Examples with TableFunctionEvaluator org.apache.hadoop.hive.ql.udf.ptf.TableFunctionEvaluator used on opensource projects

Example 1 with TableFunctionEvaluator

use of org.apache.hadoop.hive.ql.udf.ptf.TableFunctionEvaluator in project hive by apache.

the class PTFDeserializer method initializeWindowing.

public void initializeWindowing(WindowTableFunctionDef def) throws HiveException {
    ShapeDetails inpShape = def.getInput().getOutputShape();
    /*
     * 1. setup resolve, make connections
     */
    TableFunctionEvaluator tEval = def.getTFunction();
    WindowingTableFunctionResolver tResolver = (WindowingTableFunctionResolver) constructResolver(def.getResolverClassName());
    tResolver.initialize(hConf, ptfDesc, def, tEval);
    /*
     * 2. initialize WFns.
     */
    for (WindowFunctionDef wFnDef : def.getWindowFunctions()) {
        if (wFnDef.getArgs() != null) {
            for (PTFExpressionDef arg : wFnDef.getArgs()) {
                initialize(arg, inpShape);
            }
        }
        if (wFnDef.getWindowFrame() != null) {
            WindowFrameDef wFrmDef = wFnDef.getWindowFrame();
            initialize(wFrmDef, inpShape);
        }
        setupWdwFnEvaluator(wFnDef);
    }
    ArrayList<String> aliases = new ArrayList<String>();
    ArrayList<ObjectInspector> fieldOIs = new ArrayList<ObjectInspector>();
    for (WindowFunctionDef wFnDef : def.getWindowFunctions()) {
        aliases.add(wFnDef.getAlias());
        if (wFnDef.isPivotResult()) {
            fieldOIs.add(((ListObjectInspector) wFnDef.getOI()).getListElementObjectInspector());
        } else {
            fieldOIs.add(wFnDef.getOI());
        }
    }
    PTFDeserializer.addInputColumnsToList(inpShape, aliases, fieldOIs);
    StructObjectInspector wdwOutOI = ObjectInspectorFactory.getStandardStructObjectInspector(aliases, fieldOIs);
    tResolver.setWdwProcessingOutputOI(wdwOutOI);
    initialize(def.getOutputShape(), wdwOutOI);
    tResolver.initializeOutputOI();
}

Example 2 with TableFunctionEvaluator

use of org.apache.hadoop.hive.ql.udf.ptf.TableFunctionEvaluator in project hive by apache.

the class PTFDeserializer method initialize.

protected void initialize(PartitionedTableFunctionDef def) throws HiveException {
    ShapeDetails inpShape = def.getInput().getOutputShape();
    /*
     * 1. initialize args
     */
    if (def.getArgs() != null) {
        for (PTFExpressionDef arg : def.getArgs()) {
            initialize(arg, inpShape);
        }
    }
    /*
     * 2. setup resolve, make connections
     */
    TableFunctionEvaluator tEval = def.getTFunction();
    // TableFunctionResolver tResolver = FunctionRegistry.getTableFunctionResolver(def.getName());
    TableFunctionResolver tResolver = constructResolver(def.getResolverClassName());
    tResolver.initialize(hConf, ptfDesc, def, tEval);
    /*
     * 3. give Evaluator chance to setup for RawInput execution; setup RawInput shape
     */
    if (tEval.isTransformsRawInput()) {
        tResolver.initializeRawInputOI();
        initialize(def.getRawInputShape(), tEval.getRawInputOI());
    } else {
        def.setRawInputShape(inpShape);
    }
    inpShape = def.getRawInputShape();
    /*
     * 4. give Evaluator chance to setup for Output execution; setup Output shape.
     */
    tResolver.initializeOutputOI();
    initialize(def.getOutputShape(), tEval.getOutputOI());
}

Example 3 with TableFunctionEvaluator

use of org.apache.hadoop.hive.ql.udf.ptf.TableFunctionEvaluator in project hive by apache.

the class PTFDeserializer method alterOutputOIForStreaming.

