@Override
public void process(JCas jCas) throws AnalysisEngineProcessException {
try {
String apfUri = jCas.getView(Ace2005Constants.ACE_2005_APF_URI_VIEW).getSofaDataURI();
JCas initialView = jCas.getView(CAS.NAME_DEFAULT_SOFA);
String documentText = initialView.getDocumentText();
SAXBuilder builder = new SAXBuilder();
builder.setDTDHandler(null);
URI sofaDataURI = new URI(apfUri);
Document doc = builder.build(new File(sofaDataURI));
Element apfSource = doc.getRootElement();
Element apfDocument = apfSource.getChild("document");
for (Element apfEntity : apfDocument.getChildren("entity")) {
NamedEntity namedEntity = new NamedEntity(initialView);
namedEntity.setEntityType(apfEntity.getAttributeValue("TYPE"));
namedEntity.setEntitySubtype(apfEntity.getAttributeValue("SUBTYPE"));
namedEntity.setEntityClass(apfEntity.getAttributeValue("CLASS"));
namedEntity.setEntityId(apfEntity.getAttributeValue("ID"));
namedEntity.addToIndexes();
List<NamedEntityMention> mentions = new ArrayList<NamedEntityMention>();
for (Element entityMention : apfEntity.getChildren("entity_mention")) {
int start =
Integer.parseInt(
entityMention.getChild("extent").getChild("charseq").getAttributeValue("START"));
int end =
Integer.parseInt(
entityMention.getChild("extent").getChild("charseq").getAttributeValue("END"));
String givenText = entityMention.getChild("extent").getChild("charseq").getText();
String parsedText = documentText.substring(start, end + 1);
Matcher ampMatcher = ampPattern.matcher(parsedText);
parsedText = ampMatcher.replaceAll("&");
NamedEntityMention mention = new NamedEntityMention(initialView, start, end + 1);
mention.setMentionId(entityMention.getAttributeValue("ID"));
mention.setMentionType(entityMention.getAttributeValue("TYPE"));
mention.setMentionedEntity(namedEntity);
Chunk chunk = new Chunk(initialView, start, end + 1);
mention.setAnnotation(chunk);
int headStart =
Integer.parseInt(
entityMention.getChild("head").getChild("charseq").getAttributeValue("START"));
int headEnd =
Integer.parseInt(
entityMention.getChild("head").getChild("charseq").getAttributeValue("END"));
Chunk head = new Chunk(initialView, headStart, headEnd + 1);
mention.setHead(head);
mention.addToIndexes();
mentions.add(mention);
givenText = givenText.replaceAll("\\s+", " ");
parsedText = givenText.replaceAll("\\s+", " ");
// if(!givenText.equals(parsedText))
// {
//
// System.out.println("given text and parsed text differ.");
// System.out.println(givenText);
// System.out.println(parsedText);
// System.out.println(apfDocument.getAttributeValue("DOCID"));
// System.out.println(apfEntity.getAttributeValue("ID"));
// System.out.println(documentText);
// }
}
namedEntity.setMentions(new FSArray(jCas, mentions.size()));
FSCollectionFactory.fillArrayFS(namedEntity.getMentions(), mentions);
}
} catch (CASException ce) {
throw new AnalysisEngineProcessException(ce);
} catch (IOException ioe) {
throw new AnalysisEngineProcessException(ioe);
} catch (JDOMException je) {
throw new AnalysisEngineProcessException(je);
} catch (URISyntaxException use) {
throw new AnalysisEngineProcessException(use);
}
}
