Add automatic link alternative description parsing

DEVSIX-7996

Author: Dmitry Radchuk

src/main/java/com/itextpdf/html2pdf/attach/impl/tags/ABlockTagWorker.java

@@ -71,7 +71,7 @@ public void processEnd(IElementNode element, ProcessorContext context) {
                     }
             }
             ((Div) getElementResult()).getAccessibilityProperties().setRole(StandardRoles.LINK);
-            LinkHelper.applyLinkAnnotation(getElementResult(), url, context);
+            LinkHelper.applyLinkAnnotation(getElementResult(), url, context, element);
         }
 
         if (getElementResult() != null) {

src/main/java/com/itextpdf/html2pdf/attach/impl/tags/ATagWorker.java

@@ -100,7 +100,7 @@ public void processEnd(IElementNode element, ProcessorContext context) {
                     }
                     getAllElements().set(i, simulatedDiv);
                 }
-                LinkHelper.applyLinkAnnotation(getAllElements().get(i), url, context);
+                LinkHelper.applyLinkAnnotation(getAllElements().get(i), url, context, element);
             }
         }
 

src/main/java/com/itextpdf/html2pdf/attach/util/LinkHelper.java

@@ -36,13 +36,18 @@ This file is part of the iText (R) project.
 import com.itextpdf.kernel.pdf.annot.PdfAnnotation;
 import com.itextpdf.kernel.pdf.annot.PdfLinkAnnotation;
 import com.itextpdf.kernel.pdf.tagging.StandardRoles;
+import com.itextpdf.kernel.pdf.tagutils.AccessibilityProperties;
 import com.itextpdf.layout.tagging.IAccessibleElement;
 import com.itextpdf.layout.IPropertyContainer;
 import com.itextpdf.layout.element.ILeafElement;
 import com.itextpdf.layout.properties.Property;
 import com.itextpdf.html2pdf.html.AttributeConstants;
 import com.itextpdf.styledxmlparser.node.IElementNode;
 import java.util.List;
+
+import com.itextpdf.styledxmlparser.node.INode;
+import com.itextpdf.styledxmlparser.node.impl.jsoup.node.JsoupElementNode;
+import com.itextpdf.styledxmlparser.node.impl.jsoup.node.JsoupNode;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
@@ -80,7 +85,34 @@ public static void applyLinkAnnotation(IPropertyContainer container, String url)
      * @param url       the destination.
      * @param context   the processor context.
      */
+    @Deprecated
     public static void applyLinkAnnotation(IPropertyContainer container, String url, ProcessorContext context) {
+        applyLinkAnnotation(container, url, context, "");
+    }
+
+    private static String retrieveAlternativeDescription(IElementNode element) {
+        List<INode> children = element.childNodes();
+        //if there is an img tag under the link then prefer the alt attribute as a link description
+        if (children.size() == 1
+                && children.get(0).childNodes().isEmpty()
+                && children.get(0) instanceof JsoupElementNode
+                && ((JsoupElementNode)children.get(0)).getAttribute(AttributeConstants.ALT) != null) {
+            return ((JsoupElementNode)children.get(0)).getAttribute(AttributeConstants.ALT);
+        }
+        //return title attribute value in case of regular link
+        return element.getAttribute(AttributeConstants.TITLE);
+    }
+
+    /**
+     * Applies a link annotation.
+     *
+     * @param container the containing object.
+     * @param url       the destination.
+     * @param context   the processor context.
+     * @param alternateDescription description for a link.
+     */
+    @Deprecated
+    public static void applyLinkAnnotation(IPropertyContainer container, String url, ProcessorContext context, String alternateDescription) {
         if (container != null) {
             PdfLinkAnnotation linkAnnotation;
             if (url.startsWith("#")) {
@@ -94,6 +126,9 @@ public static void applyLinkAnnotation(IPropertyContainer container, String url,
             } else {
                 linkAnnotation = (PdfLinkAnnotation) new PdfLinkAnnotation(new Rectangle(0, 0, 0, 0)).setAction(PdfAction.createURI(url)).setFlags(PdfAnnotation.PRINT);
             }
+            if (container instanceof IAccessibleElement && alternateDescription != null) {
+                ((IAccessibleElement) container).getAccessibilityProperties().setAlternateDescription(alternateDescription);
+            }
             linkAnnotation.setBorder(new PdfArray(new float[]{0, 0, 0}));
             container.setProperty(Property.LINK_ANNOTATION, linkAnnotation);
             if (container instanceof ILeafElement && container instanceof IAccessibleElement) {
@@ -102,6 +137,18 @@ public static void applyLinkAnnotation(IPropertyContainer container, String url,
         }
     }
 
