Search in sources :

Example 1 with GenericUDFOPMod

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

the class TestVectorizationContext method testArithmeticExpressionVectorization.

@Test
public void testArithmeticExpressionVectorization() throws HiveException {
    /**
     * Create original expression tree for following
     * (plus (minus (plus col1 col2) col3) (multiply col4 (mod col5 col6)) )
     */
    GenericUDFOPPlus udf1 = new GenericUDFOPPlus();
    GenericUDFOPMinus udf2 = new GenericUDFOPMinus();
    GenericUDFOPMultiply udf3 = new GenericUDFOPMultiply();
    GenericUDFOPPlus udf4 = new GenericUDFOPPlus();
    GenericUDFOPMod udf5 = new GenericUDFOPMod();
    ExprNodeGenericFuncDesc sumExpr = new ExprNodeGenericFuncDesc();
    sumExpr.setTypeInfo(TypeInfoFactory.intTypeInfo);
    sumExpr.setGenericUDF(udf1);
    ExprNodeGenericFuncDesc minusExpr = new ExprNodeGenericFuncDesc();
    minusExpr.setTypeInfo(TypeInfoFactory.intTypeInfo);
    minusExpr.setGenericUDF(udf2);
    ExprNodeGenericFuncDesc multiplyExpr = new ExprNodeGenericFuncDesc();
    multiplyExpr.setTypeInfo(TypeInfoFactory.intTypeInfo);
    multiplyExpr.setGenericUDF(udf3);
    ExprNodeGenericFuncDesc sum2Expr = new ExprNodeGenericFuncDesc();
    sum2Expr.setTypeInfo(TypeInfoFactory.intTypeInfo);
    sum2Expr.setGenericUDF(udf4);
    ExprNodeGenericFuncDesc modExpr = new ExprNodeGenericFuncDesc();
    modExpr.setTypeInfo(TypeInfoFactory.intTypeInfo);
    modExpr.setGenericUDF(udf5);
    ExprNodeColumnDesc col1Expr = new ExprNodeColumnDesc(Long.class, "col1", "table", false);
    ExprNodeColumnDesc col2Expr = new ExprNodeColumnDesc(Long.class, "col2", "table", false);
    ExprNodeColumnDesc col3Expr = new ExprNodeColumnDesc(Long.class, "col3", "table", false);
    ExprNodeColumnDesc col4Expr = new ExprNodeColumnDesc(Long.class, "col4", "table", false);
    ExprNodeColumnDesc col5Expr = new ExprNodeColumnDesc(Long.class, "col5", "table", false);
    ExprNodeColumnDesc col6Expr = new ExprNodeColumnDesc(Long.class, "col6", "table", false);
    List<ExprNodeDesc> children1 = new ArrayList<ExprNodeDesc>(2);
    List<ExprNodeDesc> children2 = new ArrayList<ExprNodeDesc>(2);
    List<ExprNodeDesc> children3 = new ArrayList<ExprNodeDesc>(2);
    List<ExprNodeDesc> children4 = new ArrayList<ExprNodeDesc>(2);
    List<ExprNodeDesc> children5 = new ArrayList<ExprNodeDesc>(2);
    children1.add(minusExpr);
    children1.add(multiplyExpr);
    sumExpr.setChildren(children1);
    children2.add(sum2Expr);
    children2.add(col3Expr);
    minusExpr.setChildren(children2);
    children3.add(col1Expr);
    children3.add(col2Expr);
    sum2Expr.setChildren(children3);
    children4.add(col4Expr);
    children4.add(modExpr);
    multiplyExpr.setChildren(children4);
    children5.add(col5Expr);
    children5.add(col6Expr);
    modExpr.setChildren(children5);
    VectorizationContext vc = new VectorizationContext("name");
    vc.addInitialColumn("col1");
    vc.addInitialColumn("col2");
    vc.addInitialColumn("col3");
    vc.addInitialColumn("col4");
    vc.addInitialColumn("col5");
    vc.addInitialColumn("col6");
    vc.finishedAddingInitialColumns();
    //Generate vectorized expression
    VectorExpression ve = vc.getVectorExpression(sumExpr, VectorExpressionDescriptor.Mode.PROJECTION);
    //Verify vectorized expression
    assertTrue(ve instanceof LongColAddLongColumn);
    assertEquals(2, ve.getChildExpressions().length);
    VectorExpression childExpr1 = ve.getChildExpressions()[0];
    VectorExpression childExpr2 = ve.getChildExpressions()[1];
    System.out.println(ve.toString());
    assertEquals(6, ve.getOutputColumn());
    assertTrue(childExpr1 instanceof LongColSubtractLongColumn);
    assertEquals(1, childExpr1.getChildExpressions().length);
    assertTrue(childExpr1.getChildExpressions()[0] instanceof LongColAddLongColumn);
    assertEquals(7, childExpr1.getOutputColumn());
    assertEquals(6, childExpr1.getChildExpressions()[0].getOutputColumn());
    assertTrue(childExpr2 instanceof LongColMultiplyLongColumn);
    assertEquals(1, childExpr2.getChildExpressions().length);
    assertTrue(childExpr2.getChildExpressions()[0] instanceof LongColModuloLongColumn);
    assertEquals(8, childExpr2.getOutputColumn());
    assertEquals(6, childExpr2.getChildExpressions()[0].getOutputColumn());
}
Also used : GenericUDFOPMultiply(org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPMultiply) GenericUDFOPMod(org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPMod) ArrayList(java.util.ArrayList) ExprNodeGenericFuncDesc(org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc) LongColSubtractLongColumn(org.apache.hadoop.hive.ql.exec.vector.expressions.gen.LongColSubtractLongColumn) GenericUDFOPPlus(org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPPlus) GenericUDFOPMinus(org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPMinus) ExprNodeColumnDesc(org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc) VectorExpression(org.apache.hadoop.hive.ql.exec.vector.expressions.VectorExpression) DynamicValueVectorExpression(org.apache.hadoop.hive.ql.exec.vector.expressions.DynamicValueVectorExpression) ExprNodeDesc(org.apache.hadoop.hive.ql.plan.ExprNodeDesc) LongColAddLongColumn(org.apache.hadoop.hive.ql.exec.vector.expressions.gen.LongColAddLongColumn) LongColModuloLongColumn(org.apache.hadoop.hive.ql.exec.vector.expressions.gen.LongColModuloLongColumn) LongColMultiplyLongColumn(org.apache.hadoop.hive.ql.exec.vector.expressions.gen.LongColMultiplyLongColumn) Test(org.junit.Test)

