Examples with AnnotationObserver com.joliciel.talismane.AnnotationObserver used on opensource projects

Example 1 with AnnotationObserver

use of com.joliciel.talismane.AnnotationObserver in project talismane by joliciel-informatique.

the class RawTextProcessor method getDetectedSentences.

/**
 * Get a list of sentences currently detected. All sentences will be complete
 * - if the list ends with an incomplete sentence it is kept for another
 * round.
 *
 * @return
 */
public final List<Sentence> getDetectedSentences() {
    SentenceHolder prevHolder = this.getPreviousSentenceHolder();
    SentenceHolder currentHolder = this.getCurrentSentenceHolder();
    for (Annotation<SentenceBoundary> sentenceBoundary : sentenceBoundaries) {
        currentHolder.addSentenceBoundary(sentenceBoundary.getStart() - prevHolder.getProcessedText().length());
        currentHolder.addSentenceBoundary(sentenceBoundary.getEnd() - prevHolder.getProcessedText().length());
    }
    List<Sentence> sentences = currentHolder.getDetectedSentences(leftover);
    leftover = null;
    if (sentences.size() > 0) {
        Sentence lastSentence = sentences.get(sentences.size() - 1);
        if (!lastSentence.isComplete()) {
            leftover = lastSentence;
            if (LOG.isTraceEnabled())
                LOG.trace("Set leftover to: " + leftover.toString());
            sentences.remove(sentences.size() - 1);
        }
    }
    // ensure that sentence annotations get added to the raw text as well
    for (Sentence sentence : sentences) {
        sentence.addObserver(new AnnotationObserver() {

            int myOrigin = RawTextProcessor.this.originalStartIndex;

            @Override
            public <T extends Serializable> void beforeAddAnnotations(AnnotatedText subject, List<Annotation<T>> annotations) {
                List<Annotation<T>> newAnnotations = new ArrayList<>();
                for (Annotation<T> annotation : annotations) {
                    int originalStart = sentence.getOriginalIndex(annotation.getStart());
                    int originalEnd = sentence.getOriginalIndex(annotation.getEnd());
                    Annotation<T> newAnnotation = annotation.getAnnotation(originalStart - myOrigin, originalEnd - myOrigin);
                    newAnnotations.add(newAnnotation);
                }
                RawTextProcessor.this.addAnnotations(newAnnotations);
            }

            @Override
            public <T extends Serializable> void afterAddAnnotations(AnnotatedText subject) {
            }
        });
    }
    // position 0.
    if (currentHolder.getOriginalTextSegments().size() > 0) {
        if (leftover == null) {
            leftover = new Sentence("", currentFile, sessionId);
        }
        StringBuilder segmentsToInsert = new StringBuilder();
        if (leftover.getLeftoverOriginalText().length() > 0)
            segmentsToInsert.append(TalismaneSession.get(sessionId).getOutputDivider());
        for (String originalTextSegment : currentHolder.getOriginalTextSegments().values()) {
            segmentsToInsert.append(originalTextSegment);
        }
        leftover.setLeftoverOriginalText(leftover.getLeftoverOriginalText() + segmentsToInsert.toString());
    }
    return sentences;
}

Example 2 with AnnotationObserver

use of com.joliciel.talismane.AnnotationObserver in project talismane by joliciel-informatique.

the class RollingTextBlock method getRawTextBlock.

/**
 * Get a raw text block for annotation by filters. This covers blocks 3 and 4
 * only of the current RollingTextBlock, with analysis end at the end of
 * block3. It is assumed that annotations crossing block 2 and 3 were already
 * added by a predecessor.
 */
public AnnotatedText getRawTextBlock() {
    AnnotatedText rawTextBlock = new AnnotatedText(this.block3 + this.block4, 0, this.block3.length());
    rawTextBlock.addObserver(new AnnotationObserver() {

        @Override
        public <T extends Serializable> void beforeAddAnnotations(AnnotatedText subject, List<Annotation<T>> annotations) {
            if (annotations.size() > 0) {
                int offset = RollingTextBlock.this.block1.length() + RollingTextBlock.this.block2.length();
                List<Annotation<T>> newAnnotations = new ArrayList<>();
                for (Annotation<T> annotation : annotations) {
                    Annotation<T> newAnnotation = annotation.getAnnotation(annotation.getStart() + offset, annotation.getEnd() + offset);
                    newAnnotations.add(newAnnotation);
                }
                RollingTextBlock.this.addAnnotations(newAnnotations);
                if (LOG.isTraceEnabled()) {
                    LOG.trace("RawTextBlock Annotations received: " + annotations);
                    LOG.trace("RawTextBlock Annotations added: " + newAnnotations);
                }
            }
        }

        @Override
        public <T extends Serializable> void afterAddAnnotations(AnnotatedText subject) {
        }
    });
    return rawTextBlock;
}

Example 3 with AnnotationObserver

use of com.joliciel.talismane.AnnotationObserver in project talismane by joliciel-informatique.

the class RawTextProcessor method getProcessedText.