/*
   * If the final PTF in a PTFChain can stream its output, then set the OI of its OutputShape
   * to the OI returned by the TableFunctionEvaluator.
   */
public static void alterOutputOIForStreaming(PTFDesc ptfDesc) {
    PartitionedTableFunctionDef tDef = ptfDesc.getFuncDef();
    TableFunctionEvaluator tEval = tDef.getTFunction();
    if (tEval.canIterateOutput()) {
        tDef.getOutputShape().setOI(tEval.getOutputOI());
    }
}

Example 4 with TableFunctionEvaluator

use of org.apache.hadoop.hive.ql.udf.ptf.TableFunctionEvaluator in project hive by apache.

the class PTFTranslator method translate.

private PartitionedTableFunctionDef translate(PartitionedTableFunctionSpec spec, PTFInputDef inpDef, int inpNum) throws SemanticException {
    TableFunctionResolver tFn = FunctionRegistry.getTableFunctionResolver(spec.getName());
    if (tFn == null) {
        throw new SemanticException(String.format("Unknown Table Function %s", spec.getName()));
    }
    PartitionedTableFunctionDef def = new PartitionedTableFunctionDef();
    def.setInput(inpDef);
    def.setName(spec.getName());
    def.setResolverClassName(tFn.getClass().getName());
    def.setAlias(spec.getAlias() == null ? "ptf_" + inpNum : spec.getAlias());
    def.setExpressionTreeString(spec.getAstNode().toStringTree());
    def.setTransformsRawInput(tFn.transformsRawInput());
    /*
     * translate args
     */
    List<ASTNode> args = spec.getArgs();
    if (args != null) {
        for (ASTNode expr : args) {
            PTFExpressionDef argDef = null;
            try {
                argDef = buildExpressionDef(inpDef.getOutputShape(), expr);
            } catch (HiveException he) {
                throw new SemanticException(he);
            }
            def.addArg(argDef);
        }
    }
    tFn.initialize(hCfg, ptfDesc, def);
    TableFunctionEvaluator tEval = tFn.getEvaluator();
    def.setTFunction(tEval);
    def.setCarryForwardNames(tFn.carryForwardNames());
    tFn.setupRawInputOI();
    if (tFn.transformsRawInput()) {
        StructObjectInspector rawInOutOI = tEval.getRawInputOI();
        List<String> rawInOutColNames = tFn.getRawInputColumnNames();
        RowResolver rawInRR = buildRowResolverForPTF(def.getName(), spec.getAlias(), rawInOutOI, rawInOutColNames, inpDef.getOutputShape().getRr());
        ShapeDetails rawInpShape = setupTableFnShape(def.getName(), inpDef.getOutputShape(), rawInOutOI, rawInOutColNames, rawInRR);
        def.setRawInputShape(rawInpShape);
    } else {
        def.setRawInputShape(inpDef.getOutputShape());
    }
    translatePartitioning(def, spec);
    tFn.setupOutputOI();
    StructObjectInspector outputOI = tEval.getOutputOI();
    List<String> outColNames = tFn.getOutputColumnNames();
    RowResolver outRR = buildRowResolverForPTF(def.getName(), spec.getAlias(), outputOI, outColNames, def.getRawInputShape().getRr());
    ShapeDetails outputShape = setupTableFnShape(def.getName(), inpDef.getOutputShape(), outputOI, outColNames, outRR);
    def.setOutputShape(outputShape);
    def.setReferencedColumns(tFn.getReferencedColumns());
    return def;
}

Example 5 with TableFunctionEvaluator

use of org.apache.hadoop.hive.ql.udf.ptf.TableFunctionEvaluator in project hive by apache.

the class PTFTranslator method translate.

