Examples with PretokenisedSequence com.joliciel.talismane.tokeniser.PretokenisedSequence used on opensource projects

Example 1 with PretokenisedSequence

use of com.joliciel.talismane.tokeniser.PretokenisedSequence in project talismane by joliciel-informatique.

the class StandoffReader method hasNextSentence.

@Override
public boolean hasNextSentence() throws TalismaneException, IOException {
    if (this.getMaxSentenceCount() > 0 && sentenceCount >= this.getMaxSentenceCount()) {
    // we've reached the end, do nothing
    } else {
        if (configuration == null && sentenceIndex < sentences.size()) {
            List<StandoffToken> tokens = sentences.get(sentenceIndex++);
            LinguisticRules rules = TalismaneSession.get(sessionId).getLinguisticRules();
            if (rules == null)
                throw new RuntimeException("Linguistic rules have not been set.");
            String text = "";
            for (StandoffToken standoffToken : tokens) {
                String word = standoffToken.text;
                if (rules.shouldAddSpace(text, word))
                    text += " ";
                text += word;
            }
            Sentence sentence = new Sentence(text, sessionId);
            for (SentenceAnnotator annotator : TalismaneSession.get(sessionId).getSentenceAnnotators()) {
                annotator.annotate(sentence);
            }
            PretokenisedSequence tokenSequence = new PretokenisedSequence(sentence, sessionId);
            PosTagSequence posTagSequence = new PosTagSequence(tokenSequence);
            Map<String, PosTaggedToken> idTokenMap = new HashMap<String, PosTaggedToken>();
            for (StandoffToken standoffToken : tokens) {
                Token token = tokenSequence.addToken(standoffToken.text);
                Decision posTagDecision = new Decision(standoffToken.posTag.getCode());
                PosTaggedToken posTaggedToken = new PosTaggedToken(token, posTagDecision, sessionId);
                if (LOG.isTraceEnabled()) {
                    LOG.trace(posTaggedToken.toString());
                }
                posTaggedToken.setComment(standoffToken.comment);
                posTagSequence.addPosTaggedToken(posTaggedToken);
                idTokenMap.put(standoffToken.id, posTaggedToken);
                LOG.debug("Found token " + standoffToken.id + ", " + posTaggedToken);
            }
            tokenSequence.setWithRoot(true);
            configuration = new ParseConfiguration(posTagSequence);
            for (StandoffToken standoffToken : tokens) {
                StandoffRelation relation = relationMap.get(standoffToken.id);
                if (relation != null) {
                    PosTaggedToken head = idTokenMap.get(relation.fromToken);
                    PosTaggedToken dependent = idTokenMap.get(relation.toToken);
                    if (head == null) {
                        throw new TalismaneException("No token found for head id: " + relation.fromToken);
                    }
                    if (dependent == null) {
                        throw new TalismaneException("No token found for dependent id: " + relation.toToken);
                    }
                    DependencyArc arc = configuration.addDependency(head, dependent, relation.label, null);
                    arc.setComment(relation.comment);
                } else if (standoffToken.posTag.getOpenClassIndicator() == PosTagOpenClassIndicator.PUNCTUATION) {
                    if (punctuationDepLabel != null) {
                        PosTaggedToken dependent = idTokenMap.get(standoffToken.id);
                        for (int i = dependent.getIndex() - 1; i >= 0; i--) {
                            PosTaggedToken head = posTagSequence.get(i);
                            if (head.getTag().getOpenClassIndicator() == PosTagOpenClassIndicator.PUNCTUATION)
                                continue;
                            configuration.addDependency(head, dependent, punctuationDepLabel, null);
                            break;
                        }
                    }
                }
            }
        }
    }
    return (configuration != null);
}