Merge branch 'RED-9942-2' into 'main'

RED-9942: File only with images not recognised

See merge request fforesight/layout-parser!218

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

@@ -167,7 +167,7 @@ public class TOCEnrichmentService {
             }
         }
 
-        if (!startBlocks.isEmpty()) {
+        if (!startBlocks.isEmpty() || !startImages.isEmpty()) {
             TableOfContentItem unassigned = new TableOfContentItem(null);
             unassigned.setSectionBlocks(startBlocks);
             unassigned.setImages(startImages);

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/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-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}")