Merge branch 'RED-9976-main' into 'main'

RED-9976: Removed sorting that scrambles text in PDFTextStripper

See merge request fforesight/layout-parser!222

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

@@ -16,7 +16,7 @@ dependencies {
         exclude("org.springframework.boot", "spring-boot-starter-security")
         exclude("org.springframework.boot", "spring-boot-starter-validation")
     }
-    implementation("com.knecon.fforesight:tenant-commons:0.29.0")
+    implementation("com.knecon.fforesight:tenant-commons:0.30.0")
     implementation("com.iqser.red.commons:storage-commons:2.45.0")
 
     implementation("org.apache.pdfbox:pdfbox:${pdfBoxVersion}")
@@ -25,6 +25,9 @@ dependencies {
     implementation("com.fasterxml.jackson.datatype:jackson-datatype-jsr310:${jacksonVersion}")
     implementation("org.springframework.boot:spring-boot-starter-web:3.1.3")
     implementation("org.jgrapht:jgrapht-core:1.5.2")
+    implementation("org.apache.pdfbox:jbig2-imageio:3.0.4")
+    implementation("com.github.jai-imageio:jai-imageio-core:1.4.0")
+    implementation("com.github.jai-imageio:jai-imageio-jpeg2000:1.4.0")
     implementation("org.tinspin:tinspin-indexes:2.1.3")
     implementation("org.commonmark:commonmark:0.22.0")
     implementation("org.commonmark:commonmark-ext-gfm-tables:0.22.0")

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/graph/nodes/Page.java

@@ -68,6 +68,7 @@ public class Page {
     public TextBlock getMainBodyTextBlock() {
 
         return textBlocksOnPage.stream()
+                .filter(atb -> !atb.isEmpty())
                 .collect(new TextBlockCollector());
     }
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/image/ClassifiedImage.java

@@ -22,11 +22,10 @@ public class ClassifiedImage {
     private boolean isAppendedToSection;
     private boolean hasTransparency;
     private int page;
-    @NonNull
     private String representation;
 
 
-    public ClassifiedImage(@NonNull Rectangle2D position, @NonNull ImageType imageType, boolean hasTransparency, int page, @NonNull String representation) {
+    public ClassifiedImage(@NonNull Rectangle2D position, @NonNull ImageType imageType, boolean hasTransparency, int page, String representation) {
 
         this.position = position;
         this.imageType = imageType;

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/model/outline/TOCEnrichmentService.java

@@ -1,6 +1,7 @@
 package com.knecon.fforesight.service.layoutparser.processor.model.outline;
 
 import java.util.ArrayList;
+import java.util.Collection;
 import java.util.Collections;
 import java.util.Iterator;
 import java.util.List;
@@ -167,7 +168,7 @@ public class TOCEnrichmentService {
             }
         }
 
-        if (!startBlocks.isEmpty()) {
+        if (!startBlocks.isEmpty() || !startImages.isEmpty()) {
             TableOfContentItem unassigned = new TableOfContentItem(null);
             unassigned.setSectionBlocks(startBlocks);
             unassigned.setImages(startImages);
@@ -185,12 +186,8 @@ public class TOCEnrichmentService {
             List<Cell> previousTableNonHeaderRow = getRowWithNonHeaderCells(previousTable);
             List<Cell> tableNonHeaderRow = getRowWithNonHeaderCells(currentTable);
             // Allow merging of tables if header row is separated from first logical non-header row
-            if (previousTableNonHeaderRow.isEmpty()
-                && previousTable.getRowCount() == 1
-                && previousTable.getRows()
-                           .get(0).size() == tableNonHeaderRow.size()) {
-                previousTableNonHeaderRow = previousTable.getRows()
-                        .get(0)
+            if (previousTableNonHeaderRow.isEmpty() && previousTable.getRowCount() == 1 && previousTable.getRows().get(0).size() == tableNonHeaderRow.size()) {
+                previousTableNonHeaderRow = previousTable.getRows().get(0)
                         .stream()
                         .map(cell -> {
                             Cell fakeCell = Cell.copy(cell);
@@ -201,8 +198,7 @@ public class TOCEnrichmentService {
             }
             if (previousTableNonHeaderRow.size() == tableNonHeaderRow.size()) {
                 for (int i = currentTable.getRowCount() - 1; i >= 0; i--) { // Non header rows are most likely at bottom of table
-                    List<Cell> row = currentTable.getRows()
-                            .get(i);
+                    List<Cell> row = currentTable.getRows().get(i);
                     if (row.size() == tableNonHeaderRow.size() && row.stream()
                             .allMatch(cell -> cell.getHeaderCells().isEmpty())) {
                         for (int j = 0; j < row.size(); j++) {
@@ -225,18 +221,15 @@ public class TOCEnrichmentService {
 