Aggregations

ArrayList (java.util.ArrayList)1 DynamicValueVectorExpression (org.apache.hadoop.hive.ql.exec.vector.expressions.DynamicValueVectorExpression)1 VectorExpression (org.apache.hadoop.hive.ql.exec.vector.expressions.VectorExpression)1 LongColAddLongColumn (org.apache.hadoop.hive.ql.exec.vector.expressions.gen.LongColAddLongColumn)1 LongColModuloLongColumn (org.apache.hadoop.hive.ql.exec.vector.expressions.gen.LongColModuloLongColumn)1 LongColMultiplyLongColumn (org.apache.hadoop.hive.ql.exec.vector.expressions.gen.LongColMultiplyLongColumn)1 LongColSubtractLongColumn (org.apache.hadoop.hive.ql.exec.vector.expressions.gen.LongColSubtractLongColumn)1 ExprNodeColumnDesc (org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc)1 ExprNodeDesc (org.apache.hadoop.hive.ql.plan.ExprNodeDesc)1 ExprNodeGenericFuncDesc (org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc)1 GenericUDFOPMinus (org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPMinus)1 GenericUDFOPMod (org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPMod)1 GenericUDFOPMultiply (org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPMultiply)1 GenericUDFOPPlus (org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPPlus)1 Test (org.junit.Test)1