Examples of org.apache.uima.collection.metadata.CpeDescription

Package org.apache.uima.collection.metadata

Examples of org.apache.uima.collection.metadata.CpeDescription

org.apache.uima.collection.metadata.CpeDescription
Java representation of a Collection Processing Engine (CPE) XML descriptor. Generate an instance of this class by calling either the {@link org.apache.uima.util.XMLParser#parseCpeDescription(XMLInputSource)} or{@link org.apache.uima.collection.impl.metadata.cpe.CpeDescriptorFactory#produceDescriptor()}. A CPE instance can then be created by calling {@link org.apache.uima.UIMAFramework#produceCollectionProcessingEngine(CpeDescription)}.

   * 
   * @return CollectionProcessingEngine - initialized cpe
   */
  private CollectionProcessingEngine setupCpm(int documentCount, int threadCount,
          boolean useSlowAnnotator) throws Exception {
    CpeDescription cpeDesc = null;
    CollectionProcessingEngine cpe = null;


    String colReaderBase = JUnitExtension.getFile("CpmTests" + separator
            + "ErrorTestCollectionReader.xml").getAbsolutePath();
    String taeBase = JUnitExtension.getFile("CpmTests" + separator + "ErrorTestAnnotator.xml").getAbsolutePath();
    String casConsumerBase = JUnitExtension.getFile("CpmTests" + separator
            + "ErrorTestCasConsumer.xml").getAbsolutePath();


    // created needed descriptors
    String colReaderDesc = DescriptorMakeUtil.makeCollectionReader(colReaderBase, documentCount);
    String taeDesc = DescriptorMakeUtil.makeAnalysisEngine(taeBase);
    String casConsumerDesc = DescriptorMakeUtil.makeCasConsumer(casConsumerBase);


    // create cpm descriptor
    cpeDesc = CpeDescriptorFactory.produceDescriptor();
    cpeDesc.setInputQueueSize(2);
    cpeDesc.setOutputQueueSize(2);
    cpeDesc.setProcessingUnitThreadCount(threadCount);


    // add tae
    CpeIntegratedCasProcessor integratedProcessor = CpeDescriptorFactory
            .produceCasProcessor("ErrorTestAnnotator");
    integratedProcessor.setDescriptor(taeDesc);
    cpeDesc.addCasProcessor(integratedProcessor);


    // add slow annotator if requested
    if (useSlowAnnotator) {
      CpeIntegratedCasProcessor slowProcessor = CpeDescriptorFactory
              .produceCasProcessor("SlowAnnotator");
      slowProcessor.setDescriptor(JUnitExtension.getFile("CpmTests" + separator + "SlowAnnotator.xml").getAbsolutePath());
      cpeDesc.addCasProcessor(slowProcessor);
    }


    // add casConsumer
    CpeIntegratedCasProcessor casConsumer = CpeDescriptorFactory
            .produceCasProcessor("ErrorTest CasConsumer");
    casConsumer.setDescriptor(casConsumerDesc);
    cpeDesc.addCasProcessor(casConsumer);


    // add collectionReader
    cpeDesc.addCollectionReader(colReaderDesc);


    // produce cpe
    cpe = UIMAFramework.produceCollectionProcessingEngine(cpeDesc, null, null);


    return cpe;

View Full Code Here

   *          how many threads are used by the cpm
   * 
   * @return CollectionProcessingEngine - initialized cpe
   */
  private CollectionProcessingEngine setupCpm(int documentCount, int threadCount) throws Exception {
    CpeDescription cpeDesc = null;
    CollectionProcessingEngine cpe = null;


    try {
      String colReaderBase = JUnitExtension.getFile("CpmTests" + separator
              + "ErrorTestCollectionReader.xml").getAbsolutePath();
      String taeBase = JUnitExtension.getFile("CpmTests" + separator + "ErrorTestAnnotator.xml").getAbsolutePath();
      String casConsumerBase = JUnitExtension.getFile("CpmTests" + separator
              + "ErrorTestCasConsumer.xml").getAbsolutePath();


