private void Parse()
{
int num = 0;
if (this.OptionComputeChecksum)
this._crc32 = new Crc32();
this.Lastnodes = new Dictionary<string, HtmlNode>();
this._c = 0;
this._fullcomment = false;
this._parseerrors = new List<HtmlParseError>();
this._line = 1;
this._lineposition = 1;
this._maxlineposition = 1;
this._state = HtmlDocument.ParseState.Text;
this._oldstate = this._state;
this._documentnode._innerlength = this.Text.Length;
this._documentnode._outerlength = this.Text.Length;
this._remainderOffset = this.Text.Length;
this._lastparentnode = this._documentnode;
this._currentnode = this.CreateNode(HtmlNodeType.Text, 0);
this._currentattribute = (HtmlAttribute) null;
this._index = 0;
this.PushNodeStart(HtmlNodeType.Text, 0);
while (this._index < this.Text.Length)
{
this._c = (int) this.Text[this._index];
this.IncrementPosition();
switch (this._state)
{
case HtmlDocument.ParseState.Text:
if (!this.NewCheck())
continue;
continue;
case HtmlDocument.ParseState.WhichTag:
if (!this.NewCheck())
{
if (this._c == 47)
{
this.PushNodeNameStart(false, this._index);
}
else
{
this.PushNodeNameStart(true, this._index - 1);
this.DecrementPosition();
}
this._state = HtmlDocument.ParseState.Tag;
continue;
}
continue;
case HtmlDocument.ParseState.Tag:
if (!this.NewCheck())
{
if (HtmlDocument.IsWhiteSpace(this._c))
{
this.PushNodeNameEnd(this._index - 1);
if (this._state == HtmlDocument.ParseState.Tag)
{
this._state = HtmlDocument.ParseState.BetweenAttributes;
continue;
}
continue;
}
if (this._c == 47)
{
this.PushNodeNameEnd(this._index - 1);
if (this._state == HtmlDocument.ParseState.Tag)
{
this._state = HtmlDocument.ParseState.EmptyTag;
continue;
}
continue;
}
if (this._c == 62)
{
this.PushNodeNameEnd(this._index - 1);
if (this._state == HtmlDocument.ParseState.Tag)
{
if (!this.PushNodeEnd(this._index, false))
{
this._index = this.Text.Length;
continue;
}
if (this._state == HtmlDocument.ParseState.Tag)
{
this._state = HtmlDocument.ParseState.Text;
this.PushNodeStart(HtmlNodeType.Text, this._index);
continue;
}
continue;
}
continue;
}
continue;
}
continue;
case HtmlDocument.ParseState.BetweenAttributes:
if (!this.NewCheck() && !HtmlDocument.IsWhiteSpace(this._c))
{
if (this._c == 47 || this._c == 63)
{
this._state = HtmlDocument.ParseState.EmptyTag;
continue;
}
if (this._c == 62)
{
if (!this.PushNodeEnd(this._index, false))
{
this._index = this.Text.Length;
continue;
}
if (this._state == HtmlDocument.ParseState.BetweenAttributes)
{
this._state = HtmlDocument.ParseState.Text;
this.PushNodeStart(HtmlNodeType.Text, this._index);
continue;
}
continue;
}
this.PushAttributeNameStart(this._index - 1);
this._state = HtmlDocument.ParseState.AttributeName;
continue;
}
continue;
case HtmlDocument.ParseState.EmptyTag:
if (!this.NewCheck())
{
if (this._c == 62)
{
if (!this.PushNodeEnd(this._index, true))
{
this._index = this.Text.Length;
continue;
}
if (this._state == HtmlDocument.ParseState.EmptyTag)
{
this._state = HtmlDocument.ParseState.Text;
this.PushNodeStart(HtmlNodeType.Text, this._index);
continue;
}
continue;
}
this._state = HtmlDocument.ParseState.BetweenAttributes;
continue;
}
continue;
case HtmlDocument.ParseState.AttributeName:
if (!this.NewCheck())
{
if (HtmlDocument.IsWhiteSpace(this._c))
{
this.PushAttributeNameEnd(this._index - 1);
this._state = HtmlDocument.ParseState.AttributeBeforeEquals;
continue;
}
if (this._c == 61)
{
this.PushAttributeNameEnd(this._index - 1);
this._state = HtmlDocument.ParseState.AttributeAfterEquals;
continue;
}
if (this._c == 62)
{
this.PushAttributeNameEnd(this._index - 1);
if (!this.PushNodeEnd(this._index, false))
{
this._index = this.Text.Length;
continue;
}
if (this._state == HtmlDocument.ParseState.AttributeName)
{
this._state = HtmlDocument.ParseState.Text;
this.PushNodeStart(HtmlNodeType.Text, this._index);
continue;
}
continue;
}
continue;
}
continue;
case HtmlDocument.ParseState.AttributeBeforeEquals:
if (!this.NewCheck() && !HtmlDocument.IsWhiteSpace(this._c))
{
if (this._c == 62)
{
if (!this.PushNodeEnd(this._index, false))
{
this._index = this.Text.Length;
continue;
}
if (this._state == HtmlDocument.ParseState.AttributeBeforeEquals)
{
this._state = HtmlDocument.ParseState.Text;
this.PushNodeStart(HtmlNodeType.Text, this._index);
continue;
}
continue;
}
if (this._c == 61)
{
this._state = HtmlDocument.ParseState.AttributeAfterEquals;
continue;
}
this._state = HtmlDocument.ParseState.BetweenAttributes;
this.DecrementPosition();
continue;
}
continue;
