Compare commits

...

2 Commits

Author SHA1 Message Date
maverickstuder
7d6caabbfb tests 2024-02-14 14:54:10 +01:00
maverickstuder
ec035aca2f tests 2024-02-13 18:04:22 +01:00
6 changed files with 204 additions and 74 deletions

View File

@ -19,4 +19,5 @@ public class PageContents {
Rectangle2D cropBox;
Rectangle2D mediaBox;
List<Ruling> rulings;
List<Rectangle2D> positions;
}

View File

@ -39,6 +39,8 @@ public class PageContentExtractor {
stripper.setEndPage(pageNumber);
stripper.setPdpage(pdPage);
stripper.getText(pdDocument);
var positions = stripper.getExactPositions();
Map<Float, List<TextPositionSequence>> sortedTextPositionSequencesPerDir = stripper.getTextPositionSequences()
.stream()
@ -49,7 +51,7 @@ public class PageContentExtractor {
textPositionSequencesPerPage.add(new PageContents(sortedTextPositionSequences,
RectangleTransformations.toRectangle2D(pdPage.getCropBox()),
RectangleTransformations.toRectangle2D(pdPage.getMediaBox()),
stripper.getRulings()));
stripper.getRulings(), positions));
}
}

View File

@ -1,6 +1,5 @@
package com.knecon.fforesight.service.layoutparser.processor.services.blockification;
// TODO: figure out, why this fails the build
// import static com.knecon.fforesight.service.layoutparser.processor.services.factory.SearchTextWithTextPositionFactory.HEIGHT_PADDING;
@ -11,6 +10,7 @@ import com.knecon.fforesight.service.layoutparser.processor.model.table.Ruling;
import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPageBlock;
import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
import com.knecon.fforesight.service.layoutparser.processor.utils.RulingTextDirAdjustUtil;
import org.springframework.stereotype.Service;
import java.util.*;
@ -83,13 +83,13 @@ public class TaasBlockificationService {
continue;
}
Matcher listIdentifierPattern = listIdentifier.matcher(currentTextBlock.getText());
boolean isListIdentifier = listIdentifierPattern.find();
boolean yGap = Math.abs(currentTextBlock.getPdfMaxY() - previousTextBlock.getPdfMinY()) < previousTextBlock.getMostPopularWordHeight() * Y_GAP_SPLIT_HEIGHT_MODIFIER;
boolean sameFont = previousTextBlock.getMostPopularWordFont().equals(currentTextBlock.getMostPopularWordFont()) && previousTextBlock.getMostPopularWordFontSize() == currentTextBlock.getMostPopularWordFontSize();
boolean sameFont = previousTextBlock.getMostPopularWordFont()
.equals(currentTextBlock.getMostPopularWordFont()) && previousTextBlock.getMostPopularWordFontSize() == currentTextBlock.getMostPopularWordFontSize();
// boolean yGap = previousTextBlock != null && currentTextBlock.getMinYDirAdj() - maxY > Math.min(word.getHeight(), prev.getHeight()) * Y_GAP_SPLIT_HEIGHT_MODIFIER;
boolean alignsXRight = Math.abs(currentTextBlock.getPdfMaxX() - previousTextBlock.getPdfMaxX()) < X_ALIGNMENT_THRESHOLD;
@ -119,8 +119,10 @@ public class TaasBlockificationService {
}
alreadyMerged.add(textPageBlock);
textBlocksToMerge.add(Stream.concat(Stream.of(textPageBlock),
textPageBlocks.stream().filter(textPageBlock2 -> textPageBlock.almostIntersects(textPageBlock2, INTERSECTS_Y_THRESHOLD, 0) && !alreadyMerged.contains(textPageBlock2)).peek(alreadyMerged::add))
.toList());
textPageBlocks.stream()
.filter(textPageBlock2 -> textPageBlock.almostIntersects(textPageBlock2, INTERSECTS_Y_THRESHOLD, 0) && !alreadyMerged.contains(textPageBlock2))
.peek(alreadyMerged::add))//
.toList());
}
return textBlocksToMerge.stream().map(TextPageBlock::merge).toList();
}
@ -163,8 +165,7 @@ public class TaasBlockificationService {
while (itty.hasNext()) {
TextPageBlock block = (TextPageBlock) itty.next();
if (previous != null && previous.getOrientation().equals(Orientation.LEFT) && block.getOrientation().equals(Orientation.LEFT) && equalsWithThreshold(
block.getMaxY(),
if (previous != null && previous.getOrientation().equals(Orientation.LEFT) && block.getOrientation().equals(Orientation.LEFT) && equalsWithThreshold(block.getMaxY(),
previous.getMaxY()) || previous != null && previous.getOrientation().equals(Orientation.LEFT) && block.getOrientation()
.equals(Orientation.RIGHT) && equalsWithThreshold(block.getMaxY(), previous.getMaxY())) {
previous.add(block);
@ -189,7 +190,6 @@ public class TaasBlockificationService {
TextPositionSequence prev = null;
// TODO: make static final constant
boolean wasSplitted = false;
Float splitX1 = null;
for (TextPositionSequence word : textPositions) {

