Search in sources :

Example 31 with VectorRandomRowSource

use of org.apache.hadoop.hive.ql.exec.vector.VectorRandomRowSource in project hive by apache.

the class TestVectorArithmetic method doTestsWithDiffColumnScalar.

private void doTestsWithDiffColumnScalar(Random random, TypeInfo typeInfo1, TypeInfo typeInfo2, ColumnScalarMode columnScalarMode, Arithmetic arithmetic, boolean tryDecimal64) throws Exception {
    String typeName1 = typeInfo1.getTypeName();
    PrimitiveCategory primitiveCategory1 = ((PrimitiveTypeInfo) typeInfo1).getPrimitiveCategory();
    String typeName2 = typeInfo2.getTypeName();
    PrimitiveCategory primitiveCategory2 = ((PrimitiveTypeInfo) typeInfo2).getPrimitiveCategory();
    List<GenerationSpec> generationSpecList = new ArrayList<GenerationSpec>();
    List<DataTypePhysicalVariation> explicitDataTypePhysicalVariationList = new ArrayList<DataTypePhysicalVariation>();
    List<String> columns = new ArrayList<String>();
    int columnNum = 1;
    ExprNodeDesc col1Expr;
    Object scalar1Object = null;
    final boolean decimal64Enable1 = checkDecimal64(tryDecimal64, typeInfo1);
    if (columnScalarMode == ColumnScalarMode.COLUMN_COLUMN || columnScalarMode == ColumnScalarMode.COLUMN_SCALAR) {
        generationSpecList.add(GenerationSpec.createSameType(typeInfo1));
        explicitDataTypePhysicalVariationList.add(decimal64Enable1 ? DataTypePhysicalVariation.DECIMAL_64 : DataTypePhysicalVariation.NONE);
        String columnName = "col" + (columnNum++);
        col1Expr = new ExprNodeColumnDesc(typeInfo1, columnName, "table", false);
        columns.add(columnName);
    } else {
        scalar1Object = VectorRandomRowSource.randomPrimitiveObject(random, (PrimitiveTypeInfo) typeInfo1);
        // Adjust the decimal type to the scalar's type...
        if (typeInfo1 instanceof DecimalTypeInfo) {
            typeInfo1 = getDecimalScalarTypeInfo(scalar1Object);
        }
        col1Expr = new ExprNodeConstantDesc(typeInfo1, scalar1Object);
    }
    ExprNodeDesc col2Expr;
    Object scalar2Object = null;
    final boolean decimal64Enable2 = checkDecimal64(tryDecimal64, typeInfo2);
    if (columnScalarMode == ColumnScalarMode.COLUMN_COLUMN || columnScalarMode == ColumnScalarMode.SCALAR_COLUMN) {
        generationSpecList.add(GenerationSpec.createSameType(typeInfo2));
        explicitDataTypePhysicalVariationList.add(decimal64Enable2 ? DataTypePhysicalVariation.DECIMAL_64 : DataTypePhysicalVariation.NONE);
        String columnName = "col" + (columnNum++);
        col2Expr = new ExprNodeColumnDesc(typeInfo2, columnName, "table", false);
        columns.add(columnName);
    } else {
        scalar2Object = VectorRandomRowSource.randomPrimitiveObject(random, (PrimitiveTypeInfo) typeInfo2);
        // Adjust the decimal type to the scalar's type...
        if (typeInfo2 instanceof DecimalTypeInfo) {
            typeInfo2 = getDecimalScalarTypeInfo(scalar2Object);
        }
        col2Expr = new ExprNodeConstantDesc(typeInfo2, scalar2Object);
    }
    List<ObjectInspector> objectInspectorList = new ArrayList<ObjectInspector>();
    objectInspectorList.add(VectorRandomRowSource.getObjectInspector(typeInfo1));
    objectInspectorList.add(VectorRandomRowSource.getObjectInspector(typeInfo2));
    List<ExprNodeDesc> children = new ArrayList<ExprNodeDesc>();
    children.add(col1Expr);
    children.add(col2Expr);
    // ----------------------------------------------------------------------------------------------
    String[] columnNames = columns.toArray(new String[0]);
    VectorRandomRowSource rowSource = new VectorRandomRowSource();
    rowSource.initGenerationSpecSchema(random, generationSpecList, /* maxComplexDepth */
    0, /* allowNull */
    true, /* isUnicodeOk */
    true, explicitDataTypePhysicalVariationList);
    Object[][] randomRows = rowSource.randomRows(100000);
    VectorRandomBatchSource batchSource = VectorRandomBatchSource.createInterestingBatches(random, rowSource, randomRows, null);
    GenericUDF genericUdf;
    switch(arithmetic) {
        case ADD:
            genericUdf = new GenericUDFOPPlus();
            break;
        case SUBTRACT:
            genericUdf = new GenericUDFOPMinus();
            break;
        case MULTIPLY:
            genericUdf = new GenericUDFOPMultiply();
            break;
        case DIVIDE:
            genericUdf = new GenericUDFOPDivide();
            break;
        case MODULUS:
            genericUdf = new GenericUDFOPMod();
            break;
        default:
            throw new RuntimeException("Unexpected arithmetic " + arithmetic);
    }
    ObjectInspector[] objectInspectors = objectInspectorList.toArray(new ObjectInspector[objectInspectorList.size()]);
    ObjectInspector outputObjectInspector = null;
    try {
        outputObjectInspector = genericUdf.initialize(objectInspectors);
    } catch (Exception e) {
        Assert.fail(e.toString());
    }
    TypeInfo outputTypeInfo = TypeInfoUtils.getTypeInfoFromObjectInspector(outputObjectInspector);
