Examples with JochreDocument com.joliciel.jochre.doc.JochreDocument used on opensource projects

Example 1 with JochreDocument

use of com.joliciel.jochre.doc.JochreDocument in project jochre by urieli.

the class SegmentationTest method testSegmentation.

@Test
public void testSegmentation() throws Exception {
    // TODO: Note currently this requires high thresholds to work
    // Need to decide if this is valid in general, or only for these samples
    System.setProperty("config.file", "src/test/resources/testHighThresholds.conf");
    ConfigFactory.invalidateCaches();
    Config config = ConfigFactory.load();
    JochreSession jochreSession = new JochreSession(config);
    boolean writePixelsToLog = true;
    for (int imageNumber = 1; imageNumber <= 4; imageNumber++) {
        if (imageNumber != 1)
            continue;
        String imageName = "";
        String suffix = "";
        String text = "";
        String fileName = "";
        String userFileName;
        int rowCount = 2;
        int shapeCountRow1;
        int shapeCountRow2;
        int groupCountRow1;
        int groupCountRow2;
        int groupCountRow3 = 0;
        int shapeCountRow1Group1;
        int shapeCountRow2Group1;
        if (imageNumber == 1) {
            imageName = "MotlPeysiDemKhazns2RowsShort2";
            suffix = "jpg";
            text = "איך געה מיט אייך קיינער אין דער וועלט";
            fileName = "MotlPeysiDemKhazns2RowsShort2.pdf";
            userFileName = "Motl Peysi Dem Khazns";
            shapeCountRow1 = 13;
            shapeCountRow2 = 17;
            groupCountRow1 = 4;
            groupCountRow2 = 4;
            shapeCountRow1Group1 = 3;
            shapeCountRow2Group1 = 6;
        } else if (imageNumber == 2) {
            imageName = "MegileLiderZeresh";
            suffix = "png";
            text = "זרש, די מכשפה, װאָס שעלט ווי אַ מגפה";
            fileName = "MegileLiderManger.pdf";
            userFileName = "Megile Lider";
            shapeCountRow1 = 12;
            shapeCountRow2 = 17;
            groupCountRow1 = 3;
            groupCountRow2 = 5;
            shapeCountRow1Group1 = 4;
            shapeCountRow2Group1 = 4;
        } else if (imageNumber == 3) {
            imageName = "MendeleMoykherSforimVol1_41_0Excerpt";
            suffix = "png";
            text = "ער הייסט יאַנקיל, בעריל,";
            fileName = "MendeleMoykherSforimVol1_41_0.png";
            userFileName = "MendeleMoykherSforimVol1_41_0";
            shapeCountRow1 = 20;
            shapeCountRow2 = 0;
            groupCountRow1 = 4;
            groupCountRow2 = 0;
            shapeCountRow1Group1 = 2;
            shapeCountRow2Group1 = 0;
        } else {
            imageName = "JoinedLetterTest";
            suffix = "png";
            text = "Joined Letter Test";
            fileName = "JoinedLetterTest.png";
            userFileName = "JoinedLetterTest";
            rowCount = 2;
            shapeCountRow1 = 23;
            shapeCountRow2 = 23;
            groupCountRow1 = 4;
            groupCountRow2 = 4;
            groupCountRow3 = 5;
            shapeCountRow1Group1 = 6;
            shapeCountRow2Group1 = 5;
        }
        LOG.debug("######### imageName: " + imageName);
        // String fileName = "data/Zelmenyaners3Words.gif";
        InputStream imageFileStream = getClass().getResourceAsStream("/com/joliciel/jochre/test/resources/" + imageName + "." + suffix);
        assertNotNull(imageFileStream);
        BufferedImage image = ImageIO.read(imageFileStream);
        JochreDocument doc = new JochreDocument(jochreSession);
        doc.setFileName(fileName);
        doc.setName(userFileName);
        JochrePage page = doc.newPage();
        SourceImage sourceImage = page.newJochreImage(image, imageName);
        sourceImage.setWhiteGapFillFactor(5);
        sourceImage.setImageStatus(ImageStatus.AUTO_NEW);
        if (writePixelsToLog) {
            LOG.debug("i012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789");
            for (int y = 0; y < sourceImage.getHeight(); y++) {
                String line = "" + y;
                for (int x = 0; x < sourceImage.getWidth(); x++) {
                    if (sourceImage.isPixelBlack(x, y, sourceImage.getBlackThreshold()))
                        line += "x";
                    else
                        line += "o";
                }
                LOG.debug(line);
            }
        }
        Segmenter segmenter = new Segmenter(sourceImage, jochreSession);
        segmenter.segment();
        if (segmenter.isDrawSegmentation()) {
            BufferedImage segmentedImage = segmenter.getSegmentedImage();
            File tempDir = new File(System.getProperty("java.io.tmpdir"));
            ImageIO.write(segmentedImage, "PNG", new File(tempDir, imageName + "_seg.png"));
        }
        JochreImage jochreImage = sourceImage;
        int i = 0;
        boolean firstShape = true;
        int midPixelFirstShape = 0;
        int midPixelFirstShapeRaw = 0;
        for (Paragraph paragraph : jochreImage.getParagraphs()) {
            for (RowOfShapes row : paragraph.getRows()) {
                int j = 0;
                LOG.debug("============= Row " + i + " ================");
                for (Shape shape : row.getShapes()) {
                    LOG.debug("Shape (" + i + "," + j + "). Left = " + shape.getLeft() + ". Top = " + shape.getTop() + ". Right = " + shape.getRight() + ". Bottom = " + shape.getBottom() + ". Group: " + shape.getGroup().getIndex());
                    if (firstShape) {
                        midPixelFirstShape = shape.getPixel(3, 3);
                        midPixelFirstShapeRaw = shape.getRawPixel(3, 3);
                        firstShape = false;
                    }
                    if (writePixelsToLog) {
                        for (int y = 0; y < shape.getHeight(); y++) {
                            String line = "";
                            if (y == shape.getMeanLine())
                                line += "M";
                            else if (y == shape.getBaseLine())
                                line += "B";
                            else
                                line += y;
                            for (int x = 0; x < shape.getWidth(); x++) {