View File

@ -16,13 +16,20 @@
*/
package com.knecon.fforesight.service.layoutparser.processor.services.parsing;
import java.awt.geom.AffineTransform;
import java.awt.geom.Area;
import java.awt.geom.Rectangle2D;
import java.io.IOException;
import java.io.InputStream;
import java.util.LinkedList;
import java.util.List;
import java.util.Map;
import java.util.WeakHashMap;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.fontbox.FontBoxFont;
import org.apache.fontbox.ttf.GlyphData;
import org.apache.fontbox.ttf.TrueTypeFont;
import org.apache.fontbox.util.BoundingBox;
import org.apache.pdfbox.contentstream.PDFStreamEngine;
@ -52,12 +59,14 @@ import org.apache.pdfbox.cos.COSDictionary;
import org.apache.pdfbox.pdmodel.PDPage;
import org.apache.pdfbox.pdmodel.common.PDRectangle;
import org.apache.pdfbox.pdmodel.font.PDCIDFont;
import org.apache.pdfbox.pdmodel.font.PDCIDFontType0;
import org.apache.pdfbox.pdmodel.font.PDCIDFontType2;
import org.apache.pdfbox.pdmodel.font.PDFont;
import org.apache.pdfbox.pdmodel.font.PDFontDescriptor;
import org.apache.pdfbox.pdmodel.font.PDSimpleFont;
import org.apache.pdfbox.pdmodel.font.PDTrueTypeFont;
import org.apache.pdfbox.pdmodel.font.PDType0Font;
import org.apache.pdfbox.pdmodel.font.PDType1Font;
import org.apache.pdfbox.pdmodel.font.PDType3Font;
import org.apache.pdfbox.pdmodel.font.encoding.GlyphList;
import org.apache.pdfbox.pdmodel.graphics.state.PDGraphicsState;
@ -65,6 +74,8 @@ import org.apache.pdfbox.text.TextPosition;
import org.apache.pdfbox.util.Matrix;
import org.apache.pdfbox.util.Vector;
import lombok.Getter;
/**
* LEGACY text calculations which are known to be incorrect but are depended on by PDFTextStripper.
* <p>
@ -86,6 +97,9 @@ public class LegacyPDFStreamEngine extends PDFStreamEngine {
private final GlyphList glyphList;
private final Map<COSDictionary, Float> fontHeightMap = new WeakHashMap<COSDictionary, Float>();
@Getter
private List<Rectangle2D> exactPositions = new LinkedList<>();
/**
* Constructor.
@ -163,6 +177,51 @@ public class LegacyPDFStreamEngine extends PDFStreamEngine {
float horizontalScaling = state.getTextState().getHorizontalScaling() / 100f;
Matrix textMatrix = getTextMatrix();
TrueTypeFont ttf = null;
if (font instanceof PDTrueTypeFont) {
ttf = ((PDTrueTypeFont) font).getTrueTypeFont();
} else if (font instanceof PDType0Font) {
PDCIDFont cidFont = ((PDType0Font) font).getDescendantFont();
if (cidFont instanceof PDCIDFontType2) {
ttf = ((PDCIDFontType2) cidFont).getTrueTypeFont();
}
} else if (font instanceof PDType1Font) {
FontBoxFont fontBoxFont = ((PDType1Font) font).getFontBoxFont();
if (fontBoxFont instanceof TrueTypeFont) {
ttf = (TrueTypeFont) fontBoxFont;
} else {
System.out.println("What do?");
}
}
if (ttf != null) {
Integer glyphId = null;
if (font instanceof PDTrueTypeFont) {
glyphId = ((PDTrueTypeFont) font).codeToGID(code);
} else if (font instanceof PDType0Font) {
glyphId = ((PDType0Font) font).codeToGID(code);
} else if (font instanceof PDType1Font) {
FontBoxFont fontBoxFont = ((PDType1Font) font).getFontBoxFont();
if (fontBoxFont instanceof TrueTypeFont) {
glyphId = ((TrueTypeFont) fontBoxFont).getUnicodeCmapLookup().getGlyphId(code);
}
}
if (glyphId != null) {
GlyphData glyph = ttf.getGlyph().getGlyph(glyphId);
if (glyph != null) {
BoundingBox boundingBox = glyph.getBoundingBox();
Rectangle2D rect = new Rectangle2D.Double(boundingBox.getLowerLeftX(), boundingBox.getLowerLeftY(), boundingBox.getWidth(), boundingBox.getHeight());
Area area = new Area(rect);
AffineTransform affineTransform = textRenderingMatrix.createAffineTransform();
float factor = 1f / ttf.getUnitsPerEm();
affineTransform.scale(factor, factor);
