Merge branch 'renovate/main-com.iqser.red.commons-storage-commons-2.x' into 'main'

Update dependency com.iqser.red.commons:storage-commons to v2.36.0 (main)

See merge request fforesight/layout-parser!50

.gitignore

@@ -38,3 +38,7 @@ build/
 
 ### VS Code ###
 .vscode/
+gradlew.bat
+gradlew
+gradle.properties
+gradle/

layoutparser-service/layoutparser-service-processor/build.gradle.kts

@@ -11,12 +11,12 @@ val pdfBoxVersion = "3.0.0-RC1"
 dependencies {
     implementation(project(":layoutparser-service-internal-api"))
 
-    implementation("com.iqser.red.service:persistence-service-shared-api-v1:2.126.0") {
+    implementation("com.iqser.red.service:persistence-service-shared-api-v1:2.138.0") {
         exclude("org.springframework.boot", "spring-boot-starter-security")
         exclude("org.springframework.boot", "spring-boot-starter-validation")
     }
     implementation("com.knecon.fforesight:tenant-commons:0.10.0")
-    implementation("com.iqser.red.commons:storage-commons:2.27.0")
+    implementation("com.iqser.red.commons:storage-commons:2.36.0")
 
     implementation("org.apache.pdfbox:pdfbox:${pdfBoxVersion}")
     implementation("org.apache.pdfbox:pdfbox-tools:${pdfBoxVersion}")

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/parsing/PDFLinesTextStripper.java

@@ -4,6 +4,7 @@ import java.awt.color.CMMException;
 import java.awt.geom.Point2D;
 import java.io.IOException;
 import java.util.ArrayList;
+import java.util.Arrays;
 import java.util.List;
 
 import org.apache.pdfbox.contentstream.operator.Operator;
@@ -219,7 +220,11 @@ public class PDFLinesTextStripper extends PDFTextStripper {
             // This is a quick and dirt hack
             // Happens for file 216.pdf
             log.debug(e.getMessage());
-            return color.getComponents()[0] == 0 && color.getComponents()[1] == 0 && color.getComponents()[2] == 0 && color.getComponents()[1] == 1;
+            var result = true;
+            for (var component : color.getComponents()) {
+                result = result && component == 0;
+            }
+            return result;
         }
     }
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/visualization/LayoutGridService.java

@@ -8,10 +8,13 @@ import java.awt.geom.Point2D;
 import java.awt.geom.Rectangle2D;
 import java.awt.geom.RectangularShape;
 import java.util.ArrayList;
+import java.util.Collections;
 import java.util.Comparator;
 import java.util.List;
 import java.util.Map;
+import java.util.Optional;
 import java.util.stream.Collectors;
+import java.util.stream.Stream;
 
 import org.springframework.stereotype.Service;
 
@@ -33,7 +36,7 @@ public class LayoutGridService {
     private static final Color INNER_LINES_COLOR = new Color(255, 175, 175);
     private static final Color PARAGRAPH_COLOR = new Color(70, 130, 180);
     public static final Color TABLE_COLOR = new Color(102, 205, 170);
-    public static final Color SECTION_COLOR = new Color(23, 23, 23);
+    public static final Color SECTION_COLOR = new Color(50, 50, 50);
     public static final Color HEADLINE_COLOR = new Color(162, 56, 56);
     public static final Color HEADER_COLOR = new Color(171, 131, 6);
     public static final Color IMAGE_COLOR = new Color(253, 63, 146);
@@ -60,7 +63,7 @@ public class LayoutGridService {
             }
             if (semanticNode.getType().equals(NodeType.TABLE)) {
                 Table table = (Table) semanticNode;
-                addInnerTableLines(table, layoutGrid, INNER_LINES_COLOR);
+                addInnerTableLines(table, layoutGrid);
             }
         });
 
@@ -68,26 +71,51 @@ public class LayoutGridService {
     }
 
 
-    private void addInnerTableLines(Table table, LayoutGrid layoutGrid, Color color) {
+    private void addInnerTableLines(Table table, LayoutGrid layoutGrid) {
 
