Add SemanticAnnot

veraPDF · Dec 5, 2023 · b6f4c57 · b6f4c57
1 parent 7b1be7a
commit b6f4c57
Show file tree

Hide file tree

Showing 11 changed files with 335 additions and 19 deletions.
diff --git a/src/main/java/org/verapdf/wcag/algorithms/entities/AnnotationNode.java b/src/main/java/org/verapdf/wcag/algorithms/entities/AnnotationNode.java
@@ -1,16 +1,16 @@
 package org.verapdf.wcag.algorithms.entities;
 
-import org.verapdf.wcag.algorithms.entities.enums.SemanticType;
+import org.verapdf.wcag.algorithms.entities.content.InfoChunk;
 import org.verapdf.wcag.algorithms.entities.geometry.BoundingBox;
 
-public class AnnotationNode extends SemanticNode implements IAnnotation {
+public class AnnotationNode extends InfoChunk implements IAnnotation {
 	private final String annotationType;
 	private final Integer destinationPageNumber;
 	private final Integer destinationObjectKeyNumber;
 
 	public AnnotationNode(String annotationType, BoundingBox boundingBox, Integer destinationPageNumber,
 						  Integer destinationObjectKeyNumber) {
-		super(boundingBox, SemanticType.ANNOT, SemanticType.ANNOT);
+		super(boundingBox);
 		this.annotationType = annotationType;
 		this.destinationPageNumber = destinationPageNumber;
 		this.destinationObjectKeyNumber = destinationObjectKeyNumber;

diff --git a/src/main/java/org/verapdf/wcag/algorithms/entities/IAnnotation.java b/src/main/java/org/verapdf/wcag/algorithms/entities/IAnnotation.java
@@ -1,6 +1,8 @@
 package org.verapdf.wcag.algorithms.entities;
 
-public interface IAnnotation extends INode {
+import org.verapdf.wcag.algorithms.entities.content.IChunk;
+
+public interface IAnnotation extends IChunk {
 	String getAnnotationType();
 
 	Integer getDestinationPageNumber();

diff --git a/src/main/java/org/verapdf/wcag/algorithms/entities/SemanticAnnot.java b/src/main/java/org/verapdf/wcag/algorithms/entities/SemanticAnnot.java
@@ -0,0 +1,66 @@
+package org.verapdf.wcag.algorithms.entities;
+
+import org.verapdf.wcag.algorithms.entities.enums.SemanticType;
+
+import java.util.LinkedList;
+import java.util.List;
+
+public class SemanticAnnot extends SemanticNode {
+
+	protected final List<AnnotationNode> annotationNodes = new LinkedList<>();
+
+    public SemanticAnnot(SemanticAnnot annotationNode) {
+        this.addAnnots(annotationNode.getAnnots());
+        setSemanticType(SemanticType.ANNOT);
+    }
+
+    public SemanticAnnot(AnnotationNode annotationNode) {
+        super(annotationNode.getBoundingBox());
+        this.annotationNodes.add(annotationNode);
+        setSemanticType(SemanticType.ANNOT);
+    }
+
+    public void addAnnots(List<AnnotationNode> annots) {
+        this.annotationNodes.addAll(annots);
+        for (AnnotationNode annotationNode : annots) {
+            getBoundingBox().union(annotationNode.getBoundingBox());
+        }
+    }
+
+	public List<AnnotationNode> getAnnots() {
+		return annotationNodes;
+	}
+
+    @Override
+    public boolean equals(Object o) {
+        if (!super.equals(o)) {
+            return false;
+        }
+        if (!(o instanceof SemanticAnnot)) {
+            return false;
+        }
+        SemanticAnnot that = (SemanticAnnot) o;
+        return this.annotationNodes.equals(that.getAnnots());
+    }
+
+    @Override
+    public int hashCode() {
+        int result = super.hashCode();
+        result = 31 * result + annotationNodes.size();
+        for (AnnotationNode annotationNode : annotationNodes) {
+            result = 31 * result + annotationNode.hashCode();
+        }
+        return result;
+    }
+
+    @Override
+    public String toString() {
+        StringBuilder result = new StringBuilder("SemanticAnnot{");
+        result.append("pageNumber=");
+        result.append(getBoundingBox().getPageNumber());
+        result.append(", boundingBox=");
+        result.append(getBoundingBox());
+        result.append("}");
+        return result.toString();
+    }
+}
diff --git a/...ava/org/verapdf/wcag/algorithms/semanticalgorithms/consumers/AccumulatedNodeConsumer.java b/...ava/org/verapdf/wcag/algorithms/semanticalgorithms/consumers/AccumulatedNodeConsumer.java
@@ -35,11 +35,13 @@ public void accept(INode node) {
 		boolean isLeafChild = node.getChildren()
 				.stream()
 				.allMatch(child -> ((child instanceof SemanticSpan) ||
-						(child instanceof SemanticFigure) ||
+						(child instanceof SemanticFigure) || (child instanceof SemanticAnnot) ||
 						child.getSemanticType() == null));
 
