private void Parse()
{
int lastquote = 0;
if (OptionComputeChecksum)
{
_crc32 = new Crc32();
}
_lastnodes = new Hashtable();
_c = 0;
_fullcomment = false;
_parseerrors = new List<HtmlParseError>();
_line = 1;
_lineposition = 1;
_maxlineposition = 1;
_state = ParseState.Text;
_oldstate = _state;
_documentnode._innerlength = _text.Length;
_documentnode._outerlength = _text.Length;
_remainderOffset = _text.Length;
_lastparentnode = _documentnode;
_currentnode = CreateNode(HtmlNodeType.Text, 0);
_currentattribute = null;
_index = 0;
PushNodeStart(HtmlNodeType.Text, 0);
while (_index < _text.Length)
{
_c = _text[_index];
IncrementPosition();
switch (_state)
{
case ParseState.Text:
if (NewCheck())
continue;
break;
case ParseState.WhichTag:
if (NewCheck())
continue;
if (_c == '/')
{
PushNodeNameStart(false, _index);
}
else
{
PushNodeNameStart(true, _index - 1);
DecrementPosition();
}
_state = ParseState.Tag;
break;
case ParseState.Tag:
if (NewCheck())
continue;
if (IsWhiteSpace(_c))
{
PushNodeNameEnd(_index - 1);
if (_state != ParseState.Tag)
continue;
_state = ParseState.BetweenAttributes;
continue;
}
if (_c == '/')
{
PushNodeNameEnd(_index - 1);
if (_state != ParseState.Tag)
continue;
_state = ParseState.EmptyTag;
continue;
}
if (_c == '>')
{
PushNodeNameEnd(_index - 1);
if (_state != ParseState.Tag)
continue;
if (!PushNodeEnd(_index, false))
{
// stop parsing
_index = _text.Length;
break;
}
if (_state != ParseState.Tag)
continue;
_state = ParseState.Text;
PushNodeStart(HtmlNodeType.Text, _index);
}
break;
case ParseState.BetweenAttributes:
if (NewCheck())
continue;
if (IsWhiteSpace(_c))
continue;
if ((_c == '/') || (_c == '?'))
{
_state = ParseState.EmptyTag;
continue;
}
if (_c == '>')
{
if (!PushNodeEnd(_index, false))
{
// stop parsing
_index = _text.Length;
break;
}
if (_state != ParseState.BetweenAttributes)
continue;
_state = ParseState.Text;
PushNodeStart(HtmlNodeType.Text, _index);
continue;
}
PushAttributeNameStart(_index - 1);
_state = ParseState.AttributeName;
break;
case ParseState.EmptyTag:
if (NewCheck())
continue;
if (_c == '>')
{
if (!PushNodeEnd(_index, true))
{
// stop parsing
_index = _text.Length;
break;
}
if (_state != ParseState.EmptyTag)
continue;
_state = ParseState.Text;
PushNodeStart(HtmlNodeType.Text, _index);
continue;
}
_state = ParseState.BetweenAttributes;
break;
case ParseState.AttributeName:
if (NewCheck())
continue;
if (IsWhiteSpace(_c))
{
PushAttributeNameEnd(_index - 1);
_state = ParseState.AttributeBeforeEquals;
continue;
}
if (_c == '=')
{
PushAttributeNameEnd(_index - 1);
_state = ParseState.AttributeAfterEquals;
continue;
}
if (_c == '>')
{
PushAttributeNameEnd(_index - 1);
if (!PushNodeEnd(_index, false))
{
// stop parsing
_index = _text.Length;
break;
}
if (_state != ParseState.AttributeName)
continue;
_state = ParseState.Text;
PushNodeStart(HtmlNodeType.Text, _index);
continue;
}
break;
case ParseState.AttributeBeforeEquals:
if (NewCheck())
continue;
if (IsWhiteSpace(_c))
continue;
if (_c == '>')
{
if (!PushNodeEnd(_index, false))
{
// stop parsing
_index = _text.Length;
break;
}
if (_state != ParseState.AttributeBeforeEquals)
continue;
_state = ParseState.Text;
