Support property on link element in html5

[geoffmcl]

This issue originally raised by @da2x, in April, 2019, as PR #814 -

<link itemprop="subject" href="../example"> in Microdata is equivalent to <link property="subject" href="../example"> in RDFa. This HTML+RDFa syntax doesn’t trigger any errors or warnings from the W3C Nu Html Checker in HTML5 mode.

This change changes the requirement from link[itemprop|rel] to link[itemprop|property|rel] to support the HTML5+RDFa syntax.

See also: https://www.w3.org/TR/html-rdfa/#extensions-to-the-html5-syntax

While the PR #814 has been closed, since it was against master, it raises several important questions... maybe bugs!

Have added 2 tests - in_814.html, for html5, and in_814-1.html, for legacy html4... which can be put to the W3C validator... should get pass and fail respectively...

Running current tidy, 5.7.27, on either sample will give a warning, Warning: <link> lacks "rel" attribute! That seems wrong in both cases...

Presently experimenting with the following patch -

diff --git a/src/tags.c b/src/tags.c
index af2b7d9..ecc0ee7 100644
--- a/src/tags.c
+++ b/src/tags.c
@@ -1008,21 +1008,44 @@ void CheckTABLE( TidyDocImpl* doc, Node *node )
     }
 }
 
-/* report missing href attribute; report missing rel attribute */
+/* report missing href attribute; report missing rel attribute
+   Is. #814 - change from link[itemprop|rel] to link[itemprop|property|rel]
+   to support the HTML5+RDFa syntax.
+   Ref: https://www.w3.org/TR/html-rdfa/#extensions-to-the-html5-syntax
+ */
 void CheckLINK( TidyDocImpl* doc, Node *node )
 {
     Bool HasHref = TY_(AttrGetById)(node, TidyAttr_HREF) != NULL;
     Bool HasRel = TY_(AttrGetById)(node, TidyAttr_REL) != NULL;
     Bool HasItemprop = TY_(AttrGetById)(node, TidyAttr_ITEMPROP) != NULL;
+    AttVal *prop = TY_(AttrGetById)(node, TidyAttr_PROPERTY);
+    Bool HasProperty = (prop == NULL) ? no : yes; /* Is #814 */
 
     if (!HasHref)
     {
       TY_(ReportMissingAttr)( doc, node, "href" );
     }
 
-    if (!HasItemprop && !HasRel)
+    if (TY_(IsHTML5Mode)(doc))
+    {
+        /* Is. #814 - @da2x patch - add 'property' attr */
+        if (!HasItemprop && !HasRel && !HasProperty)
+        {
+            TY_(ReportMissingAttr)(doc, node, "rel");
+        }
+    }
+    else 
     {
-      TY_(ReportMissingAttr)( doc, node, "rel" );
+        /* legacy doc parsing */
+        if (HasProperty)
+        {
+            TY_(ReportAttrError)(doc, node, prop, MISMATCHED_ATTRIBUTE_WARN);
+        }
+        else if (!HasItemprop && !HasRel)
+        {
+            TY_(ReportMissingAttr)(doc, node, "rel");
+        }
+
     }
 }
 

Running the resultant tidy 5.7.27.I814 on the samples appears to now do the right thing... the in_814.html, the html5, has no errors/warnings, while the in_814-1.html, the html4, outputs Warning: <link> attribute "property" not allowed for HTML 4.01 Transitional...

After as little more testing, will try to set this up as a PR... unless someone beats me to it...

Look forward to further feedback... thanks...