RED-5381 Test 2

This commit is contained in:
deiflaender 2022-10-13 15:43:32 +02:00
parent dafc140f11
commit 29f47fb17a
11 changed files with 349 additions and 127 deletions

View File

@ -1,5 +1,8 @@
package com.iqser.red.service.redaction.v1.model; package com.iqser.red.service.redaction.v1.model;
import com.dslplatform.json.JsonAttribute;
import com.fasterxml.jackson.annotation.JsonIgnore;
import lombok.AllArgsConstructor; import lombok.AllArgsConstructor;
import lombok.Data; import lombok.Data;
import lombok.NoArgsConstructor; import lombok.NoArgsConstructor;
@ -14,4 +17,101 @@ public class Rectangle {
private float height; private float height;
private int page; private int page;
@JsonIgnore
@JsonAttribute(ignore = true)
public float getX1(int dir, float pageWidth, float pageHeight) {
if (dir == 90) {
return topLeft.getX();
}
else if (dir == 180) {
return pageWidth - topLeft.getX() + width;
}
else if (dir == 270) {
return topLeft.getY() + height;
}
else {
return topLeft.getX();
}
}
@JsonIgnore
@JsonAttribute(ignore = true)
public float getX2(int dir, float pageWidth, float pageHeight) {
if (dir == 90) {
return topLeft.getX() + width;
}
else if (dir == 180) {
return pageWidth - topLeft.getX();
}
else if (dir == 270) {
return topLeft.getY() ;
}
else {
return topLeft.getX() + width;
}
}
@JsonIgnore
@JsonAttribute(ignore = true)
public float getY1(int dir, float pageWidth, float pageHeight) {
if (dir == 90 ) {
return topLeft.getY();
}
else if (dir == 180) {
return pageHeight - topLeft.getY() + height;
}
else if (dir == 270) {
return pageHeight - topLeft.getX() ;
}
else {
return topLeft.getY();
}
}
@JsonIgnore
@JsonAttribute(ignore = true)
public float getY2(int dir, float pageWidth, float pageHeight) {
if (dir == 90) {
return topLeft.getY() + height;
}
else if (dir == 180 ) {
return pageHeight - topLeft.getY();
}
else if (dir == 270 ) {
return pageHeight -topLeft.getX() - width;
}
else {
return topLeft.getY() + height;
}
}
} }

View File

@ -1,8 +1,8 @@
package com.iqser.red.service.redaction.v1.server.classification.model; package com.iqser.red.service.redaction.v1.server.classification.model;
import com.iqser.red.service.redaction.v1.model.Rectangle;
import com.iqser.red.service.redaction.v1.server.redaction.model.PdfImage; import com.iqser.red.service.redaction.v1.server.redaction.model.PdfImage;
import com.iqser.red.service.redaction.v1.server.tableextraction.model.AbstractTextContainer; import com.iqser.red.service.redaction.v1.server.tableextraction.model.AbstractTextContainer;
import com.iqser.red.service.redaction.v1.server.tableextraction.model.Rectangle;
import lombok.Data; import lombok.Data;
import lombok.NonNull; import lombok.NonNull;
import lombok.RequiredArgsConstructor; import lombok.RequiredArgsConstructor;
@ -10,6 +10,8 @@ import lombok.RequiredArgsConstructor;
import java.util.ArrayList; import java.util.ArrayList;
import java.util.List; import java.util.List;
import org.apache.pdfbox.pdmodel.common.PDRectangle;
@Data @Data
@RequiredArgsConstructor @RequiredArgsConstructor
public class Page { public class Page {
@ -32,6 +34,7 @@ public class Page {
private StringFrequencyCounter fontStyleCounter = new StringFrequencyCounter(); private StringFrequencyCounter fontStyleCounter = new StringFrequencyCounter();
private double cropBoxArea; private double cropBoxArea;
private PDRectangle cropBox;
public boolean isRotated() { public boolean isRotated() {