    ExprNodeGenericFuncDesc exprDesc = new ExprNodeGenericFuncDesc(outputTypeInfo, genericUdf, children);
    final int rowCount = randomRows.length;
    Object[][] resultObjectsArray = new Object[ArithmeticTestMode.count][];
    for (int i = 0; i < ArithmeticTestMode.count; i++) {
        Object[] resultObjects = new Object[rowCount];
        resultObjectsArray[i] = resultObjects;
        ArithmeticTestMode arithmeticTestMode = ArithmeticTestMode.values()[i];
        switch(arithmeticTestMode) {
            case ROW_MODE:
                doRowArithmeticTest(typeInfo1, typeInfo2, columns, children, exprDesc, arithmetic, randomRows, columnScalarMode, rowSource.rowStructObjectInspector(), outputTypeInfo, resultObjects);
                break;
            case ADAPTOR:
            case VECTOR_EXPRESSION:
                doVectorArithmeticTest(typeInfo1, typeInfo2, columns, columnNames, rowSource.typeInfos(), rowSource.dataTypePhysicalVariations(), children, exprDesc, arithmetic, arithmeticTestMode, columnScalarMode, batchSource, exprDesc.getWritableObjectInspector(), outputTypeInfo, resultObjects);
                break;
            default:
                throw new RuntimeException("Unexpected IF statement test mode " + arithmeticTestMode);
        }
    }
    for (int i = 0; i < rowCount; i++) {
        // Row-mode is the expected value.
        Object expectedResult = resultObjectsArray[0][i];
        for (int v = 1; v < ArithmeticTestMode.count; v++) {
            Object vectorResult = resultObjectsArray[v][i];
            if (expectedResult == null || vectorResult == null) {
                if (expectedResult != null || vectorResult != null) {
                    Assert.fail("Row " + i + " typeName1 " + typeName1 + " typeName2 " + typeName2 + " outputTypeName " + outputTypeInfo.getTypeName() + " " + arithmetic + " " + ArithmeticTestMode.values()[v] + " " + columnScalarMode + " result is NULL " + (vectorResult == null) + " does not match row-mode expected result is NULL " + (expectedResult == null) + (columnScalarMode == ColumnScalarMode.SCALAR_COLUMN ? " scalar1 " + scalar1Object.toString() : "") + " row values " + Arrays.toString(randomRows[i]) + (columnScalarMode == ColumnScalarMode.COLUMN_SCALAR ? " scalar2 " + scalar2Object.toString() : ""));
                }
            } else {
                if (!expectedResult.equals(vectorResult)) {
                    Assert.fail("Row " + i + " typeName1 " + typeName1 + " typeName2 " + typeName2 + " outputTypeName " + outputTypeInfo.getTypeName() + " " + arithmetic + " " + ArithmeticTestMode.values()[v] + " " + columnScalarMode + " result " + vectorResult.toString() + " (" + vectorResult.getClass().getSimpleName() + ")" + " does not match row-mode expected result " + expectedResult.toString() + " (" + expectedResult.getClass().getSimpleName() + ")" + (columnScalarMode == ColumnScalarMode.SCALAR_COLUMN ? " scalar1 " + scalar1Object.toString() : "") + " row values " + Arrays.toString(randomRows[i]) + (columnScalarMode == ColumnScalarMode.COLUMN_SCALAR ? " scalar2 " + scalar2Object.toString() : ""));
                }
            }
        }
    }
}
Also used : GenericUDFOPMultiply(org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPMultiply) ArrayList(java.util.ArrayList) PrimitiveTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo) DataTypePhysicalVariation(org.apache.hadoop.hive.common.type.DataTypePhysicalVariation) ExprNodeColumnDesc(org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc) ExprNodeDesc(org.apache.hadoop.hive.ql.plan.ExprNodeDesc) PrimitiveCategory(org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector.PrimitiveCategory) ExprNodeConstantDesc(org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc) ObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector) VectorRandomBatchSource(org.apache.hadoop.hive.ql.exec.vector.VectorRandomBatchSource) GenericUDFOPMod(org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPMod) ExprNodeGenericFuncDesc(org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc) PrimitiveTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo) DecimalTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.DecimalTypeInfo) TypeInfo(org.apache.hadoop.hive.serde2.typeinfo.TypeInfo) SemanticException(org.apache.hadoop.hive.ql.parse.SemanticException) HiveException(org.apache.hadoop.hive.ql.metadata.HiveException) GenerationSpec(org.apache.hadoop.hive.ql.exec.vector.VectorRandomRowSource.GenerationSpec) DecimalTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.DecimalTypeInfo) GenericUDF(org.apache.hadoop.hive.ql.udf.generic.GenericUDF) GenericUDFOPDivide(org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPDivide) GenericUDFOPPlus(org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPPlus) GenericUDFOPMinus(org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPMinus) VectorRandomRowSource(org.apache.hadoop.hive.ql.exec.vector.VectorRandomRowSource)

