/**
* Adds a word as an Item to this WordRelation object.
*
* @param word the word to add
*/
public void addWord(String word) {
Item tokenItem = tokenToWords.getTokenItem();
Item wordItem = tokenItem.createDaughter();
FeatureSet featureSet = wordItem.getFeatures();
featureSet.setString("name", word);
relation.appendItem(wordItem);
}
/** Adds a break as a feature to the last item in the list. */
public void addBreak() {
Item wordItem = (Item) relation.getTail();
if (wordItem != null) {
FeatureSet featureSet = wordItem.getFeatures();
featureSet.setString("break", "1");
}
}
/** generate mention annotations (with entity numbers) based on the ACE entities and mentions. */
static void addMentionTags(Document doc, AceDocument aceDoc) {
ArrayList<AceEntity> entities = aceDoc.entities;
for (int i = 0; i < entities.size(); i++) {
AceEntity entity = entities.get(i);
ArrayList<AceEntityMention> mentions = entity.mentions;
for (int j = 0; j < mentions.size(); j++) {
AceEntityMention mention = (AceEntityMention) mentions.get(j);
// we compute a jetSpan not including trailing whitespace
Span aceSpan = mention.head;
// skip mentions in ChEnglish APF not aligned to any English text
if (aceSpan.start() < 0) continue;
Span jetSpan = new Span(aceSpan.start(), aceSpan.end() + 1);
FeatureSet features = new FeatureSet("entity", new Integer(i));
if (flags.contains("types")) {
features.put("type", entity.type.substring(0, 3));
if (entity.subtype != null) features.put("subtype", entity.subtype);
}
if (flags.contains("extents")) {
String cleanExtent = mention.text.replaceAll("\n", " ");
features.put("extent", AceEntityMention.addXmlEscapes(cleanExtent));
}
doc.annotate("mention", jetSpan, features);
}
}
}
/**
* performs the action, adding the specified Annotation. Returns the position of the end of the
* Annotation.
*/
@Override
public int perform(Document doc, PatternApplication patap) {
Span span;
HashMap bindings = patap.bestBindings;
// System.out.println ("bindings (for new annotation): " + bindings);
if (spanVariable == null) {
span = new Span(patap.startPosition, patap.bestPosition);
} else if (spanVariable.name.toString() == "0") {
span = new Span(patap.startPosition, patap.startPosition);
} else {
Object value = bindings.get(spanVariable.name);
if (value instanceof Span) {
span = (Span) value;
} else if (value instanceof Annotation) {
span = ((Annotation) value).span();
} else {
System.out.println("Value of " + spanVariable.toString() + " is not a span.or annotation");
return -1;
}
}
if (Pat.trace)
Console.println(
"Annotating "
+ doc.text(span)
+ " as "
+ type
+ " "
+ features.substitute(bindings).toSGMLString());
hideAnnotations(doc, type, span);
hideAnnotations(doc, "token", span);
Annotation newAnnotation = new Annotation(type, span, features.substitute(bindings));
doc.addAnnotation(newAnnotation);
if (bindingVariable != null) bindings.put(bindingVariable.name, newAnnotation);
return span.end();
}
static void addTimexTags(Document doc, AceDocument aceDoc) {
List<AceTimex> timeExpressions = aceDoc.timeExpressions;
for (AceTimex timex : timeExpressions) {
AceTimexMention mention = (AceTimexMention) timex.mentions.get(0);
Span aceSpan = mention.extent;
Span jetSpan = new Span(aceSpan.start(), aceSpan.end() + 1);
FeatureSet features = new FeatureSet();
if (timex.val != null && !timex.val.equals("")) features.put("val", timex.val);
if (timex.anchorVal != null && !timex.anchorVal.equals(""))
features.put("anchor_val", timex.anchorVal);
if (timex.anchorDir != null && !timex.anchorDir.equals(""))
features.put("anchor_dir", timex.anchorDir);
if (timex.set != null && !timex.set.equals("")) features.put("set", timex.set);
if (timex.mod != null && !timex.mod.equals("")) features.put("mod", timex.mod);
doc.annotate("timex2", jetSpan, features);
}
}
/** generate mention annotations (with entity numbers) based on the ACE entities and mentions. */
static void addMentionTags(Document doc, AceDocument aceDoc) {
ArrayList<AceEntity> entities = aceDoc.entities;
for (int i = 0; i < entities.size(); i++) {
AceEntity entity = (AceEntity) entities.get(i);
ArrayList<AceEntityMention> mentions = entity.mentions;
for (int j = 0; j < mentions.size(); j++) {
AceEntityMention mention = mentions.get(j);
// we compute a jetSpan not including trailing whitespace
Span aceSpan = mention.head;
Span jetSpan = new Span(aceSpan.start(), aceSpan.end() + 1);
FeatureSet features = new FeatureSet("entity", new Integer(i));
if (showTypes) {
features.put("type", entity.type.substring(0, 3));
if (entity.subtype != null) features.put("subtype", entity.subtype);
}
doc.annotate("mention", jetSpan, features);
}
}
}
/**
* Sets the token list for this utterance. Note that this could be optimized by turning the token
* list directly into the token relation.
*
* @param tokenList the tokenList
*/
private void setTokenList(List tokenList) {
setInputText(tokenList);
Relation relation = createRelation(Relation.TOKEN);
for (Iterator i = tokenList.iterator(); i.hasNext(); ) {
Token token = (Token) i.next();
String tokenWord = token.getWord();
if (tokenWord != null && tokenWord.length() > 0) {
Item item = relation.appendItem();
FeatureSet featureSet = item.getFeatures();
featureSet.setString("name", tokenWord);
featureSet.setString("whitespace", token.getWhitespace());
featureSet.setString("prepunctuation", token.getPrepunctuation());
featureSet.setString("punc", token.getPostpunctuation());
featureSet.setString("file_pos", String.valueOf(token.getPosition()));
featureSet.setString("line_number", String.valueOf(token.getLineNumber()));
}
}
}
/**
* Sets the last Item in this WordRelation to the given word.
*
* @param word the word to set
*/
public void setLastWord(String word) {
Item lastItem = relation.getTail();
FeatureSet featureSet = lastItem.getFeatures();
featureSet.setString("name", word);
}
@Override
public String toString() {
if (spanVariable == null) return "add [" + type + features.toSGMLString() + "]";
else return "add [" + type + features.toSGMLString() + "] over " + spanVariable.toString();
}
