Examples with DocumentMetaData de.tudarmstadt.ukp.dkpro.core.api.metadata.type.DocumentMetaData used on opensource projects

Example 26 with DocumentMetaData

use of de.tudarmstadt.ukp.dkpro.core.api.metadata.type.DocumentMetaData in project dkpro-tc by dkpro.

the class TestFoldUtil method createSequenceCas.

private void createSequenceCas() throws IOException, UIMAException {
    tmpFoldSeq = new TemporaryFolder();
    tmpFoldSeq.create();
    jcasSequence = JCasFactory.createJCas();
    jcasSequence.setDocumentText("One A Two B Three C Four D Five E Six F Seven G Eight H Nine I Ten J");
    setSeq(jcasSequence, 0, 5);
    setSeq(jcasSequence, 6, 11);
    setSeq(jcasSequence, 12, 19);
    setSeq(jcasSequence, 20, 26);
    setSeq(jcasSequence, 27, 33);
    setSeq(jcasSequence, 34, 39);
    setSeq(jcasSequence, 40, 47);
    setSeq(jcasSequence, 48, 55);
    setSeq(jcasSequence, 56, 62);
    setSeq(jcasSequence, 63, 68);
    DocumentMetaData dmd = new DocumentMetaData(jcasSequence);
    dmd.setDocumentId("id");
    dmd.addToIndexes();
    createJCasIdAnnotation(jcasSequence);
    AnalysisEngine xmiWriter = AnalysisEngineFactory.createEngine(BinaryCasWriter.class, BinaryCasWriter.PARAM_TARGET_LOCATION, tmpFoldSeq.getRoot(), BinaryCasWriter.PARAM_FORMAT, "6+");
    xmiWriter.process(jcasSequence);
}

Example 27 with DocumentMetaData

use of de.tudarmstadt.ukp.dkpro.core.api.metadata.type.DocumentMetaData in project dkpro-tc by dkpro.

the class FoldClassificationUnitCasMultiplier method next.

@Override
public AbstractCas next() throws AnalysisEngineProcessException {
    // Create an empty CAS as a destination for a copy.
    JCas emptyJCas = this.getEmptyJCas();
    DocumentMetaData.create(emptyJCas);
    emptyJCas.setDocumentText(jCas.getDocumentText());
    CAS emptyCas = emptyJCas.getCas();
    // Copy current CAS to the empty CAS.
    CasCopier.copyCas(jCas.getCas(), emptyCas, false);
    JCas copyJCas;
    try {
        copyJCas = emptyCas.getJCas();
    } catch (CASException e) {
        throw new AnalysisEngineProcessException("Exception while creating JCas", null, e);
    }
    // Check for multiple DocumentMetaData annotations (issue #266)
    Collection<DocumentMetaData> metaDataAnnotations = JCasUtil.select(copyJCas, DocumentMetaData.class);
    List<DocumentMetaData> metaDataAnnotationsToDelete = new ArrayList<>();
    if (metaDataAnnotations.size() > 1)
        for (DocumentMetaData metaDataAnnotation : metaDataAnnotations) if ("x-unspecified".equals(metaDataAnnotation.getLanguage()) && metaDataAnnotation.getDocumentTitle() == null && metaDataAnnotation.getDocumentId() == null && metaDataAnnotation.getDocumentUri() == null && metaDataAnnotation.getDocumentBaseUri() == null && metaDataAnnotation.getCollectionId() == null)
            metaDataAnnotationsToDelete.add(metaDataAnnotation);
    for (DocumentMetaData metaDataAnnotation : metaDataAnnotationsToDelete) copyJCas.removeFsFromIndexes(metaDataAnnotation);
    // Set new ids and URIs for copied cases.
    // The counting variable keeps track of how many new CAS objects are created from the
    // original CAS, a CAS relative counter.
    // NOTE: As it may cause confusion: If in sequence classification several or all CAS
    // contains only a single sequence this counter would be zero in all cases - this is not a
    // bug, but a cosmetic flaw
    String currentDocId = DocumentMetaData.get(jCas).getDocumentId();
    DocumentMetaData.get(copyJCas).setDocumentId(currentDocId + "_" + subCASCounter);
    String currentDocUri = DocumentMetaData.get(jCas).getDocumentUri() + "_" + subCASCounter;
    DocumentMetaData.get(copyJCas).setDocumentUri(currentDocUri);
    deleteAllTextClassificationAnnotation(copyJCas);
    setTargetAnnotation(copyJCas);
    assignNewId(copyJCas);
    subCASCounter++;
    // issue #261
    DocumentMetaData.get(copyJCas).setIsLastSegment(subCASCounter == annotations.size());
    getLogger().debug("Creating CAS " + subCASCounter + " of " + annotations.size());
    return copyJCas;
}