Java Code Examples for de.tudarmstadt.ukp.dkpro.core.api.metadata.type.DocumentMetaData#setDocumentId()

private void writeAndAssertEquals(JCas curatorCas)
    throws Exception
{
    String targetFolder = "target/test-output/" + testContext.getClassName() + "/"
            + referenceFolder.getName();
    
    DocumentMetaData dmd = DocumentMetaData.get(curatorCas);
    dmd.setDocumentId("curator");
    runPipeline(curatorCas, createEngineDescription(WebannoTsv3XWriter.class,
            WebannoTsv3XWriter.PARAM_TARGET_LOCATION, targetFolder,
            WebannoTsv3XWriter.PARAM_OVERWRITE, true));
    
    File referenceFile = new File(referenceFolder, "curator.tsv");
    assumeTrue("No reference data available for this test.", referenceFile.exists());
    
    File actualFile = new File(targetFolder, "curator.tsv");
    
    String reference = FileUtils.readFileToString(referenceFile, "UTF-8");
    String actual = FileUtils.readFileToString(actualFile, "UTF-8");
    
    assertEquals(reference, actual);
}
 

private void convert(CAS aCAS, int fact) throws CollectionException, SQLException {
  JCas jcas;
  try {
    jcas = aCAS.getJCas();
  } catch (CASException e) {
    throw new CollectionException(e);
  }
  JCasBuilder doc = new JCasBuilder(jcas);
  DocumentMetaData md = JCasUtil.selectSingle(jcas, DocumentMetaData.class);
  md.setDocumentId(Integer.toString(fact));
  doc.add("fact");
  doc.close();
}
 

public void addSettingstoJcas(JCas jcas) throws IOException {
  AidaDocumentSettings ads = new AidaDocumentSettings(jcas);
  if (this.getLanguage() != null) {
    if (jcas.getDocumentLanguage() != null && !jcas.getDocumentLanguage().equals("x-unspecified") && !jcas.getDocumentLanguage()
        .equals(this.getLanguage().toString())) {
      throw new IllegalArgumentException("Language in JCas and language in settings are different");
    }
    ads.setLanguage(this.getLanguage().toString());
    jcas.setDocumentLanguage(ads.getLanguage());
  }
  if (this.getDocChunkStrategy() != null) {
    ads.setDocChunkStrategy(this.getDocChunkStrategy().toString());
  }
  ads.setDocumentId(this.getDocumentId());
  if (ads.getDocumentInputFormat() != null) {
    ads.setDocumentInputFormat(this.getDocumentInputFormat().toString());
  }
  ads.setEncoding(this.getEncoding());
  if (disambiguationSettings != null) {
    disambiguationSettings.addToJCas(ads, jcas);
  }
  ads.addToIndexes();
  if (annotations != null) {
    annotations.addMentionsToJCas(jcas);
  }
  if (!exists(jcas, DocumentMetaData.class)) {
    DocumentMetaData md = new DocumentMetaData(jcas);
    md.setDocumentId(ads.getDocumentId());
    md.addToIndexes();
  }
}
 

private static JCas initializeJCas(StandaloneArgument argument)
        throws UIMAException
{
    JCas jCas = JCasFactory.createJCas();
    jCas.setDocumentLanguage("en");
    jCas.setDocumentText(argument.getText());

    DocumentMetaData documentMetaData = DocumentMetaData.create(jCas);
    documentMetaData.setDocumentId(argument.getId());
    documentMetaData.setDocumentTitle(argument.getDebateMetaData().getTitle());
    documentMetaData.addToIndexes();
    return jCas;
}
 

@Override
public void getNext(CAS cas) throws IOException, CollectionException {
  Resource res = nextFile();
  String name = res.getPath();
  JsonNode json = objectMapper.readTree(res.getInputStream());

  String title = null;
  if(titleKey != null) {
    title = json.get(titleKey).asText();
  }
  String content = json.get(contentKey).asText();

  StringBuilder sb = new StringBuilder();
  if (title != null) {
    sb.append(title).append(System.lineSeparator()).append(System.lineSeparator());
  } else {
    logger.debug(res.getPath() + " does not have a title field.");
  }

  if (content != null) {
    sb.append(content);
  } else {
    logger.debug(res.getPath() + " does not have a content field.");
  }

