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

Example 1 with StringColumnStatsData

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

the class TestHBaseStoreIntegration method partitionStatistics.

@Test
public void partitionStatistics() throws Exception {
    long now = System.currentTimeMillis();
    String dbname = "default";
    String tableName = "statspart";
    String[] partNames = { "ds=today", "ds=yesterday" };
    String[] partVals = { "today", "yesterday" };
    String boolcol = "boolcol";
    String longcol = "longcol";
    String doublecol = "doublecol";
    String stringcol = "stringcol";
    String binarycol = "bincol";
    String decimalcol = "deccol";
    long trues = 37;
    long falses = 12;
    long booleanNulls = 2;
    long strMaxLen = 1234;
    double strAvgLen = 32.3;
    long strNulls = 987;
    long strDVs = 906;
    List<FieldSchema> cols = new ArrayList<FieldSchema>();
    cols.add(new FieldSchema(boolcol, "boolean", "nocomment"));
    cols.add(new FieldSchema(longcol, "long", "nocomment"));
    cols.add(new FieldSchema(doublecol, "double", "nocomment"));
    cols.add(new FieldSchema(stringcol, "varchar(32)", "nocomment"));
    cols.add(new FieldSchema(binarycol, "binary", "nocomment"));
    cols.add(new FieldSchema(decimalcol, "decimal(5, 3)", "nocomment"));
    SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
    StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, emptyParameters);
    List<FieldSchema> partCols = new ArrayList<FieldSchema>();
    partCols.add(new FieldSchema("ds", "string", ""));
    Table table = new Table(tableName, dbname, "me", (int) now / 1000, (int) now / 1000, 0, sd, partCols, emptyParameters, null, null, null);
    store.createTable(table);
    for (String partVal : partVals) {
        Partition part = new Partition(Arrays.asList(partVal), dbname, tableName, (int) now / 1000, (int) now / 1000, sd, emptyParameters);
        store.addPartition(part);
    }
    for (int i = 0; i < partNames.length; i++) {
        ColumnStatistics stats = new ColumnStatistics();
        ColumnStatisticsDesc desc = new ColumnStatisticsDesc();
        desc.setLastAnalyzed(now);
        desc.setDbName(dbname);
        desc.setTableName(tableName);
        desc.setIsTblLevel(false);
        desc.setPartName(partNames[i]);
        stats.setStatsDesc(desc);
        ColumnStatisticsObj obj = new ColumnStatisticsObj();
        obj.setColName(boolcol);
        obj.setColType("boolean");
        ColumnStatisticsData data = new ColumnStatisticsData();
        BooleanColumnStatsData boolData = new BooleanColumnStatsData();
        boolData.setNumTrues(trues);
        boolData.setNumFalses(falses);
        boolData.setNumNulls(booleanNulls);
        data.setBooleanStats(boolData);
        obj.setStatsData(data);
        stats.addToStatsObj(obj);
        store.updatePartitionColumnStatistics(stats, Arrays.asList(partVals[i]));
    }
    List<ColumnStatistics> statsList = store.getPartitionColumnStatistics(dbname, tableName, Arrays.asList(partNames), Arrays.asList(boolcol));
    Assert.assertEquals(2, statsList.size());
    for (int i = 0; i < partNames.length; i++) {
        Assert.assertEquals(1, statsList.get(i).getStatsObjSize());
    }
    for (int i = 0; i < partNames.length; i++) {
        ColumnStatistics stats = new ColumnStatistics();
        ColumnStatisticsDesc desc = new ColumnStatisticsDesc();
        desc.setLastAnalyzed(now);
        desc.setDbName(dbname);
        desc.setTableName(tableName);
        desc.setIsTblLevel(false);
        desc.setPartName(partNames[i]);
        stats.setStatsDesc(desc);
        ColumnStatisticsObj obj = new ColumnStatisticsObj();
        obj.setColName(stringcol);
        obj.setColType("string");
        ColumnStatisticsData data = new ColumnStatisticsData();
        StringColumnStatsData strData = new StringColumnStatsData();
        strData.setMaxColLen(strMaxLen);
        strData.setAvgColLen(strAvgLen);
        strData.setNumNulls(strNulls);
        strData.setNumDVs(strDVs);
        data.setStringStats(strData);
        obj.setStatsData(data);
        stats.addToStatsObj(obj);
        store.updatePartitionColumnStatistics(stats, Arrays.asList(partVals[i]));
    }
    // Make sure when we ask for one we only get one
    statsList = store.getPartitionColumnStatistics(dbname, tableName, Arrays.asList(partNames), Arrays.asList(boolcol));
    Assert.assertEquals(2, statsList.size());
    for (int i = 0; i < partNames.length; i++) {
        Assert.assertEquals(1, statsList.get(i).getStatsObjSize());
    }
    statsList = store.getPartitionColumnStatistics(dbname, tableName, Arrays.asList(partNames), Arrays.asList(boolcol, stringcol));
    Assert.assertEquals(2, statsList.size());
    for (int i = 0; i < partNames.length; i++) {
        Assert.assertEquals(2, statsList.get(i).getStatsObjSize());
        // Just check one piece of the data, I don't need to check it all again
        Assert.assertEquals(booleanNulls, statsList.get(i).getStatsObj().get(0).getStatsData().getBooleanStats().getNumNulls());
        Assert.assertEquals(strDVs, statsList.get(i).getStatsObj().get(1).getStatsData().getStringStats().getNumDVs());
    }
}

