private void TagEnd(string tagName)
{
if (_normalizeXml)
{
HtmlTagType tagType = HtmlTags.GetHtmlTagType(tagName);
switch (tagType)
{
case HtmlTagType.Html:
case HtmlTagType.Head:
case HtmlTagType.Body:
return;
case HtmlTagType.Title:
_currentNode = _currentTreeNode;
return;
case HtmlTagType.Table:
if (_table == null)
{
return;
}
//_currentNode = _currentTreeNode = GetParentXXNode(_table.Table);
_currentNode = _currentTreeNode = _table.Table.Parent;
_table = null;
if (_tableStack.Count != 0)
{
_table = _tableStack.Pop();
}
return;
case HtmlTagType.DL:
if (_definitionList == null)
{
return;
}
//_currentNode = _currentTreeNode = GetParentXXNode(_definitionList);
_currentNode = _currentTreeNode = _definitionList.Parent;
_definitionList = null;
if (_definitionListStack.Count != 0)
{
_definitionList = _definitionListStack.Pop();
}
return;
}
if (_table != null)
{
switch (tagType)
{
case HtmlTagType.THead:
case HtmlTagType.TBody:
case HtmlTagType.TFoot:
_currentNode = _currentTreeNode = _table.Table;
_table.Body = null;
return;
case HtmlTagType.ColGroup:
_currentNode = _currentTreeNode = _table.Table;
_table.ColGroup = null;
return;
case HtmlTagType.Col:
if (_table.Col != null)
{
//_currentNode = _currentTreeNode = GetParentXXNode(_table.Col);
_currentNode = _currentTreeNode = _table.Col.Parent;
_table.Col = null;
}
return;
case HtmlTagType.TR:
if (_table.Row != null)
{
//_currentNode = _currentTreeNode = GetParentXXNode(_table.Row);
_currentNode = _currentTreeNode = _table.Row.Parent;
_table.Row = null;
}
return;
case HtmlTagType.TH:
case HtmlTagType.TD:
if (_table.Data != null)
{
//_currentNode = _currentTreeNode = GetParentXXNode(_table.Data);
_currentNode = _currentTreeNode = _table.Data.Parent;
_table.Data = null;
}
return;
}
}
}
XNode node = GetParentNodeByName(_currentTreeNode, tagName);
if (node != null)
{
//_currentTreeNode = GetParentXXNode(node);
_currentTreeNode = node.Parent;
}
_currentNode = _currentTreeNode;
}
public XDocument CreateXml()
{
// ATTENTION HtmlReader_v4 dont manage ReadCommentInText
//_htmlReader.ReadCommentInText = _readCommentInText;
// need close tag
//_htmlReader.GenerateCloseTag = true;
//if (!_htmlReader.GenerateCloseTag)
// throw new PBException("html reader must have option GenerateCloseTag");
//_xdocument = new XDocument();
_xdCreator = new XDocumentCreator();
//_documentNode = _xdocument;
InitXml();
_tableStack = new Stack <HtmlTable_v3>();
_table = null;
_definitionListStack = new Stack <XElement>();
_definitionList = null;
_noTag = false;
_body = false;
_title = false;
foreach (HtmlNode htmlNode in _htmlReader.Read())
{
if (htmlNode.Type == HtmlNodeType.Text || htmlNode.Type == HtmlNodeType.Comment)
{
// $$pb modif le 11/01/2015
//if (_htmlReader.IsText && !_htmlReader.IsTextSeparator && !_body)
//if (_htmlReader.IsText && !_htmlReader.IsTextSeparator && !_htmlReader.IsScript && !_body)
//{
// _body = true;
// _currentNode = _currentTreeNode = _bodyNode;
//}
if (!_generateXmlNodeOnly)
{
if (_readCommentInText)
{
if (htmlNode.Type == HtmlNodeType.Text)
{
AddText(_currentNode, ((HtmlNodeText)htmlNode).Text);
}
else //if (htmlNode.Type == HtmlNodeType.Comment)
{
AddText(_currentNode, ((HtmlNodeComment)htmlNode).Comment);
}
}
else
{
if (htmlNode.Type == HtmlNodeType.Text)
{
AddText(_currentNode, ((HtmlNodeText)htmlNode).Text);
}
else //if (htmlNode.Type == HtmlNodeType.Comment)
{
string s = ((HtmlNodeComment)htmlNode).Comment;
s = _commentCorrection.Replace(s, "-");
if (s.EndsWith("-"))
{
s += " ";
}
//AddComment(_currentNode, s);
_xdCreator.AddComment(_currentNode, s);
}
}
}
}
else if (htmlNode.Type == HtmlNodeType.Script)
{
AddText(_currentNode, ((HtmlNodeScript)htmlNode).Script);
}
else if (htmlNode.Type == HtmlNodeType.DocumentType)
{
//AddAttribute(_htmlNode, "doctype", ((HtmlNodeDocType)htmlNode).DocType);
_xdCreator.AddAttribute(_htmlNode, "doctype", ((HtmlNodeDocType)htmlNode).DocType);
}
else if (htmlNode.Type == HtmlNodeType.Property)
{
if (_generateXmlNodeOnly || _noTag)
{
continue;
}
HtmlNodeProperty htmlNodeProperty = (HtmlNodeProperty)htmlNode;
try
{
string propertyName = htmlNodeProperty.Name;
propertyName = _nameCorrection.Replace(propertyName, "");
propertyName = propertyName.ToLower();
if (propertyName == "")
{
propertyName = "__value";
}
// modif le 28/01/2014
// hexadecimal value 0x03, is an invalid character