      // created needed descriptors
      String colReaderDesc = DescriptorMakeUtil.makeCollectionReader(colReaderBase, documentCount);
      String taeDesc = DescriptorMakeUtil.makeAnalysisEngine(taeBase);
      String casConsumerDesc = DescriptorMakeUtil.makeCasConsumer(casConsumerBase);


      // create cpm descriptor
      cpeDesc = CpeDescriptorFactory.produceDescriptor();
      cpeDesc.setInputQueueSize(2);
      cpeDesc.setOutputQueueSize(2);
      cpeDesc.setProcessingUnitThreadCount(threadCount);


      // add tae
      CpeIntegratedCasProcessor integratedProcessor = CpeDescriptorFactory
              .produceCasProcessor("ErrorTestAnnotator");
      integratedProcessor.setDescriptor(taeDesc);
      cpeDesc.addCasProcessor(integratedProcessor);


      // add casConsumer
      CpeIntegratedCasProcessor casConsumer = CpeDescriptorFactory
              .produceCasProcessor("ErrorTest CasConsumer");
      casConsumer.setDescriptor(casConsumerDesc);
      cpeDesc.addCasProcessor(casConsumer);


      // add collectionReader
      cpeDesc.addCollectionReader(colReaderDesc);


      // produce cpe
      cpe = UIMAFramework.produceCollectionProcessingEngine(cpeDesc, null, null);
    } catch (Exception e) {
      e.printStackTrace();

View Full Code Here


  public void equal(File aReferenceDescriptorFile, CpeDescription aGeneratedDescriptor) {
    try {
      
      XMLInputSource in = new XMLInputSource(aReferenceDescriptorFile);
      CpeDescription referenceDesc = UIMAFramework.getXMLParser().parseCpeDescription(in);


      // First Check Collection Reader
      CpeCollectionReader[] readers = referenceDesc.getAllCollectionCollectionReaders();
      if (readers[0] != null) {
        if (aGeneratedDescriptor.getAllCollectionCollectionReaders() == null
                || aGeneratedDescriptor.getAllCollectionCollectionReaders()[0] == null) {


          fail("<collectionReader> element not found in generated CPE Descriptor");
        } else {
          // Collection Reader element is present in the generated descriptor
          if (readers[0].getCollectionIterator() != null) {
            assertTrue(isCollectionIteratorTheSame(readers[0].getCollectionIterator(),
                    aGeneratedDescriptor.getAllCollectionCollectionReaders()[0]
                            .getCollectionIterator()));
          }


          if (readers[0].getCasInitializer() != null) {
            assertTrue(isCasInitializerTheSame(readers[0].getCasInitializer(), aGeneratedDescriptor
                    .getAllCollectionCollectionReaders()[0].getCasInitializer()));


          }
        }
      }
      CpeConfiguration refCpeConfig = referenceDesc.getCpeConfiguration();
      if (refCpeConfig != null) {
        if (aGeneratedDescriptor.getCpeConfiguration() == null) {
          fail("<cpeConfig> element not found in generated CPE Descriptor");
        } else {
          isConfigTheSame(refCpeConfig, aGeneratedDescriptor.getCpeConfiguration());


        }
      }
      CpeCasProcessors refCasProcessors = referenceDesc.getCpeCasProcessors();
      if (refCasProcessors != null) {
        if (aGeneratedDescriptor.getCpeCasProcessors() == null) {
          fail("<casProcessors> element not found in generated CPE Descriptor");
        } else {
          compareCasProcessors(referenceDesc.getCpeCasProcessors(), aGeneratedDescriptor
                  .getCpeCasProcessors());
        }
      }
    } catch (Exception e) {
      e.printStackTrace();