exactPositions.add(area.createTransformedArea(affineTransform).getBounds2D());
}
}
}
float displacementX = displacement.getX();
// the sorting algorithm is based on the width of the character. As the displacement
// for vertical characters doesn't provide any suitable value for it, we have to
@ -170,15 +229,6 @@ public class LegacyPDFStreamEngine extends PDFStreamEngine {
if (font.isVertical()) {
displacementX = font.getWidth(code) / 1000;
// there may be an additional scaling factor for true type fonts
TrueTypeFont ttf = null;
if (font instanceof PDTrueTypeFont) {
ttf = ((PDTrueTypeFont) font).getTrueTypeFont();
} else if (font instanceof PDType0Font) {
PDCIDFont cidFont = ((PDType0Font) font).getDescendantFont();
if (cidFont instanceof PDCIDFontType2) {
ttf = ((PDCIDFontType2) cidFont).getTrueTypeFont();
}
}
if (ttf != null && ttf.getUnitsPerEm() != 1000) {
displacementX *= 1000f / ttf.getUnitsPerEm();
}
@ -201,11 +251,11 @@ public class LegacyPDFStreamEngine extends PDFStreamEngine {
// (modified) text rendering matrix
Matrix nextTextRenderingMatrix = td.multiply(textMatrix).multiply(ctm); // text space -> device space
float nextX = nextTextRenderingMatrix.getTranslateX();
float nextY = nextTextRenderingMatrix.getTranslateY();
float endX = nextTextRenderingMatrix.getTranslateX();
float endY = nextTextRenderingMatrix.getTranslateY();
// (modified) width and height calculations
float dxDisplay = nextX - textRenderingMatrix.getTranslateX();
float dxDisplay = endX - textRenderingMatrix.getTranslateX();
Float fontHeight = fontHeightMap.get(font.getCOSObject());
if (fontHeight == null) {
fontHeight = computeFontHeight(font);
@ -271,56 +321,56 @@ public class LegacyPDFStreamEngine extends PDFStreamEngine {
translatedTextRenderingMatrix = textRenderingMatrix;
} else {
translatedTextRenderingMatrix = Matrix.concatenate(translateMatrix, textRenderingMatrix);
nextX -= pageSize.getLowerLeftX();
nextY -= pageSize.getLowerLeftY();
endX -= pageSize.getLowerLeftX();
endY -= 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(),
translatedTextRenderingMatrix,
nextX,
nextY,
Math.abs(dyDisplay),
dxDisplay,
Math.abs(spaceWidthDisplay),
Character.toString(unicodeMapping.charAt(0)),
new int[]{code},
font,
fontSize,
(int) (fontSize * textMatrix.getScalingFactorX())));
pageSize.getWidth(),
pageSize.getHeight(),
translatedTextRenderingMatrix,
endX,
endY,
Math.abs(dyDisplay),
dxDisplay,
Math.abs(spaceWidthDisplay),
Character.toString(unicodeMapping.charAt(0)),
new int[]{code},
font,
fontSize,
(int) (fontSize * textMatrix.getScalingFactorX())));
processTextPosition(new TextPosition(pageRotation,
pageSize.getWidth(),
pageSize.getHeight(),
translatedTextRenderingMatrix,
nextX,
nextY,
Math.abs(dyDisplay),
dxDisplay,
Math.abs(spaceWidthDisplay),
Character.toString(unicodeMapping.charAt(1)),
new int[]{code},
font,
fontSize,
(int) (fontSize * textMatrix.getScalingFactorX())));
pageSize.getWidth(),
pageSize.getHeight(),
translatedTextRenderingMatrix,
endX,
endY,
Math.abs(dyDisplay),
dxDisplay,
Math.abs(spaceWidthDisplay),
Character.toString(unicodeMapping.charAt(1)),
new int[]{code},
font,
fontSize,
(int) (fontSize * textMatrix.getScalingFactorX())));
} else {
processTextPosition(new TextPosition(pageRotation,
pageSize.getWidth(),
pageSize.getHeight(),
translatedTextRenderingMatrix,
nextX,
nextY,
Math.abs(dyDisplay),
dxDisplay,
Math.abs(spaceWidthDisplay),
unicodeMapping,
new int[]{code},
font,
fontSize,
(int) (fontSize * textMatrix.getScalingFactorX())));
pageSize.getWidth(),
pageSize.getHeight(),
translatedTextRenderingMatrix,
endX,
endY,
Math.abs(dyDisplay),
dxDisplay,
Math.abs(spaceWidthDisplay),
unicodeMapping,
new int[]{code},
font,
fontSize,
(int) (fontSize * textMatrix.getScalingFactorX())));
}
}