Example 32 with VectorRandomRowSource

use of org.apache.hadoop.hive.ql.exec.vector.VectorRandomRowSource in project hive by apache.

the class TestVectorIndex method doVectorCastTest.

private boolean doVectorCastTest(TypeInfo typeInfo, List<String> columns, String[] columnNames, TypeInfo[] typeInfos, DataTypePhysicalVariation[] dataTypePhysicalVariations, List<ExprNodeDesc> children, GenericUDF udf, ExprNodeGenericFuncDesc exprDesc, IndexTestMode indexTestMode, VectorRandomBatchSource batchSource, ObjectInspector objectInspector, TypeInfo outputTypeInfo, Object[] resultObjects) throws Exception {
    HiveConf hiveConf = new HiveConf();
    if (indexTestMode == IndexTestMode.ADAPTOR) {
        hiveConf.setBoolVar(HiveConf.ConfVars.HIVE_TEST_VECTOR_ADAPTOR_OVERRIDE, true);
    }
    VectorizationContext vectorizationContext = new VectorizationContext("name", columns, Arrays.asList(typeInfos), Arrays.asList(dataTypePhysicalVariations), hiveConf);
    VectorExpression vectorExpression = vectorizationContext.getVectorExpression(exprDesc, VectorExpressionDescriptor.Mode.PROJECTION);
    vectorExpression.transientInit(hiveConf);
    if (indexTestMode == IndexTestMode.VECTOR_EXPRESSION && vectorExpression instanceof VectorUDFAdaptor) {
        System.out.println("*NO NATIVE VECTOR EXPRESSION* typeInfo " + typeInfo.toString() + " indexTestMode " + indexTestMode + " vectorExpression " + vectorExpression.toString());
    }
    System.out.println("*VECTOR EXPRESSION* " + vectorExpression.getClass().getSimpleName());
    /*
    System.out.println(
        "*DEBUG* typeInfo " + typeInfo.toString() +
        " indexTestMode " + indexTestMode +
        " vectorExpression " + vectorExpression.toString());
    */
    VectorRandomRowSource rowSource = batchSource.getRowSource();
    VectorizedRowBatchCtx batchContext = new VectorizedRowBatchCtx(columnNames, rowSource.typeInfos(), rowSource.dataTypePhysicalVariations(), /* dataColumnNums */
    null, /* partitionColumnCount */
    0, /* virtualColumnCount */
    0, /* neededVirtualColumns */
    null, vectorizationContext.getScratchColumnTypeNames(), vectorizationContext.getScratchDataTypePhysicalVariations());
    VectorizedRowBatch batch = batchContext.createVectorizedRowBatch();
    VectorExtractRow resultVectorExtractRow = new VectorExtractRow();
    resultVectorExtractRow.init(new TypeInfo[] { outputTypeInfo }, new int[] { vectorExpression.getOutputColumnNum() });
    Object[] scrqtchRow = new Object[1];
    /*
    System.out.println(
        "*DEBUG* typeInfo1 " + typeInfo1.toString() +
        " typeInfo2 " + typeInfo2.toString() +
        " arithmeticTestMode " + arithmeticTestMode +
        " columnScalarMode " + columnScalarMode +
        " vectorExpression " + vectorExpression.toString());
    */
    batchSource.resetBatchIteration();
    int rowIndex = 0;
    while (true) {
        if (!batchSource.fillNextBatch(batch)) {
            break;
        }
        vectorExpression.evaluate(batch);
        extractResultObjects(batch, rowIndex, resultVectorExtractRow, scrqtchRow, objectInspector, resultObjects);
        rowIndex += batch.size;
    }
    return true;
}
Also used : VectorizedRowBatchCtx(org.apache.hadoop.hive.ql.exec.vector.VectorizedRowBatchCtx) VectorizedRowBatch(org.apache.hadoop.hive.ql.exec.vector.VectorizedRowBatch) HiveConf(org.apache.hadoop.hive.conf.HiveConf) VectorizationContext(org.apache.hadoop.hive.ql.exec.vector.VectorizationContext) VectorUDFAdaptor(org.apache.hadoop.hive.ql.exec.vector.udf.VectorUDFAdaptor) VectorExtractRow(org.apache.hadoop.hive.ql.exec.vector.VectorExtractRow) VectorRandomRowSource(org.apache.hadoop.hive.ql.exec.vector.VectorRandomRowSource)