View File

@ -3,6 +3,7 @@ package com.iqser.red.service.redaction.v1.server.classification.model;
import com.dslplatform.json.CompiledJson; import com.dslplatform.json.CompiledJson;
import com.dslplatform.json.JsonAttribute; import com.dslplatform.json.JsonAttribute;
import com.fasterxml.jackson.annotation.JsonIgnore; import com.fasterxml.jackson.annotation.JsonIgnore;
import com.iqser.red.service.redaction.v1.server.parsing.model.TextDirection;
import com.iqser.red.service.redaction.v1.server.parsing.model.TextPositionSequence; import com.iqser.red.service.redaction.v1.server.parsing.model.TextPositionSequence;
import com.iqser.red.service.redaction.v1.server.redaction.utils.TextNormalizationUtilities; import com.iqser.red.service.redaction.v1.server.redaction.utils.TextNormalizationUtilities;
import com.iqser.red.service.redaction.v1.server.tableextraction.model.AbstractTextContainer; import com.iqser.red.service.redaction.v1.server.tableextraction.model.AbstractTextContainer;
@ -50,6 +51,115 @@ public class TextBlock extends AbstractTextContainer {
private String classification; private String classification;
private TextDirection getDir(){
return sequences.get(0).getDir();
}
private float getPageHeight(){
return sequences.get(0).getPageHeight();
}
private float getPageWidth(){
return sequences.get(0).getPageWidth();
}
@JsonIgnore
@JsonAttribute(ignore = true)
public float getX1() {
if (getDir().getDegrees() == 90) {
return minX;
}
else if (getDir().getDegrees() == 180) {
return getPageWidth() - maxX;
}
else if (getDir().getDegrees() == 270) {
return getPageWidth() - maxY ;
}
else {
return minX;
}
}
@JsonIgnore
@JsonAttribute(ignore = true)
public float getX2() {
if (getDir().getDegrees() == 90) {
return maxX;
}
else if (getDir().getDegrees() == 180) {
return getPageWidth() - minX;
}
else if (getDir().getDegrees() == 270) {
return getPageWidth() - minY ;
}
else {
return maxX;
}
}
@JsonIgnore
@JsonAttribute(ignore = true)
public float getY1() {
if (getDir().getDegrees() == 90 ) {
return minY;
}
else if (getDir().getDegrees() == 180) {
return getPageHeight() - maxY;
}
else if (getDir().getDegrees() == 270) {
return getPageHeight() - maxX;
}
else {
return minY;
}
}
@JsonIgnore
@JsonAttribute(ignore = true)
public float getY2() {
if (getDir().getDegrees() == 90) {
return maxY;
}
else if (getDir().getDegrees() == 180 ) {
return getPageHeight() - minY;
}
else if (getDir().getDegrees() == 270 ) {
return getPageHeight() - minX;
}
else {
return maxY;
}
}
public TextBlock(float minX, float maxX, float minY, float maxY, List<TextPositionSequence> sequences, int rotation) { public TextBlock(float minX, float maxX, float minY, float maxY, List<TextPositionSequence> sequences, int rotation) {
this.minX = minX; this.minX = minX;
this.maxX = maxX; this.maxX = maxX;

