internal HtmlElement(HtmlTagInfo tagInfo)
{
Name = tagInfo.TagName;
if (!string.IsNullOrEmpty(tagInfo.Text))
{
Contents.Add(tagInfo.Text);
}
}
private HtmlTagInfo ParseTag()
{
var getDelimiterByLength = new Func<IEnumerable<HtmlDelimiter>, int, HtmlDelimiter>((delims, length) =>
{
return delims.Single(d => d.Length == length);
});
// read left tag delimiter
var delimiterLength = 0;
var currentDelimiters = HtmlGrammar.Delimiters.All;
do
{
var delimiters = HtmlGrammar.Delimiters.Find(currentDelimiters, Html.Current, ++delimiterLength).ToList();
if (delimiters.Count == 0)
{
--delimiterLength;
break;
}
currentDelimiters = delimiters;
} while (Html.MoveNext());
var leftDelimiter = getDelimiterByLength(currentDelimiters, delimiterLength);
if (!Html.Current.IsTagNameChar())
{
throw new Exception("Invalid html!");
}
// read tag name
var tagName = new StringBuilder();
do
{
tagName.Append(Html.Current);
} while (Html.MoveNext() && Html.Current.IsTagNameChar());
// read attributes
var attributes = new StringBuilder();
if (Html.Current.IsSpace())
{
while (Html.MoveNext() && !Html.Current.IsClosingAngleBracket())
{
attributes.Append(Html.Current);
}
}
// read right tag delimiter
delimiterLength = 0;
currentDelimiters = leftDelimiter.NextDelimiters.ToList();
do
{
var delimiters = HtmlGrammar.Delimiters.Find(currentDelimiters, Html.Current, ++delimiterLength).ToList();
if (delimiters.Count == 0)
{
--delimiterLength;
break;
}
currentDelimiters = delimiters;
} while (Html.MoveNext());
var rightDelimiter = getDelimiterByLength(currentDelimiters, delimiterLength);
if (!leftDelimiter.IsNext(rightDelimiter))
{
throw new Exception("Invalid html");
}
var tagInfo = new HtmlTagInfo
{
TagName = tagName.ToString(),
TagType = HtmlGrammar.IdentifyTagType(leftDelimiter, rightDelimiter),
// remove / on invalid self closing tags
Attributes = attributes.ToString().TrimEnd('/')
};
return tagInfo;
}