Marmelator/layout-parser
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 289 Wiki Activity

Merge branch 'RED-8800' into 'main'

Browse Source 
RED-8800: adjust coordinates in BE to ignore cropbox

See merge request fforesight/layout-parser!179
This commit is contained in:
Kilian Schüttler 2024-07-15 17:45:13 +02:00
commit ea18d3d307
3 changed files with 68 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/parsing/LegacyPDFStreamEngine.java
View File

@ -82,6 +82,7 @@ public class LegacyPDFStreamEngine extends PDFStreamEngine {
private int pageRotation;
private PDRectangle pageSize;
private Matrix translateMatrix;
private final GlyphList glyphList;
private final Map<COSDictionary, Float> fontHeightMap = new WeakHashMap<COSDictionary, Float>();
@ -133,6 +134,12 @@ public class LegacyPDFStreamEngine extends PDFStreamEngine {
this.pageRotation = page.getRotation();
this.pageSize = page.getCropBox();
if (pageSize.getLowerLeftX() == 0 && pageSize.getLowerLeftY() == 0) {
translateMatrix = null;
} else {
// translation matrix for cropbox
translateMatrix = Matrix.getTranslateInstance(-pageSize.getLowerLeftX(), -pageSize.getLowerLeftY());
}
super.processPage(page);
}
@ -257,13 +264,22 @@ public class LegacyPDFStreamEngine extends PDFStreamEngine {
return;
}
}
// adjust for cropbox if needed
Matrix translatedTextRenderingMatrix;
if (translateMatrix == null) {
translatedTextRenderingMatrix = textRenderingMatrix;
} else {
translatedTextRenderingMatrix = Matrix.concatenate(translateMatrix, textRenderingMatrix);
nextX -= pageSize.getLowerLeftX();
nextY -= pageSize.getLowerLeftY();
}
// This is a hack for unicode letter with 2 chars e.g. RA see unicodeProblem.pdf
if (unicodeMapping.length() == 2) {
processTextPosition(new TextPosition(pageRotation,
pageSize.getWidth(),
pageSize.getHeight(),
textRenderingMatrix,
translatedTextRenderingMatrix,
nextX,
nextY,
Math.abs(dyDisplay),
@ -277,7 +293,7 @@ public class LegacyPDFStreamEngine extends PDFStreamEngine {
processTextPosition(new TextPosition(pageRotation,
pageSize.getWidth(),
pageSize.getHeight(),
textRenderingMatrix,
translatedTextRenderingMatrix,
nextX,
nextY,
Math.abs(dyDisplay),
@ -293,7 +309,7 @@ public class LegacyPDFStreamEngine extends PDFStreamEngine {
processTextPosition(new TextPosition(pageRotation,
pageSize.getWidth(),
pageSize.getHeight(),
textRenderingMatrix,
translatedTextRenderingMatrix,
nextX,
nextY,
Math.abs(dyDisplay),

4
layoutparser-service/layoutparser-service-server/src/test/java/com/knecon/fforesight/service/layoutparser/server/LayoutparserEnd2EndTest.java
View File

@ -30,11 +30,11 @@ public class LayoutparserEnd2EndTest extends AbstractTest {
@Autowired
private LayoutParsingPipeline layoutParsingPipeline;
@Disabled
@Test
@Disabled
public void testLayoutParserEndToEnd() {
String filePath = "files/Minimal Examples/RotateTextWithRulingsTestFile.pdf";
String filePath = "/tmp/OCR_TEST/10.SYN524464 FS (A16148C) - Absorção cutânea.pdf/document.pdf";
runForFile(filePath);
}

64
layoutparser-service/viewer-doc-processor/src/main/java/com/knecon/fforesight/service/viewerdoc/service/VisualizationWriter.java
View File