 		acceptSpanParagraphPart(node, isLeafChild);
 
+		acceptSemanticAnnot(node);
+
 		acceptSemanticImage(node);
 
 		checkSemanticSpanChildren(node);
@@ -105,13 +107,34 @@ private void acceptSemanticImage(INode node) {
 		}
 	}
 
+	private void acceptSemanticAnnot(INode node) {
+		SemanticAnnot annot = null;
+		for (INode child : node.getChildren()) {
+			INode accumulatedChild = StaticContainers.getAccumulatedNodeMapper().get(child);
+			if (accumulatedChild instanceof SemanticTextNode) {
+				if (!((SemanticTextNode)accumulatedChild).isEmpty() && !((SemanticTextNode)accumulatedChild).isSpaceNode()) {
+					return;
+				}
+			} else if (accumulatedChild instanceof SemanticAnnot) {
+				if (annot == null) {
+					annot = new SemanticAnnot((SemanticAnnot)accumulatedChild);
+				} else {
+					annot.addAnnots(((SemanticAnnot)accumulatedChild).getAnnots());
+				}
+			}
+		}
+		if (annot != null) {
+			StaticContainers.getAccumulatedNodeMapper().updateNode(node, annot, 1.0, SemanticType.ANNOT);
+		}
+	}
+
 	private void acceptSpanParagraphPart(INode node, boolean isLeafChild) {
 		double probability = 1;
 		SemanticPart part = null;
 		MultiBoundingBox boundingBox = new MultiBoundingBox();
 		for (INode child : node.getChildren()) {
 			if (child.getSemanticType() == null || SemanticType.isIgnoredStandardType(child.getInitialSemanticType()) ||
-					 (child instanceof IAnnotation)) {
+					 (child instanceof SemanticAnnot)) {
 				continue;
 			}
 			INode accumulatedChild = StaticContainers.getAccumulatedNodeMapper().get(child);
@@ -143,7 +166,7 @@ private void acceptSpanParagraphPart(INode node, boolean isLeafChild) {
 		SemanticType semanticType = SemanticType.PART;
 		INode accumulatedNode = part;
 		if (part != null && part.getColumns().stream().allMatch(TextColumn::hasOnlyOneBlock)) {
-			boolean isSpan  = SemanticType.SPAN.equals(node.getInitialSemanticType()) &&
+			boolean isSpan = SemanticType.SPAN.equals(node.getInitialSemanticType()) &&
 					(isLeafChild || node.getChildren().stream()
 							.allMatch(AccumulatedNodeConsumer::isAppropriateSpanChild));
 			if (isSpan) {
@@ -158,7 +181,7 @@ private void acceptSpanParagraphPart(INode node, boolean isLeafChild) {
 	}
 
 	private static boolean isAppropriateSpanChild(INode child) {
-		return (child instanceof SemanticSpan) || (child instanceof SemanticFigure) ||
+		return (child instanceof SemanticSpan) || (child instanceof SemanticFigure) || (child instanceof SemanticAnnot) ||
 				child.getInitialSemanticType() == SemanticType.LINK ||
 				child.getSemanticType() == SemanticType.SPAN || child.getSemanticType() == null;
 	}

diff --git a/...n/java/org/verapdf/wcag/algorithms/semanticalgorithms/consumers/ClusterTableConsumer.java b/...n/java/org/verapdf/wcag/algorithms/semanticalgorithms/consumers/ClusterTableConsumer.java
@@ -552,9 +552,6 @@ public static boolean isNodeInsideTable(INode node, Long id, BoundingBox boundin
         if (node.getRecognizedStructureId() == id) {
             return true;
         }
-        if (node instanceof SemanticFigure) {
-            return true;
-        }
         if (node instanceof SemanticTextNode) {
             SemanticTextNode textNode = (SemanticTextNode) node;
             return isTextNodeInsideTable(textNode, boundingBox);

