Examples with Paragraph com.joliciel.jochre.graphics.Paragraph used on opensource projects

Example 11 with Paragraph

use of com.joliciel.jochre.graphics.Paragraph in project jochre by urieli.

the class ImageController method save.

void save() {
    try {
        Comboitem selectedItem = cmbStatus.getSelectedItem();
        ImageStatus imageStatus = ImageStatus.forId((Integer) selectedItem.getValue());
        currentImage.setImageStatus(imageStatus);
        if (currentUser.getRole().equals(UserRole.ADMIN)) {
            User owner = (User) cmbOwner.getSelectedItem().getValue();
            currentImage.setOwner(owner);
        }
        GraphicsDao graphicsDao = GraphicsDao.getInstance(jochreSession);
        graphicsDao.saveJochreImage(currentImage);
        for (Paragraph paragraph : currentImage.getParagraphs()) {
            LOG.trace("Paragraph " + paragraph.getIndex() + ", " + paragraph.getRows().size() + " rows");
            for (RowOfShapes row : paragraph.getRows()) {
                List<List<String>> letterGroups = this.getLetterGroups(row);
                LOG.trace("Row " + row.getIndex() + ", " + row.getGroups().size() + " groups, " + letterGroups.size() + " letter groups");
                Iterator<List<String>> iLetterGroups = letterGroups.iterator();
                for (GroupOfShapes group : row.getGroups()) {
                    LOG.trace("Group " + group.getIndex() + " text : " + group.getWord());
                    boolean hasChange = false;
                    List<String> letters = null;
                    if (iLetterGroups.hasNext())
                        letters = iLetterGroups.next();
                    else
                        letters = new ArrayList<String>();
                    LOG.trace("Found " + letters.size() + " letters in text");
                    Iterator<String> iLetters = letters.iterator();
                    for (Shape shape : group.getShapes()) {
                        String currentLetter = shape.getLetter();
                        if (currentLetter == null)
                            currentLetter = "";
                        String newLetter = "";
                        if (iLetters.hasNext())
                            newLetter = iLetters.next();
                        if (newLetter.startsWith("[") && newLetter.endsWith("]")) {
                            newLetter = newLetter.substring(1, newLetter.length() - 1);
                        }
                        LOG.trace("currentLetter:  " + currentLetter + ", newLetter: " + newLetter);
                        if (!currentLetter.equals(newLetter)) {
                            LOG.trace("newLetter: " + newLetter);
                            shape.setLetter(newLetter);
                            shape.save();
                            hasChange = true;
                        }
                    }
                    if (hasChange)
                        LOG.trace("Group text after : " + group.getWord());
                }
            // next group
            }
        // next row
        }
        // next paragraph
        Messagebox.show(Labels.getLabel("button.saveComplete"));
    } catch (Exception e) {
        LOG.error("Failure in save", e);
        throw new RuntimeException(e);
    }
}

Example 12 with Paragraph

use of com.joliciel.jochre.graphics.Paragraph in project jochre by urieli.

the class UnknownWordListWriter method onImageComplete.

@Override
public void onImageComplete(JochreImage image) {
    try {
        for (Paragraph paragraph : image.getParagraphs()) {
            if (!paragraph.isJunk()) {
                for (RowOfShapes row : paragraph.getRows()) {
                    for (GroupOfShapes group : row.getGroups()) {
                        if (group.getBestLetterSequence() != null) {
                            for (LetterSequence subsequence : group.getBestLetterSequence().getSubsequences()) {
                                for (CountedOutcome<String> wordFrequency : subsequence.getWordFrequencies()) {
                                    if (wordFrequency.getCount() == 0) {
                                        writer.write(wordFrequency.getOutcome() + "\n");
                                        writer.flush();
                                    }
                                }
                            }
                        }
                    }
                }
            }
        }
    } catch (IOException e) {
        LOG.error("Failed to write to UnknownWordListWriter", e);
        throw new RuntimeException(e);
    }
}

Example 13 with Paragraph

use of com.joliciel.jochre.graphics.Paragraph in project jochre by urieli.

the class CorpusLexiconBuilder method buildLexicon.