Example 2 with StringColumnStatsData

use of org.apache.hadoop.hive.metastore.api.StringColumnStatsData 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());
}

Example 3 with StringColumnStatsData

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

the class TestHBaseAggregateStatsExtrapolation method allPartitionsHaveBitVectorStatusString.

@Test
public void allPartitionsHaveBitVectorStatusString() throws Exception {
    String dbName = "default";
    String tableName = "snp";
    long now = System.currentTimeMillis();
    List<FieldSchema> cols = new ArrayList<>();
    cols.add(new FieldSchema("col1_string", "string", "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("col1_string");
        obj.setColType("string");
        ColumnStatisticsData data = new ColumnStatisticsData();
        StringColumnStatsData dcsd = new StringColumnStatsData();
        dcsd.setAvgColLen(i + 1);
        dcsd.setMaxColLen(i + 10);
        dcsd.setNumNulls(i);
        dcsd.setNumDVs(10 * i + 1);
        dcsd.setBitVectors(bitVectors);
        data.setStringStats(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("col1_string", cso.getColName());
            Assert.assertEquals("string", cso.getColType());
            StringColumnStatsData lcsd = cso.getStatsData().getStringStats();
            Assert.assertEquals(10, lcsd.getAvgColLen(), 0.01);
            Assert.assertEquals(19, lcsd.getMaxColLen(), 0.01);
            Assert.assertEquals(45, lcsd.getNumNulls());
            Assert.assertEquals(3, 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("col1_string"));
    statChecker.checkStats(aggrStats);
}

Example 4 with StringColumnStatsData

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

the class TestHBaseStore method mockStringStats.

private static ColumnStatisticsObj mockStringStats(int i) {
    long maxLen = 1234 + 10 * i;
    double avgLen = 32.3 + i;
    long nulls = 987 + 10 * i;
    long dVs = 906 + i;
    ColumnStatisticsObj colStatsObj = new ColumnStatisticsObj();
    colStatsObj.setColName(STRING_COL);
    colStatsObj.setColType(STRING_TYPE);
    ColumnStatisticsData data = new ColumnStatisticsData();
    StringColumnStatsData stringData = new StringColumnStatsData();
    stringData.setMaxColLen(maxLen);
    stringData.setAvgColLen(avgLen);
    stringData.setNumNulls(nulls);
    stringData.setNumDVs(dVs);
    data.setStringStats(stringData);
    colStatsObj.setStatsData(data);
    return colStatsObj;
}

Example 5 with StringColumnStatsData

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

the class ColumnStatsAggregatorFactory method newColumnStaticsObj.

public static ColumnStatisticsObj newColumnStaticsObj(String colName, String colType, _Fields type) {
    ColumnStatisticsObj cso = new ColumnStatisticsObj();
    ColumnStatisticsData csd = new ColumnStatisticsData();
    cso.setColName(colName);
    cso.setColType(colType);
    switch(type) {
        case BOOLEAN_STATS:
            csd.setBooleanStats(new BooleanColumnStatsData());
            break;
        case LONG_STATS:
            csd.setLongStats(new LongColumnStatsData());
            break;
        case DOUBLE_STATS:
            csd.setDoubleStats(new DoubleColumnStatsData());
            break;
        case STRING_STATS:
            csd.setStringStats(new StringColumnStatsData());
            break;
        case BINARY_STATS:
            csd.setBinaryStats(new BinaryColumnStatsData());
            break;
        case DECIMAL_STATS:
            csd.setDecimalStats(new DecimalColumnStatsData());
            break;
        default:
            throw new RuntimeException("Woh, bad.  Unknown stats type!");
    }
    cso.setStatsData(csd);
    return cso;
}