Search in sources :

Example 56 with Category

use of org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector.Category in project hive by apache.

the class Vectorizer method validateStructInExpression.

private boolean validateStructInExpression(ExprNodeDesc desc, String expressionTitle, VectorExpressionDescriptor.Mode mode) {
    for (ExprNodeDesc d : desc.getChildren()) {
        TypeInfo typeInfo = d.getTypeInfo();
        if (typeInfo.getCategory() != Category.STRUCT) {
            return false;
        }
        StructTypeInfo structTypeInfo = (StructTypeInfo) typeInfo;
        ArrayList<TypeInfo> fieldTypeInfos = structTypeInfo.getAllStructFieldTypeInfos();
        ArrayList<String> fieldNames = structTypeInfo.getAllStructFieldNames();
        final int fieldCount = fieldTypeInfos.size();
        for (int f = 0; f < fieldCount; f++) {
            TypeInfo fieldTypeInfo = fieldTypeInfos.get(f);
            Category category = fieldTypeInfo.getCategory();
            if (category != Category.PRIMITIVE) {
                setExpressionIssue(expressionTitle, "Cannot vectorize struct field " + fieldNames.get(f) + " of type " + fieldTypeInfo.getTypeName());
                return false;
            }
            PrimitiveTypeInfo fieldPrimitiveTypeInfo = (PrimitiveTypeInfo) fieldTypeInfo;
            InConstantType inConstantType = VectorizationContext.getInConstantTypeFromPrimitiveCategory(fieldPrimitiveTypeInfo.getPrimitiveCategory());
            // For now, limit the data types we support for Vectorized Struct IN().
            if (inConstantType != InConstantType.INT_FAMILY && inConstantType != InConstantType.FLOAT_FAMILY && inConstantType != InConstantType.STRING_FAMILY) {
                setExpressionIssue(expressionTitle, "Cannot vectorize struct field " + fieldNames.get(f) + " of type " + fieldTypeInfo.getTypeName());
                return false;
            }
        }
    }
    return true;
}
Also used : PrimitiveCategory(org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector.PrimitiveCategory) Category(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector.Category) InConstantType(org.apache.hadoop.hive.ql.exec.vector.VectorizationContext.InConstantType) StructTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.StructTypeInfo) ExprNodeDesc(org.apache.hadoop.hive.ql.plan.ExprNodeDesc) UDFToString(org.apache.hadoop.hive.ql.udf.UDFToString) StructTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.StructTypeInfo) PrimitiveTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo) TypeInfo(org.apache.hadoop.hive.serde2.typeinfo.TypeInfo) PrimitiveTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo)

Example 57 with Category

use of org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector.Category in project hive by apache.

the class HCatSchemaUtils method getHCatFieldSchema.

private static HCatFieldSchema getHCatFieldSchema(String fieldName, TypeInfo fieldTypeInfo, String comment) throws HCatException {
    Category typeCategory = fieldTypeInfo.getCategory();
    HCatFieldSchema hCatFieldSchema;
    if (Category.PRIMITIVE == typeCategory) {
        hCatFieldSchema = new HCatFieldSchema(fieldName, (PrimitiveTypeInfo) fieldTypeInfo, comment);
    } else if (Category.STRUCT == typeCategory) {
        HCatSchema subSchema = constructHCatSchema((StructTypeInfo) fieldTypeInfo);
        hCatFieldSchema = new HCatFieldSchema(fieldName, HCatFieldSchema.Type.STRUCT, subSchema, comment);
    } else if (Category.LIST == typeCategory) {
        HCatSchema subSchema = getHCatSchema(((ListTypeInfo) fieldTypeInfo).getListElementTypeInfo());
        hCatFieldSchema = new HCatFieldSchema(fieldName, HCatFieldSchema.Type.ARRAY, subSchema, comment);
    } else if (Category.MAP == typeCategory) {
        HCatSchema subSchema = getHCatSchema(((MapTypeInfo) fieldTypeInfo).getMapValueTypeInfo());
        hCatFieldSchema = HCatFieldSchema.createMapTypeFieldSchema(fieldName, (PrimitiveTypeInfo) ((MapTypeInfo) fieldTypeInfo).getMapKeyTypeInfo(), subSchema, comment);
    } else {
        throw new TypeNotPresentException(fieldTypeInfo.getTypeName(), null);
    }
    return hCatFieldSchema;
}
Also used : Category(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector.Category) MapTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.MapTypeInfo) StructTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.StructTypeInfo) PrimitiveTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo)

