    } else if (statsObj.getStatsData().isSetDoubleStats()) {
        DoubleColumnStatsData doubleStats = statsObj.getStatsData().getDoubleStats();
        mColStats.setDoubleStats(doubleStats.isSetNumNulls() ? doubleStats.getNumNulls() : null, doubleStats.isSetNumDVs() ? doubleStats.getNumDVs() : null, doubleStats.isSetLowValue() ? doubleStats.getLowValue() : null, doubleStats.isSetHighValue() ? doubleStats.getHighValue() : null);
    } else if (statsObj.getStatsData().isSetDecimalStats()) {
        DecimalColumnStatsData decimalStats = statsObj.getStatsData().getDecimalStats();
        String low = decimalStats.isSetLowValue() ? createJdoDecimalString(decimalStats.getLowValue()) : null;
        String high = decimalStats.isSetHighValue() ? createJdoDecimalString(decimalStats.getHighValue()) : null;
        mColStats.setDecimalStats(decimalStats.isSetNumNulls() ? decimalStats.getNumNulls() : null, decimalStats.isSetNumDVs() ? decimalStats.getNumDVs() : null, low, high);
    } else if (statsObj.getStatsData().isSetStringStats()) {
        StringColumnStatsData stringStats = statsObj.getStatsData().getStringStats();
        mColStats.setStringStats(stringStats.isSetNumNulls() ? stringStats.getNumNulls() : null, stringStats.isSetNumDVs() ? stringStats.getNumDVs() : null, stringStats.isSetMaxColLen() ? stringStats.getMaxColLen() : null, stringStats.isSetAvgColLen() ? stringStats.getAvgColLen() : null);
    } else if (statsObj.getStatsData().isSetBinaryStats()) {
        BinaryColumnStatsData binaryStats = statsObj.getStatsData().getBinaryStats();
        mColStats.setBinaryStats(binaryStats.isSetNumNulls() ? binaryStats.getNumNulls() : null, binaryStats.isSetMaxColLen() ? binaryStats.getMaxColLen() : null, binaryStats.isSetAvgColLen() ? binaryStats.getAvgColLen() : null);
    } else if (statsObj.getStatsData().isSetDateStats()) {
        DateColumnStatsData dateStats = statsObj.getStatsData().getDateStats();
        mColStats.setDateStats(dateStats.isSetNumNulls() ? dateStats.getNumNulls() : null, dateStats.isSetNumDVs() ? dateStats.getNumDVs() : null, dateStats.isSetLowValue() ? dateStats.getLowValue().getDaysSinceEpoch() : null, dateStats.isSetHighValue() ? dateStats.getHighValue().getDaysSinceEpoch() : null);
    }
    return mColStats;
}

Example 13 with StringColumnStatsData

use of org.apache.hadoop.hive.metastore.api.StringColumnStatsData in project hive by apache.

the class StatObjectConverter method getTableColumnStatisticsObj.

