use of org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj in project hive by apache.
the class TestHBaseStore method decimalPartitionStatistics.
@Test
public void decimalPartitionStatistics() throws Exception {
createMockTableAndPartition(DECIMAL_TYPE, DECIMAL_VAL);
// Add partition stats for: DECIMAL_COL and partition: {PART_KEY, DECIMAL_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, DECIMAL_VAL);
stats.setStatsDesc(desc);
// Get one of the pre-created ColumnStatisticsObj
ColumnStatisticsObj obj = decimalColStatsObjs.get(0);
DecimalColumnStatsData decimalData = obj.getStatsData().getDecimalStats();
// Add to DB
stats.addToStatsObj(obj);
List<String> parVals = new ArrayList<String>();
parVals.add(DECIMAL_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.DECIMAL_STATS, dataFromDB.getSetField());
// Compare DecimalColumnStatsData
DecimalColumnStatsData decimalDataFromDB = dataFromDB.getDecimalStats();
Assert.assertEquals(decimalData.getHighValue(), decimalDataFromDB.getHighValue());
Assert.assertEquals(decimalData.getLowValue(), decimalDataFromDB.getLowValue());
Assert.assertEquals(decimalData.getNumNulls(), decimalDataFromDB.getNumNulls());
Assert.assertEquals(decimalData.getNumDVs(), decimalDataFromDB.getNumDVs());
}
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;
}
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());
}
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;
}
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());
}
Aggregations