use of org.apache.hadoop.hive.serde2.objectinspector.StandardConstantListObjectInspector in project hive by apache.
the class StatsUtils method getSizeOfComplexTypes.
/**
* Get the size of complex data types
* @param conf
* - hive conf
* @param oi
* - object inspector
* @return raw data size
*/
public static long getSizeOfComplexTypes(HiveConf conf, ObjectInspector oi) {
long result = 0;
int length = 0;
int listEntries = HiveConf.getIntVar(conf, HiveConf.ConfVars.HIVE_STATS_LIST_NUM_ENTRIES);
int mapEntries = HiveConf.getIntVar(conf, HiveConf.ConfVars.HIVE_STATS_MAP_NUM_ENTRIES);
switch(oi.getCategory()) {
case PRIMITIVE:
String colTypeLowerCase = oi.getTypeName().toLowerCase();
if (colTypeLowerCase.equals(serdeConstants.STRING_TYPE_NAME) || colTypeLowerCase.startsWith(serdeConstants.VARCHAR_TYPE_NAME) || colTypeLowerCase.startsWith(serdeConstants.CHAR_TYPE_NAME)) {
int avgColLen = (int) getAvgColLenOf(conf, oi, colTypeLowerCase);
result += JavaDataModel.get().lengthForStringOfLength(avgColLen);
} else if (colTypeLowerCase.equals(serdeConstants.BINARY_TYPE_NAME)) {
int avgColLen = (int) getAvgColLenOf(conf, oi, colTypeLowerCase);
result += JavaDataModel.get().lengthForByteArrayOfSize(avgColLen);
} else {
result += getAvgColLenOfFixedLengthTypes(colTypeLowerCase);
}
break;
case LIST:
if (oi instanceof StandardConstantListObjectInspector) {
// constant list projection of known length
StandardConstantListObjectInspector scloi = (StandardConstantListObjectInspector) oi;
length = scloi.getWritableConstantValue().size();
// check if list elements are primitive or Objects
ObjectInspector leoi = scloi.getListElementObjectInspector();
if (leoi.getCategory().equals(ObjectInspector.Category.PRIMITIVE)) {
result += getSizeOfPrimitiveTypeArraysFromType(leoi.getTypeName(), length);
} else {
result += JavaDataModel.get().lengthForObjectArrayOfSize(length);
}
} else {
StandardListObjectInspector sloi = (StandardListObjectInspector) oi;
// list overhead + (configured number of element in list * size of element)
long elemSize = getSizeOfComplexTypes(conf, sloi.getListElementObjectInspector());
result += JavaDataModel.get().arrayList() + (listEntries * elemSize);
}
break;
case MAP:
if (oi instanceof StandardConstantMapObjectInspector) {
// constant map projection of known length
StandardConstantMapObjectInspector scmoi = (StandardConstantMapObjectInspector) oi;
result += getSizeOfMap(scmoi);
} else {
StandardMapObjectInspector smoi = (StandardMapObjectInspector) oi;
result += getSizeOfComplexTypes(conf, smoi.getMapKeyObjectInspector());
result += getSizeOfComplexTypes(conf, smoi.getMapValueObjectInspector());
// hash map overhead
result += JavaDataModel.get().hashMap(mapEntries);
}
break;
case STRUCT:
if (oi instanceof StandardConstantStructObjectInspector) {
// constant map projection of known length
StandardConstantStructObjectInspector scsoi = (StandardConstantStructObjectInspector) oi;
result += getSizeOfStruct(scsoi);
} else {
StructObjectInspector soi = (StructObjectInspector) oi;
// add constant object overhead for struct
result += JavaDataModel.get().object();
// add constant struct field names references overhead
result += soi.getAllStructFieldRefs().size() * JavaDataModel.get().ref();
for (StructField field : soi.getAllStructFieldRefs()) {
result += getSizeOfComplexTypes(conf, field.getFieldObjectInspector());
}
}
break;
case UNION:
UnionObjectInspector uoi = (UnionObjectInspector) oi;
// add constant object overhead for union
result += JavaDataModel.get().object();
// add constant size for unions tags
result += uoi.getObjectInspectors().size() * JavaDataModel.get().primitive1();
for (ObjectInspector foi : uoi.getObjectInspectors()) {
result += getSizeOfComplexTypes(conf, foi);
}
break;
default:
break;
}
return result;
}
Aggregations