                                if (shape.isPixelBlack(x, y, sourceImage.getBlackThreshold()))
                                    line += "x";
                                else
                                    line += "o";
                            }
                            LOG.debug(line);
                        }
                    }
                    j++;
                }
                // next shape
                i++;
            }
        // next row
        }
        // next paragraph
        i = 0;
        for (Paragraph paragraph : jochreImage.getParagraphs()) {
            for (RowOfShapes row : paragraph.getRows()) {
                for (GroupOfShapes group : row.getGroups()) {
                    for (Shape shape : group.getShapes()) {
                        if (i < text.length()) {
                            String letter = text.substring(i, i + 1);
                            String nextLetter = "";
                            if (i + 1 < text.length())
                                nextLetter = text.substring(i + 1, i + 2);
                            if (nextLetter.equals("ָֹ") || nextLetter.equals("ַ")) {
                                letter += nextLetter;
                                i++;
                            }
                            LOG.debug("Letter: " + letter);
                            shape.setLetter(letter);
                        }
                        i++;
                    }
                    // to skip the space
                    i++;
                    LOG.debug("Space");
                }
            // next group
            }
        // next row
        }
        // next paragraph
        List<ShapeFeature<?>> features = new ArrayList<ShapeFeature<?>>();
        features.add(new VerticalElongationFeature());
        features.add(new VerticalSizeFeature());
        features.add(new TouchesBaseLineFeature());
        features.add(new TouchesMeanLineFeature());
        features.add(new EmptyCentreFeature());
        i = 0;
        DecimalFormat df = new DecimalFormat("0.00");
        firstShape = true;
        int totalRowCount = 0;
        for (Paragraph paragraph : jochreImage.getParagraphs()) {
            for (RowOfShapes row : paragraph.getRows()) {
                totalRowCount++;
                LOG.debug("============= Row " + i + " ================");
                int j = 0;
                for (GroupOfShapes group : row.getGroups()) {
                    for (Shape shape : group.getShapes()) {
                        LOG.debug("============= Shape (" + i + "," + j + ") ================");
                        LOG.debug("Left = " + shape.getLeft() + ". Top = " + shape.getTop() + ". Right = " + shape.getRight() + ". Bottom = " + shape.getBottom());
                        LOG.debug("Letter " + shape.getLetter());
                        if (firstShape) {
                            LOG.debug("mid pixel: " + midPixelFirstShape);
                            assertEquals(midPixelFirstShape, shape.getPixel(3, 3));
                            LOG.debug("mid pixel raw: " + midPixelFirstShapeRaw);
                            assertEquals(midPixelFirstShapeRaw, shape.getRawPixel(3, 3));
                            firstShape = false;
                        }
                        if (writePixelsToLog) {
                            for (int y = 0; y < shape.getHeight(); y++) {
                                String line = "";
                                if (y == shape.getMeanLine())
                                    line += "M";
                                else if (y == shape.getBaseLine())
                                    line += "B";
                                else
                                    line += y;
                                for (int x = 0; x < shape.getWidth(); x++) {
                                    if (shape.isPixelBlack(x, y, sourceImage.getBlackThreshold()))
                                        line += "x";
                                    else
                                        line += "o";
                                }
                                LOG.debug(line);
                            }
                        }
                        double[][] totals = shape.getBrightnessBySection(5, 5, 1, SectionBrightnessMeasurementMethod.RAW);
                        LOG.debug("Brightness counts");
                        for (int y = 0; y < totals[0].length; y++) {
                            String line = "";
                            for (int x = 0; x < totals.length; x++) {
                                line += df.format(totals[x][y]) + "\t";
                            }
                            LOG.debug(line);
                        }
                        for (ShapeFeature<?> feature : features) {
                            RuntimeEnvironment env = new RuntimeEnvironment();
                            FeatureResult<?> outcome = feature.check(shape, env);
                            LOG.debug(outcome.toString());
                        }
                    }
                    if (i == 0) {
                        if (j == 0)
                            assertEquals(shapeCountRow1Group1, group.getShapes().size());
                    } else if (i == 1) {
                        if (j == 0)
                            assertEquals(shapeCountRow2Group1, group.getShapes().size());
                    }
                    j++;
                }
                if (i == 0)
                    assertEquals(groupCountRow1, row.getGroups().size());
                else if (i == 1)
                    assertEquals(groupCountRow2, row.getGroups().size());
                else if (i == 2)
                    assertEquals(groupCountRow3, row.getGroups().size());
                if (i == 0)
                    assertEquals(shapeCountRow1, row.getShapes().size());
                else if (i == 1)
                    assertEquals(shapeCountRow2, row.getShapes().size());
                i++;
            }
        // next row
        }
        // next paragraph
        assertEquals(rowCount, totalRowCount);
    }
    // next test image
    LOG.debug("************** Finished ***********");
}

