/**
* Builds Jet.Tipster.Document object from Penn treebank corpus.
*
* @param in
* @return
* @throws IOException
* @throws InvalidFormatException
*/
public Treebank load(Reader in) throws IOException, InvalidFormatException {
List<ParseTreeNode> trees = new ArrayList<ParseTreeNode>();
PushbackReader input = new PushbackReader(in);
int start = 0;
while (true) {
skipWhitespace(input);
if (lookAhead(input) == -1) {
break;
}
ParseTreeNode tree = readNode(input);
trees.add(tree);
determineSpans(tree, start);
setAnnotations(tree, null);
start = tree.end;
}
String text = buildDocumentString(trees);
Document doc = new Document(text);
for (ParseTreeNode tree : trees) {
doc.annotate("sentence", new Span(tree.start, tree.end), new FeatureSet());
annotate(doc, tree);
}
return new Treebank(doc, trees);
}
public void apply(Document doc, List<Object> values, Span span, DateTime ref) {
Map params = getParameters();
String value = (String) params.get("value");
String diff = (String) params.get("diff");
String dir = (String) params.get("dir");
DateTime val = ref;
if (value != null) {
value = assignValues(value, values);
val = new DateTime(value);
} else if (diff != null) {
diff = assignValues(diff, values);
Period period = new Period(diff);
if (dir == null || dir.equals("plus")) {
val = ref.plus(period);
} else if (dir.equals("minus")) {
val = ref.minus(period);
}
} else {
val = ref;
// use set_xxx
for (Map.Entry entry : (Set<Map.Entry>) params.entrySet()) {
Matcher m = Pattern.compile("set_(.*)").matcher((String) entry.getKey());
if (m.matches()) {
String field = assignValues((String) entry.getValue(), values);
String fieldName = m.group(1);
if (fieldName.equals("month")) {
int month = Integer.parseInt(field);
val = getTimeAnnotator().normalizeMonth(val, month);
} else if (fieldName.equals("day")) {
int day = Integer.parseInt(field);
val = val.withField(DateTimeFieldType.dayOfMonth(), day);
} else {
throw new InternalError();
}
}
}
}
String formattedDate = formatter.print(val);
FeatureSet attrs = new FeatureSet();
attrs.put("VAL", formattedDate);
doc.annotate("TIMEX2", span, attrs);
}
private void annotate(Document doc, ParseTreeNode node) {
doc.addAnnotation(node.ann);
if (node.children != null) {
Annotation[] children = new Annotation[node.children.length];
for (int i = 0; i < node.children.length; i++) {
children[i] = node.children[i].ann;
}
node.ann.put("children", children);
for (ParseTreeNode child : node.children) {
annotate(doc, child);
}
}
if (node.children == null && isAddingTokens) {
// TODO: adds `case' property
doc.annotate("token", node.ann.span(), new FeatureSet());
}
}
