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

Red 8800 bp

Browse Source
This commit is contained in:
Kilian Schüttler 2024-07-15 17:58:59 +02:00
parent 44277572ab
commit d4500b879b
2 changed files with 54 additions and 12 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),

44
layoutparser-service/viewer-doc-processor/src/main/java/com/knecon/fforesight/service/viewerdoc/service/ViewerDocumentService.java
View File

@ -1,6 +1,7 @@
package com.knecon.fforesight.service.viewerdoc.service;
import java.awt.geom.AffineTransform;
import java.awt.geom.Line2D;
import java.awt.geom.Rectangle2D;
import java.io.File;
import java.io.FileOutputStream;
@ -84,6 +85,7 @@ public class ViewerDocumentService {
pdPage.setContents(ContentStreamUtility.removeLayerFromContentStreams(allLayers, classifiers));
AffineTransform textDeRotationMatrix = getTextDeRotationTransform(pdPage);
AffineTransform pageTransformationMatrix = getPageTransformationMatrix(pdPage);
if (!ContentStreamClassifier.areAllContentStreamsEscaped(classifiers)) {
// We need to save the graphics state before, such that our appended content cannot be affected by previous content streams with side effects,
@ -106,7 +108,11 @@ public class ViewerDocumentService {
contentStream.saveGraphicsState();
drawVisualizationsToContentStream(pdDocument, visualization.getVisualizationsOnPages().get(pageNumber), contentStream, textDeRotationMatrix);
drawVisualizationsToContentStream(pdDocument,
visualization.getVisualizationsOnPages().get(pageNumber),
contentStream,
textDeRotationMatrix,
pageTransformationMatrix);
contentStream.restoreGraphicsState();
@ -133,6 +139,12 @@ public class ViewerDocumentService {
}
private AffineTransform getPageTransformationMatrix(PDPage pdPage) {
return new AffineTransform(1, 0, 0, 1, pdPage.getCropBox().getLowerLeftX(), pdPage.getCropBox().getLowerLeftY());
}
private static Map<ContentStreams.Identifier, PDOptionalContentGroup> addLayersToDocument(List<Visualizations> visualizations, PDDocument pdDocument) {
Map<ContentStreams.Identifier, PDOptionalContentGroup> optionalContentGroupMap = new HashMap<>();
@ -147,7 +159,8 @@ public class ViewerDocumentService {
private static void drawVisualizationsToContentStream(PDDocument pdDocument,
VisualizationsOnPage visualizationsOnPage,
PDPageContentStream contentStream,
AffineTransform textDeRotationMatrix) throws IOException {
AffineTransform textDeRotationMatrix,
AffineTransform pageTransformationMatrix) throws IOException {
if (visualizationsOnPage.isMakePathsInvisible()) {
contentStream.addRect(0, 0, 1, 1);
@ -155,17 +168,18 @@ public class ViewerDocumentService {
}
for (ColoredLine coloredLine : visualizationsOnPage.getColoredLines()) {
Line2D line = transformLine(coloredLine.line(), pageTransformationMatrix);
contentStream.setLineWidth(coloredLine.lineWidth());
contentStream.setStrokingColor(coloredLine.color());
contentStream.moveTo((float) coloredLine.line().getX1(), (float) coloredLine.line().getY1());
contentStream.lineTo((float) coloredLine.line().getX2(), (float) coloredLine.line().getY2());
contentStream.moveTo((float) line.getX1(), (float) line.getY1());
contentStream.lineTo((float) line.getX2(), (float) line.getY2());
contentStream.stroke();
}
for (ColoredRectangle coloredRectangle : visualizationsOnPage.getColoredRectangles()) {
contentStream.setLineWidth(coloredRectangle.lineWidth());
contentStream.setStrokingColor(coloredRectangle.color());
Rectangle2D r = coloredRectangle.rectangle2D();
Rectangle2D r = transformRect(coloredRectangle.rectangle2D(), pageTransformationMatrix);
contentStream.addRect((float) r.getX(), (float) r.getY(), (float) r.getWidth(), (float) r.getHeight());
contentStream.stroke();
}
@ -175,7 +189,7 @@ public class ViewerDocumentService {
PDExtendedGraphicsState graphicsState = new PDExtendedGraphicsState();
graphicsState.setNonStrokingAlphaConstant(filledRectangle.alpha());
contentStream.setGraphicsStateParameters(graphicsState);
Rectangle2D r = filledRectangle.rectangle2D();
Rectangle2D r = transformRect(filledRectangle.rectangle2D(), pageTransformationMatrix);
contentStream.addRect((float) r.getX(), (float) r.getY(), (float) r.getWidth(), (float) r.getHeight());
contentStream.fill();
}
@ -190,7 +204,7 @@ public class ViewerDocumentService {
} else {
contentStream.setRenderingMode(RenderingMode.FILL);
}
Matrix textMatrix = getTextMatrix(placedText, textDeRotationMatrix);
Matrix textMatrix = getTextMatrix(placedText, textDeRotationMatrix, pageTransformationMatrix);
contentStream.setTextMatrix(textMatrix);
contentStream.showText(placedText.text());
contentStream.endText();
@ -223,7 +237,7 @@ public class ViewerDocumentService {
}
private static Matrix getTextMatrix(PlacedText placedText, AffineTransform textDeRotationMatrix) {
private static Matrix getTextMatrix(PlacedText placedText, AffineTransform textDeRotationMatrix, AffineTransform pageTransformationMatrix) {
Matrix textMatrix;
if (placedText.textMatrix().isEmpty()) {
@ -236,7 +250,7 @@ public class ViewerDocumentService {
} else {
textMatrix = placedText.textMatrix().get();
}
return textMatrix;
return new Matrix(pageTransformationMatrix).multiply(textMatrix);
}
@ -313,4 +327,16 @@ public class ViewerDocumentService {
});
}
private static Line2D transformLine(Line2D line, AffineTransform pageTransformation) {
return new Line2D.Double(pageTransformation.transform(line.getP1(), null), pageTransformation.transform(line.getP2(), null));
}
private static Rectangle2D transformRect(Rectangle2D r, AffineTransform pageTransformation) {
return pageTransformation.createTransformedShape(r).getBounds2D();
}
}