+    /**
+     * Applies a link annotation.
+     *
+     * @param container the containing object.
+     * @param url       the destination.
+     * @param context   the processor context.
+     * @param element the element node.
+     */
+    public static void applyLinkAnnotation(IPropertyContainer container, String url, ProcessorContext context, IElementNode element) {
+        LinkHelper.applyLinkAnnotation(container, url, context, retrieveAlternativeDescription(element));
+    }
+
     /**
      * Creates a destination
      *

src/test/java/com/itextpdf/html2pdf/HtmlConverterPdfUA2Test.java

@@ -76,10 +76,9 @@ public void simpleLinkTest() throws IOException, InterruptedException, XMPExcept
         converterProperties.setFontProvider(fontProvider);
         HtmlConverter.convertToPdf(new FileInputStream(sourceHtml), pdfDocument, converterProperties);
 
-        /* TODO: DEVSIX-7996 - Links created from html2pdf are not ua-2 compliant
-         * Two verapdf errors are generated here:
-         * 1. clause="8.9.4.1", Link annotation neither has a Contents entry nor alternate description.
-         * 2. clause="8.5.1", Real content that does not possess the semantics of text objects and does not have
+        /* TODO: DEVSIX-5700 - Links created from html2pdf are not ua-2 compliant
+         * One verapdf error is generated here:
+         * 1. clause="8.5.1", Real content that does not possess the semantics of text objects and does not have
          *    an alternate textual representation is not enclosed within Figure or Formula structure elements.
          */
         compareAndCheckCompliance(destinationPdf, cmpPdf, false);
@@ -100,10 +99,33 @@ public void backwardLinkTest() throws IOException, InterruptedException, XMPExce
         converterProperties.setFontProvider(fontProvider);
         HtmlConverter.convertToPdf(new FileInputStream(sourceHtml), pdfDocument, converterProperties);
 
-        /* TODO: DEVSIX-7996 - Links created from html2pdf are not ua-2 compliant
-         * Two verapdf errors are generated here:
-         * 1. clause="8.9.4.1", Link annotation neither has a Contents entry nor alternate description.
-         * 2. clause="8.5.1", Real content that does not possess the semantics of text objects and does not have
+        /* TODO: DEVSIX-5700 - Links created from html2pdf are not ua-2 compliant
+         * One verapdf error is generated here:
+         * 1. clause="8.5.1", Real content that does not possess the semantics of text objects and does not have
+         *    an alternate textual representation is not enclosed within Figure or Formula structure elements.
+         */
+        compareAndCheckCompliance(destinationPdf, cmpPdf, false);
+    }
+
+    @Test
+    public void imageLinkTest() throws IOException, InterruptedException, XMPException {
+        String sourceHtml = SOURCE_FOLDER + "imageLink.html";
+        String cmpPdf = SOURCE_FOLDER + "cmp_imageLink.pdf";
+        String destinationPdf = DESTINATION_FOLDER + "imageLink.pdf";
+
+        PdfDocument pdfDocument = new PdfDocument(new PdfWriter(destinationPdf, new WriterProperties().setPdfVersion(
+                PdfVersion.PDF_2_0)));
+        createSimplePdfUA2Document(pdfDocument);
+
+        ConverterProperties converterProperties = new ConverterProperties();
+        FontProvider fontProvider = new DefaultFontProvider(false, true, false);
+        converterProperties.setFontProvider(fontProvider);
+        converterProperties.setBaseUri(SOURCE_FOLDER);
+        HtmlConverter.convertToPdf(new FileInputStream(sourceHtml), pdfDocument, converterProperties);
+
+        /* TODO: DEVSIX-5700 - Links created from html2pdf are not ua-2 compliant
+         * One verapdf error is generated here:
+         * 1. clause="8.5.1", Real content that does not possess the semantics of text objects and does not have
          *    an alternate textual representation is not enclosed within Figure or Formula structure elements.
          */
         compareAndCheckCompliance(destinationPdf, cmpPdf, false);

src/test/resources/com/itextpdf/html2pdf/HtmlConverterPdfUA2Test/backwardLink.html

@@ -7,7 +7,7 @@
 
 <div id="15=15">Target div</div>
 <div style="height:1000px">Some content</div>
-<a href="#15=15">Long link text to bottom div. Long link text to bottom div. Long link text to bottom div.
+<a href="#15=15" title="alternative description">Long link text to bottom div. Long link text to bottom div. Long link text to bottom div.
     Long link text to bottom div. Long link text to bottom div. Long link text to bottom div.</a>
 
 </body>

src/test/resources/com/itextpdf/html2pdf/HtmlConverterPdfUA2Test/blue96x96.png

@@ -0,0 +1,13 @@
+<!DOCTYPE html>
+<html>
+<head>
+
+</head>
+<body>
+
+<a href="#15=15" title="alternative description"><img alt="blue96x96" src="blue96x96.png"></a>
+<div style="height:1000px">Some content</div>
+<div id="15=15">Target div</div>
+
+</body>
+</html>