         if (table.getNumberOfCols() < 1 || table.getNumberOfRows() < 1) {
             return;
         }
-        // only draw inner lines -> remove first values
-        List<Double> xs = table.streamRow(0).map(TableCell::getBBox).map(map -> map.values().stream().findAny().get()).map(RectangularShape::getMinX).collect(Collectors.toList());
-        xs.remove(0);
-        List<Double> ys = table.streamCol(0).map(TableCell::getBBox).map(map -> map.values().stream().findAny().get()).map(RectangularShape::getMaxY).collect(Collectors.toList());
-        ys.remove(0);
-        Rectangle2D tableBBox = table.getBBox().get(table.getFirstPage());
-        List<ColoredLine> coloredLines = layoutGrid.getVisualizationsPerPages().get(table.getFirstPage().getNumber() - 1).getColoredLines();
-        xs.forEach(x -> {
-            Line2D line = new Line2D.Double(new Point2D.Double(x, tableBBox.getMaxY()), new Point2D.Double(x, tableBBox.getMinY()));
-            coloredLines.add(new ColoredLine(line, color));
-        });
-        ys.forEach(y -> {
-            Line2D line = new Line2D.Double(new Point2D.Double(tableBBox.getMinX(), y), new Point2D.Double(tableBBox.getMaxX(), y));
-            coloredLines.add(new ColoredLine(line, color));
-        });
+        for (Page page : table.getPages()) {
+            Optional<Integer> optionalFirstRowOnPage = table.streamCol(0).filter(tableCell -> tableCell.isOnPage(page.getNumber())).map(TableCell::getRow).findFirst();
+            if (optionalFirstRowOnPage.isEmpty()) {
+                continue;
+            }
+            int firstRowOnPage = optionalFirstRowOnPage.get();
+            Stream<Double> xStream = switch (page.getRotation()) {
+                case 90 -> streamBBoxOfCellsOnPage(table.streamCol(0), page).map(RectangularShape::getMinX);
+                case 180 -> streamBBoxOfCellsOnPage(table.streamRow(firstRowOnPage), page).map(RectangularShape::getMaxX);
+                case 270 -> streamBBoxOfCellsOnPage(table.streamCol(0), page).map(RectangularShape::getMaxX);
+                default -> streamBBoxOfCellsOnPage(table.streamRow(firstRowOnPage), page).map(RectangularShape::getMinX);
+            };
+            List<Double> xs = xStream.collect(Collectors.toList());
+            xs.remove(0);
+            Stream<Double> yStream = switch (page.getRotation()) {
+                case 90 -> streamBBoxOfCellsOnPage(table.streamRow(firstRowOnPage), page).map(RectangularShape::getMinY);
+                case 180 -> streamBBoxOfCellsOnPage(table.streamCol(0), page).map(RectangularShape::getMinY);
+                case 270 -> streamBBoxOfCellsOnPage(table.streamRow(firstRowOnPage), page).map(RectangularShape::getMaxY);
+                default -> streamBBoxOfCellsOnPage(table.streamCol(0), page).map(RectangularShape::getMaxY);
+            };
+            List<Double> ys = yStream.collect(Collectors.toList());
+            ys.remove(0);
+
+            Rectangle2D tableBBox = table.getBBox().get(table.getFirstPage());
+            List<ColoredLine> coloredLines = layoutGrid.getVisualizationsPerPages().get(page.getNumber() - 1).getColoredLines();
+            xs.forEach(x -> {
+                Line2D line = new Line2D.Double(new Point2D.Double(x, tableBBox.getMaxY()), new Point2D.Double(x, tableBBox.getMinY()));
+                coloredLines.add(new ColoredLine(line, INNER_LINES_COLOR));
+            });
+            ys.forEach(y -> {
+                Line2D line = new Line2D.Double(new Point2D.Double(tableBBox.getMinX(), y), new Point2D.Double(tableBBox.getMaxX(), y));
+                coloredLines.add(new ColoredLine(line, INNER_LINES_COLOR));
+            });
+        }
+    }
+
+
+    private static Stream<Rectangle2D> streamBBoxOfCellsOnPage(Stream<TableCell> table, Page page) {
+
+        return table.filter(tableCell -> tableCell.isOnPage(page.getNumber())).map(TableCell::getBBox).map(bBoxMap -> bBoxMap.get(page));
     }
 