public static ColumnStatisticsObj getTableColumnStatisticsObj(MTableColumnStatistics mStatsObj) {
    ColumnStatisticsObj statsObj = new ColumnStatisticsObj();
    statsObj.setColType(mStatsObj.getColType());
    statsObj.setColName(mStatsObj.getColName());
    String colType = mStatsObj.getColType().toLowerCase();
    ColumnStatisticsData colStatsData = new ColumnStatisticsData();
    if (colType.equals("boolean")) {
        BooleanColumnStatsData boolStats = new BooleanColumnStatsData();
        boolStats.setNumFalses(mStatsObj.getNumFalses());
        boolStats.setNumTrues(mStatsObj.getNumTrues());
        boolStats.setNumNulls(mStatsObj.getNumNulls());
        colStatsData.setBooleanStats(boolStats);
    } else if (colType.equals("string") || colType.startsWith("varchar") || colType.startsWith("char")) {
        StringColumnStatsData stringStats = new StringColumnStatsData();
        stringStats.setNumNulls(mStatsObj.getNumNulls());
        stringStats.setAvgColLen(mStatsObj.getAvgColLen());
        stringStats.setMaxColLen(mStatsObj.getMaxColLen());
        stringStats.setNumDVs(mStatsObj.getNumDVs());
        colStatsData.setStringStats(stringStats);
    } else if (colType.equals("binary")) {
        BinaryColumnStatsData binaryStats = new BinaryColumnStatsData();
        binaryStats.setNumNulls(mStatsObj.getNumNulls());
        binaryStats.setAvgColLen(mStatsObj.getAvgColLen());
        binaryStats.setMaxColLen(mStatsObj.getMaxColLen());
        colStatsData.setBinaryStats(binaryStats);
    } else if (colType.equals("bigint") || colType.equals("int") || colType.equals("smallint") || colType.equals("tinyint") || colType.equals("timestamp")) {
        LongColumnStatsData longStats = new LongColumnStatsData();
        longStats.setNumNulls(mStatsObj.getNumNulls());
        Long longHighValue = mStatsObj.getLongHighValue();
        if (longHighValue != null) {
            longStats.setHighValue(longHighValue);
        }
        Long longLowValue = mStatsObj.getLongLowValue();
        if (longLowValue != null) {
            longStats.setLowValue(longLowValue);
        }
        longStats.setNumDVs(mStatsObj.getNumDVs());
        colStatsData.setLongStats(longStats);
    } else if (colType.equals("double") || colType.equals("float")) {
        DoubleColumnStatsData doubleStats = new DoubleColumnStatsData();
        doubleStats.setNumNulls(mStatsObj.getNumNulls());
        Double doubleHighValue = mStatsObj.getDoubleHighValue();
        if (doubleHighValue != null) {
            doubleStats.setHighValue(doubleHighValue);
        }
        Double doubleLowValue = mStatsObj.getDoubleLowValue();
        if (doubleLowValue != null) {
            doubleStats.setLowValue(doubleLowValue);
        }
        doubleStats.setNumDVs(mStatsObj.getNumDVs());
        colStatsData.setDoubleStats(doubleStats);
    } else if (colType.startsWith("decimal")) {
        DecimalColumnStatsData decimalStats = new DecimalColumnStatsData();
        decimalStats.setNumNulls(mStatsObj.getNumNulls());
        String decimalHighValue = mStatsObj.getDecimalHighValue();
        if (decimalHighValue != null) {
            decimalStats.setHighValue(createThriftDecimal(decimalHighValue));
        }
        String decimalLowValue = mStatsObj.getDecimalLowValue();
        if (decimalLowValue != null) {
            decimalStats.setLowValue(createThriftDecimal(decimalLowValue));
        }
        decimalStats.setNumDVs(mStatsObj.getNumDVs());
        colStatsData.setDecimalStats(decimalStats);
    } else if (colType.equals("date")) {
        DateColumnStatsData dateStats = new DateColumnStatsData();
        dateStats.setNumNulls(mStatsObj.getNumNulls());
        Long highValue = mStatsObj.getLongHighValue();
        if (highValue != null) {
            dateStats.setHighValue(new Date(highValue));
        }
        Long lowValue = mStatsObj.getLongLowValue();
        if (lowValue != null) {
            dateStats.setLowValue(new Date(lowValue));
        }
        dateStats.setNumDVs(mStatsObj.getNumDVs());
        colStatsData.setDateStats(dateStats);
    }
    statsObj.setStatsData(colStatsData);
    return statsObj;
}

Example 14 with StringColumnStatsData

use of org.apache.hadoop.hive.metastore.api.StringColumnStatsData in project hive by apache.

the class StatObjectConverter method getPartitionColumnStatisticsObj.