View File

@ -9,6 +9,8 @@ import java.util.List;
import org.springframework.stereotype.Service; import org.springframework.stereotype.Service;
import com.iqser.red.service.redaction.v1.model.Point;
import com.iqser.red.service.redaction.v1.model.Rectangle;
import com.iqser.red.service.redaction.v1.server.classification.model.FloatFrequencyCounter; import com.iqser.red.service.redaction.v1.server.classification.model.FloatFrequencyCounter;
import com.iqser.red.service.redaction.v1.server.classification.model.Orientation; import com.iqser.red.service.redaction.v1.server.classification.model.Orientation;
import com.iqser.red.service.redaction.v1.server.classification.model.Page; import com.iqser.red.service.redaction.v1.server.classification.model.Page;
@ -18,7 +20,6 @@ import com.iqser.red.service.redaction.v1.server.classification.utils.PositionUt
import com.iqser.red.service.redaction.v1.server.parsing.model.TextPositionSequence; import com.iqser.red.service.redaction.v1.server.parsing.model.TextPositionSequence;
import com.iqser.red.service.redaction.v1.server.tableextraction.model.AbstractTextContainer; import com.iqser.red.service.redaction.v1.server.tableextraction.model.AbstractTextContainer;
import com.iqser.red.service.redaction.v1.server.tableextraction.model.Cell; import com.iqser.red.service.redaction.v1.server.tableextraction.model.Cell;
import com.iqser.red.service.redaction.v1.server.tableextraction.model.Rectangle;
import com.iqser.red.service.redaction.v1.server.tableextraction.model.Ruling; import com.iqser.red.service.redaction.v1.server.tableextraction.model.Ruling;
import com.iqser.red.service.redaction.v1.server.tableextraction.model.Table; import com.iqser.red.service.redaction.v1.server.tableextraction.model.Table;
@ -32,7 +33,7 @@ public class BlockificationService {
public Page blockify(List<TextPositionSequence> textPositions, List<Ruling> horizontalRulingLines, public Page blockify(List<TextPositionSequence> textPositions, List<Ruling> horizontalRulingLines,
List<Ruling> verticalRulingLines) { List<Ruling> verticalRulingLines) {
sortRotatedSequences(textPositions); // sortRotatedSequences(textPositions);
List<TextPositionSequence> chunkWords = new ArrayList<>(); List<TextPositionSequence> chunkWords = new ArrayList<>();
List<AbstractTextContainer> chunkBlockList1 = new ArrayList<>(); List<AbstractTextContainer> chunkBlockList1 = new ArrayList<>();
@ -108,55 +109,55 @@ public class BlockificationService {
chunkBlockList1.add(cb1); chunkBlockList1.add(cb1);
} }
Iterator<AbstractTextContainer> itty = chunkBlockList1.iterator(); // Iterator<AbstractTextContainer> itty = chunkBlockList1.iterator();
//
TextBlock previousLeft = null; // TextBlock previousLeft = null;
TextBlock previousRight = null; // TextBlock previousRight = null;
while (itty.hasNext()) { // while (itty.hasNext()) {
TextBlock block = (TextBlock) itty.next(); // TextBlock block = (TextBlock) itty.next();
//
if (previousLeft != null && block.getOrientation().equals(Orientation.LEFT)) { // if (previousLeft != null && block.getOrientation().equals(Orientation.LEFT)) {
if (previousLeft.getMinY() > block.getMinY() && block.getMaxY() + block.getMostPopularWordHeight() > previousLeft // if (previousLeft.getMinY() > block.getMinY() && block.getMaxY() + block.getMostPopularWordHeight() > previousLeft
.getMinY()) { // .getMinY()) {
previousLeft.add(block); // previousLeft.add(block);
itty.remove(); // itty.remove();
continue; // continue;
} // }
} // }
//
if (previousRight != null && block.getOrientation().equals(Orientation.RIGHT)) { // if (previousRight != null && block.getOrientation().equals(Orientation.RIGHT)) {
