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

Example 16 with AnnotatedText

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

the class RegexMarkerFilterTest method testApplyWithReplacement.

@Test
public void testApplyWithReplacement() throws Exception {
    RawTextRegexAnnotator filter = new RawTextRegexAnnotator(RawTextMarkType.REPLACE, "<skip>(.*?)</skip>", 0, 1000);
    filter.setReplacement("Skipped:$1");
    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<RawTextReplaceMarker>> replaces = text.getAnnotations(RawTextReplaceMarker.class);
    assertEquals(2, replaces.size());
    int i = 0;
    for (Annotation<RawTextReplaceMarker> replace : replaces) {
        if (i == 0) {
            assertEquals("J'ai du ".length(), replace.getStart());
            assertEquals("J'ai du <skip>skip me</skip>".length(), replace.getEnd());
            assertEquals("Skipped:skip me", replace.getData().getInsertionText());
        } else if (i == 2) {
            assertEquals("J'ai du <skip>skip me</skip>mal à le croire.".length(), replace.getStart());
            assertEquals("J'ai du <skip>skip me</skip>mal à le croire.<skip>skip this</skip>".length(), replace.getEnd());
            assertEquals("Skipped:skip this", replace.getData().getInsertionText());
        }
        i++;
    }
}

Example 17 with AnnotatedText

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

the class RegexMarkerFilterTest method testUnaryOperatorsStop.

@Test
public void testUnaryOperatorsStop() throws Exception {
    RawTextRegexAnnotator filter = new RawTextRegexAnnotator(RawTextMarkType.STOP, "<skip>", 0, 1000);
    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<RawTextMarker>> markers = text.getAnnotations(RawTextMarker.class);
    assertEquals(2, markers.size());
    int i = 0;
    for (Annotation<RawTextMarker> textMarker : markers) {
        if (i == 0) {
            assertEquals(RawTextMarkType.STOP, textMarker.getData().getType());
            assertEquals("J'ai du ".length(), textMarker.getStart());
            assertEquals("J'ai du <skip>".length(), textMarker.getEnd());
        } else if (i == 1) {
            assertEquals(RawTextMarkType.STOP, textMarker.getData().getType());
            assertEquals("J'ai du <skip>skip me</skip>mal à le croire.".length(), textMarker.getStart());
            assertEquals("J'ai du <skip>skip me</skip>mal à le croire.<skip>".length(), textMarker.getEnd());
        }
        i++;
    }
}

Example 18 with AnnotatedText

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

the class RollingTextBlockTest method testGetRawTextBlock.

@Test
public void testGetRawTextBlock() 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];
    RollingTextBlock textBlock = new RollingTextBlock(true, null, sessionId);
    textBlock = textBlock.roll("1 ");
    textBlock = textBlock.roll("2 ");
    textBlock = textBlock.roll("3<skip>skip</skip> 4<sk");
    textBlock = textBlock.roll("ip>skip</skip> 5");
    // the rawTextBlock always contains the last two added sub-blocks
    // so annotations are relative to these sub-blocks
    AnnotatedText rawTextBlock = textBlock.getRawTextBlock();
    List<Annotation<RawTextSkipMarker>> skips = new ArrayList<>();
    skips.add(new Annotation<>("3".length(), "3<skip>skip</skip>".length(), new RawTextSkipMarker("me"), labels));
    skips.add(new Annotation<>("3<skip>skip</skip> 4".length(), "3<skip>skip</skip> 4<skip>skip</skip>".length(), new RawTextSkipMarker("me"), labels));
    rawTextBlock.addAnnotations(skips);
    List<Annotation<RawTextSkipMarker>> sourceSkips = textBlock.getAnnotations(RawTextSkipMarker.class);
    System.out.println(sourceSkips.toString());
    assertEquals(2, sourceSkips.size());
    int i = 0;
    // account blocks 1 and 2
    for (Annotation<RawTextSkipMarker> skip : sourceSkips) {
        if (i == 0) {
            assertEquals("1 2 3".length(), skip.getStart());
            assertEquals("1 2 3<skip>skip</skip>".length(), skip.getEnd());
        } else if (i == 1) {
            assertEquals("1 2 3<skip>skip</skip> 4".length(), skip.getStart());
            assertEquals("1 2 3<skip>skip</skip> 4<skip>skip</skip>".length(), skip.getEnd());
        }
        i++;
    }
}

Example 19 with AnnotatedText

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

the class RollingTextBlockTest method testGetProcessedTextBlock.

@Test
public void testGetProcessedTextBlock() 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];
    RollingTextBlock textBlock = new RollingTextBlock(true, null, sessionId);
    textBlock = textBlock.roll("1 ");
    textBlock = textBlock.roll("2 ");
    textBlock = textBlock.roll("3<skip>skip</skip> 4<sk");
    textBlock = textBlock.roll("ip>skip</skip> five");
    // the rawTextBlock always contains the last two added sub-blocks
    // so annotations are relative to these sub-blocks
    AnnotatedText rawTextBlock = textBlock.getRawTextBlock();
    List<Annotation<RawTextSkipMarker>> skips = new ArrayList<>();
    skips.add(new Annotation<>("3".length(), "3<skip>skip</skip>".length(), new RawTextSkipMarker("me"), labels));
    skips.add(new Annotation<>("3<skip>skip</skip> 4".length(), "3<skip>skip</skip> 4<skip>skip</skip>".length(), new RawTextSkipMarker("me"), labels));
    rawTextBlock.addAnnotations(skips);
    textBlock = textBlock.roll(" 6");
    rawTextBlock = textBlock.getRawTextBlock();
    List<Annotation<RawTextReplaceMarker>> replaces = new ArrayList<>();
    replaces.add(new Annotation<>("ip>skip</skip> ".length(), "ip>skip</skip> five".length(), new RawTextReplaceMarker("me", "5"), labels));
    rawTextBlock.addAnnotations(replaces);
    AnnotatedText processedTextBlock = textBlock.getProcessedText();
    // the processed text always concerns sub-blocks 1, 2 and 3
    // at this point, sub-block 1 has already been flushed
    assertEquals("2 3 4 5", processedTextBlock.getText());
}

