static void ReadNext(ParserContext context)
{
while (!context.EndOfString)
{
Match match;
if (context.InScriptTag)
{
match = RxNextScriptCloseToken.Match(context.Html, context.Index);
context.InScriptTag = false;
}
else
{
match = RxNextToken.Match(context.Html, context.Index);
}
if (match.Success)
{
var len = match.Index - context.Index;
if (len > 0)
{
var str = context.Html.Substring(context.Index, len);
context.Tokens.Add(new HtmlParserToken {
Type = TokenType.Text, A = context.AdjustForWhitespace(str), Raw = str
});
context.Index += len;
}
if (match.Groups["xmldecl"].Success)
{
ReadXmlDeclaration(context);
}
else if (match.Groups["doctype"].Success)
{
ReadDocTypeDeclaration(context);
}
else if (match.Groups["comment"].Success)
{
ReadComment(context);
}
else if (match.Groups["close"].Success)
{
ReadCloseElement(context);
}
else if (match.Groups["cdata"].Success)
{
ReadCdata(context);
}
else
{
ReadElement(context);
}
}
else
{
var str = context.Html.Substring(context.Index);
context.Tokens.Add(new HtmlParserToken {
Type = TokenType.Text, A = context.AdjustForWhitespace(str), Raw = str
});
return;
}
}
}
private static void ReadCloseElement(ParserContext context)
{
var match = RxReadCloseAttribute.Match(context.Html, context.Index);
if (!match.Success)
{
var str = context.Html.Substring(context.Index);
context.Tokens.Add(new HtmlParserToken {
Type = TokenType.Text, A = context.AdjustForWhitespace(str), Raw = str
});
context.Index = context.Html.Length;
}
else
{
var newToken = new HtmlParserToken {
Type = TokenType.CloseElement, Raw = match.Value, A = match.Groups["name"].Value
};
context.Tokens.Add(newToken);
//HACK there might be a tag inside of a tag (Incorrectly closed tag) like </strong</td>
//If we find this, we are going to adjust
if (newToken.A.IndexOf("<") > -1)
{
var index = match.Value.Substring(2).IndexOf("<");
newToken.A = newToken.A.Substring(0, index);
context.Index += index + 2;
}
else
{
context.Index += match.Length;
}
}
}
static void ReadNext(ParserContext context)
{
while(!context.EndOfString)
{
Match match;
if (context.InScriptTag)
{
match = RxNextScriptToken.Match(context.Html, context.Index);
if (match.Groups["close"].Success)
{
context.InScriptTag = false;
}
}
else
{
match = RxNextToken.Match(context.Html, context.Index);
}
if(match.Success)
{
var len = match.Index - context.Index;
if(len > 0)
{
var str = context.Html.Substring(context.Index, len);
context.Tokens.Add(new HtmlParserToken { Type = TokenType.Text, A = context.AdjustForWhitespace(str), Raw = str });
context.Index += len;
}
if(match.Groups["xmldecl"].Success)
ReadXmlDeclaration(context);
else if(match.Groups["doctype"].Success)
ReadDocTypeDeclaration(context);
else if(match.Groups["comment"].Success)
ReadComment(context);
else if(match.Groups["close"].Success)
ReadCloseElement(context);
else if(match.Groups["cdata"].Success)
ReadCdata(context);
else if (match.Groups["conditional"].Success)
ReadConditional(context);
else
ReadElement(context);
}
else
{
var str = context.Html.Substring(context.Index);
context.Tokens.Add(new HtmlParserToken { Type = TokenType.Text, A = context.AdjustForWhitespace(str), Raw = str });
return;
}
}
}
private static void ReadCloseElement(ParserContext context)
{
var match = RxReadCloseAttribute.Match(context.Html, context.Index);
if (!match.Success)
{
var str = context.Html.Substring(context.Index);
context.Tokens.Add(new HtmlParserToken {
Type = TokenType.Text, A = context.AdjustForWhitespace(str), Raw = str
});
context.Index = context.Html.Length;
}
else
{
context.Tokens.Add(new HtmlParserToken {
Type = TokenType.CloseElement, Raw = match.Value, A = match.Groups["name"].Value
});
context.Index += match.Length;
}
}
private static void ReadCloseElement(ParserContext context)
{
var match = RxReadCloseAttribute.Match(context.Html, context.Index);
if(!match.Success)
{
var str = context.Html.Substring(context.Index);
context.Tokens.Add(new HtmlParserToken { Type = TokenType.Text, A = context.AdjustForWhitespace(str), Raw = str });
context.Index = context.Html.Length;
}
else
{
var newToken = new HtmlParserToken { Type = TokenType.CloseElement, Raw = match.Value, A = match.Groups["name"].Value };
context.Tokens.Add(newToken);
//HACK there might be a tag inside of a tag (Incorrectly closed tag) like </strong</td>
//If we find this, we are going to adjust
if (newToken.A.IndexOf("<") > -1)
{
var index = match.Value.Substring(2).IndexOf("<");
newToken.A = newToken.A.Substring(0, index);
context.Index += index + 2;
}
else
{
context.Index += match.Length;
}
}
}
private static void ReadCloseElement(ParserContext context)
{
var match = RxReadCloseAttribute.Match(context.Html, context.Index);
if(!match.Success)
{
var str = context.Html.Substring(context.Index);
context.Tokens.Add(new HtmlParserToken { Type = TokenType.Text, A = context.AdjustForWhitespace(str), Raw = str });
context.Index = context.Html.Length;
}
else
{
context.Tokens.Add(new HtmlParserToken { Type = TokenType.CloseElement, Raw = match.Value, A = match.Groups["name"].Value });
context.Index += match.Length;
}
}