  JCas jcas;
  try {
    jcas = cas.getJCas();
  }
  catch (CASException e) {
    throw new CollectionException(e);
  }

  // Set doc id.
  String id = null;
  if(idKey != null) {
    id = json.get(idKey).asText();
  }
  if(id == null) {
    id = name;
  }
  DocumentMetaData dmd = new DocumentMetaData(jcas);
  dmd.addToIndexes();
  dmd.setDocumentId(id);
  jcas.setDocumentLanguage(getLanguage());
  jcas.setDocumentText(sb.toString());
}
 

private void parseSubDocument(JCas jcas) throws XMLStreamException, IOException, CollectionException {
  if (this.language != null) {
    jcas.setDocumentLanguage(this.language);
  }

  LinkedList openTagStack = new LinkedList();
  String docTag = this.seekSubDocumentRoot();
  StringBuilder documentText = new StringBuilder();
  String docId = null;

  String fileName;
  String docUri;
  while (this.xmlReader.hasNext() && this.xmlReader.getDepth() > 1) {
    String dotPlace;
    if (this.xmlReader.isStartElement()) {
      if (!this.xmlReader.getPrefix().isEmpty()) {
        fileName = this.xmlReader.getPrefix() + ":" + this.xmlReader.getName().getLocalPart();
      } else {
        fileName = this.xmlReader.getName().getLocalPart();
      }
      openTagStack.push(fileName);
      dotPlace = null;
      if (this.isDocIdElement(fileName) && this.docIdAttributeName != null) {
        dotPlace = this.xmlReader.getAttributeValue((String) null, this.docIdAttributeName);
      }

      this.xmlReader.next();
      docUri = this.collectText();
      if (docUri.length() > 0) {
        if (this.isDocIdElement(fileName) && this.docIdAttributeName == null) {
          dotPlace = docUri;
        }
        this.processText(jcas, fileName, docUri, documentText);
      }

      if (dotPlace != null) {
        if (docId != null) {
          throw new CollectionException("multiple_doc_id_error", new Object[] { this.docIdTag });
        }

        if (dotPlace.length() == 0) {
          throw new CollectionException("empty_doc_id_error", new Object[] { this.docIdTag });
        }

        docId = dotPlace;
      }
    } else if (this.xmlReader.isCharacters()) {
      fileName = (String) openTagStack.peek();
      dotPlace = this.collectText();
      if (dotPlace.length() != 0) {
        this.processText(jcas, fileName, dotPlace, documentText);
      }
    } else if (this.xmlReader.isEndElement()) {
      fileName = this.xmlReader.getName().getLocalPart();
      if (docTag.equals(fileName)) {
        this.xmlReader.nextTag();
        break;
      }
      openTagStack.poll();
      this.xmlReader.next();
    } else if (this.xmlReader.getEventType() == XMLStreamConstants.CDATA) {
      fileName = (String) openTagStack.peek();
      dotPlace = this.xmlReader.getText();
      if (dotPlace.length() != 0) {
        this.processText(jcas, fileName, dotPlace, documentText);
      }
      this.xmlReader.next();
    }
  }

  jcas.setDocumentText(documentText.toString());
  fileName = ((File) this.xmlFiles.get(this.currentParsedFile)).getName();
  int dotPlace1 = fileName.lastIndexOf(46);
  if (this.docIdTag != null) {
    if (docId == null) {
      throw new CollectionException("de.tudarmstadt.ukp.dkpro.core.io.xml.XmlReader_Messages", "missing_doc_id_error",
          new Object[] { this.docIdTag });
    }
  } else if (dotPlace1 >= 0) {
    docId = fileName.substring(0, dotPlace1) + "-" + this.iDoc;
  }

  docUri = ((File) this.xmlFiles.get(this.currentParsedFile)).toURI().toString();
  DocumentMetaData docMetaData = DocumentMetaData.create(jcas);
  docMetaData.setDocumentId(docId);
  docMetaData.setDocumentUri(docUri + "#" + docId);
  docMetaData.setCollectionId(this.collectionId);
}