Example 33 with VectorRandomRowSource

use of org.apache.hadoop.hive.ql.exec.vector.VectorRandomRowSource in project hive by apache.

the class TestVectorCastStatement method doVectorCastTest.

private boolean doVectorCastTest(TypeInfo typeInfo, TypeInfo targetTypeInfo, List<String> columns, String[] columnNames, TypeInfo[] typeInfos, DataTypePhysicalVariation[] dataTypePhysicalVariations, List<ExprNodeDesc> children, CastStmtTestMode castStmtTestMode, VectorRandomBatchSource batchSource, Object[] resultObjects) throws Exception {
    GenericUDF udf;
    try {
        udf = VectorizationContext.getGenericUDFForCast(targetTypeInfo);
    } catch (HiveException e) {
        return false;
    }
    ExprNodeGenericFuncDesc exprDesc = new ExprNodeGenericFuncDesc(targetTypeInfo, udf, children);
    HiveConf hiveConf = new HiveConf();
    if (castStmtTestMode == CastStmtTestMode.ADAPTOR) {
        hiveConf.setBoolVar(HiveConf.ConfVars.HIVE_TEST_VECTOR_ADAPTOR_OVERRIDE, true);
    }
    VectorizationContext vectorizationContext = new VectorizationContext("name", columns, Arrays.asList(typeInfos), Arrays.asList(dataTypePhysicalVariations), hiveConf);
    VectorExpression vectorExpression = vectorizationContext.getVectorExpression(exprDesc);
    vectorExpression.transientInit(hiveConf);
    if (castStmtTestMode == CastStmtTestMode.VECTOR_EXPRESSION && vectorExpression instanceof VectorUDFAdaptor) {
        System.out.println("*NO NATIVE VECTOR EXPRESSION* typeInfo " + typeInfo.toString() + " castStmtTestMode " + castStmtTestMode + " vectorExpression " + vectorExpression.toString());
    }
    // System.out.println("*VECTOR EXPRESSION* " + vectorExpression.getClass().getSimpleName());
    /*
    System.out.println(
        "*DEBUG* typeInfo " + typeInfo.toString() +
        " targetTypeInfo " + targetTypeInfo +
        " castStmtTestMode " + castStmtTestMode +
        " vectorExpression " + vectorExpression.toString());
    */
    VectorRandomRowSource rowSource = batchSource.getRowSource();
    VectorizedRowBatchCtx batchContext = new VectorizedRowBatchCtx(columnNames, rowSource.typeInfos(), rowSource.dataTypePhysicalVariations(), /* dataColumnNums */
    null, /* partitionColumnCount */
    0, /* virtualColumnCount */
    0, /* neededVirtualColumns */
    null, vectorizationContext.getScratchColumnTypeNames(), vectorizationContext.getScratchDataTypePhysicalVariations());
    VectorizedRowBatch batch = batchContext.createVectorizedRowBatch();
    VectorExtractRow resultVectorExtractRow = new VectorExtractRow();
    resultVectorExtractRow.init(new TypeInfo[] { targetTypeInfo }, new int[] { vectorExpression.getOutputColumnNum() });
    Object[] scrqtchRow = new Object[1];
    batchSource.resetBatchIteration();
    int rowIndex = 0;
    while (true) {
        if (!batchSource.fillNextBatch(batch)) {
            break;
        }
        vectorExpression.evaluate(batch);
        extractResultObjects(batch, rowIndex, resultVectorExtractRow, scrqtchRow, resultObjects);
        rowIndex += batch.size;
    }
    return true;
}
Also used : HiveException(org.apache.hadoop.hive.ql.metadata.HiveException) ExprNodeGenericFuncDesc(org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc) VectorizationContext(org.apache.hadoop.hive.ql.exec.vector.VectorizationContext) VectorUDFAdaptor(org.apache.hadoop.hive.ql.exec.vector.udf.VectorUDFAdaptor) VectorExtractRow(org.apache.hadoop.hive.ql.exec.vector.VectorExtractRow) VectorizedRowBatchCtx(org.apache.hadoop.hive.ql.exec.vector.VectorizedRowBatchCtx) VectorizedRowBatch(org.apache.hadoop.hive.ql.exec.vector.VectorizedRowBatch) GenericUDF(org.apache.hadoop.hive.ql.udf.generic.GenericUDF) HiveConf(org.apache.hadoop.hive.conf.HiveConf) VectorExpression(org.apache.hadoop.hive.ql.exec.vector.expressions.VectorExpression) VectorRandomRowSource(org.apache.hadoop.hive.ql.exec.vector.VectorRandomRowSource)