View File

@ -1,12 +1,22 @@
package com.knecon.fforesight.service.layoutparser.server.services;
import java.awt.geom.AffineTransform;
import java.awt.geom.Rectangle2D;
import java.nio.file.Path;
import java.util.List;
import org.apache.pdfbox.pdmodel.PDDocument;
import org.apache.pdfbox.pdmodel.PDPage;
import org.apache.pdfbox.pdmodel.PDPageContentStream;
import org.apache.pdfbox.pdmodel.font.PDFont;
import org.apache.pdfbox.pdmodel.font.PDType1Font;
import org.apache.pdfbox.pdmodel.font.Standard14Fonts;
import org.junit.jupiter.api.Disabled;
import org.junit.jupiter.api.Test;
import com.knecon.fforesight.service.layoutparser.processor.model.PageContents;
import com.knecon.fforesight.service.layoutparser.processor.model.text.RedTextPosition;
import com.knecon.fforesight.service.layoutparser.processor.model.text.TextDirection;
import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
import com.knecon.fforesight.service.layoutparser.processor.utils.RectangleTransformations;
import com.knecon.fforesight.service.layoutparser.processor.services.PageContentExtractor;
@ -20,21 +30,88 @@ class PageContentExtractorTest {
@SneakyThrows
public void testTextPositionSequenceExtraction() {
String fileName = "files/SOLICITA_VICTRATO-GOLD-II_Item 21_Mutacao_Genica (4).pdf";
//String fileName = "files/CLEAN-II-4.2.2.01_家畜残留分析法.pdf";
//String fileName = "files/BASF/2013-1110704.pdf";
//String fileName = "files/ImportRedactionTestFile_highlighted.pdf";
String fileName = "files/HelloWorldHelvetica.pdf";
var tmpFileName = Path.of("/tmp/").resolve(Path.of(fileName).getFileName() + "_TEXT_POSITION_SEQUENCES.pdf").toString();
List<PageContents> textPositionPerPage = PageContentExtractor.getSortedPageContents(fileName);
PdfDraw.drawRectanglesPerPageNumberedByLine(fileName,
PdfDraw.drawRectanglesPerPage(fileName,
textPositionPerPage.stream()
.map(t -> t.getSortedTextPositionSequences()
.stream()
.map(TextPositionSequence::getRectangle)
.map(RectangleTransformations::toRectangle2D)
//.map(t -> t.getSortedTextPositionSequences()
.map(t -> t.getPositions())
// .stream().flatMap(sequence -> sequence.getTextPositions().stream().map(textPosition -> mapRedTextPositionToInitialUserSpace(textPosition, sequence)))
//.map(textPositionSequence -> (Rectangle2D) new Rectangle2D.Double(textPositionSequence.getMaxXDirAdj(), textPositionSequence.getMaxYDirAdj(), textPositionSequence.getWidth(), textPositionSequence.getHeight()))
.map(List::of)
.toList())
//.map(List::of)
//.toList())
.toList(), tmpFileName);
}
public final int HEIGHT_PADDING = 2;
private Rectangle2D mapRedTextPositionToInitialUserSpace(RedTextPosition textPosition, TextPositionSequence sequence) {
float textHeight = sequence.getTextHeight() + HEIGHT_PADDING;
Rectangle2D rectangle2D = new Rectangle2D.Double(textPosition.getXDirAdj(),
textPosition.getYDirAdj() - textHeight,
textPosition.getWidthDirAdj(),
