Examples with GenericUDFIf org.apache.hadoop.hive.ql.udf.generic.GenericUDFIf used on opensource projects

Example 1 with GenericUDFIf

use of org.apache.hadoop.hive.ql.udf.generic.GenericUDFIf in project hive by apache.

the class TestVectorIfStatement method doVectorIfTest.

private void doVectorIfTest(TypeInfo typeInfo, IfVariation ifVariation, List<String> columns, String[] columnNames, TypeInfo[] typeInfos, DataTypePhysicalVariation[] dataTypePhysicalVariations, List<ExprNodeDesc> children, IfStmtTestMode ifStmtTestMode, ColumnScalarMode columnScalarMode, VectorRandomBatchSource batchSource, Object[] resultObjects) throws Exception {
    final boolean isFilter = ifVariation.isFilter;
    GenericUDF udf;
    switch(ifStmtTestMode) {
        case VECTOR_EXPRESSION:
            udf = new GenericUDFIf();
            break;
        case ADAPTOR_WHEN:
            udf = new GenericUDFWhen();
            break;
        default:
            throw new RuntimeException("Unexpected IF statement test mode " + ifStmtTestMode);
    }
    ExprNodeGenericFuncDesc exprDesc = new ExprNodeGenericFuncDesc(typeInfo, udf, children);
    String ifExprMode = (ifStmtTestMode != IfStmtTestMode.VECTOR_EXPRESSION ? "adaptor" : "good");
    HiveConf hiveConf = new HiveConf();
    hiveConf.setVar(HiveConf.ConfVars.HIVE_VECTORIZED_IF_EXPR_MODE, ifExprMode);
    VectorizationContext vectorizationContext = new VectorizationContext("name", columns, Arrays.asList(typeInfos), Arrays.asList(dataTypePhysicalVariations), hiveConf);
    VectorExpression vectorExpression = vectorizationContext.getVectorExpression(exprDesc, (isFilter ? VectorExpressionDescriptor.Mode.FILTER : VectorExpressionDescriptor.Mode.PROJECTION));
    final TypeInfo outputTypeInfo;
    final ObjectInspector objectInspector;
    if (!isFilter) {
        outputTypeInfo = vectorExpression.getOutputTypeInfo();
        objectInspector = TypeInfoUtils.getStandardWritableObjectInspectorFromTypeInfo(outputTypeInfo);
    } else {
        outputTypeInfo = null;
        objectInspector = null;
    }
    if (ifStmtTestMode == IfStmtTestMode.VECTOR_EXPRESSION && vectorExpression instanceof VectorUDFAdaptor) {
        System.out.println("*NO NATIVE VECTOR EXPRESSION* typeInfo " + typeInfo.toString() + " ifStmtTestMode " + ifStmtTestMode + " ifVariation " + ifVariation + " columnScalarMode " + columnScalarMode + " vectorExpression " + vectorExpression.toString());
    }
    String[] outputScratchTypeNames = vectorizationContext.getScratchColumnTypeNames();
    DataTypePhysicalVariation[] outputDataTypePhysicalVariations = vectorizationContext.getScratchDataTypePhysicalVariations();
    VectorizedRowBatchCtx batchContext = new VectorizedRowBatchCtx(columnNames, typeInfos, dataTypePhysicalVariations, /* dataColumnNums */
    null, /* partitionColumnCount */
    0, /* virtualColumnCount */
    0, /* neededVirtualColumns */
    null, outputScratchTypeNames, outputDataTypePhysicalVariations);
    VectorizedRowBatch batch = batchContext.createVectorizedRowBatch();
    // System.out.println("*VECTOR EXPRESSION* " + vectorExpression.getClass().getSimpleName());
    /*
    System.out.println(
        "*DEBUG* typeInfo " + typeInfo.toString() +