View Full Code Here


      // Enable schema validation (omit this to speed up initialization)
      // UIMAFramework.getXMLParser().enableSchemaValidation(true);


      // build a Collection Processing Engine descriptor that will drive processing
      CpeDescription cpeDesc = CpeDescriptorFactory.produceDescriptor();


      // add collection reader that will read input docs
      cpeDesc.addCollectionReader(FileSystemCollectionReader.getDescriptorURL().toString());
      // specify configuration parameters for collection reader
      CasProcessorConfigurationParameterSettings crSettings = CpeDescriptorFactory
              .produceCasProcessorConfigurationParameterSettings();
      CpeCollectionReader cpeCollRdr = cpeDesc.getAllCollectionCollectionReaders()[0];
      cpeCollRdr.setConfigurationParameterSettings(crSettings);
      crSettings.setParameterValue(FileSystemCollectionReader.PARAM_INPUTDIR, inputDir
              .getAbsolutePath());
      crSettings.setParameterValue(FileSystemCollectionReader.PARAM_ENCODING, encoding);
      crSettings.setParameterValue(FileSystemCollectionReader.PARAM_LANGUAGE, language);
      if (xcasInput) {
        crSettings.setParameterValue(FileSystemCollectionReader.PARAM_XCAS, "XCAS");
      } else if (xmiInput) {
        crSettings.setParameterValue(FileSystemCollectionReader.PARAM_XCAS, "XMI");
      }
      if (xLenient) {
        crSettings.setParameterValue(FileSystemCollectionReader.PARAM_LENIENT, "true");
      }


      // if XML tag was specified, configure XmlDetagger annotator and add to CPE
      CpeCasProcessor xmlDetaggerCasProc = null;
      if (xmlTagName != null && xmlTagName.length() > 0) {
        xmlDetaggerCasProc = CpeDescriptorFactory.produceCasProcessor("XmlDetagger");
        CpeComponentDescriptor cpeComponentDescriptor = 
          CpeDescriptorFactory.produceComponentDescriptor(XmlDetagger.getDescriptorURL().toString());
        xmlDetaggerCasProc.setCpeComponentDescriptor(cpeComponentDescriptor);
        CasProcessorConfigurationParameterSettings detaggerSettings = CpeDescriptorFactory
                .produceCasProcessorConfigurationParameterSettings();
        xmlDetaggerCasProc.setConfigurationParameterSettings(detaggerSettings);
        detaggerSettings.setParameterValue(XmlDetagger.PARAM_TEXT_TAG, xmlTagName);
        xmlDetaggerCasProc.setMaxErrorCount(0);
        cpeDesc.addCasProcessor(xmlDetaggerCasProc);
      }


      // add user's AE to CPE
      CpeCasProcessor casProc = CpeDescriptorFactory.produceCasProcessor("UserAE");
      CpeComponentDescriptor cpeComponentDescriptor = 
        CpeDescriptorFactory.produceComponentDescriptor(aeSpecifierFile.getAbsolutePath());
      casProc.setCpeComponentDescriptor(cpeComponentDescriptor);
      casProc.setMaxErrorCount(0);
      cpeDesc.addCasProcessor(casProc);


      // add CAS Consumer that will write the output
      // create and configure CAS consumer that will write the output
      CpeCasProcessor casCon = null;
      if (outputDir != null) {
        casCon = CpeDescriptorFactory.produceCasProcessor("CasConsumer");
        cpeComponentDescriptor = 
          CpeDescriptorFactory.produceComponentDescriptor(InlineXmlCasConsumer.getDescriptorURL().toString());
        casCon.setCpeComponentDescriptor(cpeComponentDescriptor);
        CasProcessorConfigurationParameterSettings consumerSettings = CpeDescriptorFactory
                .produceCasProcessorConfigurationParameterSettings();
        casCon.setConfigurationParameterSettings(consumerSettings);
        consumerSettings.setParameterValue(InlineXmlCasConsumer.PARAM_OUTPUTDIR, outputDir
                .getAbsolutePath());
        if (xcasInput) {
          consumerSettings.setParameterValue(InlineXmlCasConsumer.PARAM_XCAS, "XCAS");
        } else if (xmiInput) {
          consumerSettings.setParameterValue(InlineXmlCasConsumer.PARAM_XCAS, "XMI");
        }
        casCon.setMaxErrorCount(0);
        cpeDesc.addCasProcessor(casCon);
      }