diff --git a/...java/org/verapdf/wcag/algorithms/semanticalgorithms/consumers/HeadingCaptionConsumer.java b/...java/org/verapdf/wcag/algorithms/semanticalgorithms/consumers/HeadingCaptionConsumer.java
@@ -20,7 +20,7 @@ public void accept(INode node) {
 		boolean isLeafChild = node.getChildren()
 				.stream()
 				.allMatch(child -> ((child instanceof SemanticSpan) ||
-						(child instanceof SemanticFigure) ||
+						(child instanceof SemanticFigure) || (child instanceof SemanticAnnot) ||
 						child.getSemanticType() == null));
 		if (isLeafChild) {
 			return;

diff --git a/.../wcag/algorithms/semanticalgorithms/consumers/SemanticDocumentPostprocessingConsumer.java b/.../wcag/algorithms/semanticalgorithms/consumers/SemanticDocumentPostprocessingConsumer.java
@@ -30,7 +30,7 @@ public void updateBoundingBoxes(ITree tree) {
 		for (INode node : tree) {
 			MultiBoundingBox boundingBox = new MultiBoundingBox();
 			for (INode child : node.getChildren()) {
-				if (!(child instanceof IAnnotation)) {
+				if (!(child instanceof SemanticAnnot)) {
 					boundingBox.union(child.getBoundingBox());
 				}
 			}

diff --git a/...f/wcag/algorithms/semanticalgorithms/consumers/SemanticDocumentPreprocessingConsumer.java b/...f/wcag/algorithms/semanticalgorithms/consumers/SemanticDocumentPreprocessingConsumer.java
@@ -75,7 +75,7 @@ public void accept(INode node) {
         structElementsNumber++;
         MultiBoundingBox boundingBox = new MultiBoundingBox(node.getBoundingBox());
         for (INode child : node.getChildren()) {
-            if (!(child instanceof SemanticFigure) && !(child instanceof IAnnotation)) {
+            if (!(child instanceof SemanticFigure) && !(child instanceof SemanticAnnot)) {
                 boundingBox.union(child.getBoundingBox());
             }
         }

diff --git a/...n/java/org/verapdf/wcag/algorithms/semanticalgorithms/consumers/TOCDetectionConsumer.java b/...n/java/org/verapdf/wcag/algorithms/semanticalgorithms/consumers/TOCDetectionConsumer.java
@@ -514,8 +514,8 @@ private List<TextChunk> getTextChunks(INode node, Integer pageNumber) {
     }
 
     private static List<IAnnotation> getInheritorAnnotations(INode node) {
-        if (node instanceof IAnnotation) {
-            return Collections.singletonList((IAnnotation)node);
+        if (node instanceof SemanticAnnot) {
+            return Collections.singletonList(((SemanticAnnot) node).getAnnots().get(0));
         }
         List<IAnnotation> annotations = new LinkedList<>();
         for (INode child : node.getChildren()) {

diff --git a/src/test/java/org/verapdf/wcag/algorithms/entities/JsonToPdfTree.java b/src/test/java/org/verapdf/wcag/algorithms/entities/JsonToPdfTree.java
@@ -55,9 +55,9 @@ private static LineArtChunk getLineArtChunk(JsonLineArtChunk jsonNode) {
 		return new LineArtChunk(new BoundingBox(jsonNode.getPageNumber(), jsonNode.getBoundingBox()), lineChunks);
 	}
 
-	private static AnnotationNode getAnnotationNode(JsonAnnotationNode jsonNode) {
-		return new AnnotationNode(jsonNode.getAnnotationType(), new BoundingBox(jsonNode.getBoundingBox()),
-		                          jsonNode.getDestinationPageNumber(), jsonNode.getDestinationObjectKeyNumber());
+	private static SemanticAnnot getAnnotationNode(JsonAnnotationNode jsonNode) {
+		return new SemanticAnnot(new AnnotationNode(jsonNode.getAnnotationType(), new BoundingBox(jsonNode.getBoundingBox()),
+				jsonNode.getDestinationPageNumber(), jsonNode.getDestinationObjectKeyNumber()));
 	}
 
 	private static LineChunk getLineChunk(JsonLineChunk jsonNode) {