 
@@ -95,17 +123,16 @@ public class LayoutGridService {
 
         Map<Page, Rectangle2D> bBoxMap = semanticNode.getBBox();
         List<SemanticNode> subSections = semanticNode.streamAllSubNodesOfType(NodeType.SECTION).toList();
+        Page firstPage = semanticNode.getFirstPage();
         if (!subSections.isEmpty()) {
-            Page firstPage = semanticNode.getFirstPage();
             addPlacedText(firstPage, bBoxMap.get(firstPage), buildTreeIdString(semanticNode), layoutGrid);
         } else {
             bBoxMap.forEach(((page, textBBox) -> addPlacedText(page, textBBox, buildTreeIdString(semanticNode), layoutGrid)));
         }
         if (bBoxMap.values().size() == 1) {
-            Rectangle2D r = RectangleTransformations.pad(bBoxMap.values().stream().findFirst().get(), LINE_WIDTH, LINE_WIDTH);
-            int pageNumber = bBoxMap.keySet().stream().findFirst().get().getNumber() - 1;
-            List<ColoredLine> coloredLines = layoutGrid.getVisualizationsPerPages().get(pageNumber).getColoredLines();
-            List<Line2D> lines = createLinesFromRectangle(r);
+            Rectangle2D r = RectangleTransformations.pad(bBoxMap.get(firstPage), LINE_WIDTH, LINE_WIDTH);
+            List<ColoredLine> coloredLines = layoutGrid.getVisualizationsPerPages().get(firstPage.getNumber() - 1).getColoredLines();
+            List<Line2D> lines = createLinesFromRectangle(r, firstPage.getRotation());
             // add string to top line
             var firstLine = lines.remove(0);
             coloredLines.add(new ColoredLine(firstLine, color));
@@ -115,7 +142,7 @@ public class LayoutGridService {
             return;
         }
         List<Page> pagesInOrder = bBoxMap.keySet().stream().sorted(Comparator.comparingInt(Page::getNumber)).collect(Collectors.toList());
-        var firstPage = pagesInOrder.remove(0);
+        pagesInOrder.remove(0);
         addLinesForFirstPageOfSection(semanticNode, color, firstPage, layoutGrid);
         var lastPage = pagesInOrder.remove(pagesInOrder.size() - 1);
         addLinesForLastPageOfSection(semanticNode, color, lastPage, layoutGrid);
@@ -127,8 +154,14 @@ public class LayoutGridService {
 
     private void addPlacedText(Page page, Rectangle2D textBBox, String s, LayoutGrid layoutGrid) {
 
+        Point2D.Float upperLeftCorner = switch (page.getRotation()) {
+            case 90 -> new Point2D.Float((float) (textBBox.getMinX()), (float) textBBox.getMinY());
+            case 180 -> new Point2D.Float((float) (textBBox.getMaxX()), (float) textBBox.getMinY());
+            case 270 -> new Point2D.Float((float) (textBBox.getMaxX()), (float) textBBox.getMaxY());
+            default -> new Point2D.Float((float) (textBBox.getMinX()), (float) textBBox.getMaxY());
+        };
         var placedTexts = layoutGrid.getVisualizationsPerPages().get(page.getNumber() - 1).getPlacedTexts();
-        placedTexts.add(new PlacedText(s, new Point2D.Float((float) (textBBox.getMinX()), (float) textBBox.getMaxY())));
+        placedTexts.add(new PlacedText(s, upperLeftCorner));
     }
 
 
@@ -136,7 +169,7 @@ public class LayoutGridService {
 
         List<ColoredLine> coloredLines = layoutGrid.getVisualizationsPerPages().get(middlePage.getNumber() - 1).getColoredLines();
         Rectangle2D r = RectangleTransformations.pad(semanticNode.getBBox().get(middlePage), LINE_WIDTH, LINE_WIDTH);
-        var midPageLines = createLinesFromRectangle(r);
+        var midPageLines = createLinesFromRectangle(r, middlePage.getRotation());
         // remove top line
         midPageLines.remove(0);
         // remove top line
@@ -154,7 +187,7 @@ public class LayoutGridService {
 
         List<ColoredLine> coloredLines = layoutGrid.getVisualizationsPerPages().get(lastPage.getNumber() - 1).getColoredLines();
         Rectangle2D r = RectangleTransformations.pad(semanticNode.getBBox().get(lastPage), LINE_WIDTH, LINE_WIDTH);
-        var lastPageLines = createLinesFromRectangle(r);
+        var lastPageLines = createLinesFromRectangle(r, lastPage.getRotation());
         // remove top line
         lastPageLines.remove(0);
         // add string to left line
@@ -170,7 +203,7 @@ public class LayoutGridService {
 
         List<ColoredLine> coloredLines = layoutGrid.getVisualizationsPerPages().get(firstPage.getNumber() - 1).getColoredLines();
         Rectangle2D r = RectangleTransformations.pad(semanticNode.getBBox().get(firstPage), LINE_WIDTH, LINE_WIDTH);
-        var firstPageLines = createLinesFromRectangle(r);
+        var firstPageLines = createLinesFromRectangle(r, firstPage.getRotation());
         // remove bottom line
         firstPageLines.remove(2);
         // add string to top line
@@ -196,8 +229,9 @@ public class LayoutGridService {
      |                  |
     D|__________________| C
     The returned List are the lines [AB, BC, DC, AD]
+    The List is reordered, such that the order of the returned lines are always as viewed on the page.
      */
-    private List<Line2D> createLinesFromRectangle(Rectangle2D r) {
+    private List<Line2D> createLinesFromRectangle(Rectangle2D r, int pageRotation) {
         // +0.5 to join the lines
         List<Line2D> lines = new ArrayList<>(4);
         float lineWidthCorrection = LINE_WIDTH * 0.5f;
@@ -213,7 +247,23 @@ public class LayoutGridService {
         lines.add(new Line2D.Float(b, c));
         lines.add(new Line2D.Float(d1, c1));
         lines.add(new Line2D.Float(a, d));
-        return lines;
+
+        return switch (pageRotation) {
+            case 90 -> {
+                Collections.rotate(lines, 1);
+                yield lines;
+            }
+            case 180 -> {
+                Collections.rotate(lines, 2);
+                yield lines;
+            }
+            case 270 -> {
+                Collections.rotate(lines, 3);
+                yield lines;
+            }
+
+            default -> lines;
+        };
     }
 