Example 2 with JochreDocument

use of com.joliciel.jochre.doc.JochreDocument in project jochre by urieli.

the class TextGetterImplTest method testGetTextFontSizes.

@Test
public void testGetTextFontSizes() {
    final JochreDocument doc = mock(JochreDocument.class);
    final JochrePage page = mock(JochrePage.class);
    final JochreImage jochreImage = mock(JochreImage.class);
    final Paragraph paragraph = mock(Paragraph.class);
    final RowOfShapes row = mock(RowOfShapes.class);
    final GroupOfShapes group1 = mock(GroupOfShapes.class);
    final GroupOfShapes group2 = mock(GroupOfShapes.class);
    final GroupOfShapes group3 = mock(GroupOfShapes.class);
    final GroupOfShapes group4 = mock(GroupOfShapes.class);
    final Shape shape1 = mock(Shape.class);
    final Shape shape2 = mock(Shape.class);
    final Shape shape3 = mock(Shape.class);
    final Shape shape4 = mock(Shape.class);
    final List<Paragraph> paragraphs = new ArrayList<>();
    paragraphs.add(paragraph);
    final List<RowOfShapes> rows = new ArrayList<>();
    rows.add(row);
    when(jochreImage.getPage()).thenReturn(page);
    when(page.getDocument()).thenReturn(doc);
    when(doc.isLeftToRight()).thenReturn(true);
    when(jochreImage.getParagraphs()).thenReturn(paragraphs);
    when(paragraph.getRows()).thenReturn(rows);
    List<GroupOfShapes> groups = new ArrayList<>();
    groups.add(group1);
    groups.add(group2);
    groups.add(group3);
    groups.add(group4);
    when(row.getGroups()).thenReturn(groups);
    List<Shape> shapes1 = new ArrayList<>();
    shapes1.add(shape1);
    when(group1.getShapes()).thenReturn(shapes1);
    when(group1.getXHeight()).thenReturn(10);
    List<Shape> shapes2 = new ArrayList<>();
    shapes2.add(shape2);
    when(group2.getShapes()).thenReturn(shapes2);
    when(group2.getXHeight()).thenReturn(20);
    List<Shape> shapes3 = new ArrayList<>();
    shapes3.add(shape3);
    when(group3.getShapes()).thenReturn(shapes3);
    when(group3.getXHeight()).thenReturn(10);
    List<Shape> shapes4 = new ArrayList<>();
    shapes4.add(shape4);
    when(group4.getShapes()).thenReturn(shapes4);
    when(group4.getXHeight()).thenReturn(5);
    when(shape1.getLetter()).thenReturn("A");
    when(shape1.getXHeight()).thenReturn(10);
    when(shape2.getLetter()).thenReturn("B");
    when(shape2.getXHeight()).thenReturn(20);
    when(shape3.getLetter()).thenReturn("C");
    when(shape3.getXHeight()).thenReturn(10);
    when(shape4.getLetter()).thenReturn("D");
    when(shape4.getXHeight()).thenReturn(5);
    StringWriter writer = new StringWriter();
    TextGetter textGetter = new TextGetter(writer, TextFormat.XHTML);
    textGetter.onImageComplete(jochreImage);
    String result = writer.toString();
    LOG.debug(result);
    assertEquals("<p dir=\"rtl\">A <big>B </big>C <small>D </small></p>", result);
}

