feat: disallow remote links in toc/landmarks/page-list nav

A new message `NAV_010` (`ERROR`) is reported when a `toc`, `landmarks`,
or `page-list` nav contains links to remote resources.

Fix #890

src/main/java/com/adobe/epubcheck/messages/DefaultSeverities.java

@@ -155,6 +155,7 @@ private void initialize()
     severities.put(MessageId.NAV_007, Severity.USAGE);
     severities.put(MessageId.NAV_008, Severity.USAGE);
     severities.put(MessageId.NAV_009, Severity.ERROR);
+    severities.put(MessageId.NAV_010, Severity.ERROR);
 
     // NCX
     severities.put(MessageId.NCX_001, Severity.ERROR);

src/main/java/com/adobe/epubcheck/messages/MessageId.java

@@ -148,6 +148,7 @@ public enum MessageId implements Comparable<MessageId>
   NAV_007("NAV-007"),
   NAV_008("NAV-008"),
   NAV_009("NAV-009"),
+  NAV_010("NAV-010"),
 
   // Epub2 based table of content messages
   NCX_001("NCX-001"),

src/main/java/com/adobe/epubcheck/nav/NavHandler.java

@@ -1,18 +1,44 @@
 package com.adobe.epubcheck.nav;
 
+import java.util.EnumSet;
 import java.util.Set;
 
+import com.adobe.epubcheck.api.EPUBLocation;
+import com.adobe.epubcheck.messages.MessageId;
 import com.adobe.epubcheck.opf.ValidationContext;
 import com.adobe.epubcheck.ops.OPSHandler30;
+import com.adobe.epubcheck.util.EpubConstants;
 import com.adobe.epubcheck.util.FeatureEnum;
+import com.adobe.epubcheck.util.PathUtil;
 import com.adobe.epubcheck.vocab.StructureVocab.EPUB_TYPES;
 import com.adobe.epubcheck.xml.XMLElement;
 import com.adobe.epubcheck.xml.XMLParser;
+import com.google.common.base.CaseFormat;
+import com.google.common.base.Converter;
 
 public class NavHandler extends OPSHandler30
 {
 
-  private boolean inToc = false;
+  private NavType currentNavType = NavType.NONE;
+  private boolean isNavTypes = false;
+
+  private static enum NavType
+  {
+    NONE,
+    TOC,
+    LANDMARKS,
+    PAGE_LIST,
+    OTHER;
+
+    private Converter<String, String> formatter = CaseFormat.UPPER_UNDERSCORE
+        .converterTo(CaseFormat.LOWER_HYPHEN);
+
+    @Override
+    public String toString()
+    {
+      return formatter.convert(this.name());
+    }
+  }
 
   NavHandler(ValidationContext context, XMLParser parser)
   {
@@ -24,10 +50,29 @@ public void startElement()
   {
     super.startElement();
     XMLElement e = parser.getCurrentElement();
-    String name = e.getName();
-    if (inToc && "a".equals(name))
+    if (EpubConstants.HtmlNamespaceUri.equals(e.getNamespace()) && e.getName().equals("a"))
     {
-      context.featureReport.report(FeatureEnum.TOC_LINKS, parser.getLocation());
+      String href = e.getAttribute("href");
+      if (href != null)
+      {
+        // Feature reporting
+        if (currentNavType == NavType.TOC)
+        {
+          context.featureReport.report(FeatureEnum.TOC_LINKS, parser.getLocation());
+        }
+
+        // For 'toc', 'landmarks', and 'page-list' nav:
+        // the link MUST resolve to a Top-level Content Document
+        // Note: links to out-of-spine in-container items are already reported
+        // (RSC_011), so we only need to report links to remote resources
+        if (EnumSet.of(NavType.TOC, NavType.LANDMARKS, NavType.PAGE_LIST).contains(currentNavType)
+            && PathUtil.isRemote(href))
+        {
+          report.message(MessageId.NAV_010,
+              EPUBLocation.create(path, parser.getLineNumber(), parser.getColumnNumber()),
+              currentNavType, href);
+        }
+      }
     }
   }
 
@@ -36,39 +81,56 @@ public void endElement()
   {
     super.endElement();
     XMLElement e = parser.getCurrentElement();
-    String name = e.getName();
-    if (inToc && "nav".equals(name))
+    if (EpubConstants.HtmlNamespaceUri.equals(e.getNamespace()) && e.getName().equals("nav"))
     {
-      inToc = false;
+      currentNavType = NavType.NONE;
     }
   }
 