Example 58 with Category

use of org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector.Category in project hive by apache.

the class VerifyFastRow method verifyDeserializeRead.

public static void verifyDeserializeRead(DeserializeRead deserializeRead, PrimitiveTypeInfo primitiveTypeInfo, Writable writable) throws IOException {
    boolean isNull;
    isNull = !deserializeRead.readNextField();
    if (isNull) {
        if (writable != null) {
            TestCase.fail(deserializeRead.getClass().getName() + " field reports null but object is not null " + "(class " + writable.getClass().getName() + ", " + writable.toString() + ")");
        }
        return;
    } else if (writable == null) {
        TestCase.fail("Field report not null but object is null");
    }
    switch(primitiveTypeInfo.getPrimitiveCategory()) {
        case BOOLEAN:
            {
                boolean value = deserializeRead.currentBoolean;
                if (!(writable instanceof BooleanWritable)) {
                    TestCase.fail("Boolean expected writable not Boolean");
                }
                boolean expected = ((BooleanWritable) writable).get();
                if (value != expected) {
                    TestCase.fail("Boolean field mismatch (expected " + expected + " found " + value + ")");
                }
            }
            break;
        case BYTE:
            {
                byte value = deserializeRead.currentByte;
                if (!(writable instanceof ByteWritable)) {
                    TestCase.fail("Byte expected writable not Byte");
                }
                byte expected = ((ByteWritable) writable).get();
                if (value != expected) {
                    TestCase.fail("Byte field mismatch (expected " + (int) expected + " found " + (int) value + ")");
                }
            }
            break;
        case SHORT:
            {
                short value = deserializeRead.currentShort;
                if (!(writable instanceof ShortWritable)) {
                    TestCase.fail("Short expected writable not Short");
                }
                short expected = ((ShortWritable) writable).get();
                if (value != expected) {
                    TestCase.fail("Short field mismatch (expected " + expected + " found " + value + ")");
                }
            }
            break;
        case INT:
            {
                int value = deserializeRead.currentInt;
                if (!(writable instanceof IntWritable)) {
                    TestCase.fail("Integer expected writable not Integer");
                }
                int expected = ((IntWritable) writable).get();
                if (value != expected) {
                    TestCase.fail("Int field mismatch (expected " + expected + " found " + value + ")");
                }
            }
            break;
        case LONG:
            {
                long value = deserializeRead.currentLong;
                if (!(writable instanceof LongWritable)) {
                    TestCase.fail("Long expected writable not Long");
                }
                Long expected = ((LongWritable) writable).get();
                if (value != expected) {
                    TestCase.fail("Long field mismatch (expected " + expected + " found " + value + ")");
                }
            }
            break;
        case FLOAT:
            {
                float value = deserializeRead.currentFloat;
                if (!(writable instanceof FloatWritable)) {
                    TestCase.fail("Float expected writable not Float");
                }
                float expected = ((FloatWritable) writable).get();
                if (value != expected) {
                    TestCase.fail("Float field mismatch (expected " + expected + " found " + value + ")");
                }
            }
            break;
        case DOUBLE:
            {
                double value = deserializeRead.currentDouble;
                if (!(writable instanceof DoubleWritable)) {
                    TestCase.fail("Double expected writable not Double");
                }
                double expected = ((DoubleWritable) writable).get();
                if (value != expected) {
                    TestCase.fail("Double field mismatch (expected " + expected + " found " + value + ")");
                }
            }
            break;
        case STRING:
            {
                byte[] stringBytes = Arrays.copyOfRange(deserializeRead.currentBytes, deserializeRead.currentBytesStart, deserializeRead.currentBytesStart + deserializeRead.currentBytesLength);
                Text text = new Text(stringBytes);
                String string = text.toString();
                String expected = ((Text) writable).toString();
                if (!string.equals(expected)) {
                    TestCase.fail("String field mismatch (expected '" + expected + "' found '" + string + "')");
                }
            }
            break;
        case CHAR:
            {
                byte[] stringBytes = Arrays.copyOfRange(deserializeRead.currentBytes, deserializeRead.currentBytesStart, deserializeRead.currentBytesStart + deserializeRead.currentBytesLength);
                Text text = new Text(stringBytes);
                String string = text.toString();
                HiveChar hiveChar = new HiveChar(string, ((CharTypeInfo) primitiveTypeInfo).getLength());
                HiveChar expected = ((HiveCharWritable) writable).getHiveChar();
                if (!hiveChar.equals(expected)) {
                    TestCase.fail("Char field mismatch (expected '" + expected + "' found '" + hiveChar + "')");
                }