Example 3 with JochreDocument

use of com.joliciel.jochre.doc.JochreDocument in project jochre by urieli.

the class TextGetterImplTest method testGetText.

@Test
public void testGetText() {
    final JochreDocument doc = mock(JochreDocument.class);
    final JochrePage page = mock(JochrePage.class);
    final JochreImage jochreImage = mock(JochreImage.class);
    final Paragraph paragraph = mock(Paragraph.class);
    final RowOfShapes row = mock(RowOfShapes.class);
    final GroupOfShapes group = mock(GroupOfShapes.class);
    final Shape shape1 = mock(Shape.class);
    final Shape shape2 = mock(Shape.class);
    final Shape shape3 = mock(Shape.class);
    final Shape shape4 = mock(Shape.class);
    final Shape shape5 = mock(Shape.class);
    final Shape shape6 = mock(Shape.class);
    final Shape shape7 = mock(Shape.class);
    final Shape shape8 = mock(Shape.class);
    final Shape shape9 = mock(Shape.class);
    final List<Paragraph> paragraphs = new ArrayList<>();
    paragraphs.add(paragraph);
    final List<RowOfShapes> rows = new ArrayList<>();
    rows.add(row);
    final List<GroupOfShapes> groups = new ArrayList<>();
    groups.add(group);
    when(jochreImage.getPage()).thenReturn(page);
    when(page.getDocument()).thenReturn(doc);
    when(doc.isLeftToRight()).thenReturn(false);
    when(jochreImage.getParagraphs()).thenReturn(paragraphs);
    when(paragraph.getRows()).thenReturn(rows);
    when(row.getGroups()).thenReturn(groups);
    List<Shape> shapes = new ArrayList<>();
    shapes.add(shape1);
    shapes.add(shape2);
    shapes.add(shape3);
    shapes.add(shape4);
    shapes.add(shape5);
    shapes.add(shape6);
    shapes.add(shape7);
    shapes.add(shape8);
    shapes.add(shape9);
    when(group.getShapes()).thenReturn(shapes);
    when(group.getXHeight()).thenReturn(10);
    when(shape1.getLetter()).thenReturn(",");
    when(shape2.getLetter()).thenReturn(",");
    when(shape3.getLetter()).thenReturn("|אַ");
    when(shape4.getLetter()).thenReturn("אַ|");
    when(shape5.getLetter()).thenReturn("|m");
    when(shape6.getLetter()).thenReturn("m|");
    when(shape7.getLetter()).thenReturn("|ש");
    when(shape8.getLetter()).thenReturn("ע|");
    when(shape9.getLetter()).thenReturn(",");
    StringWriter writer = new StringWriter();
    TextGetter textGetter = new TextGetter(writer, TextFormat.PLAIN);
    textGetter.onImageComplete(jochreImage);
    String result = writer.toString();
    LOG.debug(result);
    assertEquals("„אַm|שע|, \n", result);
}

Example 4 with JochreDocument

use of com.joliciel.jochre.doc.JochreDocument in project jochre by urieli.

the class PdfDocumentProcessor method process.

