Examples with ICompilerFactory org.apache.hyracks.algebricks.compiler.api.ICompilerFactory used on opensource projects

Example 1 with ICompilerFactory

use of org.apache.hyracks.algebricks.compiler.api.ICompilerFactory in project asterixdb by apache.

the class APIFramework method compileQuery.

public JobSpecification compileQuery(IClusterInfoCollector clusterInfoCollector, MetadataProvider metadataProvider, Query rwQ, int varCounter, String outputDatasetName, SessionOutput output, ICompiledDmlStatement statement) throws AlgebricksException, RemoteException, ACIDException {
    SessionConfig conf = output.config();
    if (!conf.is(SessionConfig.FORMAT_ONLY_PHYSICAL_OPS) && conf.is(SessionConfig.OOB_REWRITTEN_EXPR_TREE)) {
        output.out().println();
        printPlanPrefix(output, "Rewritten expression tree");
        if (rwQ != null) {
            rwQ.accept(astPrintVisitorFactory.createLangVisitor(output.out()), 0);
        }
        printPlanPostfix(output);
    }
    org.apache.asterix.common.transactions.JobId asterixJobId = JobIdFactory.generateJobId();
    metadataProvider.setJobId(asterixJobId);
    ILangExpressionToPlanTranslator t = translatorFactory.createExpressionToPlanTranslator(metadataProvider, varCounter);
    ILogicalPlan plan;
    // statement = null when it's a query
    if (statement == null || statement.getKind() != Statement.Kind.LOAD) {
        plan = t.translate(rwQ, outputDatasetName, statement);
    } else {
        plan = t.translateLoad(statement);
    }
    if (!conf.is(SessionConfig.FORMAT_ONLY_PHYSICAL_OPS) && conf.is(SessionConfig.OOB_LOGICAL_PLAN)) {
        output.out().println();
        printPlanPrefix(output, "Logical plan");
        if (rwQ != null || (statement != null && statement.getKind() == Statement.Kind.LOAD)) {
            LogicalOperatorPrettyPrintVisitor pvisitor = new LogicalOperatorPrettyPrintVisitor(output.out());
            PlanPrettyPrinter.printPlan(plan, pvisitor, 0);
        }
        printPlanPostfix(output);
    }
    CompilerProperties compilerProperties = metadataProvider.getApplicationContext().getCompilerProperties();
    int frameSize = compilerProperties.getFrameSize();
    Map<String, String> querySpecificConfig = metadataProvider.getConfig();
    // Validates the user-overridden query parameters.
    validateConfig(querySpecificConfig);
    int sortFrameLimit = getFrameLimit(CompilerProperties.COMPILER_SORTMEMORY_KEY, querySpecificConfig.get(CompilerProperties.COMPILER_SORTMEMORY_KEY), compilerProperties.getSortMemorySize(), frameSize, MIN_FRAME_LIMIT_FOR_SORT);
    int groupFrameLimit = getFrameLimit(CompilerProperties.COMPILER_GROUPMEMORY_KEY, querySpecificConfig.get(CompilerProperties.COMPILER_GROUPMEMORY_KEY), compilerProperties.getGroupMemorySize(), frameSize, MIN_FRAME_LIMIT_FOR_GROUP_BY);
    int joinFrameLimit = getFrameLimit(CompilerProperties.COMPILER_JOINMEMORY_KEY, querySpecificConfig.get(CompilerProperties.COMPILER_JOINMEMORY_KEY), compilerProperties.getJoinMemorySize(), frameSize, MIN_FRAME_LIMIT_FOR_JOIN);
    OptimizationConfUtil.getPhysicalOptimizationConfig().setFrameSize(frameSize);
    OptimizationConfUtil.getPhysicalOptimizationConfig().setMaxFramesExternalSort(sortFrameLimit);
    OptimizationConfUtil.getPhysicalOptimizationConfig().setMaxFramesExternalGroupBy(groupFrameLimit);
    OptimizationConfUtil.getPhysicalOptimizationConfig().setMaxFramesForJoin(joinFrameLimit);
    HeuristicCompilerFactoryBuilder builder = new HeuristicCompilerFactoryBuilder(OptimizationContextFactory.INSTANCE);
    builder.setPhysicalOptimizationConfig(OptimizationConfUtil.getPhysicalOptimizationConfig());
    builder.setLogicalRewrites(ruleSetFactory.getLogicalRewrites(metadataProvider.getApplicationContext()));
    builder.setPhysicalRewrites(ruleSetFactory.getPhysicalRewrites(metadataProvider.getApplicationContext()));
    IDataFormat format = metadataProvider.getFormat();
    ICompilerFactory compilerFactory = builder.create();
    builder.setExpressionEvalSizeComputer(format.getExpressionEvalSizeComputer());
    builder.setIMergeAggregationExpressionFactory(new MergeAggregationExpressionFactory());
    builder.setPartialAggregationTypeComputer(new PartialAggregationTypeComputer());
    builder.setExpressionTypeComputer(ExpressionTypeComputer.INSTANCE);
    builder.setMissableTypeComputer(MissableTypeComputer.INSTANCE);
    builder.setConflictingTypeResolver(ConflictingTypeResolver.INSTANCE);
    int parallelism = getParallelism(querySpecificConfig.get(CompilerProperties.COMPILER_PARALLELISM_KEY), compilerProperties.getParallelism());
