// 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();
}
public WordsToSentencesAnnotator(
boolean verbose,
String boundaryTokenRegex,
Set<String> boundaryToDiscard,
Set<String> htmlElementsToDiscard,
String newlineIsSentenceBreak) {
this(
verbose,
false,
new WordToSentenceProcessor<CoreLabel>(
boundaryTokenRegex,
boundaryToDiscard,
htmlElementsToDiscard,
WordToSentenceProcessor.stringToNewlineIsSentenceBreak(newlineIsSentenceBreak)));
}
public WordsToSentencesAnnotator(
boolean verbose,
String boundaryTokenRegex,
Set<String> boundaryToDiscard,
Set<String> htmlElementsToDiscard,
String newlineIsSentenceBreak,
String boundaryMultiTokenRegex,
Set<String> tokenRegexesToDiscard) {
this(
verbose,
false,
new WordToSentenceProcessor<CoreLabel>(
boundaryTokenRegex,
boundaryToDiscard,
htmlElementsToDiscard,
WordToSentenceProcessor.stringToNewlineIsSentenceBreak(newlineIsSentenceBreak),
(boundaryMultiTokenRegex != null)
? TokenSequencePattern.compile(boundaryMultiTokenRegex)
: null,
tokenRegexesToDiscard));
}
/**
* If setCountLineNumbers is set to true, we count line numbers by telling the underlying splitter
* to return empty lists of tokens and then treating those empty lists as empty lines. We don't
* actually include empty sentences in the annotation, though.
*/
@Override
public void annotate(Annotation annotation) {
if (VERBOSE) {
System.err.print("Sentence splitting ...");
}
if (!annotation.has(CoreAnnotations.TokensAnnotation.class)) {
throw new IllegalArgumentException(
"WordsToSentencesAnnotator: unable to find words/tokens in: " + annotation);
}
// get text and tokens from the document
String text = annotation.get(CoreAnnotations.TextAnnotation.class);
List<CoreLabel> tokens = annotation.get(CoreAnnotations.TokensAnnotation.class);
String docID = annotation.get(CoreAnnotations.DocIDAnnotation.class);
// System.err.println("Tokens are: " + tokens);
// assemble the sentence annotations
int tokenOffset = 0;
int lineNumber = 0;
// section annotations to mark sentences with
CoreMap sectionAnnotations = null;
List<CoreMap> sentences = new ArrayList<>();
for (List<CoreLabel> sentenceTokens : wts.process(tokens)) {
if (countLineNumbers) {
++lineNumber;
}
if (sentenceTokens.isEmpty()) {
if (!countLineNumbers) {
throw new IllegalStateException("unexpected empty sentence: " + sentenceTokens);
} else {
continue;
}
}
// get the sentence text from the first and last character offsets
int begin = sentenceTokens.get(0).get(CoreAnnotations.CharacterOffsetBeginAnnotation.class);
int last = sentenceTokens.size() - 1;
int end = sentenceTokens.get(last).get(CoreAnnotations.CharacterOffsetEndAnnotation.class);
String sentenceText = text.substring(begin, end);
// create a sentence annotation with text and token offsets
Annotation sentence = new Annotation(sentenceText);
sentence.set(CoreAnnotations.CharacterOffsetBeginAnnotation.class, begin);
sentence.set(CoreAnnotations.CharacterOffsetEndAnnotation.class, end);
sentence.set(CoreAnnotations.TokensAnnotation.class, sentenceTokens);
sentence.set(CoreAnnotations.TokenBeginAnnotation.class, tokenOffset);
tokenOffset += sentenceTokens.size();
sentence.set(CoreAnnotations.TokenEndAnnotation.class, tokenOffset);
sentence.set(CoreAnnotations.SentenceIndexAnnotation.class, sentences.size());
if (countLineNumbers) {
sentence.set(CoreAnnotations.LineNumberAnnotation.class, lineNumber);
}
// Annotate sentence with section information.
// Assume section start and end appear as first and last tokens of sentence
CoreLabel sentenceStartToken = sentenceTokens.get(0);
CoreLabel sentenceEndToken = sentenceTokens.get(sentenceTokens.size() - 1);
CoreMap sectionStart = sentenceStartToken.get(CoreAnnotations.SectionStartAnnotation.class);
if (sectionStart != null) {
// Section is started
sectionAnnotations = sectionStart;
}
if (sectionAnnotations != null) {
// transfer annotations over to sentence
ChunkAnnotationUtils.copyUnsetAnnotations(sectionAnnotations, sentence);
}
String sectionEnd = sentenceEndToken.get(CoreAnnotations.SectionEndAnnotation.class);
if (sectionEnd != null) {
sectionAnnotations = null;
}
if (docID != null) {
sentence.set(CoreAnnotations.DocIDAnnotation.class, docID);
}
int index = 1;
for (CoreLabel token : sentenceTokens) {
token.setIndex(index++);
token.setSentIndex(sentences.size());
if (docID != null) {
token.setDocID(docID);
}
}
// add the sentence to the list
sentences.add(sentence);
}
// the condition below is possible if sentenceBoundaryToDiscard is initialized!
/*
if (tokenOffset != tokens.size()) {
throw new RuntimeException(String.format(
"expected %d tokens, found %d", tokens.size(), tokenOffset));
}
*/
// add the sentences annotations to the document
annotation.set(CoreAnnotations.SentencesAnnotation.class, sentences);
}