         return table.getRows()
                 .stream()
-                .flatMap(row -> row.stream()
-                        .filter(cell -> !cell.getHeaderCells().isEmpty()))
-                .findAny().isEmpty();
-
+                .flatMap(Collection::stream)
+                .allMatch(cell -> cell.getHeaderCells().isEmpty());
     }
 
 
     private List<Cell> getRowWithNonHeaderCells(TablePageBlock table) {
 
         for (int i = table.getRowCount() - 1; i >= 0; i--) { // Non header rows are most likely at bottom of table
-            List<Cell> row = table.getRows()
-                    .get(i);
+            List<Cell> row = table.getRows().get(i);
             if (row.size() == 1) {
                 continue;
             }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/factory/DocumentGraphFactory.java

@@ -65,7 +65,7 @@ public class DocumentGraphFactory {
 
         document.getPages()
                 .forEach(context::buildAndAddPageWithCounter);
-        addSections(layoutParsingType, document, context, documentGraph);
+        addSectionsForToC(layoutParsingType, document, context, documentGraph);
         addHeaderAndFooterToEachPage(document, context);
 
         documentGraph.setNumberOfPages(context.pages.size());
@@ -84,6 +84,7 @@ public class DocumentGraphFactory {
                 .filter(SemanticNode::isLeaf)
                 .filter(node -> !node.getType().equals(NodeType.HEADER))
                 .filter(node -> !node.getType().equals(NodeType.FOOTER))
+                .filter(node -> !node.getType().equals(NodeType.IMAGE))
                 .map(SemanticNode::getTextBlock)
                 .map(TextBlock::getAtomicTextBlocks)
                 .flatMap(Collection::stream)
@@ -91,7 +92,7 @@ public class DocumentGraphFactory {
     }
 
 
-    private void addSections(LayoutParsingType layoutParsingType, ClassificationDocument classificationDocument, Context context, Document document) {
+    private void addSectionsForToC(LayoutParsingType layoutParsingType, ClassificationDocument classificationDocument, Context context, Document document) {
 
         for (TableOfContentItem tocItem : classificationDocument.getTableOfContents()) {
             var parent = tocItem.getParent() == null ? null : tocItem.getParent().getSection();
@@ -259,7 +260,6 @@ public class DocumentGraphFactory {
         DocumentTree documentTree;
         Map<Page, Integer> pages;
         List<AbstractSemanticNode> sections;
-        List<ClassifiedImage> images;
         TextBlockFactory textBlockFactory;
 
 
@@ -268,7 +268,6 @@ public class DocumentGraphFactory {
             documentTree = new DocumentTree(document);
             pages = new HashMap<>();
             sections = new LinkedList<>();
-            images = new LinkedList<>();
             textBlockFactory = new TextBlockFactory();
         }
 

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/factory/TableNodeFactory.java

@@ -36,10 +36,7 @@ public class TableNodeFactory {
                          Document document) {
 
         setPageNumberInCells(tablesToMerge);
-        Set<Page> pages = tablesToMerge.stream()
-                .map(AbstractPageBlock::getPage)
-                .map(context::getPage)
-                .collect(Collectors.toSet());
+
         List<List<Cell>> mergedRows = tablesToMerge.stream()
                 .map(TablePageBlock::getRows)
                 .flatMap(Collection::stream)

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/mapper/DocumentGraphMapper.java

@@ -84,6 +84,7 @@ public class DocumentGraphMapper {
                 switch (entryData.getType()) {
                     case HEADER -> pages.forEach(page -> page.setHeader((Header) node));
                     case FOOTER -> pages.forEach(page -> page.setFooter((Footer) node));
+                    case IMAGE -> pages.forEach(page -> page.getImages().add((Image) node));
                     default -> textBlock.getAtomicTextBlocks()
                             .forEach(atb -> atb.getPage().getTextBlocksOnPage().add(atb));
                 }