    AlgebricksAbsolutePartitionConstraint computationLocations = chooseLocations(clusterInfoCollector, parallelism, metadataProvider.getClusterLocations());
    builder.setClusterLocations(computationLocations);
    ICompiler compiler = compilerFactory.createCompiler(plan, metadataProvider, t.getVarCounter());
    if (conf.isOptimize()) {
        compiler.optimize();
        if (conf.is(SessionConfig.OOB_OPTIMIZED_LOGICAL_PLAN)) {
            if (conf.is(SessionConfig.FORMAT_ONLY_PHYSICAL_OPS)) {
                // For Optimizer tests.
                AlgebricksAppendable buffer = new AlgebricksAppendable(output.out());
                PlanPrettyPrinter.printPhysicalOps(plan, buffer, 0);
            } else {
                printPlanPrefix(output, "Optimized logical plan");
                if (rwQ != null || (statement != null && statement.getKind() == Statement.Kind.LOAD)) {
                    LogicalOperatorPrettyPrintVisitor pvisitor = new LogicalOperatorPrettyPrintVisitor(output.out());
                    PlanPrettyPrinter.printPlan(plan, pvisitor, 0);
                }
                printPlanPostfix(output);
            }
        }
    }
    if (rwQ != null && rwQ.isExplain()) {
        try {
            LogicalOperatorPrettyPrintVisitor pvisitor = new LogicalOperatorPrettyPrintVisitor();
            PlanPrettyPrinter.printPlan(plan, pvisitor, 0);
            ResultUtil.printResults(metadataProvider.getApplicationContext(), pvisitor.get().toString(), output, new Stats(), null);
            return null;
        } catch (IOException e) {
            throw new AlgebricksException(e);
        }
    }
    if (!conf.isGenerateJobSpec()) {
        return null;
    }
    builder.setBinaryBooleanInspectorFactory(format.getBinaryBooleanInspectorFactory());
    builder.setBinaryIntegerInspectorFactory(format.getBinaryIntegerInspectorFactory());
    builder.setComparatorFactoryProvider(format.getBinaryComparatorFactoryProvider());
    builder.setExpressionRuntimeProvider(new ExpressionRuntimeProvider(QueryLogicalExpressionJobGen.INSTANCE));
    builder.setHashFunctionFactoryProvider(format.getBinaryHashFunctionFactoryProvider());
    builder.setHashFunctionFamilyProvider(format.getBinaryHashFunctionFamilyProvider());
    builder.setMissingWriterFactory(format.getMissingWriterFactory());
    builder.setPredicateEvaluatorFactoryProvider(format.getPredicateEvaluatorFactoryProvider());
    final SessionConfig.OutputFormat outputFormat = conf.fmt();
    switch(outputFormat) {
        case LOSSLESS_JSON:
            builder.setPrinterProvider(format.getLosslessJSONPrinterFactoryProvider());
            break;
        case CSV:
            builder.setPrinterProvider(format.getCSVPrinterFactoryProvider());
            break;
        case ADM:
            builder.setPrinterProvider(format.getADMPrinterFactoryProvider());
            break;
        case CLEAN_JSON:
            builder.setPrinterProvider(format.getCleanJSONPrinterFactoryProvider());
            break;
        default:
            throw new AlgebricksException("Unexpected OutputFormat: " + outputFormat);
    }
    builder.setSerializerDeserializerProvider(format.getSerdeProvider());
    builder.setTypeTraitProvider(format.getTypeTraitProvider());
    builder.setNormalizedKeyComputerFactoryProvider(format.getNormalizedKeyComputerFactoryProvider());
    JobEventListenerFactory jobEventListenerFactory = new JobEventListenerFactory(asterixJobId, metadataProvider.isWriteTransaction());
    JobSpecification spec = compiler.createJob(metadataProvider.getApplicationContext(), jobEventListenerFactory);
    // When the top-level statement is a query, the statement parameter is null.
    if (statement == null) {
        // Sets a required capacity, only for read-only queries.
        // DDLs and DMLs are considered not that frequent.
        spec.setRequiredClusterCapacity(ResourceUtils.getRequiredCompacity(plan, computationLocations, sortFrameLimit, groupFrameLimit, joinFrameLimit, frameSize));
    }
    if (conf.is(SessionConfig.OOB_HYRACKS_JOB)) {
        printPlanPrefix(output, "Hyracks job");
        if (rwQ != null) {
            try {
                output.out().println(new ObjectMapper().writerWithDefaultPrettyPrinter().writeValueAsString(spec.toJSON()));
            } catch (IOException e) {
                throw new AlgebricksException(e);
            }
            output.out().println(spec.getUserConstraints());
        }
        printPlanPostfix(output);
    }
    return spec;
}