PushNodeStart(HtmlNodeType.Text, _index);
continue;
}
if (_c == '=')
{
_state = ParseState.AttributeAfterEquals;
continue;
}
// no equals, no whitespace, it's a new attrribute starting
_state = ParseState.BetweenAttributes;
DecrementPosition();
break;
case ParseState.AttributeAfterEquals:
if (NewCheck())
continue;
if (IsWhiteSpace(_c))
continue;
if ((_c == '\'') || (_c == '"'))
{
_state = ParseState.QuotedAttributeValue;
PushAttributeValueStart(_index, _c);
lastquote = _c;
continue;
}
if (_c == '>')
{
if (!PushNodeEnd(_index, false))
{
// stop parsing
_index = _text.Length;
break;
}
if (_state != ParseState.AttributeAfterEquals)
continue;
_state = ParseState.Text;
PushNodeStart(HtmlNodeType.Text, _index);
continue;
}
PushAttributeValueStart(_index - 1);
_state = ParseState.AttributeValue;
break;
case ParseState.AttributeValue:
if (NewCheck())
continue;
if (IsWhiteSpace(_c))
{
PushAttributeValueEnd(_index - 1);
_state = ParseState.BetweenAttributes;
continue;
}
if (_c == '>')
{
PushAttributeValueEnd(_index - 1);
if (!PushNodeEnd(_index, false))
{
// stop parsing
_index = _text.Length;
break;
}
if (_state != ParseState.AttributeValue)
continue;
_state = ParseState.Text;
PushNodeStart(HtmlNodeType.Text, _index);
continue;
}
break;
case ParseState.QuotedAttributeValue:
if (_c == lastquote)
{
PushAttributeValueEnd(_index - 1);
_state = ParseState.BetweenAttributes;
continue;
}
if (_c == '<')
{
if (_index < _text.Length)
{
if (_text[_index] == '%')
{
_oldstate = _state;
_state = ParseState.ServerSideCode;
continue;
}
}
}
break;
case ParseState.Comment:
if (_c == '>')
{
if (_fullcomment)
{
if ((_text[_index - 2] != '-') ||
(_text[_index - 3] != '-'))
{
continue;
}
}
if (!PushNodeEnd(_index, false))
{
// stop parsing
_index = _text.Length;
break;
}
_state = ParseState.Text;
PushNodeStart(HtmlNodeType.Text, _index);
continue;
}
break;
case ParseState.ServerSideCode:
if (_c == '%')
{
if (_index < _text.Length)
{
if (_text[_index] == '>')
{
switch (_oldstate)
{
case ParseState.AttributeAfterEquals:
_state = ParseState.AttributeValue;
break;
case ParseState.BetweenAttributes:
PushAttributeNameEnd(_index + 1);
_state = ParseState.BetweenAttributes;
break;
default:
_state = _oldstate;
break;
}
IncrementPosition();
}
}
}
break;
case ParseState.PcData:
// look for </tag + 1 char
// check buffer end
if ((_currentnode._namelength + 3) <= (_text.Length - (_index - 1)))
{
if (string.Compare(_text.Substring(_index - 1, _currentnode._namelength + 2),
"</" + _currentnode.Name, true) == 0)
{
int c = _text[_index - 1 + 2 + _currentnode.Name.Length];
if ((c == '>') || (IsWhiteSpace(c)))
{
// add the script as a text node
HtmlNode script = CreateNode(HtmlNodeType.Text,
_currentnode._outerstartindex +
_currentnode._outerlength);
script._outerlength = _index - 1 - script._outerstartindex;
_currentnode.AppendChild(script);
PushNodeStart(HtmlNodeType.Element, _index - 1);
PushNodeNameStart(false, _index - 1 + 2);
_state = ParseState.Tag;
IncrementPosition();
}
}
}
break;
}
}
// finish the current work
if (_currentnode._namestartindex > 0)
{
PushNodeNameEnd(_index);
}
PushNodeEnd(_index, false);
// we don't need this anymore
_lastnodes.Clear();
}