if (previousRight.getMinY() > block.getMinY() && block.getMaxY() + block.getMostPopularWordHeight() > previousRight // if (previousRight.getMinY() > block.getMinY() && block.getMaxY() + block.getMostPopularWordHeight() > previousRight
.getMinY()) { // .getMinY()) {
previousRight.add(block); // previousRight.add(block);
itty.remove(); // itty.remove();
continue; // continue;
} // }
} // }
//
if (block.getOrientation().equals(Orientation.LEFT)) { // if (block.getOrientation().equals(Orientation.LEFT)) {
previousLeft = block; // previousLeft = block;
} else if (block.getOrientation().equals(Orientation.RIGHT)) { // } else if (block.getOrientation().equals(Orientation.RIGHT)) {
previousRight = block; // previousRight = block;
} // }
} // }
//
itty = chunkBlockList1.iterator(); // itty = chunkBlockList1.iterator();
TextBlock previous = null; // TextBlock previous = null;
while (itty.hasNext()) { // while (itty.hasNext()) {
TextBlock block = (TextBlock) itty.next(); // TextBlock block = (TextBlock) itty.next();
//
if (previous != null && previous.getOrientation().equals(Orientation.LEFT) && block.getOrientation() // if (previous != null && previous.getOrientation().equals(Orientation.LEFT) && block.getOrientation()
.equals(Orientation.LEFT) && equalsWithThreshold(block.getMaxY(), previous.getMaxY()) || previous != null && previous // .equals(Orientation.LEFT) && equalsWithThreshold(block.getMaxY(), previous.getMaxY()) || previous != null && previous
.getOrientation() // .getOrientation()
.equals(Orientation.LEFT) && block.getOrientation() // .equals(Orientation.LEFT) && block.getOrientation()
.equals(Orientation.RIGHT) && equalsWithThreshold(block.getMaxY(), previous.getMaxY())) { // .equals(Orientation.RIGHT) && equalsWithThreshold(block.getMaxY(), previous.getMaxY())) {
previous.add(block); // previous.add(block);
itty.remove(); // itty.remove();
continue; // continue;
} // }
//
previous = block; // previous = block;
} // }
return new Page(chunkBlockList1); return new Page(chunkBlockList1);
} }
@ -258,17 +259,17 @@ public class BlockificationService {
if (documentFontSizeCounter.getMostPopular() != null) { if (documentFontSizeCounter.getMostPopular() != null) {
if (textBlock.getMostPopularWordFontSize() >= documentFontSizeCounter.getMostPopular()) { if (textBlock.getMostPopularWordFontSize() >= documentFontSizeCounter.getMostPopular()) {
if (textBlock.getMinX() < minX) { if (textBlock.getX1() < minX) {
minX = textBlock.getMinX(); minX = textBlock.getX1();
} }
if (textBlock.getMaxX() > maxX) { if (textBlock.getX2() > maxX) {
maxX = textBlock.getMaxX(); maxX = textBlock.getX2();
} }
if (textBlock.getMinY() < minY) { if (textBlock.getY1() < minY) {
minY = textBlock.getMinY(); minY = textBlock.getY1();
} }
if (textBlock.getMaxY() > maxY) { if (textBlock.getY2() > maxY) {
maxY = textBlock.getMaxY(); maxY = textBlock.getY2();
} }
} }
@ -284,17 +285,17 @@ public class BlockificationService {
continue; continue;
} }
for (TextBlock textBlock : cell.getTextBlocks()) { for (TextBlock textBlock : cell.getTextBlocks()) {
if (textBlock.getMinX() < minX) { if (textBlock.getX1() < minX) {
minX = textBlock.getMinX(); minX = textBlock.getX1();
} }
if (textBlock.getMaxX() > maxX) { if (textBlock.getX2() > maxX) {
maxX = textBlock.getMaxX(); maxX = textBlock.getX2();
} }
if (textBlock.getMinY() < minY) { if (textBlock.getY1() < minY) {
minY = textBlock.getMinY(); minY = textBlock.getY1();
} }
if (textBlock.getMaxY() > maxY) { if (textBlock.getY2() > maxY) {
maxY = textBlock.getMaxY(); maxY = textBlock.getY2();
} }
} }
} }
@ -303,7 +304,7 @@ public class BlockificationService {
} }
} }
return new Rectangle(minY, minX, maxX - minX, maxY - minY); return new Rectangle(new Point(minX, minY), maxX - minX, maxY - minY, 0);
} }