Example 20 with AnnotatedText

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

the class RollingTextBlockTest 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];
    RollingTextBlock textBlock = new RollingTextBlock(true, null, sessionId);
    textBlock = textBlock.roll("Sentence 1<sent/>Sentence 2. Sentence");
    textBlock = textBlock.roll(" 3.");
    // the rawTextBlock always contains the last two added sub-blocks
    // so annotations are relative to these sub-blocks
    AnnotatedText rawTextBlock = textBlock.getRawTextBlock();
    List<Annotation<RawTextSentenceBreakMarker>> sentenceBreaks = new ArrayList<>();
    // we add a sentence break annotation (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)");
    sentenceBreaks.add(new Annotation<>("".length(), "Sentence 1<sent/>".length(), new RawTextSentenceBreakMarker("me"), labels));
    rawTextBlock.addAnnotations(sentenceBreaks);
    List<Annotation<RawTextSkipMarker>> skips = new ArrayList<>();
    skips.add(new Annotation<>("Sentence 1".length(), "Sentence 1<sent/>".length(), new RawTextSkipMarker("me"), labels));
    rawTextBlock.addAnnotations(skips);
    System.out.println("textBlock text: " + textBlock.getText());
    System.out.println("textBlock annotations: " + textBlock.getAnnotations().toString());
    textBlock = textBlock.roll(" Sentence 4.");
    AnnotatedText processedTextBlock = textBlock.getProcessedText();
    assertEquals("Sentence 1 Sentence 2. Sentence 3.", 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));
    processedTextBlock.addAnnotations(sentenceBoundaries);
    List<Sentence> sentences = textBlock.getDetectedSentences();
    System.out.println("sentences: " + sentences.toString());
    assertEquals(2, 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));
    System.out.println("textBlock text: " + textBlock.getText());
    System.out.println("textBlock annotations: " + textBlock.getAnnotations().toString());
    textBlock = textBlock.roll("");
    // we have now rolled all text up until sentence 4 into the processed
    // area
    processedTextBlock = textBlock.getProcessedText();
    assertEquals("Sentence 1 Sentence 2. Sentence 3. Sentence 4.", processedTextBlock.getText());
    // add a sentence boundary for "Sentence 3"
    System.out.println("add a sentence boundary for \"Sentence 3\", this time inside the analysis range");
    sentenceBoundaries = new ArrayList<>();
    sentenceBoundaries.add(new Annotation<>("Sentence 1 Sentence 2.".length(), "Sentence 1 Sentence 2. Sentence 3.".length(), new SentenceBoundary(), labels));
    processedTextBlock.addAnnotations(sentenceBoundaries);
    sentences = textBlock.getDetectedSentences();
    System.out.println("sentences: " + sentences.toString());
    assertEquals(1, sentences.size());
    assertEquals("Sentence 3.", sentences.get(0).getText());
    assertEquals("Sentence 1<sent/>Sentence 2. ".length(), sentences.get(0).getOriginalIndex(0));
    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);
    System.out.println(sentenceBoundaries.toString());
    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());
    // roll in a final empty block - we now have an empty block at block 3,
    // so that any leftover in block 2 should be marked as complete
    // since sentences never overlap empty blocks.
    textBlock = textBlock.roll("");
    sentences = textBlock.getDetectedSentences();
    System.out.println("sentences: " + sentences.toString());
    assertEquals(1, sentences.size());
    assertEquals("Sentence 4.", sentences.get(0).getText());
    assertEquals("Sentence 1<sent/>Sentence 2. Sentence 3. ".length(), sentences.get(0).getOriginalIndex(0));
    // note: at this point the initial two blocks have been rolled out
    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(" 3. Sentence 4.", textBlock.getText());
    sentenceBoundaries = textBlock.getAnnotations(SentenceBoundary.class);
    assertEquals(1, sentenceBoundaries.size());
    assertEquals("".length(), sentenceBoundaries.get(0).getStart());
    assertEquals(" 3.".length(), sentenceBoundaries.get(0).getEnd());
    // test that sentence annotations get added to the original raw text
    Sentence sentence4 = sentences.get(0);
    List<Annotation<String>> annotations = new ArrayList<>();
    annotations.add(new Annotation<String>("Sentence ".length(), "Sentence 4".length(), "four", labels));
    sentence4.addAnnotations(annotations);
    System.out.println("textBlock text: " + textBlock.getText());
    System.out.println("textBlock annotations: " + textBlock.getAnnotations().toString());
    annotations = textBlock.getAnnotations(String.class);
    assertEquals(1, annotations.size());
    assertEquals(" 3. Sentence ".length(), annotations.get(0).getStart());
    assertEquals(" 3. Sentence 4".length(), annotations.get(0).getEnd());
    textBlock.getProcessedText();
}