public PTFDesc translate(WindowingSpec wdwSpec, SemanticAnalyzer semAly, HiveConf hCfg, RowResolver inputRR, UnparseTranslator unparseT) throws SemanticException {
    init(semAly, hCfg, inputRR, unparseT);
    windowingSpec = wdwSpec;
    ptfDesc = new PTFDesc();
    ptfDesc.setCfg(hCfg);
    ptfDesc.setLlInfo(llInfo);
    WindowTableFunctionDef wdwTFnDef = new WindowTableFunctionDef();
    ptfDesc.setFuncDef(wdwTFnDef);
    PTFQueryInputSpec inpSpec = new PTFQueryInputSpec();
    inpSpec.setType(PTFQueryInputType.WINDOWING);
    wdwTFnDef.setInput(translate(inpSpec, 0));
    ShapeDetails inpShape = wdwTFnDef.getInput().getOutputShape();
    WindowingTableFunctionResolver tFn = (WindowingTableFunctionResolver) FunctionRegistry.getTableFunctionResolver(FunctionRegistry.WINDOWING_TABLE_FUNCTION);
    if (tFn == null) {
        throw new SemanticException(String.format("Internal Error: Unknown Table Function %s", FunctionRegistry.WINDOWING_TABLE_FUNCTION));
    }
    wdwTFnDef.setName(FunctionRegistry.WINDOWING_TABLE_FUNCTION);
    wdwTFnDef.setResolverClassName(tFn.getClass().getName());
    wdwTFnDef.setAlias("ptf_" + 1);
    wdwTFnDef.setExpressionTreeString(null);
    wdwTFnDef.setTransformsRawInput(false);
    tFn.initialize(hCfg, ptfDesc, wdwTFnDef);
    TableFunctionEvaluator tEval = tFn.getEvaluator();
    wdwTFnDef.setTFunction(tEval);
    wdwTFnDef.setCarryForwardNames(tFn.carryForwardNames());
    wdwTFnDef.setRawInputShape(inpShape);
    PartitioningSpec partiSpec = wdwSpec.getQueryPartitioningSpec();
    if (partiSpec == null) {
        throw new SemanticException("Invalid use of Windowing: there is no Partitioning associated with Windowing");
    }
    PartitionDef partDef = translate(inpShape, wdwSpec.getQueryPartitionSpec());
    OrderDef ordDef = translate(inpShape, wdwSpec.getQueryOrderSpec(), partDef);
    wdwTFnDef.setPartition(partDef);
    wdwTFnDef.setOrder(ordDef);
    /*
     * process Wdw functions
     */
    ArrayList<WindowFunctionDef> windowFunctions = new ArrayList<WindowFunctionDef>();
    if (wdwSpec.getWindowExpressions() != null) {
        for (WindowExpressionSpec expr : wdwSpec.getWindowExpressions()) {
            if (expr instanceof WindowFunctionSpec) {
                WindowFunctionDef wFnDef = translate(wdwTFnDef, (WindowFunctionSpec) expr);
                windowFunctions.add(wFnDef);
            }
        }
        wdwTFnDef.setWindowFunctions(windowFunctions);
    }
    /*
     * set outputFromWdwFnProcessing
     */
    ArrayList<String> aliases = new ArrayList<String>();
    ArrayList<ObjectInspector> fieldOIs = new ArrayList<ObjectInspector>();
    for (WindowFunctionDef wFnDef : windowFunctions) {
        aliases.add(wFnDef.getAlias());
        if (wFnDef.isPivotResult()) {
            fieldOIs.add(((ListObjectInspector) wFnDef.getOI()).getListElementObjectInspector());
        } else {
            fieldOIs.add(wFnDef.getOI());
        }
    }
    PTFTranslator.addInputColumnsToList(inpShape, aliases, fieldOIs);
    StructObjectInspector wdwOutOI = ObjectInspectorFactory.getStandardStructObjectInspector(aliases, fieldOIs);
    tFn.setWdwProcessingOutputOI(wdwOutOI);
    RowResolver wdwOutRR = buildRowResolverForWindowing(wdwTFnDef);
    ShapeDetails wdwOutShape = setupShape(wdwOutOI, null, wdwOutRR);
    wdwTFnDef.setOutputShape(wdwOutShape);
    tFn.setupOutputOI();
    PTFDeserializer.alterOutputOIForStreaming(ptfDesc);
    return ptfDesc;
}