@ -53,7 +53,7 @@ public class VisualizationWriter {
begin(page);
AffineTransform textDeRotationMatrix = getTextDeRotationTransform(page);
AffineTransform pageTransformation = getTextDeRotationTransform(page);
AffineTransform pageTransformation = getCropboxAdjustment(page);
for (LayerGroup layerGroup : layerGroups) {
@ -68,7 +68,7 @@ public class VisualizationWriter {
Element escape = builder.createGroupBegin();
writer.writeElement(escape);
writeVisualizations(pageNumber, layerGroup, textDeRotationMatrix);
writeVisualizations(pageNumber, layerGroup, textDeRotationMatrix, pageTransformation);
Element escapeEnd = builder.createGroupEnd();
writer.writeElement(escapeEnd);
@ -87,7 +87,14 @@ public class VisualizationWriter {
}
private void writeVisualizations(int pageNumber, LayerGroup layerGroup, AffineTransform textDeRotationMatrix) throws PDFNetException {
@SneakyThrows
private AffineTransform getCropboxAdjustment(Page page) {
return new AffineTransform(1, 0, 0, 1, page.getCropBox().getX1(), page.getCropBox().getY1());
}
private void writeVisualizations(int pageNumber, LayerGroup layerGroup, AffineTransform textDeRotationMatrix, AffineTransform pageTransformation) throws PDFNetException {
for (Visualizations visualization : layerGroup.getVisualizations()) {
@ -108,8 +115,7 @@ public class VisualizationWriter {
Element escape = builder.createGroupBegin();
writer.writeElement(escape);
writeVisualization(visualizationsOnPage, textDeRotationMatrix);
writeVisualization(visualizationsOnPage, textDeRotationMatrix, pageTransformation);
Element escapeEnd = builder.createGroupEnd();
writer.writeElement(escapeEnd);
@ -139,7 +145,7 @@ public class VisualizationWriter {
@SneakyThrows
private void writeVisualization(VisualizationsOnPage visualizationsOnPage, AffineTransform textDeRotationMatrix) {
private void writeVisualization(VisualizationsOnPage visualizationsOnPage, AffineTransform textDeRotationMatrix, AffineTransform pageTransformation) {
if (visualizationsOnPage.isMakePathsInvisible()) {
Element rect = builder.createRect(0, 0, 0, 0);
@ -149,28 +155,28 @@ public class VisualizationWriter {
for (ColoredLine coloredLine : visualizationsOnPage.getColoredLines()) {
drawColoredLine(coloredLine);
drawColoredLine(coloredLine, pageTransformation);
}
for (ColoredRectangle coloredRectangle : visualizationsOnPage.getColoredRectangles()) {
drawColoredRectangle(coloredRectangle);
drawColoredRectangle(coloredRectangle, pageTransformation);
}
for (FilledRectangle filledRectangle : visualizationsOnPage.getFilledRectangles()) {
drawFilledRectangle(filledRectangle);
drawFilledRectangle(filledRectangle, pageTransformation);
}
for (PlacedText placedText : visualizationsOnPage.getPlacedTexts()) {
writePlacedText(textDeRotationMatrix, placedText);
writePlacedText(textDeRotationMatrix, placedText, pageTransformation);
}
}
private void writePlacedText(AffineTransform textDeRotationMatrix, PlacedText placedText) throws PDFNetException {
private void writePlacedText(AffineTransform textDeRotationMatrix, PlacedText placedText, AffineTransform pageTransformation) throws PDFNetException {
float[] rgbComponents = placedText.color().getRGBColorComponents(null);
Font font = fontMap.get(placedText.font());
@ -180,7 +186,7 @@ public class VisualizationWriter {
text.getGState().setFillColor(color);
}
try (Matrix2D textMatrix = getTextMatrix(placedText, textDeRotationMatrix)) {
try (Matrix2D textMatrix = getTextMatrix(placedText, textDeRotationMatrix, pageTransformation)) {
text.setTextMatrix(textMatrix);
}
@ -217,11 +223,13 @@ public class VisualizationWriter {
}
private void drawFilledRectangle(FilledRectangle filledRectangle) throws PDFNetException {
private void drawFilledRectangle(FilledRectangle filledRectangle, AffineTransform pageTransformation) throws PDFNetException {
float[] rgbComponents = filledRectangle.color().getRGBColorComponents(null);
Rectangle2D r = filledRectangle.rectangle2D();
r = transformRect(r, pageTransformation);
Element rect = builder.createRect(r.getX(), r.getY(), r.getWidth(), r.getHeight());
rect.setPathFill(true);
@ -237,11 +245,13 @@ public class VisualizationWriter {
}
private void drawColoredRectangle(ColoredRectangle coloredRectangle) throws PDFNetException {
private void drawColoredRectangle(ColoredRectangle coloredRectangle, AffineTransform pageTransformation) throws PDFNetException {
float[] rgbComponents = coloredRectangle.color().getRGBColorComponents(null);
Rectangle2D r = coloredRectangle.rectangle2D();
r = transformRect(r, pageTransformation);
Element rect = builder.createRect(r.getX(), r.getY(), r.getWidth(), r.getHeight());
rect.setPathStroke(true);
@ -257,11 +267,19 @@ public class VisualizationWriter {
}
private void drawColoredLine(ColoredLine coloredLine) throws PDFNetException {
private Rectangle2D transformRect(Rectangle2D r, AffineTransform pageTransformation) {
return pageTransformation.createTransformedShape(r).getBounds2D();
}
private void drawColoredLine(ColoredLine coloredLine, AffineTransform pageTransformation) throws PDFNetException {
float[] rgbComponents = coloredLine.color().getRGBColorComponents(null);
Line2D l = coloredLine.line();
l = transformLine(pageTransformation, l);
builder.pathBegin();
builder.moveTo(l.getX1(), l.getY1());
builder.lineTo(l.getX2(), l.getY2());
@ -279,8 +297,15 @@ public class VisualizationWriter {
}
private static Line2D transformLine(AffineTransform pageTransformation, Line2D line) {
return new Line2D.Double(pageTransformation.transform(line.getP1(), null), pageTransformation.transform(line.getP2(), null));
}
@SneakyThrows
private static Matrix2D getTextMatrix(PlacedText placedText, AffineTransform textDeRotationMatrix) {
@SuppressWarnings("PMD.CloseResource")
private static Matrix2D getTextMatrix(PlacedText placedText, AffineTransform textDeRotationMatrix, AffineTransform pageTransformation) {
Matrix2D textMatrix;
if (placedText.textMatrix().isEmpty()) {
@ -289,7 +314,12 @@ public class VisualizationWriter {
var matrix = placedText.textMatrix().get();
textMatrix = toMatrix2D(matrix);
}
return textMatrix;
try (var pageMatrix = toMatrix2D(pageTransformation)) {
return pageMatrix.multiply(textMatrix);
} finally {
textMatrix.close();
}
}