/**
 * Visit the images and return the JochreDocument containing them.
 */
public JochreDocument process() {
    try {
        LOG.debug("PdfImageVisitorImpl.visitImages");
        if (this.currentMonitor != null)
            currentMonitor.setCurrentAction("imageMonitor.extractingNextImage");
        JochreDocument jochreDocument = this.documentProcessor.onDocumentStart();
        jochreDocument.setTotalPageCount(this.getPageCount());
        for (Entry<String, String> field : this.getFields().entrySet()) {
            jochreDocument.getFields().put(field.getKey(), field.getValue());
        }
        this.visitImages();
        JochrePage finalPage = jochreDocument.getCurrentPage();
        if (finalPage != null) {
            documentProcessor.onPageComplete(finalPage);
        }
        this.documentProcessor.onDocumentComplete(jochreDocument);
        this.documentProcessor.onAnalysisComplete();
        if (this.currentMonitor != null) {
            currentMonitor.setFinished(true);
        }
        return jochreDocument;
    } catch (Exception e) {
        if (this.currentMonitor != null)
            this.currentMonitor.setException(e);
        LOG.error("Failed processing in " + this.getClass().getSimpleName(), e);
        throw new RuntimeException(e);
    }
}

Example 5 with JochreDocument

use of com.joliciel.jochre.doc.JochreDocument in project jochre by urieli.

the class TrainingCorpusShapeSplitterTest method testSplit.

@Test
public void testSplit() throws Exception {
    System.setProperty("config.file", "src/test/resources/testDualCharacters.conf");
    ConfigFactory.invalidateCaches();
    Config config = ConfigFactory.load();
    final JochreSession jochreSession = new JochreSession(config);
    final Shape shape = mock(Shape.class);
    final Shape shape1 = mock(Shape.class);
    final Shape shape2 = mock(Shape.class);
    final Shape shape3 = mock(Shape.class);
    final Shape shape4 = mock(Shape.class);
    final GroupOfShapes group = mock(GroupOfShapes.class);
    final RowOfShapes row = mock(RowOfShapes.class);
    final Paragraph paragraph = mock(Paragraph.class);
    final JochreImage jochreImage = mock(JochreImage.class);
    final JochrePage jochrePage = mock(JochrePage.class);
    final JochreDocument jochreDocument = mock(JochreDocument.class);
    final Iterator<Split> i = (Iterator<Split>) mock(Iterator.class);
    final List<Split> splits = (List<Split>) mock(List.class);
    final Split split1 = mock(Split.class);
    final Split split2 = mock(Split.class);
    final Split split3 = mock(Split.class);
    when(shape.getLetter()).thenReturn("אָבּער");
    when(shape.getLeft()).thenReturn(100);
    when(shape.getRight()).thenReturn(200);
    when(shape.getTop()).thenReturn(100);
    when(shape.getBottom()).thenReturn(200);
    when(shape.getGroup()).thenReturn(group);
    when(shape.getJochreImage()).thenReturn(jochreImage);
    when(group.getRow()).thenReturn(row);
    when(row.getParagraph()).thenReturn(paragraph);
    when(paragraph.getImage()).thenReturn(jochreImage);
    when(jochreImage.getPage()).thenReturn(jochrePage);
    when(jochrePage.getDocument()).thenReturn(jochreDocument);
    when(jochreDocument.getLocale()).thenReturn(jochreSession.getLocale());
    when(shape.getSplits()).thenReturn(splits);
    when(splits.iterator()).thenReturn(i);
    when(i.hasNext()).thenReturn(true).thenReturn(true).thenReturn(true).thenReturn(false);
    when(i.next()).thenReturn(split1).thenReturn(split2).thenReturn(split3);
    when(split1.getPosition()).thenReturn(35);
    when(split2.getPosition()).thenReturn(59);
    when(split3.getPosition()).thenReturn(82);
    when(jochreImage.getShape(100, 100, 135, 200)).thenReturn(shape1);
    when(jochreImage.getShape(136, 100, 159, 200)).thenReturn(shape2);
    when(jochreImage.getShape(160, 100, 182, 200)).thenReturn(shape3);
    when(jochreImage.getShape(183, 100, 200, 200)).thenReturn(shape4);
    LOG.debug(shape.toString());
    LOG.debug(shape.getLetter());
    TrainingCorpusShapeSplitter splitter = new TrainingCorpusShapeSplitter(jochreSession);
    List<ShapeSequence> result = splitter.split(shape);
    ShapeSequence shapeSequence = result.get(0);
    assertEquals(4, shapeSequence.size());
    LOG.debug("Split into: " + shapeSequence.toString());
    verify(shape1).setLetter("אָ");
    verify(shape2).setLetter("בּ");
    verify(shape3).setLetter("ע");
    verify(shape4).setLetter("ר");
}