public static ColumnStatisticsObj getPartitionColumnStatisticsObj(MPartitionColumnStatistics mStatsObj) {
    ColumnStatisticsObj statsObj = new ColumnStatisticsObj();
    statsObj.setColType(mStatsObj.getColType());
    statsObj.setColName(mStatsObj.getColName());
    String colType = mStatsObj.getColType().toLowerCase();
    ColumnStatisticsData colStatsData = new ColumnStatisticsData();
    if (colType.equals("boolean")) {
        BooleanColumnStatsData boolStats = new BooleanColumnStatsData();
        boolStats.setNumFalses(mStatsObj.getNumFalses());
        boolStats.setNumTrues(mStatsObj.getNumTrues());
        boolStats.setNumNulls(mStatsObj.getNumNulls());
        colStatsData.setBooleanStats(boolStats);
    } else if (colType.equals("string") || colType.startsWith("varchar") || colType.startsWith("char")) {
        StringColumnStatsData stringStats = new StringColumnStatsData();
        stringStats.setNumNulls(mStatsObj.getNumNulls());
        stringStats.setAvgColLen(mStatsObj.getAvgColLen());
        stringStats.setMaxColLen(mStatsObj.getMaxColLen());
        stringStats.setNumDVs(mStatsObj.getNumDVs());
        colStatsData.setStringStats(stringStats);
    } else if (colType.equals("binary")) {
        BinaryColumnStatsData binaryStats = new BinaryColumnStatsData();
        binaryStats.setNumNulls(mStatsObj.getNumNulls());
        binaryStats.setAvgColLen(mStatsObj.getAvgColLen());
        binaryStats.setMaxColLen(mStatsObj.getMaxColLen());
        colStatsData.setBinaryStats(binaryStats);
    } else if (colType.equals("tinyint") || colType.equals("smallint") || colType.equals("int") || colType.equals("bigint") || colType.equals("timestamp")) {
        LongColumnStatsData longStats = new LongColumnStatsData();
        longStats.setNumNulls(mStatsObj.getNumNulls());
        if (mStatsObj.getLongHighValue() != null) {
            longStats.setHighValue(mStatsObj.getLongHighValue());
        }
        if (mStatsObj.getLongLowValue() != null) {
            longStats.setLowValue(mStatsObj.getLongLowValue());
        }
        longStats.setNumDVs(mStatsObj.getNumDVs());
        colStatsData.setLongStats(longStats);
    } else if (colType.equals("double") || colType.equals("float")) {
        DoubleColumnStatsData doubleStats = new DoubleColumnStatsData();
        doubleStats.setNumNulls(mStatsObj.getNumNulls());
        if (mStatsObj.getDoubleHighValue() != null) {
            doubleStats.setHighValue(mStatsObj.getDoubleHighValue());
        }
        if (mStatsObj.getDoubleLowValue() != null) {
            doubleStats.setLowValue(mStatsObj.getDoubleLowValue());
        }
        doubleStats.setNumDVs(mStatsObj.getNumDVs());
        colStatsData.setDoubleStats(doubleStats);
    } else if (colType.startsWith("decimal")) {
        DecimalColumnStatsData decimalStats = new DecimalColumnStatsData();
        decimalStats.setNumNulls(mStatsObj.getNumNulls());
        if (mStatsObj.getDecimalHighValue() != null) {
            decimalStats.setHighValue(createThriftDecimal(mStatsObj.getDecimalHighValue()));
        }
        if (mStatsObj.getDecimalLowValue() != null) {
            decimalStats.setLowValue(createThriftDecimal(mStatsObj.getDecimalLowValue()));
        }
        decimalStats.setNumDVs(mStatsObj.getNumDVs());
        colStatsData.setDecimalStats(decimalStats);
    } else if (colType.equals("date")) {
        DateColumnStatsData dateStats = new DateColumnStatsData();
        dateStats.setNumNulls(mStatsObj.getNumNulls());
        dateStats.setHighValue(new Date(mStatsObj.getLongHighValue()));
        dateStats.setLowValue(new Date(mStatsObj.getLongLowValue()));
        dateStats.setNumDVs(mStatsObj.getNumDVs());
        colStatsData.setDateStats(dateStats);
    }
    statsObj.setStatsData(colStatsData);
    return statsObj;
}

Example 15 with StringColumnStatsData

use of org.apache.hadoop.hive.metastore.api.StringColumnStatsData in project hive by apache.

the class TestHBaseAggrStatsCacheIntegration method hit.

