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

Example 11 with AnnotatedText

use of com.joliciel.talismane.AnnotatedText 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;
}

Example 12 with AnnotatedText

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

the class TalismaneAPIExamples method example2.

/**
 * Similar to example1, but begins with filtering and sentence detection.
 */
public static void example2(String sessionId) throws Exception {
    String text = "Les gens qui voient de travers pensent que les bancs verts qu'on voit sur les trottoirs " + "sont faits pour les impotents ou les ventripotents. " + "Mais c'est une absurdité, car, à la vérité, ils sont là, c'est notoire, " + "pour accueillir quelque temps les amours débutants.";
    RawText rawText = new RawText(text, true, sessionId);
    // issues (e.g. replace " with ")
    for (RawTextAnnotator filter : TalismaneSession.get(sessionId).getTextAnnotators()) {
        filter.annotate(rawText);
    }
    // retrieve the processed text after filters have been applied
    AnnotatedText processedText = rawText.getProcessedText();
    // detect sentences
    SentenceDetector sentenceDetector = SentenceDetector.getInstance(sessionId);
    sentenceDetector.detectSentences(processedText);
    // the detected sentences can be retrieved directly from the raw text
    // this allows annotations made on the sentences to get reflected in the
    // raw text
    List<Sentence> sentences = rawText.getDetectedSentences();
    for (Sentence sentence : sentences) {
        // assignment for a given word)
        for (SentenceAnnotator annotator : TalismaneSession.get(sessionId).getSentenceAnnotators()) {
            annotator.annotate(sentence);
        }
        // tokenise the text
        Tokeniser tokeniser = Tokeniser.getInstance(sessionId);
        TokenSequence tokenSequence = tokeniser.tokeniseSentence(sentence);
        // pos-tag the token sequence
        PosTagger posTagger = PosTaggers.getPosTagger(sessionId);
        PosTagSequence posTagSequence = posTagger.tagSentence(tokenSequence);
        System.out.println(posTagSequence);
        // parse the pos-tag sequence
        Parser parser = Parsers.getParser(sessionId);
        ParseConfiguration parseConfiguration = parser.parseSentence(posTagSequence);
        System.out.println(parseConfiguration);
        ParseTree parseTree = new ParseTree(parseConfiguration, true);
        System.out.println(parseTree);
    }
}

Example 13 with AnnotatedText

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

the class RawTextTest method testGetDetectedSentences.

@Test
public void testGetDetectedSentences() throws Exception {
    System.setProperty("config.file", "src/test/resources/test.conf");
    ConfigFactory.invalidateCaches();
    final Config config = ConfigFactory.load();
    final String sessionId = "test";
    String[] labels = new String[0];
    String text = "Sentence 1<sent/>Sentence 2. Sentence 3. Sentence 4.";
    RawText textBlock = new RawText(text, true, sessionId);
    // we add a sentence break annotation to the raw text (as if it was
    // added by a filter)
    System.out.println("we add a sentence break annotation (as if it was added by a filter)");
    List<Annotation<RawTextSentenceBreakMarker>> sentenceBreaks = new ArrayList<>();
    sentenceBreaks.add(new Annotation<>("Sentence 1".length(), "Sentence 1<sent/>".length(), new RawTextSentenceBreakMarker("me"), labels));
    textBlock.addAnnotations(sentenceBreaks);
    List<Annotation<RawTextSkipMarker>> skips = new ArrayList<>();
    skips.add(new Annotation<>("Sentence 1".length(), "Sentence 1<sent/>".length(), new RawTextSkipMarker("me"), labels));
    textBlock.addAnnotations(skips);
    System.out.println("textBlock text: " + textBlock.getText());
    System.out.println("textBlock annotations: " + textBlock.getAnnotations().toString());
    AnnotatedText processedTextBlock = textBlock.getProcessedText();
    assertEquals("Sentence 1 Sentence 2. Sentence 3. Sentence 4.", processedTextBlock.getText());
    // add sentence boundaries to the processed text (as if they were added
    // by a sentence detector)
    System.out.println("add sentence boundaries to the processed text (as if they were added by a sentence detector)");
    List<Annotation<SentenceBoundary>> sentenceBoundaries = new ArrayList<>();
    sentenceBoundaries.add(new Annotation<>("Sentence 1".length(), "Sentence 1 Sentence 2.".length(), new SentenceBoundary(), labels));
    sentenceBoundaries.add(new Annotation<>("Sentence 1 Sentence 2.".length(), "Sentence 1 Sentence 2. Sentence 3.".length(), new SentenceBoundary(), labels));
    processedTextBlock.addAnnotations(sentenceBoundaries);
    assertEquals("Sentence 1 Sentence 2. Sentence 3. Sentence 4.", processedTextBlock.getText());
    System.out.println("textBlock text: " + textBlock.getText());
    System.out.println("textBlock annotations: " + textBlock.getAnnotations().toString());
    // ensure that the sentence boundary annotations in the original text
    // are in the right place
    assertEquals("Sentence 1<sent/>Sentence 2. Sentence 3. Sentence 4.", textBlock.getText());
    sentenceBoundaries = textBlock.getAnnotations(SentenceBoundary.class);
    assertEquals(2, sentenceBoundaries.size());
    assertEquals("Sentence 1<sent/>".length(), sentenceBoundaries.get(0).getStart());
    assertEquals("Sentence 1<sent/>Sentence 2.".length(), sentenceBoundaries.get(0).getEnd());
    assertEquals("Sentence 1<sent/>Sentence 2.".length(), sentenceBoundaries.get(1).getStart());
    assertEquals("Sentence 1<sent/>Sentence 2. Sentence 3.".length(), sentenceBoundaries.get(1).getEnd());
    List<Sentence> sentences = textBlock.getDetectedSentences();
    System.out.println("sentences: " + sentences.toString());
    assertEquals(4, sentences.size());
    assertEquals("Sentence 1", sentences.get(0).getText());
    assertEquals("".length(), sentences.get(0).getOriginalIndex(0));
    assertEquals("Sentence 2.", sentences.get(1).getText());
    assertEquals("Sentence 1<sent/>".length(), sentences.get(1).getOriginalIndex(0));
    assertEquals("Sentence 3.", sentences.get(2).getText());
    assertEquals("Sentence 1<sent/>Sentence 2. ".length(), sentences.get(2).getOriginalIndex(0));
    assertEquals("Sentence 4.", sentences.get(3).getText());
    assertEquals("Sentence 1<sent/>Sentence 2. Sentence 3. ".length(), sentences.get(3).getOriginalIndex(0));
    // test that sentence annotations get added to the original raw text
    Sentence sentence4 = sentences.get(3);
    List<Annotation<String>> annotations = new ArrayList<>();
    annotations.add(new Annotation<String>("Sentence ".length(), "Sentence 4".length(), "four", labels));
    sentence4.addAnnotations(annotations);
    annotations = textBlock.getAnnotations(String.class);
    assertEquals(1, annotations.size());
    assertEquals("Sentence 1<sent/>Sentence 2. Sentence 3. Sentence ".length(), annotations.get(0).getStart());
    assertEquals("Sentence 1<sent/>Sentence 2. Sentence 3. Sentence 4".length(), annotations.get(0).getEnd());
}