Example 34 with VectorRandomRowSource

use of org.apache.hadoop.hive.ql.exec.vector.VectorRandomRowSource in project hive by apache.

the class TestVectorStringConcat method doStringConcatTestsWithDiffColumnScalar.

private void doStringConcatTestsWithDiffColumnScalar(Random random, String stringTypeName1, String stringTypeName2, ColumnScalarMode columnScalarMode) throws Exception {
    TypeInfo stringTypeInfo1 = TypeInfoUtils.getTypeInfoFromTypeString(stringTypeName1);
    PrimitiveCategory stringPrimitiveCategory1 = ((PrimitiveTypeInfo) stringTypeInfo1).getPrimitiveCategory();
    TypeInfo stringTypeInfo2 = TypeInfoUtils.getTypeInfoFromTypeString(stringTypeName2);
    PrimitiveCategory stringPrimitiveCategory2 = ((PrimitiveTypeInfo) stringTypeInfo2).getPrimitiveCategory();
    String functionName = "concat";
    List<GenerationSpec> generationSpecList = new ArrayList<GenerationSpec>();
    List<DataTypePhysicalVariation> explicitDataTypePhysicalVariationList = new ArrayList<DataTypePhysicalVariation>();
    List<String> columns = new ArrayList<String>();
    int columnNum = 1;
    ExprNodeDesc col1Expr;
    if (columnScalarMode == ColumnScalarMode.COLUMN_COLUMN || columnScalarMode == ColumnScalarMode.COLUMN_SCALAR) {
        generationSpecList.add(GenerationSpec.createSameType(stringTypeInfo1));
        explicitDataTypePhysicalVariationList.add(DataTypePhysicalVariation.NONE);
        String columnName = "col" + (columnNum++);
        col1Expr = new ExprNodeColumnDesc(stringTypeInfo1, columnName, "table", false);
        columns.add(columnName);
    } else {
        Object scalar1Object = VectorRandomRowSource.randomPrimitiveObject(random, (PrimitiveTypeInfo) stringTypeInfo1);
        col1Expr = new ExprNodeConstantDesc(stringTypeInfo1, scalar1Object);
    }
    ExprNodeDesc col2Expr;
    if (columnScalarMode == ColumnScalarMode.COLUMN_COLUMN || columnScalarMode == ColumnScalarMode.SCALAR_COLUMN) {
        generationSpecList.add(GenerationSpec.createSameType(stringTypeInfo2));
        explicitDataTypePhysicalVariationList.add(DataTypePhysicalVariation.NONE);
        String columnName = "col" + (columnNum++);
        col2Expr = new ExprNodeColumnDesc(stringTypeInfo2, columnName, "table", false);
        columns.add(columnName);
    } else {
        Object scalar2Object = VectorRandomRowSource.randomPrimitiveObject(random, (PrimitiveTypeInfo) stringTypeInfo2);
        col2Expr = new ExprNodeConstantDesc(stringTypeInfo2, scalar2Object);
    }
    List<ExprNodeDesc> children = new ArrayList<ExprNodeDesc>();
    children.add(col1Expr);
    children.add(col2Expr);
    // ----------------------------------------------------------------------------------------------
    String[] columnNames = columns.toArray(new String[0]);
    VectorRandomRowSource rowSource = new VectorRandomRowSource();
    rowSource.initGenerationSpecSchema(random, generationSpecList, /* maxComplexDepth */
    0, /* allowNull */
    true, /* isUnicodeOk */
    true, explicitDataTypePhysicalVariationList);
    Object[][] randomRows = rowSource.randomRows(100000);
    VectorRandomBatchSource batchSource = VectorRandomBatchSource.createInterestingBatches(random, rowSource, randomRows, null);
    String[] outputScratchTypeNames = new String[] { "string" };
    VectorizedRowBatchCtx batchContext = new VectorizedRowBatchCtx(columnNames, rowSource.typeInfos(), rowSource.dataTypePhysicalVariations(), /* dataColumnNums */
    null, /* partitionColumnCount */
    0, /* virtualColumnCount */
    0, /* neededVirtualColumns */
    null, outputScratchTypeNames, null);
    GenericUDF genericUdf;
    FunctionInfo funcInfo = null;
    try {
        funcInfo = FunctionRegistry.getFunctionInfo(functionName);
    } catch (SemanticException e) {
        Assert.fail("Failed to load " + functionName + " " + e);