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

@@ -25,10 +25,22 @@ import java.io.StringWriter;
 import java.io.Writer;
 import java.text.Bidi;
 import java.text.Normalizer;
-import java.util.*;
+import java.util.ArrayDeque;
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.Deque;
+import java.util.HashMap;
+import java.util.Iterator;
+import java.util.LinkedList;
+import java.util.List;
+import java.util.Map;
+import java.util.SortedMap;
+import java.util.SortedSet;
+import java.util.StringTokenizer;
+import java.util.TreeMap;
+import java.util.TreeSet;
 import java.util.regex.Pattern;
 
-import lombok.Getter;
 import org.apache.commons.logging.Log;
 import org.apache.commons.logging.LogFactory;
 import org.apache.pdfbox.cos.COSDictionary;
@@ -46,6 +58,8 @@ import org.apache.pdfbox.text.TextPositionComparator;
 
 import com.knecon.fforesight.service.layoutparser.processor.utils.QuickSort;
 
+import lombok.Getter;
+
 /**
  * This is just a copy except i only adjusted lines 594-607 cause this is a bug in Pdfbox.
  * see S416.pdf
@@ -194,40 +208,33 @@ public class PDFTextStripper extends LegacyPDFStreamEngine {
     }
 
 
+    public void beginMarkedContentSequence(COSName tag, COSDictionary properties) {
 
-    public void beginMarkedContentSequence(COSName tag, COSDictionary properties)
-    {
         PDMarkedContent markedContent = PDMarkedContent.create(tag, properties);
-        if (this.currentMarkedContents.isEmpty())
-        {
+        if (this.currentMarkedContents.isEmpty()) {
             this.markedContents.add(markedContent);
-        }
-        else
-        {
-            PDMarkedContent currentMarkedContent =
-                    this.currentMarkedContents.peek();
-            if (currentMarkedContent != null)
-            {
+        } else {
+            PDMarkedContent currentMarkedContent = this.currentMarkedContents.peek();
+            if (currentMarkedContent != null) {
                 currentMarkedContent.addMarkedContent(markedContent);
             }
         }
         this.currentMarkedContents.push(markedContent);
     }
 
+
     @Override
-    public void endMarkedContentSequence()
-    {
-        if (!this.currentMarkedContents.isEmpty())
-        {
+    public void endMarkedContentSequence() {
+
+        if (!this.currentMarkedContents.isEmpty()) {
             this.currentMarkedContents.pop();
         }
     }
 
 
-    public void xobject(PDXObject xobject)
-    {
-        if (!this.currentMarkedContents.isEmpty())
-        {
+    public void xobject(PDXObject xobject) {
+
+        if (!this.currentMarkedContents.isEmpty()) {
             this.currentMarkedContents.peek().addXObject(xobject);
         }
     }
@@ -313,7 +320,11 @@ public class PDFTextStripper extends LegacyPDFStreamEngine {
             endBookmarkPageNumber = -1;
         }
 
-        if (startBookmarkPageNumber == -1 && startBookmark != null && endBookmarkPageNumber == -1 && endBookmark != null && startBookmark.getCOSObject() == endBookmark.getCOSObject()) {
+        if (startBookmarkPageNumber == -1
+            && startBookmark != null
+            && endBookmarkPageNumber == -1
+            && endBookmark != null
+            && startBookmark.getCOSObject() == endBookmark.getCOSObject()) {
             // this is a special case where both the start and end bookmark
             // are the same but point to nothing. In this case
             // we will not extract any text.
@@ -360,7 +371,9 @@ public class PDFTextStripper extends LegacyPDFStreamEngine {
     @Override
     public void processPage(PDPage page) throws IOException {
 
-        if (currentPageNo >= startPage && currentPageNo <= endPage && (startBookmarkPageNumber == -1 || currentPageNo >= startBookmarkPageNumber) && (endBookmarkPageNumber == -1 || currentPageNo <= endBookmarkPageNumber)) {
+        if (currentPageNo >= startPage && currentPageNo <= endPage && (startBookmarkPageNumber == -1 || currentPageNo >= startBookmarkPageNumber) && (endBookmarkPageNumber == -1
+                                                                                                                                                      || currentPageNo
+                                                                                                                                                         <= endBookmarkPageNumber)) {
             startPage(page);
 
             int numberOfArticleSections = 1;
@@ -635,7 +648,6 @@ public class PDFTextStripper extends LegacyPDFStreamEngine {
                         var normalized = normalize(line);
 //                        normalized.stream().filter(l -> System.out.println(l.getText().contains("Plenarprotokoll 20/24")).findFirst().isPresent()
 