Example 14 with AnnotatedText

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

the class RegexMarkerFilterTest method testApply.

@Test
public void testApply() throws Exception {
    RawTextRegexAnnotator filter = new RawTextRegexAnnotator(RawTextMarkType.SKIP, "<skip>.*?</skip>", 0, 1000);
    AnnotatedText text = new AnnotatedText("J'ai du <skip>skip me</skip>mal à le croire.<skip>skip me</skip>");
    filter.annotate(text);
    LOG.debug(text.getAnnotations().toString());
    List<Annotation<RawTextSkipMarker>> skips = text.getAnnotations(RawTextSkipMarker.class);
    assertEquals(2, skips.size());
    int i = 0;
    for (Annotation<RawTextSkipMarker> skip : skips) {
        if (i == 0) {
            assertEquals("J'ai du ".length(), skip.getStart());
            assertEquals("J'ai du <skip>skip me</skip>".length(), skip.getEnd());
        } else if (i == 2) {
            assertEquals("J'ai du <skip>skip me</skip>mal à le croire.".length(), skip.getStart());
            assertEquals("J'ai du <skip>skip me</skip>mal à le croire.<skip>skip me</skip>".length(), skip.getEnd());
        }
        i++;
    }
}

Example 15 with AnnotatedText

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

the class RegexMarkerFilterTest method testTag.

@Test
public void testTag() throws Exception {
    RawTextRegexAnnotator filter = new RawTextRegexAnnotator(RawTextMarkType.TAG, "<skip>(.*?)</skip>", 0, 1000);
    filter.setAttribute(new StringAttribute("TAG1", "x"));
    AnnotatedText text = new AnnotatedText("J'ai du <skip>skip me</skip>mal à le croire.<skip>skip this</skip>");
    filter.annotate(text);
    LOG.debug(text.getAnnotations().toString());
    List<Annotation<StringAttribute>> attributes = text.getAnnotations(StringAttribute.class);
    assertEquals(2, attributes.size());
    int i = 0;
    for (Annotation<StringAttribute> attribute : attributes) {
        if (i == 0) {
            assertEquals("J'ai du ".length(), attribute.getStart());
            assertEquals("J'ai du <skip>skip me</skip>".length(), attribute.getEnd());
            assertEquals("TAG1", attribute.getData().getKey());
            assertEquals("x", attribute.getData().getValue());
        } else if (i == 1) {
            assertEquals("J'ai du <skip>skip me</skip>mal à le croire.".length(), attribute.getStart());
            assertEquals("J'ai du <skip>skip me</skip>mal à le croire.<skip>skip this</skip>".length(), attribute.getEnd());
            assertEquals("TAG1", attribute.getData().getKey());
            assertEquals("x", attribute.getData().getValue());
        }
        i++;
    }
}