            }
            break;
        case VARCHAR:
            {
                byte[] stringBytes = Arrays.copyOfRange(deserializeRead.currentBytes, deserializeRead.currentBytesStart, deserializeRead.currentBytesStart + deserializeRead.currentBytesLength);
                Text text = new Text(stringBytes);
                String string = text.toString();
                HiveVarchar hiveVarchar = new HiveVarchar(string, ((VarcharTypeInfo) primitiveTypeInfo).getLength());
                HiveVarchar expected = ((HiveVarcharWritable) writable).getHiveVarchar();
                if (!hiveVarchar.equals(expected)) {
                    TestCase.fail("Varchar field mismatch (expected '" + expected + "' found '" + hiveVarchar + "')");
                }
            }
            break;
        case DECIMAL:
            {
                HiveDecimal value = deserializeRead.currentHiveDecimalWritable.getHiveDecimal();
                if (value == null) {
                    TestCase.fail("Decimal field evaluated to NULL");
                }
                HiveDecimal expected = ((HiveDecimalWritable) writable).getHiveDecimal();
                if (!value.equals(expected)) {
                    DecimalTypeInfo decimalTypeInfo = (DecimalTypeInfo) primitiveTypeInfo;
                    int precision = decimalTypeInfo.getPrecision();
                    int scale = decimalTypeInfo.getScale();
                    TestCase.fail("Decimal field mismatch (expected " + expected.toString() + " found " + value.toString() + ") precision " + precision + ", scale " + scale);
                }
            }
            break;
        case DATE:
            {
                Date value = deserializeRead.currentDateWritable.get();
                Date expected = ((DateWritable) writable).get();
                if (!value.equals(expected)) {
                    TestCase.fail("Date field mismatch (expected " + expected.toString() + " found " + value.toString() + ")");
                }
            }
            break;
        case TIMESTAMP:
            {
                Timestamp value = deserializeRead.currentTimestampWritable.getTimestamp();
                Timestamp expected = ((TimestampWritable) writable).getTimestamp();
                if (!value.equals(expected)) {
                    TestCase.fail("Timestamp field mismatch (expected " + expected.toString() + " found " + value.toString() + ")");
                }
            }
            break;
        case INTERVAL_YEAR_MONTH:
            {
                HiveIntervalYearMonth value = deserializeRead.currentHiveIntervalYearMonthWritable.getHiveIntervalYearMonth();
                HiveIntervalYearMonth expected = ((HiveIntervalYearMonthWritable) writable).getHiveIntervalYearMonth();
                if (!value.equals(expected)) {
                    TestCase.fail("HiveIntervalYearMonth field mismatch (expected " + expected.toString() + " found " + value.toString() + ")");
                }
            }
            break;
        case INTERVAL_DAY_TIME:
            {
                HiveIntervalDayTime value = deserializeRead.currentHiveIntervalDayTimeWritable.getHiveIntervalDayTime();
                HiveIntervalDayTime expected = ((HiveIntervalDayTimeWritable) writable).getHiveIntervalDayTime();
                if (!value.equals(expected)) {
                    TestCase.fail("HiveIntervalDayTime field mismatch (expected " + expected.toString() + " found " + value.toString() + ")");
                }
            }
            break;
        case BINARY:
            {
                byte[] byteArray = Arrays.copyOfRange(deserializeRead.currentBytes, deserializeRead.currentBytesStart, deserializeRead.currentBytesStart + deserializeRead.currentBytesLength);
                BytesWritable bytesWritable = (BytesWritable) writable;
                byte[] expected = Arrays.copyOfRange(bytesWritable.getBytes(), 0, bytesWritable.getLength());
                if (byteArray.length != expected.length) {
                    TestCase.fail("Byte Array field mismatch (expected " + Arrays.toString(expected) + " found " + Arrays.toString(byteArray) + ")");
                }
                for (int b = 0; b < byteArray.length; b++) {
                    if (byteArray[b] != expected[b]) {
                        TestCase.fail("Byte Array field mismatch (expected " + Arrays.toString(expected) + " found " + Arrays.toString(byteArray) + ")");
                    }
                }
            }
            break;
        default:
            throw new Error("Unknown primitive category " + primitiveTypeInfo.getPrimitiveCategory());
    }
}
Also used : VarcharTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.VarcharTypeInfo) CharTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.CharTypeInfo) HiveChar(org.apache.hadoop.hive.common.type.HiveChar) DoubleWritable(org.apache.hadoop.hive.serde2.io.DoubleWritable) Text(org.apache.hadoop.io.Text) BytesWritable(org.apache.hadoop.io.BytesWritable) HiveVarchar(org.apache.hadoop.hive.common.type.HiveVarchar) ShortWritable(org.apache.hadoop.hive.serde2.io.ShortWritable) Timestamp(java.sql.Timestamp) Date(java.sql.Date) DecimalTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.DecimalTypeInfo) FloatWritable(org.apache.hadoop.io.FloatWritable) HiveIntervalYearMonth(org.apache.hadoop.hive.common.type.HiveIntervalYearMonth) BooleanWritable(org.apache.hadoop.io.BooleanWritable) HiveDecimal(org.apache.hadoop.hive.common.type.HiveDecimal) LongWritable(org.apache.hadoop.io.LongWritable) ByteWritable(org.apache.hadoop.hive.serde2.io.ByteWritable) IntWritable(org.apache.hadoop.io.IntWritable) HiveIntervalDayTime(org.apache.hadoop.hive.common.type.HiveIntervalDayTime)