-
                         lastLineStartPosition = handleLineSeparation(current, lastPosition, lastLineStartPosition, maxHeightForLine);
                         writeLine(normalized, current.isParagraphStart);
                         line.clear();
@@ -647,8 +659,8 @@ public class PDFTextStripper extends LegacyPDFStreamEngine {
                     }
                     // test if our TextPosition starts after a new word would be expected to start
                     if (expectedStartOfNextWordX != EXPECTED_START_OF_NEXT_WORD_X_RESET_VALUE && expectedStartOfNextWordX < positionX
-                            // only bother adding a word separator if the last character was not a word separator
-                            && (wordSeparator.isEmpty() || //
+                        // only bother adding a word separator if the last character was not a word separator
+                        && (wordSeparator.isEmpty() || //
                             (lastPosition.getTextPosition().getUnicode() != null && !lastPosition.getTextPosition().getUnicode().endsWith(wordSeparator)))) {
                         line.add(LineItem.getWordSeparator());
                     }
@@ -914,8 +926,7 @@ public class PDFTextStripper extends LegacyPDFStreamEngine {
                     textList.add(text);
                 }
             }
-            if (!this.currentMarkedContents.isEmpty())
-            {
+            if (!this.currentMarkedContents.isEmpty()) {
                 this.currentMarkedContents.peek().addText(text);
             }
         }
@@ -1711,7 +1722,6 @@ public class PDFTextStripper extends LegacyPDFStreamEngine {
         int numberOfStrings = line.size();
         for (int i = 0; i < numberOfStrings; i++) {
             WordWithTextPositions word = line.get(i);
-            word.getTextPositions().sort(Comparator.comparing(TextPosition::getXDirAdj));
             writeString(word.getText(), word.getTextPositions(), isParagraphEnd && i == numberOfStrings - 1);
             if (i < numberOfStrings - 1) {
                 writeWordSeparator();
@@ -2102,7 +2112,9 @@ public class PDFTextStripper extends LegacyPDFStreamEngine {
             return endParagraphWritten;
         }
 
-        public void setEndParagraphWritten(){
+
+        public void setEndParagraphWritten() {
+
             endParagraphWritten = true;
         }
 
@@ -2145,7 +2157,6 @@ public class PDFTextStripper extends LegacyPDFStreamEngine {
             this.isHangingIndent = true;
         }
 
-
     }
 
 }

layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/utils/TableMergingUtility.java

@@ -36,7 +36,7 @@ public class TableMergingUtility {
             TablePageBlock consecutiveTable = consecutiveTables.get(i);
 
             if (consecutiveTable.getColCount() == originalTablePageBlock.getColCount() //
-                && headersMatch(originalTablePageBlock, consecutiveTable) //
+                && getHeaders(consecutiveTable).isEmpty() //
                 && outerBoundaryAlignsX(originalTablePageBlock, consecutiveTable) //
                 && consecutiveOrSamePage(currentTable, consecutiveTable) //
                 && !tableBetween(currentTable, consecutiveTable, findTablesBetween(consecutiveTables, currentTableIndex, i))) {
@@ -80,12 +80,6 @@ public class TableMergingUtility {
     }
 
 
-    private static boolean headersMatch(TablePageBlock originalTable, TablePageBlock consecutiveTable) {
-
-        return getHeaders(consecutiveTable).isEmpty() || getHeaders(originalTable).equals(getHeaders(consecutiveTable));
-    }
-
-
     private static boolean outerBoundaryAlignsX(TablePageBlock originalTablePageBlock, TablePageBlock consecutiveTable) {
 
         return Math.abs(consecutiveTable.getMinX() - originalTablePageBlock.getMinX()) < TABLE_ALIGNMENT_THRESHOLD

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

@@ -30,7 +30,7 @@ dependencies {
     implementation(project(":layoutparser-service-internal-api"))
 
     implementation("com.iqser.red.commons:storage-commons:2.45.0")
-    implementation("com.knecon.fforesight:tenant-commons:0.29.0")
+    implementation("com.knecon.fforesight:tenant-commons:0.30.0")
     implementation("com.knecon.fforesight:tracing-commons:0.5.0")
     implementation("com.knecon.fforesight:lifecycle-commons:0.6.0")
     implementation("org.springframework.boot:spring-boot-starter-actuator:${springBootStarterVersion}")