Search in sources :

Example 26 with ColumnStatisticsObj

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

the class TestHBaseStore method mockDecimalStats.

private static ColumnStatisticsObj mockDecimalStats(int i) {
    Decimal high = new Decimal();
    high.setScale((short) 3);
    String strHigh = String.valueOf(3876 + 100 * i);
    high.setUnscaled(strHigh.getBytes());
    Decimal low = new Decimal();
    low.setScale((short) 3);
    String strLow = String.valueOf(38 + i);
    low.setUnscaled(strLow.getBytes());
    long nulls = 13 + i;
    long dVs = 923947293L + 100 * i;
    ColumnStatisticsObj colStatsObj = new ColumnStatisticsObj();
    colStatsObj.setColName(DECIMAL_COL);
    colStatsObj.setColType(DECIMAL_TYPE);
    ColumnStatisticsData data = new ColumnStatisticsData();
    DecimalColumnStatsData decimalData = new DecimalColumnStatsData();
    decimalData.setHighValue(high);
    decimalData.setLowValue(low);
    decimalData.setNumNulls(nulls);
    decimalData.setNumDVs(dVs);
    data.setDecimalStats(decimalData);
    colStatsObj.setStatsData(data);
    return colStatsObj;
}
Also used : ColumnStatisticsObj(org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj) DecimalColumnStatsData(org.apache.hadoop.hive.metastore.api.DecimalColumnStatsData) Decimal(org.apache.hadoop.hive.metastore.api.Decimal) ColumnStatisticsData(org.apache.hadoop.hive.metastore.api.ColumnStatisticsData)

Example 27 with ColumnStatisticsObj

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

the class TestHBaseStoreBitVector method doublePartitionStatistics.

@Test
public void doublePartitionStatistics() throws Exception {
    createMockTableAndPartition(DOUBLE_TYPE, DOUBLE_VAL);
    // Add partition stats for: DOUBLE_COL and partition: {PART_KEY, DOUBLE_VAL} to DB
    // Because of the way our mock implementation works we actually need to not create the table
    // before we set statistics on it.
    ColumnStatistics stats = new ColumnStatistics();
    // Get a default ColumnStatisticsDesc for partition level stats
    ColumnStatisticsDesc desc = getMockPartColStatsDesc(PART_KEY, DOUBLE_VAL);
    stats.setStatsDesc(desc);
    // Get one of the pre-created ColumnStatisticsObj
    ColumnStatisticsObj obj = doubleColStatsObjs.get(0);
    DoubleColumnStatsData doubleData = obj.getStatsData().getDoubleStats();
    // Add to DB
    stats.addToStatsObj(obj);
    List<String> parVals = new ArrayList<String>();
    parVals.add(DOUBLE_VAL);
    store.updatePartitionColumnStatistics(stats, parVals);
    // Get from DB
    List<String> partNames = new ArrayList<String>();
    partNames.add(desc.getPartName());
    List<String> colNames = new ArrayList<String>();
    colNames.add(obj.getColName());
    List<ColumnStatistics> statsFromDB = store.getPartitionColumnStatistics(DB, TBL, partNames, colNames);
    // Compare ColumnStatisticsDesc
    Assert.assertEquals(1, statsFromDB.size());
    Assert.assertEquals(desc.getLastAnalyzed(), statsFromDB.get(0).getStatsDesc().getLastAnalyzed());
    Assert.assertEquals(DB, statsFromDB.get(0).getStatsDesc().getDbName());
    Assert.assertEquals(TBL, statsFromDB.get(0).getStatsDesc().getTableName());
    Assert.assertFalse(statsFromDB.get(0).getStatsDesc().isIsTblLevel());
    // Compare ColumnStatisticsObj
    Assert.assertEquals(1, statsFromDB.get(0).getStatsObjSize());
    ColumnStatisticsObj objFromDB = statsFromDB.get(0).getStatsObj().get(0);
    ColumnStatisticsData dataFromDB = objFromDB.getStatsData();
    // Compare ColumnStatisticsData
    Assert.assertEquals(ColumnStatisticsData._Fields.DOUBLE_STATS, dataFromDB.getSetField());
    // Compare DoubleColumnStatsData
    DoubleColumnStatsData doubleDataFromDB = dataFromDB.getDoubleStats();
    Assert.assertEquals(doubleData.getHighValue(), doubleDataFromDB.getHighValue(), 0.01);
    Assert.assertEquals(doubleData.getLowValue(), doubleDataFromDB.getLowValue(), 0.01);
    Assert.assertEquals(doubleData.getNumNulls(), doubleDataFromDB.getNumNulls());
    Assert.assertEquals(doubleData.getNumDVs(), doubleDataFromDB.getNumDVs());
    Assert.assertEquals(doubleData.getBitVectors(), doubleDataFromDB.getBitVectors());
}
Also used : ColumnStatistics(org.apache.hadoop.hive.metastore.api.ColumnStatistics) ColumnStatisticsObj(org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj) DoubleColumnStatsData(org.apache.hadoop.hive.metastore.api.DoubleColumnStatsData) ColumnStatisticsDesc(org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc) ArrayList(java.util.ArrayList) ColumnStatisticsData(org.apache.hadoop.hive.metastore.api.ColumnStatisticsData) Test(org.junit.Test)

