Search in sources :

Example 86 with ColumnStatisticsData

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

the class TestHBaseAggregateStatsExtrapolation method noPartitionsHaveBitVectorStatus.

@Test
public void noPartitionsHaveBitVectorStatus() throws Exception {
    String dbName = "default";
    String tableName = "snp";
    long now = System.currentTimeMillis();
    List<FieldSchema> cols = new ArrayList<>();
    cols.add(new FieldSchema("col2", "long", "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);
    List<List<String>> partVals = new ArrayList<>();
    for (int i = 0; i < 10; i++) {
        List<String> partVal = Arrays.asList("" + i);
        partVals.add(partVal);
        StorageDescriptor psd = new StorageDescriptor(sd);
        psd.setLocation("file:/tmp/default/hit/ds=" + partVal);
        Partition part = new Partition(partVal, 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=" + partVal);
        cs.setStatsDesc(desc);
        ColumnStatisticsObj obj = new ColumnStatisticsObj();
        obj.setColName("col2");
        obj.setColType("long");
        ColumnStatisticsData data = new ColumnStatisticsData();
        LongColumnStatsData dcsd = new LongColumnStatsData();
        dcsd.setHighValue(1000 + i);
        dcsd.setLowValue(-1000 - i);
        dcsd.setNumNulls(i);
        dcsd.setNumDVs(10 * i);
        data.setLongStats(dcsd);
        obj.setStatsData(data);
        cs.addToStatsObj(obj);
        store.updatePartitionColumnStatistics(cs, partVal);
    }
    Checker statChecker = new Checker() {

        @Override
        public void checkStats(AggrStats aggrStats) throws Exception {
            Assert.assertEquals(10, aggrStats.getPartsFound());
            Assert.assertEquals(1, aggrStats.getColStatsSize());
            ColumnStatisticsObj cso = aggrStats.getColStats().get(0);
            Assert.assertEquals("col2", cso.getColName());
            Assert.assertEquals("long", cso.getColType());
            LongColumnStatsData lcsd = cso.getStatsData().getLongStats();
            Assert.assertEquals(1009, lcsd.getHighValue(), 0.01);
            Assert.assertEquals(-1009, lcsd.getLowValue(), 0.01);
            Assert.assertEquals(45, lcsd.getNumNulls());
            Assert.assertEquals(90, lcsd.getNumDVs());
        }
    };
    List<String> partNames = new ArrayList<>();
    for (int i = 0; i < 10; i++) {
        partNames.add("ds=" + i);
    }
    AggrStats aggrStats = store.get_aggr_stats_for(dbName, tableName, partNames, Arrays.asList("col2"));
    statChecker.checkStats(aggrStats);
}
Also used : ColumnStatistics(org.apache.hadoop.hive.metastore.api.ColumnStatistics) 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) LongColumnStatsData(org.apache.hadoop.hive.metastore.api.LongColumnStatsData) ColumnStatisticsObj(org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj) ColumnStatisticsDesc(org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc) ArrayList(java.util.ArrayList) List(java.util.List) ColumnStatisticsData(org.apache.hadoop.hive.metastore.api.ColumnStatisticsData) Test(org.junit.Test)

Example 87 with ColumnStatisticsData

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

the class TestHBaseStore method mockBinaryStats.

private static ColumnStatisticsObj mockBinaryStats(int i) {
    ;
    long maxLen = 123412987L + 10 * i;
    double avgLen = 76.98 + i;
    long nulls = 976998797L + 10 * i;
    ColumnStatisticsObj colStatsObj = new ColumnStatisticsObj();
    colStatsObj.setColName(BINARY_COL);
    colStatsObj.setColType(BINARY_TYPE);
    ColumnStatisticsData data = new ColumnStatisticsData();
    BinaryColumnStatsData binaryData = new BinaryColumnStatsData();
    binaryData.setMaxColLen(maxLen);
    binaryData.setAvgColLen(avgLen);
    binaryData.setNumNulls(nulls);
    data.setBinaryStats(binaryData);
    colStatsObj.setStatsData(data);
    return colStatsObj;
}
Also used : ColumnStatisticsObj(org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj) ColumnStatisticsData(org.apache.hadoop.hive.metastore.api.ColumnStatisticsData) BinaryColumnStatsData(org.apache.hadoop.hive.metastore.api.BinaryColumnStatsData)

Example 88 with ColumnStatisticsData

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

the class TestHBaseStore 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());
}
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 89 with ColumnStatisticsData

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

the class TestHBaseStore 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;
    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);
    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 90 with ColumnStatisticsData

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

the class TestHBaseStore method stringTableStatistics.

@Test
public void stringTableStatistics() throws Exception {
    createMockTable(STRING_TYPE);
    // Add a string table stats for STRING_COL 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 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());
}
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)

Aggregations

ColumnStatisticsData (org.apache.hadoop.hive.metastore.api.ColumnStatisticsData)108 ColumnStatisticsObj (org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj)95 ColumnStatistics (org.apache.hadoop.hive.metastore.api.ColumnStatistics)62 ColumnStatisticsDesc (org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc)56 Test (org.junit.Test)53 ArrayList (java.util.ArrayList)47 LongColumnStatsData (org.apache.hadoop.hive.metastore.api.LongColumnStatsData)35 FieldSchema (org.apache.hadoop.hive.metastore.api.FieldSchema)34 Table (org.apache.hadoop.hive.metastore.api.Table)33 StorageDescriptor (org.apache.hadoop.hive.metastore.api.StorageDescriptor)32 SerDeInfo (org.apache.hadoop.hive.metastore.api.SerDeInfo)31 BooleanColumnStatsData (org.apache.hadoop.hive.metastore.api.BooleanColumnStatsData)30 Partition (org.apache.hadoop.hive.metastore.api.Partition)30 AggrStats (org.apache.hadoop.hive.metastore.api.AggrStats)29 DoubleColumnStatsData (org.apache.hadoop.hive.metastore.api.DoubleColumnStatsData)27 StringColumnStatsData (org.apache.hadoop.hive.metastore.api.StringColumnStatsData)25 DecimalColumnStatsData (org.apache.hadoop.hive.metastore.api.DecimalColumnStatsData)23 BinaryColumnStatsData (org.apache.hadoop.hive.metastore.api.BinaryColumnStatsData)22 HashMap (java.util.HashMap)20 List (java.util.List)18