private DomElement ReadFreeText()
{
var para = new DomElement("p");
para.IsFreeText = true;
var sb = new StringBuilder();
while(true)
{
var c = reader.Peek();
if (c == '<')
{
para.content = sb.ToString();
break;
}
else if (c == -1)
{
para.content = sb.ToString();
break;
}
else
{
reader.Read();
sb.Append((char)c);
}
}
return para;
}
private DomElement ReadElementBody(DomElement element)
{
while (true)
{
var c = reader.Peek();
var d1 = (char)c;
if (c == -1)
{
//if (element.IsFreeText)
break;
}
else if (c == '<')
{
reader.Read();
ReadWhiteSpace();
c = reader.Peek();
d1 = (char)c;
if (c == '/')
{
reader.Read();
ReadWhiteSpace();
var tag = ReadWord();
if (tag != element.Tag)
throw new Exception("Closing element not found");
ReadWhiteSpace();
c = reader.Peek();
d1 = (char)c;
if (c != '>')
throw new Exception();
reader.Read();
break;
}
else if (c == '!')
{
reader.Read();
var comment = "<!--";
var sbComment = new StringBuilder();
sbComment.Append("<!");
var length = comment.Length - sbComment.Length;
for (int i = 0; i < length; ++i)
{
var c2 = reader.Read();
sbComment.Append((char)c2);
}
if (sbComment.ToString() != comment)
throw new Exception();
var para = ReadComment();
element.Children.Add(para);
}
else
{
var innerElement = ReadElementHead();
if (!innerElement.IsVoidElement)
innerElement = ReadElementBody(innerElement);
element.Children.Add(innerElement);
}
}
else
{
var para = ReadFreeText();
element.Children.Add(para);
}
}
return element;
}
private DomElement ReadElementHead()
{
var tag = ReadWord();
var element = new DomElement().GetElementFromString(tag);
element = ReadAttrs(element);
return element;
}
private DomElement ReadAttrs(DomElement element)
{
while(true)
{
var c = reader.Peek();
if (c == '>')
{
reader.Read();
break;
}
else if (element.IsVoidElement && c == '/')
{
reader.Read();
c = reader.Peek();
if (c == '>')
{
reader.Read();
break;
}
}
else if (IsWhiteSpace(c))
{
ReadWhiteSpace();
}
else if (IsTag(c))
{
//reader.Read();
var tag = ReadWord().ToLower();
Expect('=');
var c2 = Expect('"', '\'');
//Expect(c2);
if (tag == "id")
{
var word = ReadWord();
Expect(c2);
element.ID = word;
}
else if (tag == "class")
{
var spaces = new int[] { ' ', '\t', '\n', '\r' };
while(true)
{
var word = ReadWord();
element.Class.Add(word);
var expect = Expect(c2, ' ', '\t', '\n', '\r');
if (expect == c2)
{
break;
}
//else if (spaces.Contains(expect))
//{
// continue;
//}
//else
// throw new Exception("Unrecognized character " + (char)expect);
}
}
}
//else if (c == '=')
//{
//}
//else if (c == '\'' || c == '"')
//{
//}
else
{
reader.Read();
}
}
return element;
}
private DomElement parseDom()
{
var para = new DomElement("p");
para.IsFreeText = true;
var dom = ReadElementBody(para);
return dom;
}
public Dom(string html)
{
//this.html = html;
//html = "<sizzle>" + html + "</sizzle>";
this.reader = new StringReader(html);
//parseDom();
this.DomContainer = parseDom();
}
private List<DomElement> GetElements(DomElement el, List<Selector.SelectorPart> sel)
{
var matchEls = new List<DomElement>();
//var s = sel.Selectors.First();
foreach(var child in el.Children)
{
if (child.IsMatch(sel))
{
matchEls.Add(child);
}
else
{
matchEls.AddRange(GetElements(child, sel));
}
}
return matchEls;
}
private List<DomElement> BeforeAfter(List<DomElement> children, string text, bool before)
{
var newParent = new List<DomElement>();
foreach (var child in children)
{
if (before && this.Elements.Contains(child))
{
var para = new DomElement("p");
para.IsFreeText = true;
para.content = text;
newParent.Add(para);
}
var items = BeforeAfter(child.Children, text, before);
child.Children = items;
newParent.Add(child);
if (!before && this.Elements.Contains(child))
{
var para = new DomElement("p");
para.IsFreeText = true;
para.content = text;
newParent.Add(para);
}
}
return newParent;
}