        " ifStmtTestMode " + ifStmtTestMode +
        " ifVariation " + ifVariation +
        " columnScalarMode " + columnScalarMode +
        " vectorExpression " + vectorExpression.toString());
    */
    VectorExtractRow resultVectorExtractRow = null;
    Object[] scrqtchRow = null;
    if (!isFilter) {
        resultVectorExtractRow = new VectorExtractRow();
        final int outputColumnNum = vectorExpression.getOutputColumnNum();
        resultVectorExtractRow.init(new TypeInfo[] { outputTypeInfo }, new int[] { outputColumnNum });
        scrqtchRow = new Object[1];
    }
    boolean copySelectedInUse = false;
    int[] copySelected = new int[VectorizedRowBatch.DEFAULT_SIZE];
    batchSource.resetBatchIteration();
    int rowIndex = 0;
    while (true) {
        if (!batchSource.fillNextBatch(batch)) {
            break;
        }
        final int originalBatchSize = batch.size;
        if (isFilter) {
            copySelectedInUse = batch.selectedInUse;
            if (batch.selectedInUse) {
                System.arraycopy(batch.selected, 0, copySelected, 0, originalBatchSize);
            }
        }
        // In filter mode, the batch size can be made smaller.
        vectorExpression.evaluate(batch);
        if (!isFilter) {
            extractResultObjects(batch, rowIndex, resultVectorExtractRow, scrqtchRow, objectInspector, resultObjects);
        } else {
            final int currentBatchSize = batch.size;
            if (copySelectedInUse && batch.selectedInUse) {
                int selectIndex = 0;
                for (int i = 0; i < originalBatchSize; i++) {
                    final int originalBatchIndex = copySelected[i];
                    final boolean booleanResult;
                    if (selectIndex < currentBatchSize && batch.selected[selectIndex] == originalBatchIndex) {
                        booleanResult = true;
                        selectIndex++;
                    } else {
                        booleanResult = false;
                    }
                    resultObjects[rowIndex + i] = new BooleanWritable(booleanResult);
                }
            } else if (batch.selectedInUse) {
                int selectIndex = 0;
                for (int i = 0; i < originalBatchSize; i++) {
                    final boolean booleanResult;
                    if (selectIndex < currentBatchSize && batch.selected[selectIndex] == i) {
                        booleanResult = true;
                        selectIndex++;
                    } else {
                        booleanResult = false;
                    }
                    resultObjects[rowIndex + i] = new BooleanWritable(booleanResult);
                }
            } else if (currentBatchSize == 0) {
                // Whole batch got zapped.
                for (int i = 0; i < originalBatchSize; i++) {
                    resultObjects[rowIndex + i] = new BooleanWritable(false);
                }
            } else {
                // Every row kept.
                for (int i = 0; i < originalBatchSize; i++) {
                    resultObjects[rowIndex + i] = new BooleanWritable(true);
                }
            }
        }
        rowIndex += originalBatchSize;
    }
}