// found in http://www.reseau-gesat.com/Gesat/Yvelines,78/Fontenay-le-Fleury,31443/esat-cotra,e1596/
// <html><head><meta name="keywords" content="Conditionnement, travaux &agrave; fa&ccedil;onToutes activit&eacute;s en entreprise Entretien et cr&eacute;ation despaces verts" />
string propertyValue = htmlNodeProperty.Value;
if (propertyValue != null)
{
propertyValue = propertyValue.Replace("\x03", "");
}
//AddAttribute(_currentNode, propertyName, propertyValue);
_xdCreator.AddAttribute(_currentNode, propertyName, propertyValue);
//if (_htmlReader.IsMarkBeginEnd)
// TagEnd(_htmlReader.MarkName.ToLower());
}
catch (Exception ex)
{
Trace.WriteLine($"error in HtmlToXml_v2.CreateXml() : line {htmlNode.Line} column {htmlNode.Column}");
Trace.WriteLine(ex.Message);
}
}
//else if (_htmlReader.IsMarkBeginEnd)
//{
// string tagName = _htmlReader.MarkName.ToLower();
// tagName = _replace.Replace(tagName, "_");
// if (tagName == "") tagName = "_";
// TagBegin(tagName, true);
//}
//else if (_htmlReader.IsMarkBegin)
else if (htmlNode.Type == HtmlNodeType.OpenTag)
{
HtmlNodeOpenTag htmlNodeOpenTag = (HtmlNodeOpenTag)htmlNode;
string tagName = htmlNodeOpenTag.Name.ToLower();
tagName = _nameCorrection.Replace(tagName, "_");
if (tagName == "")
{
tagName = "_";
}
//TagBegin(tagName, false);
AddTagBegin(tagName);
}
//else if (htmlNode.Type == HtmlNodeType.CloseTag)
//{
// HtmlNodeCloseTag htmlNodeCloseTag = (HtmlNodeCloseTag)htmlNode;
// string tagName = htmlNodeCloseTag.Name.ToLower();
// tagName = _nameCorrection.Replace(tagName, "_");
// if (tagName == "")
// tagName = "_";
// TagEnd(tagName);
//}
//else if (_htmlReader.IsMarkEnd)
else if (htmlNode.Type == HtmlNodeType.EndTag)
{
HtmlNodeEndTag htmlNodeEndTag = (HtmlNodeEndTag)htmlNode;
string tagName = htmlNodeEndTag.Name.ToLower();
tagName = _nameCorrection.Replace(tagName, "_");
if (tagName == "")
{
tagName = "_";
}
TagEnd(tagName);
}
}
//return _xdocument;
return(_xdCreator.XDocument);
}
// bool bTagEnd
private void AddTagBegin(string tagName)
{
_noTag = false;
HtmlTagType tagType = HtmlTags.GetHtmlTagType(tagName);
HtmlTag tag = HtmlTags.GetHtmlTag(tagType);
if (_normalizeXml)
{
if (tagType == HtmlTagType.Html || tagType == HtmlTagType.Head)
{
_noTag = true;
return;
}
if (tagType == HtmlTagType.Body)
{
_noTag = true;
if (!_body)
{
_body = true;
_currentNode = _currentTreeNode = _bodyNode;
}
return;
}
if (tagType == HtmlTagType.Title)
{
if (!_title)
{
//if (!bTagEnd)
//{
_title = true;
_currentNode = _titleNode;
//}
}
else
{
_noTag = true;
}
return;
}
// $$pb modif le 11/01/2015
//if (!_body && tag.TagCategory != HtmlTagCategory.Head)
//{
// _body = true;
// _currentNode = _currentTreeNode = _bodyNode;
//}
}
//_currentNode = CreateElement(tagName);
if (_normalizeXml)
{
//if (tagType == HtmlTagType.Table && !bTagEnd)
if (tagType == HtmlTagType.Table)
{
//AddElement(_currentTreeNode, _currentNode);
//_currentTreeNode = _currentNode;
_currentTreeNode = _currentNode = _xdCreator.AddElement(_currentTreeNode, tagName);
if (_table != null)
{
_tableStack.Push(_table);
}
_table = new HtmlTable_v3();
_table.Table = _currentNode;
return;
}
//if (TagBeginTableCategory(tag, bTagEnd))
if (TagBeginTableCategory(tag, tagName))
{
return;
}
//if (tagType == HtmlTagType.DL && !bTagEnd)
if (tagType == HtmlTagType.DL)
{
//AddElement(_currentTreeNode, _currentNode);
//_currentTreeNode = _currentNode;
_currentTreeNode = _currentNode = _xdCreator.AddElement(_currentTreeNode, tagName);
if (_definitionList != null)
{
_definitionListStack.Push(_definitionList);
}
_definitionList = _currentNode;
return;
}
//if (TagBeginDefinitionListCategory(tag, bTagEnd))
if (TagBeginDefinitionListCategory(tag, tagName))
{
return;
}
// $$pb à revérifier
// il faut au moins annuler gLastPNode quand un des parents de gLastPNode se ferme
//if (tagType == HtmlTagTypeEnum.P)
//{
// // pour gérer une balise <p> qui n'a pas de fin de balise </p>
// if (gLastPNode != null)
// {
// gCurrentTreeNode = GetParentXXNode(gLastPNode);
// gLastPNode = null;
// }
// if (!bTagEnd) gLastPNode = gCurrentNode;
//}
}
//AddElement(_currentTreeNode, _currentNode);
_currentNode = _xdCreator.AddElement(_currentTreeNode, tagName);
//if (!bTagEnd && tag.EndBoundType != HtmlBoundType.Forbidden)
if (tag.EndBoundType != HtmlBoundType.Forbidden)
{
_currentTreeNode = _currentNode;
}
}