View File

@ -1,11 +1,12 @@
package com.iqser.red.service.redaction.v1.server.classification.service; package com.iqser.red.service.redaction.v1.server.classification.service;
import com.iqser.red.service.redaction.v1.model.Point;
import com.iqser.red.service.redaction.v1.model.Rectangle;
import com.iqser.red.service.redaction.v1.server.classification.model.Document; import com.iqser.red.service.redaction.v1.server.classification.model.Document;
import com.iqser.red.service.redaction.v1.server.classification.model.Page; import com.iqser.red.service.redaction.v1.server.classification.model.Page;
import com.iqser.red.service.redaction.v1.server.classification.model.TextBlock; import com.iqser.red.service.redaction.v1.server.classification.model.TextBlock;
import com.iqser.red.service.redaction.v1.server.classification.utils.PositionUtils; import com.iqser.red.service.redaction.v1.server.classification.utils.PositionUtils;
import com.iqser.red.service.redaction.v1.server.tableextraction.model.AbstractTextContainer; import com.iqser.red.service.redaction.v1.server.tableextraction.model.AbstractTextContainer;
import com.iqser.red.service.redaction.v1.server.tableextraction.model.Rectangle;
import lombok.RequiredArgsConstructor; import lombok.RequiredArgsConstructor;
import lombok.extern.slf4j.Slf4j; import lombok.extern.slf4j.Slf4j;
import org.springframework.stereotype.Service; import org.springframework.stereotype.Service;
@ -32,7 +33,23 @@ public class ClassificationService {
for (Page page : document.getPages()) { for (Page page : document.getPages()) {
Rectangle btf = page.isLandscape() ? landscapeBodyTextFrame : bodyTextFrame; Rectangle btf = page.isLandscape() ? landscapeBodyTextFrame : bodyTextFrame;
if (!page.isLandscape() && (page.getRotation() == 90 || page.getRotation() == 270)){
var width = btf.getX2(page.getRotation(), page.getCropBox().getWidth(), page.getCropBox().getHeight()) - btf.getX1(page.getRotation(), page.getCropBox().getWidth(), page.getCropBox().getHeight());
var height = btf.getY2(page.getRotation(), page.getCropBox().getWidth(), page.getCropBox().getHeight()) - btf.getY1(page.getRotation(), page.getCropBox().getWidth(), page.getCropBox().getHeight());
var x1= btf.getX1(page.getRotation(), page.getCropBox().getWidth(), page.getCropBox().getHeight());
var y1 = btf.getY1(page.getRotation(), page.getCropBox().getWidth(), page.getCropBox().getHeight());
page.setBodyTextFrame(new Rectangle(new Point(x1, y1), width, height,0));
} else if (page.getRotation() == 180) {
page.setBodyTextFrame(new Rectangle(new Point(btf.getTopLeft().getX(), page.getCropBox().getHeight() - btf.getTopLeft().getY() -btf.getHeight()), btf.getWidth(), btf.getHeight(),0));
} else {
page.setBodyTextFrame(btf); page.setBodyTextFrame(btf);
}
classifyPage(btf, page, document, headlineFontSizes); classifyPage(btf, page, document, headlineFontSizes);
} }
} }
@ -51,6 +68,8 @@ public class ClassificationService {
public void classifyBlock(TextBlock textBlock, Rectangle bodyTextFrame, Page page, Document document, public void classifyBlock(TextBlock textBlock, Rectangle bodyTextFrame, Page page, Document document,
List<Float> headlineFontSizes) { List<Float> headlineFontSizes) {
System.out.println("B" + textBlock.getSequences().get(0).getDir());
if (document.getFontSizeCounter().getMostPopular() == null) { if (document.getFontSizeCounter().getMostPopular() == null) {
textBlock.setClassification("Other"); textBlock.setClassification("Other");
return; return;

