Document, Node: Recognize HTML documents

kjellahl · kjellahl · commit f23e0ff9828c · 2021-06-09T15:46:01.000+02:00
Documents created with htmlReadDoc() have xmlElementType == XML_HTML_DOCUMENT_NODE. Handle that type like the XML_DOCUMENT_NODE type. htmlDoc* is an alias for xmlDoc*. Fixes #23
diff --git a/libxml++/document.cc b/libxml++/document.cc
@@ -56,6 +56,7 @@ void find_wrappers(xmlNode* node, NodeMap& node_map)
     case XML_ATTRIBUTE_DECL:
     case XML_ENTITY_DECL:
     case XML_DOCUMENT_NODE:
+    case XML_HTML_DOCUMENT_NODE:
       has_attributes = false;
       break;
     default:
@@ -102,6 +103,7 @@ void remove_found_wrappers(xmlNode* node, NodeMap& node_map)
     case XML_ATTRIBUTE_DECL:
     case XML_ENTITY_DECL:
     case XML_DOCUMENT_NODE:
+    case XML_HTML_DOCUMENT_NODE:
       has_attributes = false;
       break;
     default:
@@ -457,6 +459,7 @@ int Document::process_xinclude(bool generate_xinclude_nodes, bool fixup_base_uri
       delete reinterpret_cast<Dtd*>(node);
       break;
     case XML_DOCUMENT_NODE:
+    case XML_HTML_DOCUMENT_NODE:
       delete reinterpret_cast<Document*>(node);
       break;
     default:
diff --git a/libxml++/nodes/node.cc b/libxml++/nodes/node.cc
@@ -449,7 +449,9 @@ ustring Node::eval_to_string(const ustring& xpath, const PrefixNsMap& namespaces
 
 ustring Node::get_namespace_prefix() const
 {
-  if(impl_->type == XML_DOCUMENT_NODE || impl_->type == XML_ENTITY_DECL)
+  if(impl_->type == XML_DOCUMENT_NODE ||
+     impl_->type == XML_HTML_DOCUMENT_NODE ||
+     impl_->type == XML_ENTITY_DECL)
   {
     //impl_ is actually of type xmlDoc or xmlEntity, instead of just xmlNode.
     //libxml does not always use GObject-style inheritance, so xmlDoc and
@@ -475,6 +477,7 @@ ustring Node::get_namespace_prefix() const
 ustring Node::get_namespace_uri() const
 {
   if(impl_->type == XML_DOCUMENT_NODE ||
+     impl_->type == XML_HTML_DOCUMENT_NODE ||
      impl_->type == XML_ENTITY_DECL ||
      impl_->type == XML_ATTRIBUTE_DECL)
   {
@@ -591,6 +594,7 @@ void Node::create_wrapper(xmlNode* node)
       break;
     }
     case XML_DOCUMENT_NODE:
+    case XML_HTML_DOCUMENT_NODE:
     {
       // do nothing. For Documents it's the wrapper that is the owner.
       break;
@@ -639,6 +643,7 @@ void Node::free_wrappers(xmlNode* node)
       node->_private = nullptr;
       return;
     case XML_DOCUMENT_NODE:
+    case XML_HTML_DOCUMENT_NODE:
       //Do not free now. The Document is usually the one who owns the caller.
       return;
     default:

Original file line number	Diff line number	Diff line change
`@@ -449,7 +449,9 @@ ustring Node::eval_to_string(const ustring& xpath, const PrefixNsMap& namespaces`
`449`	`449`
`450`	`450`	`ustring Node::get_namespace_prefix() const`
`451`	`451`	`{`
`452`		`- if(impl_->type == XML_DOCUMENT_NODE \|\| impl_->type == XML_ENTITY_DECL)`
	`452`	`+ if(impl_->type == XML_DOCUMENT_NODE \|\|`
	`453`	`+ impl_->type == XML_HTML_DOCUMENT_NODE \|\|`
	`454`	`+ impl_->type == XML_ENTITY_DECL)`
`453`	`455`	`{`
`454`	`456`	`//impl_ is actually of type xmlDoc or xmlEntity, instead of just xmlNode.`
`455`	`457`	`//libxml does not always use GObject-style inheritance, so xmlDoc and`
`@@ -475,6 +477,7 @@ ustring Node::get_namespace_prefix() const`
`475`	`477`	`ustring Node::get_namespace_uri() const`
`476`	`478`	`{`
`477`	`479`	`if(impl_->type == XML_DOCUMENT_NODE \|\|`
	`480`	`+ impl_->type == XML_HTML_DOCUMENT_NODE \|\|`
`478`	`481`	`impl_->type == XML_ENTITY_DECL \|\|`
`479`	`482`	`impl_->type == XML_ATTRIBUTE_DECL)`
`480`	`483`	`{`
`@@ -591,6 +594,7 @@ void Node::create_wrapper(xmlNode* node)`
`591`	`594`	`break;`
`592`	`595`	`}`
`593`	`596`	`case XML_DOCUMENT_NODE:`
	`597`	`+ case XML_HTML_DOCUMENT_NODE:`
`594`	`598`	`{`
`595`	`599`	`// do nothing. For Documents it's the wrapper that is the owner.`
`596`	`600`	`break;`
`@@ -639,6 +643,7 @@ void Node::free_wrappers(xmlNode* node)`
`639`	`643`	`node->_private = nullptr;`
`640`	`644`	`return;`
`641`	`645`	`case XML_DOCUMENT_NODE:`
	`646`	`+ case XML_HTML_DOCUMENT_NODE:`
`642`	`647`	`//Do not free now. The Document is usually the one who owns the caller.`
`643`	`648`	`return;`
`644`	`649`	`default:`