Example 2 with GenericUDFIf

use of org.apache.hadoop.hive.ql.udf.generic.GenericUDFIf in project hive by apache.

the class TestVectorizationContext method testIfConditionalExprs.

/**
 * Test that correct VectorExpression classes are chosen for the
 * IF (expr1, expr2, expr3) conditional expression for integer, float,
 * boolean, timestamp and string input types. expr1 is always an input column expression
 * of type long. expr2 and expr3 can be column expressions or constants of other types
 * but must have the same type.
 */
@Test
public void testIfConditionalExprs() throws HiveException {
    // Predicate.
    ExprNodeColumnDesc col1Expr = new ExprNodeColumnDesc(Boolean.class, "col1", "table", false);
    ExprNodeColumnDesc col2Expr = new ExprNodeColumnDesc(Long.class, "col2", "table", false);
    ExprNodeColumnDesc col3Expr = new ExprNodeColumnDesc(Long.class, "col3", "table", false);
    ExprNodeConstantDesc constDesc2 = new ExprNodeConstantDesc(Integer.valueOf(1));
    ExprNodeConstantDesc constDesc3 = new ExprNodeConstantDesc(Integer.valueOf(2));
    // long column/column IF
    GenericUDFIf udf = new GenericUDFIf();
    List<ExprNodeDesc> children1 = new ArrayList<ExprNodeDesc>();
    children1.add(col1Expr);
    children1.add(col2Expr);
    children1.add(col3Expr);
    ExprNodeGenericFuncDesc exprDesc = new ExprNodeGenericFuncDesc(TypeInfoFactory.booleanTypeInfo, udf, children1);
    List<String> columns = new ArrayList<String>();
    columns.add("col0");
    columns.add("col1");
    columns.add("col2");
    columns.add("col3");
    VectorizationContext vc = new VectorizationContext("name", columns);
    exprDesc.setTypeInfo(TypeInfoFactory.longTypeInfo);
    VectorExpression ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprLongColumnLongColumn);
    // long column/scalar IF
    children1.set(2, new ExprNodeConstantDesc(1L));
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprLongColumnLongScalar);
    // long scalar/scalar IF
    children1.set(1, new ExprNodeConstantDesc(1L));
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprLongScalarLongScalar);
    // long scalar/column IF
    children1.set(2, col3Expr);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprLongScalarLongColumn);
    // test for double type
    col2Expr = new ExprNodeColumnDesc(Double.class, "col2", "table", false);
    col3Expr = new ExprNodeColumnDesc(Double.class, "col3", "table", false);
    // double column/column IF
    children1.set(1, col2Expr);
    children1.set(2, col3Expr);
    exprDesc.setTypeInfo(TypeInfoFactory.doubleTypeInfo);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprDoubleColumnDoubleColumn);
    // double column/scalar IF
    children1.set(2, new ExprNodeConstantDesc(1D));
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprDoubleColumnDoubleScalar);
    // double scalar/scalar IF
    children1.set(1, new ExprNodeConstantDesc(1D));
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprDoubleScalarDoubleScalar);
    // double scalar/column IF
    children1.set(2, col3Expr);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprDoubleScalarDoubleColumn);
    // double scalar/long column IF
    children1.set(2, new ExprNodeColumnDesc(Long.class, "col3", "table", false));
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprColumnCondExpr);
    // Additional combinations of (long,double)X(column,scalar) for each of the second
    // and third arguments are omitted. We have coverage of all the source templates
    // already.
    // test for timestamp type
    col2Expr = new ExprNodeColumnDesc(Timestamp.class, "col2", "table", false);
    col3Expr = new ExprNodeColumnDesc(Timestamp.class, "col3", "table", false);
    // timestamp column/column IF
    children1.set(1, col2Expr);
    children1.set(2, col3Expr);
    exprDesc.setTypeInfo(TypeInfoFactory.timestampTypeInfo);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprTimestampColumnColumn);
    // timestamp column/scalar IF where scalar is really a CAST of a constant to timestamp.
    ExprNodeGenericFuncDesc f = new ExprNodeGenericFuncDesc();
    f.setGenericUDF(new GenericUDFTimestamp());
    f.setTypeInfo(TypeInfoFactory.timestampTypeInfo);
    List<ExprNodeDesc> children2 = new ArrayList<ExprNodeDesc>();
    f.setChildren(children2);
    children2.add(new ExprNodeConstantDesc("2013-11-05 00:00:00.000"));
    children1.set(2, f);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprTimestampColumnScalar);
    // timestamp scalar/scalar
    children1.set(1, f);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprTimestampScalarScalar);
    // timestamp scalar/column
    children1.set(2, col3Expr);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprTimestampScalarColumn);
    // test for boolean type