/**
 * Build a lexicon from the training corpus.
 */
public TextFileLexicon buildLexicon() {
    TextFileLexicon lexicon = new TextFileLexicon();
    JochreCorpusImageReader imageReader = new JochreCorpusImageReader(jochreSession);
    imageReader.setSelectionCriteria(criteria);
    String wordText = "";
    while (imageReader.hasNext()) {
        JochreImage image = imageReader.next();
        for (Paragraph paragraph : image.getParagraphs()) {
            // rows ending in dashes can only be held-over within the same
            // paragraph.
            // to avoid strange things like a page number getting added to
            // the word,
            // if the dash is on the last row of the page.
            String holdoverWord = null;
            for (RowOfShapes row : paragraph.getRows()) {
                for (GroupOfShapes group : row.getGroups()) {
                    if (group.isBrokenWord())
                        continue;
                    wordText = "";
                    for (Shape shape : group.getShapes()) {
                        if (shape.getLetter() != null)
                            wordText += shape.getLetter();
                    }
                    if (wordText.length() == 0) {
                        lexicon.incrementEntry("");
                        continue;
                    }
                    List<String> words = jochreSession.getLinguistics().splitText(wordText);
                    int i = 0;
                    for (String word : words) {
                        if (i == 0) {
                            // first word
                            if (holdoverWord != null && holdoverWord.length() > 0) {
                                word = holdoverWord + word;
                                holdoverWord = null;
                            }
                        }
                        if (i == words.size() - 1) {
                            // last word
                            if (group.getIndex() == row.getGroups().size() - 1 && word.endsWith("-")) {
                                // a dash at the end of a line
                                if (group.isHardHyphen())
                                    holdoverWord = word;
                                else
                                    holdoverWord = word.substring(0, word.length() - 1);
                                word = "";
                            }
                        }
                        lexicon.incrementEntry(word);
                        i++;
                    }
                }
            }
        }
    }
    return lexicon;
}

Example 14 with Paragraph

use of com.joliciel.jochre.graphics.Paragraph in project jochre by urieli.

the class SegmentationTest method testAlsacien1.

/**
 * Segmentation errors reported for Alsacien.
 *
 * @throws Exception
 */
@Test
public void testAlsacien1(@Mocked final JochrePage jochrePage, @Mocked final JochreDocument jochreDoc) throws Exception {
    Map<String, Object> configMap = new HashMap<>();
    configMap.put("jochre.locale", "de");
    Config config = ConfigFactory.parseMap(configMap).withFallback(ConfigFactory.load());
    JochreSession jochreSession = new JochreSession(config);
    new Expectations() {

        {
            jochrePage.getDocument();
            result = jochreDoc;
            minTimes = 0;
            jochreDoc.isLeftToRight();
            result = true;
            minTimes = 0;
        }
    };
    String imageName = "Alsacien1.jpg";
    LOG.debug(imageName);
    InputStream imageFileStream = getClass().getResourceAsStream("/com/joliciel/jochre/segmentation/" + imageName);
    assertNotNull(imageFileStream);
    BufferedImage image = ImageIO.read(imageFileStream);
    SourceImage sourceImage = new SourceImage(jochrePage, "", image, jochreSession);
    Segmenter segmenter = new Segmenter(sourceImage, jochreSession);
    segmenter.segment();
    List<Rectangle> textPars = new ArrayList<>();
    Rectangle textPar1 = new Rectangle(715, 517, 462, 115);
    // TODO: for now it's splitting this paragraph by row, since it's assuming
    // paragraphs cannot be
    // both outdented and indented on the same page
    // Rectangle textPar2 = new Rectangle(50, 666, 1798, 1039);
    Rectangle textPar3 = new Rectangle(55, 1837, 1777, 335);
    Rectangle textPar4 = new Rectangle(50, 2211, 1765, 154);
    Rectangle textPar5 = new Rectangle(44, 2404, 1782, 511);
    Rectangle textPar6 = new Rectangle(50, 2948, 1776, 154);
    Rectangle textPar7 = new Rectangle(50, 3135, 1770, 77);
    // title paragraph
    textPars.add(textPar1);
    // textPars.add(textPar2);
    textPars.add(textPar3);
    textPars.add(textPar4);
    textPars.add(textPar5);
    textPars.add(textPar6);
    textPars.add(textPar7);
    int i = 0;
    int j = 0;
    List<Paragraph> textParagraphs = new ArrayList<>();
    for (Paragraph par : sourceImage.getParagraphs()) {
        Rectangle real = new Rectangle(par.getLeft(), par.getTop(), par.getRight() - par.getLeft(), par.getBottom() - par.getTop());
        Rectangle expected = textPars.get(i);
        Rectangle intersection = expected.intersection(real);
        double realArea = real.width * real.height;
        double expectedArea = expected.width * expected.height;
        double intersectionArea = intersection.width * intersection.height;
        double realRatio = intersectionArea / realArea;
        double expectedRatio = intersectionArea / expectedArea;
        LOG.debug("Paragraph " + j + ": " + par.toString());
        LOG.debug("realRatio: " + realRatio);
        LOG.debug("expectedRatio: " + expectedRatio);
        if (realRatio > 0.8 && expectedRatio > 0.8) {
            LOG.debug("Found");
            textParagraphs.add(par);
            i++;
        }
        j++;
    }
    assertEquals(textPars.size(), textParagraphs.size());
    int[] rowCounts = new int[] { 1, 4, 2, 6, 2, 1 };
    int[] wordCountsFirstRow = new int[] { 2, 0, 0, 0, 0, 0, 0 };
    for (i = 0; i < textParagraphs.size(); i++) {
        assertEquals("row count " + i, rowCounts[i], textParagraphs.get(i).getRows().size());
        RowOfShapes row = textParagraphs.get(i).getRows().get(0);
        if (wordCountsFirstRow[i] > 0)
            assertEquals("word count " + i, wordCountsFirstRow[i], row.getGroups().size());
    }
}

