internal Document(GumboDocumentNode node, GumboFactory factory)
: base(node, null)
{
_children = factory.CreateLazy(() => ImmutableArray.CreateRange(node.GetChildren().OrderBy(x => x.index_within_parent).Select(x => factory.CreateNode(x, this))));
HasDocType = node.document.has_doctype;
Name = NativeUtf8.StringFromNativeUtf8(node.document.name);
PublicIdentifier = NativeUtf8.StringFromNativeUtf8(node.document.public_identifier);
SystemIdentifier = NativeUtf8.StringFromNativeUtf8(node.document.system_identifier);
DocTypeQuirksMode = node.document.doc_type_quirks_mode;
}
internal Attribute(GumboAttribute attribute, Element parent)
{
Parent = parent ?? throw new ArgumentNullException(nameof(parent));
Name = NativeUtf8.StringFromNativeUtf8(attribute.name);
Value = NativeUtf8.StringFromNativeUtf8(attribute.value);
OriginalName = NativeUtf8.StringFromNativeUtf8(attribute.original_name.data, (int)attribute.original_name.length);
OriginalValue = NativeUtf8.StringFromNativeUtf8(attribute.original_value.data, (int)attribute.original_value.length);
NameStart = attribute.name_start;
NameEnd = attribute.name_end;
ValueStart = attribute.value_start;
ValueEnd = attribute.value_end;
Namespace = attribute.attr_namespace;
}
public Gumbo(string html, GumboLibraryOptions?options = null)
{
_options = CreateOptions(options);
_html = NativeUtf8.NativeUtf8FromString(html);
_outputPtr = NativeMethods.gumbo_parse(_html);
var output = Marshal.PtrToStructure <GumboOutput>(_outputPtr);
_gumboDocumentNode = output.GetDocument();
Errors = output.GetErrors();
var lazyFactory = new LazyFactory(() => _disposed, typeof(Gumbo).Name);
_gumboFactory = new GumboFactory(lazyFactory);
Document = (Document)_gumboFactory.CreateNode(_gumboDocumentNode);
}
internal Element(GumboElementNode node, Node parent, GumboFactory factory)
: base(node, parent)
{
_children = factory.CreateLazy(() => ImmutableArray.CreateRange(node.GetChildren().OrderBy(x => x.index_within_parent).Select(x => factory.CreateNode(x, this))));
_attributes = factory.CreateLazy(() => ImmutableArray.CreateRange(node.GetAttributes().Select(x => factory.CreateAttribute(x, this))));
_value = factory.CreateLazy(() => string.Concat(Children.Select(x => x is Element ? ((Element)x).Value : ((Text)x).Value)));
StartPosition = node.element.start_pos;
EndPosition = node.element.end_pos;
Tag = node.element.tag;
TagNamespace = node.element.tag_namespace;
OriginalTag = NativeUtf8.StringFromNativeUtf8(node.element.original_tag.data, (int)node.element.original_tag.length);
OriginalTagName = GetTagNameFromOriginalTag(node.element);
OriginalEndTag = NativeUtf8.StringFromNativeUtf8(node.element.original_end_tag.data, (int)node.element.original_end_tag.length);
NormalizedTagName = NativeUtf8.StringFromNativeUtf8(NativeMethods.gumbo_normalized_tagname(node.element.tag));
}
static XNode CreateXNode(GumboNode node)
{
switch (node.type)
{
case GumboNodeType.GUMBO_NODE_DOCUMENT: return(new XDocument(((GumboDocumentNode)node).GetChildren().Select(CreateXNode)));
case GumboNodeType.GUMBO_NODE_ELEMENT:
case GumboNodeType.GUMBO_NODE_TEMPLATE:
var elementNode = (GumboElementNode)node;
var elementName = GetName(elementNode.element.tag);
var attributes = elementNode.GetAttributes().Select(x => new XAttribute(NativeUtf8.StringFromNativeUtf8(x.name), NativeUtf8.StringFromNativeUtf8(x.value)));
return(new XElement(elementName, attributes, elementNode.GetChildren().Select(CreateXNode)));
case GumboNodeType.GUMBO_NODE_TEXT: return(new XText(NativeUtf8.StringFromNativeUtf8(((GumboTextNode)node).text.text)));
case GumboNodeType.GUMBO_NODE_CDATA: return(new XCData(NativeUtf8.StringFromNativeUtf8(((GumboTextNode)node).text.text)));
case GumboNodeType.GUMBO_NODE_COMMENT: return(new XComment(NativeUtf8.StringFromNativeUtf8(((GumboTextNode)node).text.text)));
case GumboNodeType.GUMBO_NODE_WHITESPACE: return(new XText(NativeUtf8.StringFromNativeUtf8(((GumboTextNode)node).text.text)));
default: throw new NotImplementedException($"Node type '{node.type}' is not implemented");
}
}
internal Text(GumboTextNode node, Node parent) : base(node, parent)
{
Value = NativeUtf8.StringFromNativeUtf8(node.text.text);
StartPosition = node.text.start_pos;
}
public static string MarshalToString(this GumboStringPiece stringPiece) => NativeUtf8.StringFromNativeUtf8(stringPiece.data, (int)stringPiece.length);