textHeight + HEIGHT_PADDING);
AffineTransform transform = new AffineTransform();
if (sequence.getDir() == TextDirection.ZERO || sequence.getDir() == TextDirection.HALF_CIRCLE) {
transform.rotate(sequence.getDir().getRadians(), sequence.getPageWidth() / 2f, sequence.getPageHeight() / 2f);
transform.translate(0f, sequence.getPageHeight());
} else if (sequence.getDir() == TextDirection.QUARTER_CIRCLE) {
transform.rotate(sequence.getDir().getRadians(), sequence.getPageWidth() / 2f, sequence.getPageWidth() / 2f);
transform.translate(0f, sequence.getPageWidth());
} else {
transform.rotate(sequence.getDir().getRadians(), sequence.getPageHeight() / 2f, sequence.getPageHeight() / 2f);
transform.translate(0f, sequence.getPageWidth());
}
transform.scale(1., -1.);
return transform.createTransformedShape(rectangle2D).getBounds2D();
}
@Test
@SneakyThrows
public void generatePDF() {
// Create a new PDF document
PDDocument document = new PDDocument();
// Create a blank page
PDPage page = new PDPage();
document.addPage(page);
// Load the Helvetica font
PDFont font = new PDType1Font(Standard14Fonts.FontName.HELVETICA);
// Start a content stream to write text
PDPageContentStream contentStream = new PDPageContentStream(document, page);
contentStream.beginText();
// Set font and font size
contentStream.setFont(font, 12);
// Set text position
contentStream.newLineAtOffset(50, 700);
// Write the text
contentStream.showText("Hello World in Helvetica!");
// Finish writing text
contentStream.endText();
contentStream.close();
// Save the PDF
document.save("/tmp/MyPDF.pdf");
document.close();
System.out.println("PDF created successfully!");
}
}

View File

@ -64,11 +64,11 @@ public class PdfDraw {
var rectanglesInLine = rectanglesOnPage.get(lineNumber);
PdfVisualisationUtility.drawRectangle2DList(pdDocument, pageNumber, rectanglesInLine, PdfVisualisationUtility.Options.builder().stroke(true).build());
double y = Math.min(rectanglesInLine.get(0).getMinY(), rectanglesInLine.get(0).getMaxY());
PdfVisualisationUtility.drawText(String.format("%d", lineNumber),
/**PdfVisualisationUtility.drawText(String.format("%d", lineNumber),
pdDocument,
new Point2D.Double(rectanglesInLine.get(0).getX() - (5 + (5 * countNumberOfDigits(lineNumber))), y + 2),
pageNumber,
PdfVisualisationUtility.Options.builder().stroke(true).build());
PdfVisualisationUtility.Options.builder().stroke(true).build());**/
}
}
pdDocument.save(out);
@ -252,12 +252,12 @@ public class PdfDraw {
rectangle2D = RectangleTransformations.pad(rectangle2D, 10, 10);
}
drawRectangle2DList(document, page.getNumber(), List.of(rectangle2D), options);
drawText(buildString(entry),
/**drawText(buildString(entry),
document,
new Point2D.Double(rectangle2D.getMinX(), rectangle2D.getMaxY() + 2),
page.getNumber(),
options,
entry.getType() == NodeType.TABLE_CELL);
entry.getType() == NodeType.TABLE_CELL);**/
}
}