/// <summary>
/// Parse the html part, the part from prev parsing index to the beginning of the next html tag.<br/>
/// </summary>
/// <param name="source">the html source to parse</param>
/// <param name="tagIdx">the index of the next html tag</param>
/// <param name="curBox">the current box in html tree parsing</param>
/// <returns>the end of the parsed part, the new start index</returns>
private static int ParseHtmlTag(string source, int tagIdx, ref CssBox curBox)
{
var endIdx = source.IndexOf('>', tagIdx + 1);
if (endIdx > 0)
{
string tagName;
Dictionary <string, string> tagAttributes;
var length = endIdx - tagIdx + 1 - (source[endIdx - 1] == '/' ? 1 : 0);
if (ParseHtmlTag(source, tagIdx, length, out tagName, out tagAttributes))
{
if (!HtmlUtils.IsSingleTag(tagName) && curBox.ParentBox != null)
{
// need to find the parent tag to go one level up
curBox = DomUtils.FindParent(curBox.ParentBox, tagName, curBox);
}
}
else if (!string.IsNullOrEmpty(tagName))
{
//new SubString(source, lastEnd + 1, tagmatch.Index - lastEnd - 1)
var isSingle = HtmlUtils.IsSingleTag(tagName) || source[endIdx - 1] == '/';
var tag = new HtmlTag(tagName, isSingle, tagAttributes);
if (isSingle)
{
// the current box is not changed
CssBox.CreateBox(tag, curBox);
}
else
{
// go one level down, make the new box the current box
curBox = CssBox.CreateBox(tag, curBox);
}
}
else
{
endIdx = tagIdx + 1;
}
}
return(endIdx);
}
/// <summary>
/// Parses the document
/// </summary>
public static CssBox ParseDocument(string document)
{
document = RemoveHtmlComments(document);
int lastEnd = -1;
CssBox root = null;
CssBox curBox = null;
var tags = RegexParserUtils.Match(RegexParserUtils.HtmlTag, document);
foreach (Match tagmatch in tags)
{
string text = tagmatch.Index > 0 ? document.Substring(lastEnd + 1, tagmatch.Index - lastEnd - 1) : String.Empty;
var emptyText = String.IsNullOrEmpty(text.Trim());
if (!emptyText)
{
if (curBox == null)
{
root = curBox = CssBox.CreateBlock();
}
var abox = CssBox.CreateBox(curBox);
abox.Text = text;
}
var tag = ParseHtmlTag(tagmatch.Value);
if (tag.IsClosing)
{
// handle tags that have no content but whitespace
if (emptyText && curBox != null && curBox.Boxes.Count == 0 && !string.IsNullOrEmpty(text))
{
var abox = CssBox.CreateBox(curBox);
abox.Text = " ";
}
// need to find the parent tag to go one level up
curBox = DomUtils.FindParent(root, tag.Name, curBox);
}
else if (tag.IsSingle)
{
// the current box is not changed
new CssBox(curBox, tag);
}
else
{
// go one level down, make the new box the current box
curBox = new CssBox(curBox, tag);
}
if (root == null && curBox != null)
{
root = curBox;
root.Display = CssConstants.Block;
}
lastEnd = tagmatch.Index + tagmatch.Length - 1;
}
if (root == null)
{
root = CssBox.CreateBlock();
var abox = CssBox.CreateBox(root);
abox.Text = document;
}
else if (lastEnd < document.Length)
{
var endText = document.Substring(lastEnd + 1);
if (!string.IsNullOrEmpty(endText.Trim()))
{
var abox = CssBox.CreateBox(root);
abox.Text = endText;
}
}
return(root);
}