View File

@ -1,7 +1,7 @@
package com.iqser.red.service.redaction.v1.server.classification.utils; package com.iqser.red.service.redaction.v1.server.classification.utils;
import com.iqser.red.service.redaction.v1.model.Rectangle;
import com.iqser.red.service.redaction.v1.server.classification.model.TextBlock; import com.iqser.red.service.redaction.v1.server.classification.model.TextBlock;
import com.iqser.red.service.redaction.v1.server.tableextraction.model.Rectangle;
import lombok.experimental.UtilityClass; import lombok.experimental.UtilityClass;
@UtilityClass @UtilityClass
@ -19,10 +19,10 @@ public class PositionUtils {
double threshold = textBlock.getMostPopularWordHeight() * 3; double threshold = textBlock.getMostPopularWordHeight() * 3;
if (textBlock.getMinX() + threshold > btf.getX() && if (textBlock.getMinX() + threshold > btf.getTopLeft().getX() &&
textBlock.getMaxX() - threshold < btf.getX() + btf.getWidth() && textBlock.getMaxX() - threshold < btf.getTopLeft().getX() + btf.getWidth() &&
textBlock.getMinY() + threshold > btf.getY() && textBlock.getMinY() + threshold > btf.getTopLeft().getY() &&
textBlock.getMaxY() - threshold < btf.getY() + btf.getHeight()) { textBlock.getMaxY() - threshold < btf.getTopLeft().getY() + btf.getHeight()) {
return true; return true;
} else { } else {
return false; return false;
@ -37,10 +37,10 @@ public class PositionUtils {
return false; return false;
} }
if (rotated && textBlock.getMinX() < btf.getX()) { if (rotated && textBlock.getMinX() < btf.getTopLeft().getX()) {
// Its very strange, P{0,0} is on top left in this case, instead of lower left. // Its very strange, P{0,0} is on top left in this case, instead of lower left.
return true; return true;
} else if (!rotated && textBlock.getMinY() > btf.getY() + btf.getHeight()) { } else if (!rotated && textBlock.getMinY() > btf.getTopLeft().getY() + btf.getHeight()) {
return true; return true;
} else { } else {
return false; return false;
@ -57,7 +57,7 @@ public class PositionUtils {
return false; return false;
} }
if (textBlock.getMaxY() < btf.getY()) { if (textBlock.getMaxY() < btf.getTopLeft().getY()) {
return true; return true;
} else { } else {
return false; return false;
@ -74,7 +74,7 @@ public class PositionUtils {
return false; return false;
} }
if (textBlock.getMinY() < btf.getY()) { if (textBlock.getMinY() < btf.getTopLeft().getY()) {
return true; return true;
} else { } else {
return false; return false;

View File

@ -163,14 +163,6 @@ public class TextPositionSequence implements CharSequence {
if (getDir().getDegrees() == 90) { if (getDir().getDegrees() == 90) {
return textPositions.get(0).getYDirAdj() - getTextHeight(); return textPositions.get(0).getYDirAdj() - getTextHeight();
} }
else if (getDir().getDegrees() == 180) {
return pageWidth - textPositions.get(textPositions.size() - 1).getXDirAdj() - textPositions.get(textPositions.size() - 1).getWidthDirAdj() - HEIGHT_PADDING;
}
else if (getDir().getDegrees() == 270) {
return pageWidth - textPositions.get(0).getYDirAdj() ;
}
else { else {
return textPositions.get(0).getXDirAdj(); return textPositions.get(0).getXDirAdj();
} }
@ -185,14 +177,6 @@ public class TextPositionSequence implements CharSequence {
return textPositions.get(0).getYDirAdj(); return textPositions.get(0).getYDirAdj();
} }
else if (getDir().getDegrees() == 180) {
return pageWidth - textPositions.get(0).getXDirAdj() ;
}
else if (getDir().getDegrees() == 270) {
return pageWidth - textPositions.get(0).getYDirAdj() + getTextHeight();
}
else { else {
return textPositions.get(textPositions.size() - 1).getXDirAdj() + textPositions.get(textPositions.size() - 1).getWidthDirAdj() + HEIGHT_PADDING; return textPositions.get(textPositions.size() - 1).getXDirAdj() + textPositions.get(textPositions.size() - 1).getWidthDirAdj() + HEIGHT_PADDING;
@ -211,16 +195,11 @@ public class TextPositionSequence implements CharSequence {
return textPositions.get(0).getXDirAdj(); return textPositions.get(0).getXDirAdj();
} }
else if (getDir().getDegrees() == 180) { if (getDir().getDegrees() == 270) {
return textPositions.get(0).getYDirAdj() - getTextHeight(); return textPositions.get(0).getYDirAdj() - getTextHeight();
} }
else if (getDir().getDegrees() == 270) {
return pageHeight - textPositions.get(textPositions.size() - 1).getXDirAdj() - getTextHeight() + HEIGHT_PADDING;
}
else { else {
return pageHeight - textPositions.get(0).getYDirAdj(); return pageHeight - textPositions.get(0).getYDirAdj();
} }
@ -235,15 +214,10 @@ public class TextPositionSequence implements CharSequence {
return textPositions.get(textPositions.size() - 1).getXDirAdj() + getTextHeight() - HEIGHT_PADDING; return textPositions.get(textPositions.size() - 1).getXDirAdj() + getTextHeight() - HEIGHT_PADDING;
} }
else if (getDir().getDegrees() == 180 ) { if (getDir().getDegrees() == 270) {
return textPositions.get(0).getYDirAdj(); return textPositions.get(0).getYDirAdj();
} }
else if (getDir().getDegrees() == 270 ) {
return pageHeight - textPositions.get(0).getXDirAdj();
}
else { else {
return pageHeight - textPositions.get(0).getYDirAdj() + getTextHeight(); return pageHeight - textPositions.get(0).getYDirAdj() + getTextHeight();
} }