Example 28 with ColumnStatisticsObj

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

the class TestHBaseStoreBitVector method mockDoubleStats.

private static ColumnStatisticsObj mockDoubleStats(int i) {
    double high = 123423.23423 + 100 * i;
    double low = 0.00001234233 - 50 * i;
    long nulls = 92 + i;
    long dVs = 1234123421L + 10 * i;
    String bitVectors = "{0, 1, 2, 3, 4, 5, 6, 7, 8}{0, 2, 3, 4, 5, 6, 7, 8}";
    ColumnStatisticsObj colStatsObj = new ColumnStatisticsObj();
    colStatsObj.setColName(DOUBLE_COL);
    colStatsObj.setColType(DOUBLE_TYPE);
    ColumnStatisticsData data = new ColumnStatisticsData();
    DoubleColumnStatsData doubleData = new DoubleColumnStatsData();
    doubleData.setHighValue(high);
    doubleData.setLowValue(low);
    doubleData.setNumNulls(nulls);
    doubleData.setNumDVs(dVs);
    doubleData.setBitVectors(bitVectors);
    data.setDoubleStats(doubleData);
    colStatsObj.setStatsData(data);
    return colStatsObj;
}
Also used : ColumnStatisticsObj(org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj) DoubleColumnStatsData(org.apache.hadoop.hive.metastore.api.DoubleColumnStatsData) ColumnStatisticsData(org.apache.hadoop.hive.metastore.api.ColumnStatisticsData)

Example 29 with ColumnStatisticsObj

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

the class TestHBaseStoreBitVector method stringTableStatistics.

@Test
public void stringTableStatistics() throws Exception {
    createMockTable(STRING_COL, STRING_TYPE);
    ColumnStatistics stats = new ColumnStatistics();
    // Get a default ColumnStatisticsDesc for table level stats
    ColumnStatisticsDesc desc = getMockTblColStatsDesc();
    stats.setStatsDesc(desc);
    // Get one of the pre-created ColumnStatisticsObj
    ColumnStatisticsObj obj = stringColStatsObjs.get(0);
    StringColumnStatsData stringData = obj.getStatsData().getStringStats();
    // Add to DB
    stats.addToStatsObj(obj);
    store.updateTableColumnStatistics(stats);
    // Get from DB
    ColumnStatistics statsFromDB = store.getTableColumnStatistics(DB, TBL, Arrays.asList(STRING_COL));
    // Compare ColumnStatisticsDesc
    Assert.assertEquals(desc.getLastAnalyzed(), statsFromDB.getStatsDesc().getLastAnalyzed());
    Assert.assertEquals(DB, statsFromDB.getStatsDesc().getDbName());
    Assert.assertEquals(TBL, statsFromDB.getStatsDesc().getTableName());
    Assert.assertTrue(statsFromDB.getStatsDesc().isIsTblLevel());
    // Compare ColumnStatisticsObj
    Assert.assertEquals(1, statsFromDB.getStatsObjSize());
    ColumnStatisticsObj objFromDB = statsFromDB.getStatsObj().get(0);
    ColumnStatisticsData dataFromDB = objFromDB.getStatsData();
    // Compare ColumnStatisticsData
    Assert.assertEquals(ColumnStatisticsData._Fields.STRING_STATS, dataFromDB.getSetField());
    // Compare StringColumnStatsData
    StringColumnStatsData stringDataFromDB = dataFromDB.getStringStats();
    Assert.assertEquals(stringData.getMaxColLen(), stringDataFromDB.getMaxColLen());
    Assert.assertEquals(stringData.getAvgColLen(), stringDataFromDB.getAvgColLen(), 0.01);
    Assert.assertEquals(stringData.getNumNulls(), stringDataFromDB.getNumNulls());
    Assert.assertEquals(stringData.getNumDVs(), stringDataFromDB.getNumDVs());
    Assert.assertEquals(stringData.getBitVectors(), stringDataFromDB.getBitVectors());
}
Also used : ColumnStatistics(org.apache.hadoop.hive.metastore.api.ColumnStatistics) ColumnStatisticsObj(org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj) ColumnStatisticsDesc(org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc) StringColumnStatsData(org.apache.hadoop.hive.metastore.api.StringColumnStatsData) ColumnStatisticsData(org.apache.hadoop.hive.metastore.api.ColumnStatisticsData) Test(org.junit.Test)