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/visualization/ViewerDocumentService.java

@@ -42,7 +42,7 @@ public class ViewerDocumentService {
     private static final String layerName = "Layout grid";
 
     private static final int FONT_SIZE = 10;
-    public static final float LINE_WIDTH = 1.5f;
+    public static final float LINE_WIDTH = 1f;
 
     private final LayoutGridService layoutGridService;
 
@@ -118,6 +118,7 @@ public class ViewerDocumentService {
             dictionariesToUpdate.add(pdPage.getCOSObject());
             dictionariesToUpdate.add(pdPage.getResources().getCOSObject());
         }
+        dictionariesToUpdate.add(pdDocument.getDocumentInformation().getCOSObject());
         pdDocument.saveIncremental(outputStream, dictionariesToUpdate);
         log.info("Saved Viewer Document");
     }

layoutparser-service/layoutparser-service-server/build.gradle.kts

@@ -19,10 +19,10 @@ dependencies {
     implementation(project(":layoutparser-service-processor"))
     implementation(project(":layoutparser-service-internal-api"))
 
-    implementation("com.iqser.red.commons:storage-commons:2.27.0")
+    implementation("com.iqser.red.commons:storage-commons:2.36.0")
     implementation("com.knecon.fforesight:tenant-commons:0.10.0")
     implementation("org.springframework.boot:spring-boot-starter-actuator:3.1.2")
-    implementation("com.amazonaws:aws-java-sdk-s3:1.12.529")
+    implementation("com.amazonaws:aws-java-sdk-s3:1.12.533")
     implementation("org.springframework.boot:spring-boot-starter-amqp:3.1.2")
     implementation("org.springframework.cloud:spring-cloud-starter-openfeign:4.0.4")
 
@@ -30,7 +30,7 @@ dependencies {
     testImplementation(project(":layoutparser-service-internal-api"))
 
     testImplementation("org.springframework.boot:spring-boot-starter-amqp:3.1.2")
-    testImplementation("com.iqser.red.service:persistence-service-shared-api-v1:2.126.0")
+    testImplementation("com.iqser.red.service:persistence-service-shared-api-v1:2.138.0")
     testImplementation("com.iqser.red.commons:jackson-commons:1.3.0")
     testImplementation("com.fasterxml.jackson.module:jackson-module-afterburner:${jacksonVersion}")
     testImplementation("com.fasterxml.jackson.datatype:jackson-datatype-jsr310:${jacksonVersion}")

layoutparser-service/layoutparser-service-server/src/test/java/com/knecon/fforesight/service/layoutparser/server/BdrJsonBuildTest.java

@@ -1,5 +1,7 @@
 package com.knecon.fforesight.service.layoutparser.server;
 
+import static org.assertj.core.api.AssertionsForClassTypes.assertThat;
+
 import java.awt.Color;
 import java.io.File;
 import java.io.FileInputStream;
@@ -17,6 +19,7 @@ import org.apache.pdfbox.pdmodel.PDDocument;
 import org.junit.jupiter.api.Disabled;
 import org.junit.jupiter.api.Test;
 import org.springframework.beans.factory.annotation.Autowired;
+import org.springframework.core.io.ClassPathResource;
 
 import com.fasterxml.jackson.databind.ObjectMapper;
 import com.knecon.fforesight.service.layoutparser.internal.api.data.taas.ResearchDocumentData;
@@ -56,6 +59,18 @@ public class BdrJsonBuildTest extends AbstractTest {
     }
 
 
+    @Test
+    @SneakyThrows
+    public void testBDRFile199865() {
+
+        File file = new ClassPathResource("files/bdr/Drucksache_19_9865.pdf").getFile();
+        Document document = buildGraph(file);
+        ResearchDocumentData researchDocumentData = TaasDocumentDataMapper.fromDocument(document);
+        assertThat(researchDocumentData).isNotNull();
+
+    }
+
+
     @Test
     @Disabled
     public void writeBDRDocumentData() throws IOException {