    }
    genericUdf = funcInfo.getGenericUDF();
    final int rowCount = randomRows.length;
    Object[][] resultObjectsArray = new Object[StringConcatTestMode.count][];
    for (int i = 0; i < StringConcatTestMode.count; i++) {
        Object[] resultObjects = new Object[rowCount];
        resultObjectsArray[i] = resultObjects;
        StringConcatTestMode stringConcatTestMode = StringConcatTestMode.values()[i];
        switch(stringConcatTestMode) {
            case ROW_MODE:
                doRowStringConcatTest(stringTypeInfo1, stringTypeInfo2, columns, children, randomRows, columnScalarMode, rowSource.rowStructObjectInspector(), genericUdf, resultObjects);
                break;
            case ADAPTOR:
            case VECTOR_EXPRESSION:
                doVectorStringConcatTest(stringTypeInfo1, stringTypeInfo2, columns, rowSource.typeInfos(), children, stringConcatTestMode, columnScalarMode, batchSource, batchContext, rowSource.rowStructObjectInspector(), genericUdf, resultObjects);
                break;
            default:
                throw new RuntimeException("Unexpected IF statement test mode " + stringConcatTestMode);
        }
    }
    for (int i = 0; i < rowCount; i++) {
        // Row-mode is the expected value.
        Object expectedResult = resultObjectsArray[0][i];
        for (int v = 1; v < StringConcatTestMode.count; v++) {
            Object vectorResult = resultObjectsArray[v][i];
            if (expectedResult == null || vectorResult == null) {
                if (expectedResult != null || vectorResult != null) {
                    Assert.fail("Row " + i + " " + StringConcatTestMode.values()[v] + " " + columnScalarMode + " result is NULL " + (vectorResult == null) + " does not match row-mode expected result is NULL " + (expectedResult == null) + " row values " + Arrays.toString(randomRows[i]));
                }
            } else {
                if (!expectedResult.equals(vectorResult)) {
                    Assert.fail("Row " + i + " " + StringConcatTestMode.values()[v] + " " + columnScalarMode + " result \"" + vectorResult.toString() + "\"" + " (" + vectorResult.getClass().getSimpleName() + ")" + " does not match row-mode expected result \"" + expectedResult.toString() + "\"" + " (" + expectedResult.getClass().getSimpleName() + ")" + " row values " + Arrays.toString(randomRows[i]));
                }
            }
        }
    }
}
Also used : ArrayList(java.util.ArrayList) PrimitiveTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo) DataTypePhysicalVariation(org.apache.hadoop.hive.common.type.DataTypePhysicalVariation) ExprNodeColumnDesc(org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc) ExprNodeDesc(org.apache.hadoop.hive.ql.plan.ExprNodeDesc) PrimitiveCategory(org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector.PrimitiveCategory) SemanticException(org.apache.hadoop.hive.ql.parse.SemanticException) ExprNodeConstantDesc(org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc) VectorRandomBatchSource(org.apache.hadoop.hive.ql.exec.vector.VectorRandomBatchSource) FunctionInfo(org.apache.hadoop.hive.ql.exec.FunctionInfo) PrimitiveTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo) VarcharTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.VarcharTypeInfo) DecimalTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.DecimalTypeInfo) TypeInfo(org.apache.hadoop.hive.serde2.typeinfo.TypeInfo) CharTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.CharTypeInfo) GenerationSpec(org.apache.hadoop.hive.ql.exec.vector.VectorRandomRowSource.GenerationSpec) VectorizedRowBatchCtx(org.apache.hadoop.hive.ql.exec.vector.VectorizedRowBatchCtx) GenericUDF(org.apache.hadoop.hive.ql.udf.generic.GenericUDF) VectorRandomRowSource(org.apache.hadoop.hive.ql.exec.vector.VectorRandomRowSource)