case HtmlDocument.ParseState.AttributeAfterEquals:
if (!this.NewCheck() && !HtmlDocument.IsWhiteSpace(this._c))
{
if (this._c == 39 || this._c == 34)
{
this._state = HtmlDocument.ParseState.QuotedAttributeValue;
this.PushAttributeValueStart(this._index, this._c);
num = this._c;
continue;
}
if (this._c == 62)
{
if (!this.PushNodeEnd(this._index, false))
{
this._index = this.Text.Length;
continue;
}
if (this._state == HtmlDocument.ParseState.AttributeAfterEquals)
{
this._state = HtmlDocument.ParseState.Text;
this.PushNodeStart(HtmlNodeType.Text, this._index);
continue;
}
continue;
}
this.PushAttributeValueStart(this._index - 1);
this._state = HtmlDocument.ParseState.AttributeValue;
continue;
}
continue;
case HtmlDocument.ParseState.AttributeValue:
if (!this.NewCheck())
{
if (HtmlDocument.IsWhiteSpace(this._c))
{
this.PushAttributeValueEnd(this._index - 1);
this._state = HtmlDocument.ParseState.BetweenAttributes;
continue;
}
if (this._c == 62)
{
this.PushAttributeValueEnd(this._index - 1);
if (!this.PushNodeEnd(this._index, false))
{
this._index = this.Text.Length;
continue;
}
if (this._state == HtmlDocument.ParseState.AttributeValue)
{
this._state = HtmlDocument.ParseState.Text;
this.PushNodeStart(HtmlNodeType.Text, this._index);
continue;
}
continue;
}
continue;
}
continue;
case HtmlDocument.ParseState.Comment:
if (this._c == 62 && (!this._fullcomment || (int) this.Text[this._index - 2] == 45 && (int) this.Text[this._index - 3] == 45))
{
if (!this.PushNodeEnd(this._index, false))
{
this._index = this.Text.Length;
continue;
}
this._state = HtmlDocument.ParseState.Text;
this.PushNodeStart(HtmlNodeType.Text, this._index);
continue;
}
continue;
case HtmlDocument.ParseState.QuotedAttributeValue:
if (this._c == num)
{
this.PushAttributeValueEnd(this._index - 1);
this._state = HtmlDocument.ParseState.BetweenAttributes;
continue;
}
if (this._c == 60 && this._index < this.Text.Length && (int) this.Text[this._index] == 37)
{
this._oldstate = this._state;
this._state = HtmlDocument.ParseState.ServerSideCode;
continue;
}
continue;
case HtmlDocument.ParseState.ServerSideCode:
if (this._c == 37 && this._index < this.Text.Length && (int) this.Text[this._index] == 62)
{
switch (this._oldstate)
{
case HtmlDocument.ParseState.BetweenAttributes:
this.PushAttributeNameEnd(this._index + 1);
this._state = HtmlDocument.ParseState.BetweenAttributes;
break;
case HtmlDocument.ParseState.AttributeAfterEquals:
this._state = HtmlDocument.ParseState.AttributeValue;
break;
default:
this._state = this._oldstate;
break;
}
this.IncrementPosition();
continue;
}
continue;
case HtmlDocument.ParseState.PcData:
if (this._currentnode._namelength + 3 <= this.Text.Length - (this._index - 1) && string.Compare(this.Text.Substring(this._index - 1, this._currentnode._namelength + 2), "</" + this._currentnode.Name, StringComparison.OrdinalIgnoreCase) == 0)
{
int c = (int) this.Text[this._index - 1 + 2 + this._currentnode.Name.Length];
if (c == 62 || HtmlDocument.IsWhiteSpace(c))
{
HtmlNode node = this.CreateNode(HtmlNodeType.Text, this._currentnode._outerstartindex + this._currentnode._outerlength);
node._outerlength = this._index - 1 - node._outerstartindex;
this._currentnode.AppendChild(node);
this.PushNodeStart(HtmlNodeType.Element, this._index - 1);
this.PushNodeNameStart(false, this._index - 1 + 2);
this._state = HtmlDocument.ParseState.Tag;
this.IncrementPosition();
continue;
}
continue;
}
continue;
default:
continue;
}
}
if (this._currentnode._namestartindex > 0)
this.PushNodeNameEnd(this._index);
this.PushNodeEnd(this._index, false);
this.Lastnodes.Clear();
}
private void Parse()
{
int lastquote = 0;
if (OptionComputeChecksum)
{
_crc32 = new Crc32();
}
Lastnodes = new Dictionary<string, HtmlNode>();
_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, StringComparison.OrdinalIgnoreCase) == 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();
}
private void Parse()
{
int lastquote = 0;
if (OptionComputeChecksum)
{
_crc32 = new Crc32();
}
_lastnodes = new Hashtable();
_c = 0;
_fullcomment = false;
_parseerrors = new ArrayList();
_line = 1;
_lineposition = 1;
_maxlineposition = 1;
_state = ParseState.Text;
_oldstate = _state;
_documentnode._innerlength = _text.FullLength;
_documentnode._outerlength = _text.FullLength;
_lastparentnode = _documentnode;
_currentnode = CreateNode(HtmlNodeType.Text, 0);
_currentattribute = null;
_index = 0;
PushNodeStart(HtmlNodeType.Text, 0);
DoParse (lastquote);
}