    col2Expr = new ExprNodeColumnDesc(Boolean.class, "col2", "table", false);
    col3Expr = new ExprNodeColumnDesc(Boolean.class, "col3", "table", false);
    // column/column
    children1.set(1, col2Expr);
    children1.set(2, col3Expr);
    exprDesc.setTypeInfo(TypeInfoFactory.booleanTypeInfo);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprLongColumnLongColumn);
    // column/scalar IF
    children1.set(2, new ExprNodeConstantDesc(true));
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprLongColumnLongScalar);
    // scalar/scalar IF
    children1.set(1, new ExprNodeConstantDesc(true));
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprLongScalarLongScalar);
    // scalar/column IF
    children1.set(2, col3Expr);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprLongScalarLongColumn);
    // test for string type
    constDesc2 = new ExprNodeConstantDesc("Alpha");
    constDesc3 = new ExprNodeConstantDesc("Bravo");
    col2Expr = new ExprNodeColumnDesc(String.class, "col2", "table", false);
    col3Expr = new ExprNodeColumnDesc(String.class, "col3", "table", false);
    // column/column
    children1.set(1, col2Expr);
    children1.set(2, col3Expr);
    exprDesc.setTypeInfo(TypeInfoFactory.stringTypeInfo);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprStringGroupColumnStringGroupColumn);
    // column/scalar
    children1.set(2, constDesc3);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprStringGroupColumnStringScalar);
    // scalar/scalar
    children1.set(1, constDesc2);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprStringScalarStringScalar);
    // scalar/column
    children1.set(2, col3Expr);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprStringScalarStringGroupColumn);
    // test for CHAR type
    CharTypeInfo charTypeInfo = new CharTypeInfo(10);
    constDesc2 = new ExprNodeConstantDesc(charTypeInfo, new HiveChar("Alpha", 10));
    constDesc3 = new ExprNodeConstantDesc(charTypeInfo, new HiveChar("Bravo", 10));
    col2Expr = new ExprNodeColumnDesc(charTypeInfo, "col2", "table", false);
    col3Expr = new ExprNodeColumnDesc(charTypeInfo, "col3", "table", false);
    // column/column
    children1.set(1, col2Expr);
    children1.set(2, col3Expr);
    ve = vc.getVectorExpression(exprDesc);
    exprDesc.setTypeInfo(charTypeInfo);
    assertTrue(ve instanceof IfExprCondExprCondExpr);
    // column/scalar
    children1.set(2, constDesc3);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprStringGroupColumnCharScalar);
    // scalar/scalar
    children1.set(1, constDesc2);
    // ve = vc.getVectorExpression(exprDesc);
    // assertTrue(ve instanceof IfExprCharScalarCharScalar);
    // scalar/column
    children1.set(2, col3Expr);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprCharScalarStringGroupColumn);
    // test for VARCHAR type
    VarcharTypeInfo varcharTypeInfo = new VarcharTypeInfo(10);
    constDesc2 = new ExprNodeConstantDesc(varcharTypeInfo, new HiveVarchar("Alpha", 10));
    constDesc3 = new ExprNodeConstantDesc(varcharTypeInfo, new HiveVarchar("Bravo", 10));
    col2Expr = new ExprNodeColumnDesc(varcharTypeInfo, "col2", "table", false);
    col3Expr = new ExprNodeColumnDesc(varcharTypeInfo, "col3", "table", false);
    // column/column
    children1.set(1, col2Expr);
    children1.set(2, col3Expr);
    exprDesc.setTypeInfo(varcharTypeInfo);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprStringGroupColumnStringGroupColumn);
    // column/scalar
    children1.set(2, constDesc3);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprStringGroupColumnVarCharScalar);
    // scalar/scalar
    children1.set(1, constDesc2);
    // ve = vc.getVectorExpression(exprDesc);
    // assertTrue(ve instanceof IfExprVarCharScalarVarCharScalar);
    // scalar/column
    children1.set(2, col3Expr);
    ve = vc.getVectorExpression(exprDesc);
    assertTrue(ve instanceof IfExprVarCharScalarStringGroupColumn);
}

Example 3 with GenericUDFIf

use of org.apache.hadoop.hive.ql.udf.generic.GenericUDFIf in project hive by apache.

the class TestVectorIfStatement method doRowIfTest.

private void doRowIfTest(TypeInfo typeInfo, List<String> columns, List<ExprNodeDesc> children, Object[][] randomRows, ObjectInspector rowInspector, Object[] resultObjects) throws Exception {
    GenericUDF udf = new GenericUDFIf();
    ExprNodeGenericFuncDesc exprDesc = new ExprNodeGenericFuncDesc(typeInfo, udf, children);
    HiveConf hiveConf = new HiveConf();
    ExprNodeEvaluator evaluator = ExprNodeEvaluatorFactory.get(exprDesc, hiveConf);
    evaluator.initialize(rowInspector);
    final int rowCount = randomRows.length;
    for (int i = 0; i < rowCount; i++) {
        Object[] row = randomRows[i];
        Object result = evaluator.evaluate(row);
        resultObjects[i] = result;
    }
}