Example 35 with VectorRandomRowSource

use of org.apache.hadoop.hive.ql.exec.vector.VectorRandomRowSource in project hive by apache.

the class TestVectorStringUnary method doTests.

private void doTests(Random random, String typeName, String functionName) throws Exception {
    TypeInfo typeInfo = TypeInfoUtils.getTypeInfoFromTypeString(typeName);
    TypeInfo targetTypeInfo;
    if (functionName.equals("char_length") || functionName.equals("length") || functionName.equals("octet_length")) {
        targetTypeInfo = TypeInfoFactory.intTypeInfo;
    } else {
        targetTypeInfo = typeInfo;
    }
    List<GenerationSpec> generationSpecList = new ArrayList<GenerationSpec>();
    List<DataTypePhysicalVariation> explicitDataTypePhysicalVariationList = new ArrayList<DataTypePhysicalVariation>();
    List<String> columns = new ArrayList<String>();
    int columnNum = 1;
    ExprNodeDesc col1Expr;
    StringGenerationOption stringGenerationOption = new StringGenerationOption(true, true);
    generationSpecList.add(GenerationSpec.createStringFamily(typeInfo, stringGenerationOption));
    explicitDataTypePhysicalVariationList.add(DataTypePhysicalVariation.NONE);
    String columnName = "col" + (columnNum++);
    col1Expr = new ExprNodeColumnDesc(typeInfo, columnName, "table", false);
    columns.add(columnName);
    VectorRandomRowSource rowSource = new VectorRandomRowSource();
    rowSource.initGenerationSpecSchema(random, generationSpecList, /* maxComplexDepth */
    0, /* allowNull */
    true, /* isUnicodeOk */
    true, explicitDataTypePhysicalVariationList);
    List<ExprNodeDesc> children = new ArrayList<ExprNodeDesc>();
    children.add(col1Expr);
    // ----------------------------------------------------------------------------------------------
    String[] columnNames = columns.toArray(new String[0]);
    String[] outputScratchTypeNames = new String[] { targetTypeInfo.getTypeName() };
    DataTypePhysicalVariation[] outputDataTypePhysicalVariations = new DataTypePhysicalVariation[] { DataTypePhysicalVariation.NONE };
    VectorizedRowBatchCtx batchContext = new VectorizedRowBatchCtx(columnNames, rowSource.typeInfos(), rowSource.dataTypePhysicalVariations(), /* dataColumnNums */
    null, /* partitionColumnCount */
    0, /* virtualColumnCount */
    0, /* neededVirtualColumns */
    null, outputScratchTypeNames, outputDataTypePhysicalVariations);
    Object[][] randomRows = rowSource.randomRows(100000);
    VectorRandomBatchSource batchSource = VectorRandomBatchSource.createInterestingBatches(random, rowSource, randomRows, null);
    GenericUDF genericUdf;
    FunctionInfo funcInfo = null;
    try {
        funcInfo = FunctionRegistry.getFunctionInfo(functionName);
    } catch (SemanticException e) {
        Assert.fail("Failed to load " + functionName + " " + e);
    }
    genericUdf = funcInfo.getGenericUDF();
    final int rowCount = randomRows.length;
    Object[][] resultObjectsArray = new Object[StringUnaryTestMode.count][];
    for (int i = 0; i < StringUnaryTestMode.count; i++) {
        Object[] resultObjects = new Object[rowCount];
        resultObjectsArray[i] = resultObjects;
        StringUnaryTestMode stringUnaryTestMode = StringUnaryTestMode.values()[i];
        switch(stringUnaryTestMode) {
            case ROW_MODE:
                doRowIfTest(typeInfo, targetTypeInfo, columns, children, randomRows, rowSource.rowStructObjectInspector(), genericUdf, resultObjects);
                break;
            case ADAPTOR:
            case VECTOR_EXPRESSION:
                doVectorIfTest(typeInfo, targetTypeInfo, columns, rowSource.typeInfos(), rowSource.dataTypePhysicalVariations(), children, stringUnaryTestMode, batchSource, batchContext, genericUdf, resultObjects);
                break;
            default:
                throw new RuntimeException("Unexpected STRING Unary test mode " + stringUnaryTestMode);
        }
    }
    for (int i = 0; i < rowCount; i++) {
        // Row-mode is the expected value.
        Object expectedResult = resultObjectsArray[0][i];
        for (int v = 1; v < StringUnaryTestMode.count; v++) {
            Object vectorResult = resultObjectsArray[v][i];
            if (expectedResult == null || vectorResult == null) {
                if (expectedResult != null || vectorResult != null) {
                    Assert.fail("Row " + i + " " + StringUnaryTestMode.values()[v] + " typeName " + typeName + " result is NULL " + (vectorResult == null) + " does not match row-mode expected result is NULL " + (expectedResult == null) + " functionName " + functionName + " genericUdf " + genericUdf.getClass().getSimpleName());
                }
            } else {
                if (!expectedResult.equals(vectorResult)) {
                    Assert.fail("Row " + i + " " + StringUnaryTestMode.values()[v] + " typeName " + typeName + " result \"" + vectorResult.toString() + "\"" + " (" + vectorResult.getClass().getSimpleName() + ")" + " does not match row-mode expected result \"" + expectedResult.toString() + "\"" + " (" + expectedResult.getClass().getSimpleName() + ")" + " functionName " + functionName + " genericUdf " + genericUdf.getClass().getSimpleName());
                }
            }
        }
    }
}
Also used : ArrayList(java.util.ArrayList) DataTypePhysicalVariation(org.apache.hadoop.hive.common.type.DataTypePhysicalVariation) ExprNodeColumnDesc(org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc) ExprNodeDesc(org.apache.hadoop.hive.ql.plan.ExprNodeDesc) SemanticException(org.apache.hadoop.hive.ql.parse.SemanticException) VectorRandomBatchSource(org.apache.hadoop.hive.ql.exec.vector.VectorRandomBatchSource) FunctionInfo(org.apache.hadoop.hive.ql.exec.FunctionInfo) PrimitiveTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo) DecimalTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.DecimalTypeInfo) TypeInfo(org.apache.hadoop.hive.serde2.typeinfo.TypeInfo) GenerationSpec(org.apache.hadoop.hive.ql.exec.vector.VectorRandomRowSource.GenerationSpec) VectorizedRowBatchCtx(org.apache.hadoop.hive.ql.exec.vector.VectorizedRowBatchCtx) GenericUDF(org.apache.hadoop.hive.ql.udf.generic.GenericUDF) StringGenerationOption(org.apache.hadoop.hive.ql.exec.vector.VectorRandomRowSource.StringGenerationOption) VectorRandomRowSource(org.apache.hadoop.hive.ql.exec.vector.VectorRandomRowSource)