Example 30 with ColumnStatisticsObj

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

the class TestHBaseAggregateStatsCache method allWithStats.

// Do to limitations in the Mock infrastructure we use for HBase testing we can only test
// this for a single column table and we can't really test hits in hbase, only in memory or
// build from scratch.  But it's still useful to cover many bugs.  More in depth testing with
// multiple columns and with HBase hits is done in TestHBaseAggrStatsCacheIntegration.
@Test
public void allWithStats() 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"));
    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);
        store.updatePartitionColumnStatistics(cs, partVals);
    }
    Checker statChecker = new Checker() {

        @Override
        public void checkStats(AggrStats aggrStats) throws Exception {
            Assert.assertEquals(2, aggrStats.getPartsFound());
            Assert.assertEquals(1, 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());
        }
    };
    AggrStats aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=today", "ds=yesterday"), Arrays.asList("col1"));
    statChecker.checkStats(aggrStats);
    // Check that we had to build it from the stats
    Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
    Assert.assertEquals(1, store.backdoor().getStatsCache().totalGets.getCnt());
    Assert.assertEquals(1, 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"));
    statChecker.checkStats(aggrStats);
    Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
    Assert.assertEquals(2, store.backdoor().getStatsCache().totalGets.getCnt());
    Assert.assertEquals(1, store.backdoor().getStatsCache().misses.getCnt());
}
Also used : ColumnStatistics(org.apache.hadoop.hive.metastore.api.ColumnStatistics) BooleanColumnStatsData(org.apache.hadoop.hive.metastore.api.BooleanColumnStatsData) Partition(org.apache.hadoop.hive.metastore.api.Partition) Table(org.apache.hadoop.hive.metastore.api.Table) AggrStats(org.apache.hadoop.hive.metastore.api.AggrStats) FieldSchema(org.apache.hadoop.hive.metastore.api.FieldSchema) SerDeInfo(org.apache.hadoop.hive.metastore.api.SerDeInfo) ArrayList(java.util.ArrayList) StorageDescriptor(org.apache.hadoop.hive.metastore.api.StorageDescriptor) ColumnStatisticsObj(org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj) ColumnStatisticsDesc(org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc) ColumnStatisticsData(org.apache.hadoop.hive.metastore.api.ColumnStatisticsData) Test(org.junit.Test)

Aggregations

ColumnStatisticsObj (org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj)139 ColumnStatisticsData (org.apache.hadoop.hive.metastore.api.ColumnStatisticsData)95 ArrayList (java.util.ArrayList)75 ColumnStatistics (org.apache.hadoop.hive.metastore.api.ColumnStatistics)75 ColumnStatisticsDesc (org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc)63 Test (org.junit.Test)57 FieldSchema (org.apache.hadoop.hive.metastore.api.FieldSchema)39 Table (org.apache.hadoop.hive.metastore.api.Table)39 AggrStats (org.apache.hadoop.hive.metastore.api.AggrStats)33 Partition (org.apache.hadoop.hive.metastore.api.Partition)33 StorageDescriptor (org.apache.hadoop.hive.metastore.api.StorageDescriptor)32 LongColumnStatsData (org.apache.hadoop.hive.metastore.api.LongColumnStatsData)31 SerDeInfo (org.apache.hadoop.hive.metastore.api.SerDeInfo)31 BooleanColumnStatsData (org.apache.hadoop.hive.metastore.api.BooleanColumnStatsData)26 List (java.util.List)24 DoubleColumnStatsData (org.apache.hadoop.hive.metastore.api.DoubleColumnStatsData)22 StringColumnStatsData (org.apache.hadoop.hive.metastore.api.StringColumnStatsData)22 HashMap (java.util.HashMap)21 BinaryColumnStatsData (org.apache.hadoop.hive.metastore.api.BinaryColumnStatsData)18 DecimalColumnStatsData (org.apache.hadoop.hive.metastore.api.DecimalColumnStatsData)18