Example 59 with Category

use of org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector.Category in project presto by prestodb.

the class HiveBucketing method getJavaObjectInspector.

private static ObjectInspector getJavaObjectInspector(ObjectInspector objectInspector) {
    checkArgument(objectInspector.getCategory() == Category.PRIMITIVE, "Unsupported object inspector category %s", objectInspector.getCategory());
    PrimitiveObjectInspector poi = ((PrimitiveObjectInspector) objectInspector);
    switch(poi.getPrimitiveCategory()) {
        case BOOLEAN:
            return javaBooleanObjectInspector;
        case BYTE:
            return javaByteObjectInspector;
        case SHORT:
            return javaShortObjectInspector;
        case INT:
            return javaIntObjectInspector;
        case LONG:
            return javaLongObjectInspector;
        case STRING:
            return javaStringObjectInspector;
    }
    throw new RuntimeException("Unsupported type: " + poi.getPrimitiveCategory());
}
Also used : PrimitiveObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector)

Aggregations

Category (org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector.Category)25 PrimitiveTypeInfo (org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo)25 PrimitiveCategory (org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector.PrimitiveCategory)23 PrimitiveObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector)16 ObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector)12 BytesWritable (org.apache.hadoop.io.BytesWritable)12 UDFArgumentTypeException (org.apache.hadoop.hive.ql.exec.UDFArgumentTypeException)11 StructObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector)11 DecimalTypeInfo (org.apache.hadoop.hive.serde2.typeinfo.DecimalTypeInfo)11 TypeInfo (org.apache.hadoop.hive.serde2.typeinfo.TypeInfo)11 Text (org.apache.hadoop.io.Text)11 HiveChar (org.apache.hadoop.hive.common.type.HiveChar)10 HiveDecimal (org.apache.hadoop.hive.common.type.HiveDecimal)10 HiveVarchar (org.apache.hadoop.hive.common.type.HiveVarchar)10 IntWritable (org.apache.hadoop.io.IntWritable)10 ByteWritable (org.apache.hadoop.hive.serde2.io.ByteWritable)9 DateWritable (org.apache.hadoop.hive.serde2.io.DateWritable)9 DoubleWritable (org.apache.hadoop.hive.serde2.io.DoubleWritable)9 ShortWritable (org.apache.hadoop.hive.serde2.io.ShortWritable)9 BooleanWritable (org.apache.hadoop.io.BooleanWritable)9