// todo: give options for document splitting. A line or the whole file or
// sentence splitting as now
public Iterator<List<IN>> getIterator(Reader r) {
Tokenizer<IN> tokenizer = tokenizerFactory.getTokenizer(r);
// PTBTokenizer.newPTBTokenizer(r, false, true);
List<IN> words = new ArrayList<IN>();
IN previous = tokenFactory.makeToken();
StringBuilder prepend = new StringBuilder();
/*
* This changes SGML tags into whitespace -- it should maybe be moved
* elsewhere
*/
while (tokenizer.hasNext()) {
IN w = tokenizer.next();
String word = w.get(CoreAnnotations.TextAnnotation.class);
Matcher m = sgml.matcher(word);
if (m.matches()) {
String before = StringUtils.getNotNullString(w.get(CoreAnnotations.BeforeAnnotation.class));
String after = StringUtils.getNotNullString(w.get(CoreAnnotations.AfterAnnotation.class));
prepend.append(before).append(word);
String previousTokenAfter =
StringUtils.getNotNullString(previous.get(CoreAnnotations.AfterAnnotation.class));
previous.set(AfterAnnotation.class, previousTokenAfter + word + after);
// previous.appendAfter(w.word() + w.after());
} else {
String before = StringUtils.getNotNullString(w.get(CoreAnnotations.BeforeAnnotation.class));
if (prepend.length() > 0) {
w.set(BeforeAnnotation.class, prepend.toString() + before);
// w.prependBefore(prepend.toString());
prepend = new StringBuilder();
}
words.add(w);
previous = w;
}
}
List<List<IN>> sentences = wts.process(words);
String after = "";
IN last = null;
for (List<IN> sentence : sentences) {
int pos = 0;
for (IN w : sentence) {
w.set(PositionAnnotation.class, Integer.toString(pos));
after = StringUtils.getNotNullString(w.get(CoreAnnotations.AfterAnnotation.class));
w.remove(AfterAnnotation.class);
last = w;
}
}
if (last != null) {
last.set(AfterAnnotation.class, after);
}
return sentences.iterator();
}
@Override
public Sequence<IString> process(String input) {
String tokenizerInput = toUncased(input.trim());
Tokenizer<CoreLabel> tokenizer = tf.getTokenizer(new StringReader(tokenizerInput));
List<String> outputStrings = new LinkedList<>();
while (tokenizer.hasNext()) {
String string = tokenizer.next().get(TextAnnotation.class);
outputStrings.add(string);
}
return IStrings.toIStringSequence(outputStrings);
}
public static void main(String[] args) throws IOException {
Tokenizer<String> att = new ArabicTreebankTokenizer(new FileReader(args[0]));
while (att.hasNext()) {
System.out.print(att.next());
}
}
