normalize HTML case-insensitive attributes + basic support for custom elements

src/main/java/com/adobe/epubcheck/xml/HTMLUtils.java

@@ -0,0 +1,83 @@
+package com.adobe.epubcheck.xml;
+
+import com.google.common.base.Preconditions;
+
+/**
+ * Utilities for HTML-specific logic.
+ * 
+ */
+public class HTMLUtils
+{
+  /**
+   * Returns whether an attribute is defined as having a case-insensitive value in
+   * HTML. This is notably the case of boolean attributes and enumerated
+   * attributes.
+   * 
+   * @param name
+   *          the name of an attribute defined in the HTML specification
+   * @return <code>true</code> iff the attribute value is case-insensitive
+   */
+  public static boolean isCaseInsensitiveAttribute(String name)
+  {
+    switch (Preconditions.checkNotNull(name))
+    {
+    case "align":
+    case "allowfullscreen":
+    case "allowpaymentrequest":
+    case "allowusermedia":
+    case "async":
+    case "autocapitalize":
+    case "autocomplete":
+    case "autofocus":
+    case "autoplay":
+    case "checked":
+    case "contenteditable":
+    case "controls":
+    case "crossorigin":
+    case "default":
+    case "defer":
+    case "dir":
+    case "disabled":
+    case "draggable":
+    case "formnovalidate":
+    case "hidden":
+    case "http-equiv":
+    case "ismap":
+    case "itemscope":
+    case "kind":
+    case "loop":
+    case "multiple":
+    case "muted":
+    case "nomodule":
+    case "novalidate":
+    case "open":
+    case "playsinline":
+    case "preload":
+    case "readonly":
+    case "required":
+    case "reversed":
+    case "scope":
+    case "selected":
+    case "shape":
+    case "sizes":
+    case "spellcheck":
+    case "step":
+    case "translate":
+    case "type":
+    case "typemustmatch":
+    case "valign":
+    case "value":
+    case "wrap":
+      return true;
+    default:
+      return false;
+    }
+
+  }
+
+  public static boolean isCustomElement(String namespace, String name)
+  {
+    return Namespaces.XHTML.equals(namespace) && Preconditions.checkNotNull(name).contains("-");
+  }
+
+}

src/main/java/com/adobe/epubcheck/xml/XMLParser.java

@@ -28,10 +28,8 @@
 import java.io.IOException;
 import java.io.InputStream;
 import java.io.StringReader;
-import java.util.ArrayList;
 import java.util.HashSet;
 import java.util.Hashtable;
-import java.util.List;
 import java.util.Locale;
 import java.util.Set;
 import java.util.Vector;
@@ -466,39 +464,18 @@ public void startDocument()
     }
   }
 
-  public void startElement(String namespaceURI, String localName, String qName, Attributes atts)
+  public void startElement(String namespaceURI, String localName, String qName,
+      Attributes parsedAttribs)
     throws SAXException
   {
-
-    AttributesImpl attribs = new AttributesImpl(atts);
-
+    Attributes attribs = preprocessAttributes(namespaceURI, localName, qName, parsedAttribs);
+
     if ("application/xhtml+xml".equals(context.mimeType)
-        && context.version == EPUBVersion.VERSION_3)
-    {
-      try
-      {
-        int len = attribs.getLength();
-        List<String> removals = new ArrayList<String>();
-        for (int i = 0; i < len; i++)
-        {
-          if (attribs.getLocalName(i).startsWith("data-"))
-          {
-            removals.add(attribs.getQName(i));
-          }
-          else if (isCustomNamespaceAttr(attribs.getURI(i)))
-          {
-            removals.add(attribs.getQName(i));
-          }
-        }
-        for (String remove : removals)
-        {
-          int rmv = attribs.getIndex(remove);
-          // System.out.println("removing attribute " + attribs.getQName(rmv));
-          attribs.removeAttribute(rmv);
-        }
-      } catch (Exception e)
-      {
-        System.err.println("data-* removal exception: " + e.getMessage());
+        && context.version == EPUBVersion.VERSION_3) {
+      // Pre-process HTML custom elements to set them in the proprietary namespace supported
+      // by the Nu Html Checker
+      if (HTMLUtils.isCustomElement(namespaceURI, localName)) {
+        namespaceURI = "http://n.validator.nu/custom-elements/";
       }
     }
 
@@ -564,23 +541,51 @@ else if (isCustomNamespaceAttr(attribs.getURI(i)))
     knownXHTMLContentDocsNamespaces.add(Namespaces.XLINK);
   }
 