View File

@ -99,10 +99,10 @@ public class PdfSegmentationService {
stripper.getText(pdDocument); stripper.getText(pdDocument);
PDRectangle pdr = pdPage.getMediaBox(); PDRectangle pdr = pdPage.getMediaBox();
boolean isLandscape = pdr.getWidth() > pdr.getHeight();
int rotation = pdPage.getRotation(); int rotation = pdPage.getRotation();
boolean isRotated = rotation != 0 && rotation != 360; boolean isLandscape = pdr.getWidth() > pdr.getHeight() && (rotation == 0 || rotation == 180) ||
pdr.getHeight() > pdr.getWidth() && (rotation == 90 || rotation == 270);
CleanRulings cleanRulings = rulingCleaningService.getCleanRulings(pdfTableCells.get(pageNumber), stripper.getRulings(), stripper.getMinCharWidth(), stripper.getMaxCharHeight()); CleanRulings cleanRulings = rulingCleaningService.getCleanRulings(pdfTableCells.get(pageNumber), stripper.getRulings(), stripper.getMinCharWidth(), stripper.getMaxCharHeight());
Page page = blockificationService.blockify(stripper.getTextPositionSequences(), cleanRulings.getHorizontal(), cleanRulings.getVertical()); Page page = blockificationService.blockify(stripper.getTextPositionSequences(), cleanRulings.getHorizontal(), cleanRulings.getVertical());
@ -112,8 +112,9 @@ public class PdfSegmentationService {
page.setCropBoxArea(cropboxArea); page.setCropBoxArea(cropboxArea);
page.setRotation(rotation); page.setRotation(rotation);
page.setLandscape(isLandscape || isRotated); page.setLandscape(isLandscape);
page.setPageNumber(pageNumber); page.setPageNumber(pageNumber);
page.setCropBox(cropbox);
tableExtractionService.extractTables(cleanRulings, page); tableExtractionService.extractTables(cleanRulings, page);
buildPageStatistics(page); buildPageStatistics(page);

View File

@ -31,6 +31,10 @@ public class Rectangle extends Rectangle2D.Float {
} }
}; };
public Rectangle() { public Rectangle() {
super(); super();
} }