@Test
public void hit() throws Exception {
    String dbName = "default";
    String tableName = "hit";
    List<String> partVals1 = Arrays.asList("today");
    List<String> partVals2 = Arrays.asList("yesterday");
    long now = System.currentTimeMillis();
    List<FieldSchema> cols = new ArrayList<>();
    cols.add(new FieldSchema("col1", "boolean", "nocomment"));
    cols.add(new FieldSchema("col2", "varchar", "nocomment"));
    SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
    StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, Collections.<String, String>emptyMap());
    List<FieldSchema> partCols = new ArrayList<>();
    partCols.add(new FieldSchema("ds", "string", ""));
    Table table = new Table(tableName, dbName, "me", (int) now, (int) now, 0, sd, partCols, Collections.<String, String>emptyMap(), null, null, null);
    store.createTable(table);
    for (List<String> partVals : Arrays.asList(partVals1, partVals2)) {
        StorageDescriptor psd = new StorageDescriptor(sd);
        psd.setLocation("file:/tmp/default/hit/ds=" + partVals.get(0));
        Partition part = new Partition(partVals, dbName, tableName, (int) now, (int) now, psd, Collections.<String, String>emptyMap());
        store.addPartition(part);
        ColumnStatistics cs = new ColumnStatistics();
        ColumnStatisticsDesc desc = new ColumnStatisticsDesc(false, dbName, tableName);
        desc.setLastAnalyzed(now);
        desc.setPartName("ds=" + partVals.get(0));
        cs.setStatsDesc(desc);
        ColumnStatisticsObj obj = new ColumnStatisticsObj();
        obj.setColName("col1");
        obj.setColType("boolean");
        ColumnStatisticsData data = new ColumnStatisticsData();
        BooleanColumnStatsData bcsd = new BooleanColumnStatsData();
        bcsd.setNumFalses(10);
        bcsd.setNumTrues(20);
        bcsd.setNumNulls(30);
        data.setBooleanStats(bcsd);
        obj.setStatsData(data);
        cs.addToStatsObj(obj);
        obj = new ColumnStatisticsObj();
        obj.setColName("col2");
        obj.setColType("varchar");
        data = new ColumnStatisticsData();
        StringColumnStatsData scsd = new StringColumnStatsData();
        scsd.setAvgColLen(10.3);
        scsd.setMaxColLen(2000);
        scsd.setNumNulls(3);
        scsd.setNumDVs(12342);
        data.setStringStats(scsd);
        obj.setStatsData(data);
        cs.addToStatsObj(obj);
        store.updatePartitionColumnStatistics(cs, partVals);
    }
    Checker statChecker = new Checker() {

        @Override
        public void checkStats(AggrStats aggrStats) throws Exception {
            Assert.assertEquals(2, aggrStats.getPartsFound());
            Assert.assertEquals(2, aggrStats.getColStatsSize());
            ColumnStatisticsObj cso = aggrStats.getColStats().get(0);
            Assert.assertEquals("col1", cso.getColName());
            Assert.assertEquals("boolean", cso.getColType());
            BooleanColumnStatsData bcsd = cso.getStatsData().getBooleanStats();
            Assert.assertEquals(20, bcsd.getNumFalses());
            Assert.assertEquals(40, bcsd.getNumTrues());
            Assert.assertEquals(60, bcsd.getNumNulls());
            cso = aggrStats.getColStats().get(1);
            Assert.assertEquals("col2", cso.getColName());
            Assert.assertEquals("varchar", cso.getColType());
            StringColumnStatsData scsd = cso.getStatsData().getStringStats();
            Assert.assertEquals(10.3, scsd.getAvgColLen(), 0.1);
            Assert.assertEquals(2000, scsd.getMaxColLen());
            Assert.assertEquals(6, scsd.getNumNulls());
            Assert.assertEquals(12342, scsd.getNumDVs());
        }
    };
    AggrStats aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=today", "ds=yesterday"), Arrays.asList("col1", "col2"));
    statChecker.checkStats(aggrStats);
    // Check that we had to build it from the stats
    Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
    Assert.assertEquals(2, store.backdoor().getStatsCache().totalGets.getCnt());
    Assert.assertEquals(2, store.backdoor().getStatsCache().misses.getCnt());
    // Call again, this time it should come from memory.  Also, reverse the name order this time
    // to assure that we still hit.
    aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=yesterday", "ds=today"), Arrays.asList("col1", "col2"));
    statChecker.checkStats(aggrStats);
    Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
    Assert.assertEquals(4, store.backdoor().getStatsCache().totalGets.getCnt());
    Assert.assertEquals(2, store.backdoor().getStatsCache().misses.getCnt());
    store.backdoor().getStatsCache().flushMemory();
    // Call again, this time it should come from hbase
    aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=today", "ds=yesterday"), Arrays.asList("col1", "col2"));
    statChecker.checkStats(aggrStats);
    Assert.assertEquals(2, store.backdoor().getStatsCache().hbaseHits.getCnt());
    Assert.assertEquals(6, store.backdoor().getStatsCache().totalGets.getCnt());
    Assert.assertEquals(2, store.backdoor().getStatsCache().misses.getCnt());
}