      // if XML detagger is used, we need to configure sofa mappings for the CPE
      if (xmlDetaggerCasProc != null) {
        // For XML detagger map default sofa to "xmlDocument"

View Full Code Here

   * 
   * @return CollectionProcessingEngine - initialized cpe
   */
  private CollectionProcessingEngine setupCpm(int documentCount, int threadCount, int poolSize,
      ResourceManager resourceManager) throws Exception {
    CpeDescription cpeDesc = null;
    CollectionProcessingEngine cpe = null;


    try {
      String colReaderBase = JUnitExtension.getFile("CpmTests" + separator
              + "ErrorTestCollectionReader.xml").getAbsolutePath();
      String taeBase = JUnitExtension.getFile("CpmTests" + separator + "aggrContainingPearSpecifier.xml").getAbsolutePath();
      String casConsumerBase = JUnitExtension.getFile("CpmTests" + separator
              + "ErrorTestCasConsumer.xml").getAbsolutePath();


      // created needed descriptors
      String colReaderDesc = DescriptorMakeUtil.makeCollectionReader(colReaderBase, documentCount);
//      String taeDesc = DescriptorMakeUtil.makeAnalysisEngine(taeBase);
      String taeDesc = taeBase;
      String casConsumerDesc = DescriptorMakeUtil.makeCasConsumer(casConsumerBase);


      // create cpm descriptor
      cpeDesc = CpeDescriptorFactory.produceDescriptor();
      cpeDesc.setInputQueueSize(threadCount);
      cpeDesc.setOutputQueueSize(threadCount);
      cpeDesc.setProcessingUnitThreadCount(threadCount);
      cpeDesc.getCpeCasProcessors().setPoolSize(poolSize);


      // add tae
      CpeIntegratedCasProcessor integratedProcessor = CpeDescriptorFactory
              .produceCasProcessor("ErrorTestAnnotator");
      integratedProcessor.setDescriptor(taeDesc);
      cpeDesc.addCasProcessor(integratedProcessor);


      // add casConsumer
      CpeIntegratedCasProcessor casConsumer = CpeDescriptorFactory
              .produceCasProcessor("ErrorTest CasConsumer");
      casConsumer.setDescriptor(casConsumerDesc);
      cpeDesc.addCasProcessor(casConsumer);


      // add collectionReader
      cpeDesc.addCollectionReader(colReaderDesc);


      // produce cpe
      cpe = UIMAFramework.produceCollectionProcessingEngine(cpeDesc, resourceManager, null);
    } catch (Exception e) {
      e.printStackTrace();

View Full Code Here

   * This is needed for XMLParser.parseCpeDesription() to work. Typically users should use
   * CpeDescriptorFactory.produceDescriptor() instead. - APL
   */
  public CpeDescriptionImpl(XMLInputSource aInput) throws InvalidXMLException {
    try {
      CpeDescription descriptor = CpeDescriptorFactory.produceDescriptor(aInput);
      if (descriptor.getResourceManagerConfiguration() != null) {
        resourceMgrConfig = descriptor.getResourceManagerConfiguration();
      }
      if (descriptor.getCpeConfiguration() != null) {
        setCpeConfig(descriptor.getCpeConfiguration());
      }
      CpeCollectionReader[] readers = descriptor.getAllCollectionCollectionReaders();
      if (readers != null && readers.length > 0 && readers[0] != null) {
        addCollectionReader(readers[0]);
      }
      CpeCasProcessors cps = descriptor.getCpeCasProcessors();
      if (cps != null) {
        CpeCasProcessor[] casProcessors = cps.getAllCpeCasProcessors();
        for (int i = 0; casProcessors != null && i < casProcessors.length; i++) {
          addCasProcessor(casProcessors[i]);
        }