Example 15 with Paragraph

use of com.joliciel.jochre.graphics.Paragraph in project jochre by urieli.

the class SegmentationTest method testAlsacienPlay3.

/**
 * Segmentation errors reported for Alsacien play - challenging because of the
 * unusual indentation.
 *
 * @throws Exception
 */
@Test
public void testAlsacienPlay3(@Mocked final JochrePage jochrePage, @Mocked final JochreDocument jochreDoc) throws Exception {
    Map<String, Object> configMap = new HashMap<>();
    configMap.put("jochre.locale", "de");
    Config config = ConfigFactory.parseMap(configMap).withFallback(ConfigFactory.load());
    JochreSession jochreSession = new JochreSession(config);
    new Expectations() {

        {
            jochrePage.getDocument();
            result = jochreDoc;
            minTimes = 0;
            jochreDoc.isLeftToRight();
            result = true;
            minTimes = 0;
        }
    };
    String imageName = "AlsacienPlay3.jpg";
    LOG.debug(imageName);
    InputStream imageFileStream = getClass().getResourceAsStream("/com/joliciel/jochre/segmentation/" + imageName);
    assertNotNull(imageFileStream);
    BufferedImage image = ImageIO.read(imageFileStream);
    SourceImage sourceImage = new SourceImage(jochrePage, "", image, jochreSession);
    Segmenter segmenter = new Segmenter(sourceImage, jochreSession);
    segmenter.segment();
    List<Rectangle> textPars = new ArrayList<>();
    Rectangle textPar1 = new Rectangle(712, 532, 556, 52);
    Rectangle textPar2 = new Rectangle(324, 600, 1324, 128);
    Rectangle textPar3 = new Rectangle(680, 730, 592, 50);
    Rectangle textPar4 = new Rectangle(404, 808, 684, 48);
    // title paragraph
    textPars.add(textPar1);
    textPars.add(textPar2);
    textPars.add(textPar3);
    textPars.add(textPar4);
    int i = 0;
    int j = 0;
    List<Paragraph> textParagraphs = new ArrayList<>();
    for (Paragraph par : sourceImage.getParagraphs()) {
        Rectangle real = new Rectangle(par.getLeft(), par.getTop(), par.getRight() - par.getLeft(), par.getBottom() - par.getTop());
        Rectangle expected = textPars.get(i);
        Rectangle intersection = expected.intersection(real);
        double realArea = real.width * real.height;
        double expectedArea = expected.width * expected.height;
        double intersectionArea = intersection.width * intersection.height;
        double realRatio = intersectionArea / realArea;
        double expectedRatio = intersectionArea / expectedArea;
        LOG.debug("Paragraph " + j + ": " + par.toString());
        LOG.debug("realRatio: " + realRatio);
        LOG.debug("expectedRatio: " + expectedRatio);
        if (realRatio > 0.8 && expectedRatio > 0.8) {
            LOG.debug("Found");
            textParagraphs.add(par);
            i++;
            if (i >= textPars.size())
                break;
        }
        j++;
    }
    assertEquals(textPars.size(), textParagraphs.size());
    int[] rowCounts = new int[] { 1, 2, 1, 1 };
    // TODO: words in "spaced" rows (uses spacing to emphasize instead of bold
    // or italics) get split
    // should try to detect multiple single letter words
    int[] wordCountsFirstRow = new int[] { 0, 10, 0, 5 };
    for (i = 0; i < textParagraphs.size(); i++) {
        assertEquals("row count " + i, rowCounts[i], textParagraphs.get(i).getRows().size());
        RowOfShapes row = textParagraphs.get(i).getRows().get(0);
        if (wordCountsFirstRow[i] > 0)
            assertEquals("word count " + i, wordCountsFirstRow[i], row.getGroups().size());
    }
}