/**
 * Return processed text ready for sentence detection.
 *
 * It has sentence break and non-sentence-break annotations inherited from the
 * present RawTextProcessor. Any sentence-break annotations added will
 * automatically get reflected in the current RollingTextBlock.
 *
 * @return
 */
public final AnnotatedText getProcessedText() {
    LOG.trace("getProcessedTextBlock");
    int textStartPos = this.getTextProcessingStart();
    int textEndPos = this.getTextProcessingEnd();
    SentenceHolder prevHolder = this.getPreviousSentenceHolder();
    SentenceHolder currentHolder = this.getCurrentSentenceHolder();
    SentenceHolder nextHolder = this.getNextSentenceHolder();
    StringBuilder sb = new StringBuilder();
    String processedText1 = prevHolder.getProcessedText();
    String processedText2 = currentHolder.getProcessedText();
    String processedText3 = nextHolder.getProcessedText();
    sb.append(processedText1);
    sb.append(processedText2);
    sb.append(processedText3);
    String processedText = sb.toString();
    List<Annotation<RawTextMarker>> myAnnotations = this.getAnnotations(RawTextMarker.class);
    List<Annotation<RawTextMarker>> hisAnnotations = new ArrayList<>();
    int prevHolderOriginalIndex = prevHolder.getOriginalStartIndex();
    for (Annotation<RawTextMarker> myAnnotation : myAnnotations) {
        if ((myAnnotation.getStart() >= textStartPos && myAnnotation.getStart() < textEndPos) || ((myAnnotation.getEnd() >= textStartPos && myAnnotation.getEnd() < textEndPos))) {
            int originalStart = prevHolderOriginalIndex + myAnnotation.getStart();
            int originalEnd = prevHolderOriginalIndex + myAnnotation.getEnd();
            int localStart = processedText1.length();
            if (originalStart >= currentHolder.getOriginalStartIndex())
                localStart += currentHolder.getIndex(originalStart);
            int localEnd = processedText1.length() + currentHolder.getIndex(originalEnd);
            Annotation<RawTextMarker> hisAnnotation = myAnnotation.getAnnotation(localStart, localEnd);
            hisAnnotations.add(hisAnnotation);
        }
    }
    if (LOG.isTraceEnabled()) {
        LOG.trace("raw annotations: " + myAnnotations);
        LOG.trace("processed annotations: " + hisAnnotations);
    }
    AnnotatedText processedTextBlock = new AnnotatedText(processedText, processedText1.length(), processedText1.length() + processedText2.length());
    processedTextBlock.addAnnotations(hisAnnotations);
    processedTextBlock.addObserver(new AnnotationObserver() {

        // an observer which adds any annotations added to the
        // processedTextBlock back to myself, at the correct position
        @Override
        public <T extends Serializable> void beforeAddAnnotations(AnnotatedText subject, List<Annotation<T>> annotations) {
            int offset = textStartPos;
            int length1 = prevHolder.getProcessedText().length();
            int length2 = currentHolder.getProcessedText().length();
            int sentence2HolderStart = currentHolder.getOriginalStartIndex();
            List<Annotation<T>> newAnnotations = new ArrayList<>();
            for (Annotation<T> annotation : annotations) {
                int originalStart = -1;
                if (annotation.getStart() < length1)
                    originalStart = prevHolder.getOriginalIndex(annotation.getStart());
                else if (annotation.getStart() < length1 + length2)
                    originalStart = currentHolder.getOriginalIndex(annotation.getStart() - length1);
                if (originalStart >= 0) {
                    int originalEnd = -1;
                    if (annotation.getEnd() <= length1 + length2)
                        originalEnd = currentHolder.getOriginalIndex(annotation.getEnd() - length1);
                    else
                        originalEnd = nextHolder.getOriginalIndex(annotation.getEnd() - (length1 + length2));
                    if (originalEnd >= 0) {
                        Annotation<T> newAnnotation = annotation.getAnnotation(originalStart - sentence2HolderStart + offset, originalEnd - sentence2HolderStart + offset);
                        newAnnotations.add(newAnnotation);
                        if (annotation.getData() instanceof SentenceBoundary) {
                            @SuppressWarnings("unchecked") Annotation<SentenceBoundary> sentenceBoundary = (Annotation<SentenceBoundary>) annotation;
                            sentenceBoundaries.add(sentenceBoundary);
                        }
                    }
                }
            }
            RawTextProcessor.this.addAnnotations(newAnnotations);
            if (LOG.isTraceEnabled()) {
                LOG.trace("ProcessedTextBlock Annotations received: " + annotations);
                LOG.trace("ProcessedTextBlock Annotations added: " + newAnnotations);
            }
        }

        @Override
        public <T extends Serializable> void afterAddAnnotations(AnnotatedText subject) {
        }
    });
    return processedTextBlock;
}