Aggregations

VectorRandomRowSource (org.apache.hadoop.hive.ql.exec.vector.VectorRandomRowSource)50 Random (java.util.Random)24 VerifyFastRowHashMap (org.apache.hadoop.hive.ql.exec.vector.mapjoin.fast.CheckFastRowHashMap.VerifyFastRowHashMap)24 Test (org.junit.Test)24 TypeInfo (org.apache.hadoop.hive.serde2.typeinfo.TypeInfo)20 ArrayList (java.util.ArrayList)19 DataTypePhysicalVariation (org.apache.hadoop.hive.common.type.DataTypePhysicalVariation)19 VectorRandomBatchSource (org.apache.hadoop.hive.ql.exec.vector.VectorRandomBatchSource)19 ExprNodeColumnDesc (org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc)19 ExprNodeDesc (org.apache.hadoop.hive.ql.plan.ExprNodeDesc)19 GenerationSpec (org.apache.hadoop.hive.ql.exec.vector.VectorRandomRowSource.GenerationSpec)18 PrimitiveTypeInfo (org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo)17 DecimalTypeInfo (org.apache.hadoop.hive.serde2.typeinfo.DecimalTypeInfo)16 GenericUDF (org.apache.hadoop.hive.ql.udf.generic.GenericUDF)15 VectorizedRowBatchCtx (org.apache.hadoop.hive.ql.exec.vector.VectorizedRowBatchCtx)12 ExprNodeGenericFuncDesc (org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc)12 ExprNodeConstantDesc (org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc)11 ObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector)11 PrimitiveCategory (org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector.PrimitiveCategory)8 HiveConf (org.apache.hadoop.hive.conf.HiveConf)7