View Full Code Here

   * 
   * @return CollectionProcessingEngine - initialized cpe
   */
  private CollectionProcessingEngine setupCpm(int documentCount, int threadCount,
          boolean useSlowAnnotator, boolean singleThreaded) throws Exception {
    CpeDescription cpeDesc = null;
    CollectionProcessingEngine cpe = null;


    try {
      String colReaderBase = JUnitExtension.getFile("CpmTests" + separator
              + "ErrorTestCollectionReader.xml").getAbsolutePath();
      String taeBase = JUnitExtension.getFile("CpmTests" + separator + "ErrorTestAnnotator.xml").getAbsolutePath();
      String casConsumerBase = JUnitExtension.getFile("CpmTests" + separator
              + "ErrorTestCasConsumer.xml").getAbsolutePath();


      // created needed descriptors
      String colReaderDesc = DescriptorMakeUtil.makeCollectionReader(colReaderBase, documentCount);
      String taeDesc = DescriptorMakeUtil.makeAnalysisEngine(taeBase);
      String casConsumerDesc = DescriptorMakeUtil.makeCasConsumer(casConsumerBase);


      // create cpm descriptor
      cpeDesc = CpeDescriptorFactory.produceDescriptor();
      cpeDesc.setInputQueueSize(2);
      cpeDesc.setOutputQueueSize(2);
      cpeDesc.setProcessingUnitThreadCount(threadCount);


      if (singleThreaded) {
        cpeDesc.setDeployment("single-threaded");
      }


      // add tae
      CpeIntegratedCasProcessor integratedProcessor = CpeDescriptorFactory
              .produceCasProcessor("ErrorTestAnnotator");
      integratedProcessor.setDescriptor(taeDesc);
      cpeDesc.addCasProcessor(integratedProcessor);


      // add slow annotator if requested
      if (useSlowAnnotator) {
        CpeIntegratedCasProcessor slowProcessor = CpeDescriptorFactory
                .produceCasProcessor("SlowAnnotator");
        slowProcessor.setDescriptor(JUnitExtension.getFile("CpmTests" + separator
                + "SlowAnnotator.xml").getAbsolutePath());
        cpeDesc.addCasProcessor(slowProcessor);
      }


      // add casConsumer
      CpeIntegratedCasProcessor casConsumer = CpeDescriptorFactory
              .produceCasProcessor("ErrorTest CasConsumer");
      casConsumer.setDescriptor(casConsumerDesc);
      cpeDesc.addCasProcessor(casConsumer);


      // add collectionReader
      cpeDesc.addCollectionReader(colReaderDesc);


      // produce cpe
      cpe = UIMAFramework.produceCollectionProcessingEngine(cpeDesc, null, null);
    } catch (Exception e) {
      e.printStackTrace();

View Full Code Here

   * 
   * @return CollectionProcessingEngine - initialized cpe
   */
  private CollectionProcessingEngine setupCpm(int documentCount, String exceptionName,
          int exceptionSequence, String functionName) {
    CpeDescription cpeDesc = null;
    CollectionProcessingEngine cpe = null;


    try {
      String colReaderBase = JUnitExtension.getFile("CpmTests" + FS + "ErrorTestCollectionReader.xml").getAbsolutePath();
      String taeBase = JUnitExtension.getFile("CpmTests" + FS + "ErrorTestAnnotator.xml").getAbsolutePath();
      String casConsumerBase = JUnitExtension.getFile("CpmTests" + FS + "ErrorTestCasConsumer.xml").getAbsolutePath();