+  @Override
+  protected void checkType(XMLElement e, String type)
+  {
+    isNavTypes = (EpubConstants.HtmlNamespaceUri.equals(e.getNamespace())
+        && e.getName().equals("nav"));
+    super.checkType(e, type);
+    isNavTypes = false;
+  }
+
+  @Override
   protected void checkTypes(Set<EPUB_TYPES> types)
   {
     super.checkTypes(types);
-    if (types.contains(EPUB_TYPES.TOC))
-    {
-      inToc = true;
-    }
-    if (types.contains(EPUB_TYPES.PAGE_LIST))
-    {
-      context.featureReport.report(FeatureEnum.PAGE_LIST, parser.getLocation());
-    }
-    if (types.contains(EPUB_TYPES.LOI))
-    {
-      context.featureReport.report(FeatureEnum.LOI, parser.getLocation());
-    }
-    if (types.contains(EPUB_TYPES.LOT))
-    {
-      context.featureReport.report(FeatureEnum.LOT, parser.getLocation());
-    }
-    if (types.contains(EPUB_TYPES.LOA))
-    {
-      context.featureReport.report(FeatureEnum.LOA, parser.getLocation());
-    }
-    if (types.contains(EPUB_TYPES.LOV))
+    if (isNavTypes)
     {
-      context.featureReport.report(FeatureEnum.LOV, parser.getLocation());
+      if (types.contains(EPUB_TYPES.TOC))
+      {
+        currentNavType = NavType.TOC;
+      }
+      if (types.contains(EPUB_TYPES.PAGE_LIST))
+      {
+        currentNavType = NavType.PAGE_LIST;
+        context.featureReport.report(FeatureEnum.PAGE_LIST, parser.getLocation());
+      }
+      if (types.contains(EPUB_TYPES.LANDMARKS))
+      {
+        currentNavType = NavType.LANDMARKS;
+      }
+      if (types.contains(EPUB_TYPES.LOI))
+      {
+        context.featureReport.report(FeatureEnum.LOI, parser.getLocation());
+      }
+      if (types.contains(EPUB_TYPES.LOT))
+      {
+        context.featureReport.report(FeatureEnum.LOT, parser.getLocation());
+      }
+      if (types.contains(EPUB_TYPES.LOA))
+      {
+        context.featureReport.report(FeatureEnum.LOA, parser.getLocation());
+      }
+      if (types.contains(EPUB_TYPES.LOV))
+      {
+        context.featureReport.report(FeatureEnum.LOV, parser.getLocation());
+      }
     }
   }
 

src/main/resources/com/adobe/epubcheck/messages/MessageBundle.properties

@@ -142,6 +142,7 @@ NAV_006=Content documents contain 'figure' elements but the Navigation Document
 NAV_007=Content documents contain 'table' elements but the Navigation Document does not have a listing of tables (epub:type="lot").
 NAV_008=Content documents contain 'video' elements but the Navigation Document does not have a listing of video clips (epub:type="lov").
 NAV_009=Region-based navigation links must point to Fixed-Layout Documents.
+NAV_010='%1$s' nav must not link to remote resources; found link to '%2$s'.
 
 #NCX EPUB v2 Table of Contents
 NCX_001=NCX identifier ('%1$s') does not match OPF identifier ('%2$s').

src/test/java/com/adobe/epubcheck/api/Epub30CheckExpandedTest.java

@@ -199,10 +199,24 @@ public void testValidateEPUBPInvalidLoremForeign()
   }
 
   @Test
-  public void testValidateEPUB30_navInvalid()
+  public void testValidateNav_TocMissing()
   {
     Collections.addAll(expectedErrors, MessageId.RSC_005);
-    testValidateDocument("invalid/nav-invalid/");
+    testValidateDocument("invalid/nav-toc-missing/");
+  }
+
+  @Test
+  public void testValidateNav_LinksOutOfSpine()
+  {
+    expectedErrors.add(MessageId.RSC_011);
+    testValidateDocument("invalid/nav-links-out-of-spine/");
+  }
+
+  @Test
+  public void testValidateNav_LinksRemote()
+  {
+    expectedErrors.addAll(Collections.nCopies(3, MessageId.NAV_010));
+    testValidateDocument("invalid/nav-links-remote/");
   }
 
   @Test