View File

@ -79,7 +79,10 @@ public class PdfVisualisationService {
} }
contentStream.setStrokingColor(Color.YELLOW); contentStream.setStrokingColor(Color.YELLOW);
contentStream.addRect((float) analyzedPage.getBodyTextFrame().getX(), (float) analyzedPage.getBodyTextFrame().getY(), (float) analyzedPage.getBodyTextFrame().getWidth(), (float) analyzedPage.getBodyTextFrame().getHeight()); // var widhth = analyzedPage.getBodyTextFrame().getX2(pdPage.getRotation(), pdPage.getCropBox().getWidth(), pdPage.getCropBox().getHeight()) - analyzedPage.getBodyTextFrame().getX1(pdPage.getRotation(), pdPage.getCropBox().getWidth(), pdPage.getCropBox().getHeight());
// var height = analyzedPage.getBodyTextFrame().getY2(pdPage.getRotation(), pdPage.getCropBox().getWidth(), pdPage.getCropBox().getHeight()) - analyzedPage.getBodyTextFrame().getY1(pdPage.getRotation(), pdPage.getCropBox().getWidth(), pdPage.getCropBox().getHeight());
contentStream.addRect(analyzedPage.getBodyTextFrame().getTopLeft().getX(), analyzedPage.getBodyTextFrame().getTopLeft().getY(), analyzedPage.getBodyTextFrame().getWidth(), analyzedPage.getBodyTextFrame().getHeight());
contentStream.stroke(); contentStream.stroke();
contentStream.close(); contentStream.close();
@ -91,33 +94,40 @@ public class PdfVisualisationService {
contentStream.setStrokingColor(Color.RED); contentStream.setStrokingColor(Color.RED);
contentStream.addRect(textBlock.getMinX(), textBlock.getMinY(), textBlock.getWidth(), textBlock.getHeight()); contentStream.addRect(textBlock.getX1(), textBlock.getY1(), textBlock.getX2() - textBlock.getX1(), textBlock.getY2() - textBlock.getY1());
contentStream.stroke(); contentStream.stroke();
if (textBlock.getClassification() != null) { if (textBlock.getClassification() != null) {
contentStream.beginText(); contentStream.beginText();
contentStream.setFont(PDType1Font.TIMES_ROMAN, 6f);
contentStream.newLineAtOffset(textBlock.getX1(), textBlock.getY2() + 2);
contentStream.showText(textBlock.getClassification() + textBlock.getOrientation() + "-->" + textBlock.getSequences().get(0).getDir());
contentStream.endText();
contentStream.setNonStrokingColor(Color.BLUE); contentStream.setNonStrokingColor(Color.BLUE);
contentStream.setFont(PDType1Font.TIMES_ROMAN, 2f); contentStream.setFont(PDType1Font.TIMES_ROMAN, 2f);
contentStream.newLineAtOffset(textBlock.getMinX(), textBlock.getMaxY() + 10);
contentStream.showText(textBlock.getClassification() + textBlock.getOrientation() + "-->" + textBlock.getSequences().get(0).getDir());
contentStream.beginText();
contentStream.newLineAtOffset(textBlock.getX1(), textBlock.getY1());
contentStream.showText("MinX,MinY(" + textBlock.getX1() + "," + textBlock.getY1() + ")");
contentStream.endText(); contentStream.endText();
contentStream.beginText(); contentStream.beginText();
contentStream.newLineAtOffset(textBlock.getMinX(), textBlock.getMinY()); contentStream.newLineAtOffset(textBlock.getX2(), textBlock.getY1());
contentStream.showText("MinX,MinY"); contentStream.showText("MaxX,MinY(" + textBlock.getX2() + "," + textBlock.getY1() + ")");
contentStream.endText(); contentStream.endText();
contentStream.beginText(); contentStream.beginText();
contentStream.newLineAtOffset(textBlock.getMaxX(), textBlock.getMinY()); contentStream.newLineAtOffset(textBlock.getX1(), textBlock.getY2());
contentStream.showText("MaxX,MinY"); contentStream.showText("MinX,MaxY(" + textBlock.getX1() + "," + textBlock.getY2() + ")");
contentStream.endText(); contentStream.endText();
contentStream.beginText(); contentStream.beginText();
contentStream.newLineAtOffset(textBlock.getMinX(), textBlock.getMaxY()); contentStream.newLineAtOffset(textBlock.getX2(), textBlock.getY2());
contentStream.showText("MinX,MaxY"); contentStream.showText("MaxX,MaxY(" + textBlock.getX2() + "," + textBlock.getY2() + ")");
contentStream.endText();
contentStream.beginText();
contentStream.newLineAtOffset(textBlock.getMaxX(), textBlock.getMaxY());
contentStream.showText("MaxX,MaxY");
contentStream.endText(); contentStream.endText();

View File

@ -1168,7 +1168,7 @@ public class RedactionIntegrationTest {
System.out.println("classificationTest"); System.out.println("classificationTest");
AnalyzeRequest request = prepareStorage("files/new/RotateTestFile.pdf"); AnalyzeRequest request = prepareStorage("files/RSS/02 - A22833B - Acute Oral (Up and Down) - Final Report1.pdf");
RedactionRequest redactionRequest = RedactionRequest.builder() RedactionRequest redactionRequest = RedactionRequest.builder()
.dossierId(request.getDossierId()) .dossierId(request.getDossierId())