      // first, prepare all descriptors as needed
      String colReaderDesc = DescriptorMakeUtil.makeCollectionReader(colReaderBase, documentCount);
      String taeDesc = DescriptorMakeUtil.makeAnalysisEngine(taeBase);
      String casConsumerDesc = DescriptorMakeUtil.makeCasConsumer(casConsumerBase, true,
              functionName, exceptionSequence, exceptionName);


      // secondly, create the cpm based on the descriptors
      cpeDesc = CpeDescriptorFactory.produceDescriptor();


      // managing the default behaviour of this client
      CpeIntegratedCasProcessor integratedProcessor = CpeDescriptorFactory
              .produceCasProcessor("ErrorTestAnnotator");
      integratedProcessor.setDescriptor(taeDesc);
      integratedProcessor.setActionOnMaxError("terminate");


      CpeIntegratedCasProcessor casConsumer = CpeDescriptorFactory
              .produceCasProcessor("ErrorTest CasConsumer");
      casConsumer.setDescriptor(casConsumerDesc);


      // - add all descriptors
      cpeDesc.addCollectionReader(colReaderDesc);
      cpeDesc.addCasProcessor(integratedProcessor);
      cpeDesc.addCasProcessor(casConsumer);
      cpeDesc.setInputQueueSize(2);
      cpeDesc.setOutputQueueSize(2);
      cpeDesc.setProcessingUnitThreadCount(1);
      // - Create a new CPE
      cpe = UIMAFramework.produceCollectionProcessingEngine(cpeDesc, null, null);
    } catch (Exception e) {
      e.printStackTrace();
    }

View Full Code Here


  /**
   * Test a CPE descriptor that uses the import syntax.
   */
  public void testImports() throws Exception {
    CpeDescription cpeDesc = UIMAFramework.getXMLParser().parseCpeDescription(
            new XMLInputSource(JUnitExtension.getFile("CollectionProcessingEngineImplTest/CpeImportTest.xml")));            
    CollectionProcessingEngine cpe = UIMAFramework.produceCollectionProcessingEngine(cpeDesc);


    TestStatusCallbackListener listener = new TestStatusCallbackListener();
    cpe.addStatusCallbackListener(listener);

View Full Code Here

  
  /**
   * Test a CPE descriptor using import by name and requiring data patht o be set
   */
  public void testImportsWithDataPath() throws Exception {
    CpeDescription cpeDesc = UIMAFramework.getXMLParser().parseCpeDescription(
            new XMLInputSource(JUnitExtension.getFile("CollectionProcessingEngineImplTest/CpeImportDataPathTest.xml")));
    ResourceManager resMgr = UIMAFramework.newDefaultResourceManager();
    File dataPathDir = JUnitExtension.getFile("CollectionProcessingEngineImplTest/imports");
    resMgr.setDataPath(dataPathDir.getAbsolutePath());
    CollectionProcessingEngine cpe = UIMAFramework.produceCollectionProcessingEngine(cpeDesc, resMgr, null);

View Full Code Here

0 1 2 3 4 5 6 7 8 9

TOP

Related Classes of org.apache.uima.collection.metadata.CpeDescription

org.apache.ctakes.core.cpe.CmdLineCpeRunner

org.apache.ctakes.postagger.test.CpeTests

org.apache.ctakes.regression.test.RegressionPipelineTest

org.apache.ctakes.ytex.tools.RunCPE

org.apache.uima.application.metadata.impl.UimaApplication_Impl

org.apache.uima.collection.impl.CollectionProcessingEngine_implTest

org.apache.uima.collection.impl.cpm.CasPoolTest

org.apache.uima.collection.impl.cpm.CpeDescriptorSerialization_Test

org.apache.uima.collection.impl.cpm.CpeImportTest

org.apache.uima.collection.impl.cpm.CpmAE_ErrorTest

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.