-  private boolean isCustomNamespaceAttr(String nsuri)
+  private Attributes preprocessAttributes(String elemNamespace, String elemName, String elemQName,
+      Attributes originalAttribs)
   {
-
-    if (nsuri == null || nsuri.trim().length() == 0)
-    {
-      return false;
-    }
-
-    for (String ns : knownXHTMLContentDocsNamespaces)
+    AttributesImpl attributes = new AttributesImpl(originalAttribs);
+    if ("application/xhtml+xml".equals(context.mimeType)
+        && context.version == EPUBVersion.VERSION_3)
     {
-      if (ns.equals(nsuri))
+      try
       {
-        return false;
+        for (int i = attributes.getLength() - 1; i >= 0; i--)
+        {
+          if (isDataAttribute(attributes, i) || isCustomNamespaceAttribute(attributes, i))
+          {
+            attributes.removeAttribute(i);
+          }
+          else if (Namespaces.XHTML.equals(elemNamespace)
+              && isCaseInsensitiveAttribute(attributes, i))
+          {
+            attributes.setValue(i, attributes.getValue(i).toLowerCase(Locale.ENGLISH));
+          }
+        }
+      } catch (Exception e)
+      {
+        throw new IllegalStateException("data-* removal exception", e);
       }
     }
+    return attributes;
+  }
 
-    return true;
+  private static boolean isDataAttribute(Attributes attributes, int index)
+  {
+    return attributes.getLocalName(index).startsWith("data-");
+  }
+
+  private boolean isCustomNamespaceAttribute(Attributes attributes, int index)
+  {
+    String ns = attributes.getURI(index);
+    return (ns != null && ns.trim().length() > 0
+        && !knownXHTMLContentDocsNamespaces.contains(ns.trim()));
+  }
+
+  private static boolean isCaseInsensitiveAttribute(Attributes attributes, int index)
+  {
+    return (attributes.getURI(index).isEmpty()
+        && HTMLUtils.isCaseInsensitiveAttribute(attributes.getLocalName(index)));
   }
 
   public void startPrefixMapping(String arg0, String arg1)

src/test/java/com/adobe/epubcheck/ops/OPSCheckerTest.java

@@ -891,9 +891,27 @@ public void testValidateXHTML301MDInvalid()
   }
 
   @Test
-  public void testValidateXHTML301CustomAttributes()
+  public void testAttributesInCustomNS()
   {
-    testValidateDocument("xhtml/valid/custom-ns-attrs.xhtml", "application/xhtml+xml",
+    // test that attribute in a custom namespace are ignored
+    testValidateDocument("xhtml/valid/attrs-custom-ns.xhtml", "application/xhtml+xml",
+        EPUBVersion.VERSION_3);
+  }
+
+  @Test
+  public void testAttributesCaseInsensitive()
+  {
+    // test that the value of HTML boolean attributes and enumerated attributes are
+    // parsed in a case-insensitive manner
+    testValidateDocument("xhtml/valid/attrs-case-insensitive.xhtml", "application/xhtml+xml",
+        EPUBVersion.VERSION_3);
+  }
+
+  @Test
+  public void testCustomElements()
+  {
+    // test that HTML custom elements are not rejected by the schema
+    testValidateDocument("xhtml/valid/custom-elements.xhtml", "application/xhtml+xml",
         EPUBVersion.VERSION_3);
   }
 

src/test/resources/30/single/xhtml/valid/attrs-case-insensitive.xhtml

@@ -0,0 +1,11 @@
+<!DOCTYPE html>	 
+<html xmlns="http://www.w3.org/1999/xhtml" lang="en">
+  <head>
+    <title>Test</title>
+    <meta charset="UTF-8"/>
+  </head>
+  <body>
+    <h1>Test</h1>
+    <div hidden="HIDDEN"></div>
+  </body>
+</html>

src/test/resources/30/single/xhtml/valid/attrs-custom-ns.xhtml

@@ -0,0 +1,11 @@
+<!DOCTYPE html>	 
+<html xmlns="http://www.w3.org/1999/xhtml" xmlns:foo="https://example.org" lang="en">
+    <head>
+        <title>Test</title>
+        <meta charset="UTF-8" />
+    </head>
+    <body>
+        <h1>Test</h1>
+        <p foo:bar1="baz" foo:bar2="baz" foo:bar3="baz">custom attribute!</p>
+    </body>
+</html>

src/test/resources/30/single/xhtml/valid/custom-elements.xhtml

@@ -0,0 +1,16 @@
+<!DOCTYPE html>	 
+<html xmlns="http://www.w3.org/1999/xhtml" lang="en">
+  <head>
+    <title>Test</title>
+    <meta charset="UTF-8" />
+  </head>
+  <body>
+    <h1>Test</h1>
+    <epub-switch>
+      <epub-case required-namespace="https://example.org"> oh my! </epub-case>
+      <epub-default>
+        <p>is this real?</p>
+      </epub-default>
+    </epub-switch>
+  </body>
+</html>