public XDocument ParseDocument(string text)
{
//System.Diagnostics.Stopwatch sw = System.Diagnostics.Stopwatch.StartNew();
XDocument doc = new XDocument();
try
{
StringCounter sc = new StringCounter(text);
sc.TrimStart();
if (sc.StartsWith("<"))
{
while (!sc.IsAtEnd)
{
XContainer childNode = this.ParseNode(sc, doc);
}
XElement[] enm = MyHelper.EnumToArray(doc.Elements());
if (enm.Length > 1)
{
XElement root = new XElement(XName.Get("root"));
foreach (XElement elem in enm)
{
root.Add(elem);
}
doc.Elements().Remove();
doc.Add(root);
}
}
}
catch (Exception ex)
{
System.Diagnostics.Debug.WriteLine(ex.Message);
}
//sw.Stop();
//System.Diagnostics.Debug.WriteLine(sw.Elapsed.TotalMilliseconds.ToString("N0"));
return doc;
}
public XParseDocument ParseDocument(string text)
{
//System.Diagnostics.Stopwatch sw = System.Diagnostics.Stopwatch.StartNew();
XParseDocument doc = new XParseDocument();
StringCounter sc;
try
{
sc = new StringCounter(text);
sc.TrimStart();
if (sc.StartsWith("<"))
{
while (!sc.IsAtEnd)
{
XContainer childNode = this.ParseNode(sc, doc);
}
XParseElement[] enm = MyHelper.EnumToArray(doc.Elements());
if (enm.Length > 1)
{
XParseElement root = new XParseElement(XParseName.Get("root"));
foreach (XParseElement elem in enm)
{
root.Add(elem);
}
doc.Elements().Remove();
doc.Add(root);
}
}
}
catch (Exception ex)
{
System.Diagnostics.Debug.WriteLine(ex.Message);
}
//sw.Stop();
//System.Diagnostics.Debug.WriteLine(sw.Elapsed.TotalMilliseconds.ToString("N0"));
return(doc);
}
private XContainer ParseNode(StringCounter sc, XContainer parent)
{
sc.TrimStart();
if (sc.StartValue == '<')
{
sc.Index = sc.Index + 1;
XContainer node = null;
bool isComment = false;
bool isDeclaration = false;
int breakOff = 0;
XParseName name = null;
if (sc.StartsWith("?xml"))
{
//Declaration
isDeclaration = true;
sc.Index += 4;
}
else if (sc.StartValue == '!')
{
//Comment
isComment = true;
if (sc.Value[sc.Index + 1] == '-')
{
breakOff = 1;
}
if (sc.Value[sc.Index + 2] == '-')
{
breakOff = 2;
}
sc.Index += breakOff + 1;
sc.TrimStart();
}
else if (IsValidTagChar(sc.StartValue))
{
//Name
name = ParseName(sc);
if (name != null)
{
node = new XParseElement(name);
}
else
{
throw new Exception("Invalid Node Name.");
}
}
else
{
throw new Exception("Invalid Node Name.");
}
if (node != null || isComment || isDeclaration)
{
//Attributes
bool elementAtEnd = name != null ? (name.LocalName == "br") : false;
string comment = string.Empty;
string declVer = string.Empty;
string declEnc = string.Empty;
string declSta = string.Empty;
for (int i = sc.Index; i < sc.Value.Length; i++)
{
if (!isComment && !isDeclaration)
{
//Node Attributes
if (sc.Value[i] != ' ')
{
if (sc.Value[i] == '>')
{
sc.Index = i + 1;
break;
}
else if (sc.Value[i] == '/' && sc.Value[i + 1] == '>')
{
elementAtEnd = true;
sc.Index += 2;
break;
}
else if (IsValidTagChar(sc.Value[i]))
{
XParseAttribute att = ParseAttribute(sc.NewIndex(i));
i = sc.Index - 1;
if (att != null)
{
node.Add(att);
}
}
}
}
else
{
if (isComment)
{
//Comment
if ((breakOff == 2 && sc.Value[i] == '-' && sc.Value[i + 1] == '-' && sc.Value[i + 2] == '>') || (breakOff == 0 && sc.Value[i] == '>') || (breakOff == 1 && sc.Value[i] == '-' && sc.Value[i + 1] == '>'))
{
if (parent != null)
{
parent.Add(new XComment(comment));
}
sc.Index = i + breakOff + 1;
break;
}
else
{
comment += sc.Value[i];
}
}
else if (isDeclaration)
{
//Declaration
if (sc.Value[i] == '?' && sc.Value[i + 1] == '>')
{
if (parent != null && parent is XParseDocument)
{
((XParseDocument)parent).Declaration = new XDeclaration(declVer, declEnc, declSta);
}
sc.Index = i + 2;
break;
}
else if (IsValidTagChar(sc.Value[i]))
{
XParseAttribute att = ParseAttribute(sc.NewIndex(i));
i = sc.Index - 1;
if (att != null)
{
if (att.Name.LocalName.ToLower() == "version")
{
declVer = att.Value;
}
else if (att.Name.LocalName.ToLower() == "encoding")
{
declEnc = att.Value;
}
else if (att.Name.LocalName.ToLower() == "standalone")
{
declSta = att.Value;
}
}
}
}
}
}
if (node != null)
{
parent.Add(node);
}
//Content
if (node != null && !elementAtEnd)
{
string innerText = string.Empty;
string fullName = name.ToString();
for (int i = sc.Index; i < sc.Value.Length; i++)
{
if (sc.Value[i] == '<')
{
if (innerText != string.Empty)
{
node.Add(XmlParser.DecodeXml(innerText));
innerText = string.Empty;
}
if (sc.Value[i + 1] == '/')
{
//End Tag
XParseName endName = ParseName(sc.NewIndex(i + 2));
if (endName != null)
{
if (CompareXName(endName, name))
{
//Actual End Name
sc.Index = i + 3 + fullName.Length;
break;
}
else
{
if (name.LocalName != "script")
{
//Other End Name
XContainer par = ((parent is XParseElement && XmlParser.CompareXName(((XParseElement)parent).Name, endName)) ? parent : (XContainer)this.FindParent(parent, endName));
if (par != null)
{
if (name.LocalName != "form")
{
XParseElement[] enm = MyHelper.EnumToArray(node.Elements());
if (enm.Length > 0)
{
foreach (XParseElement elem in enm)
{
parent.Add(elem);
}
}
node.Elements().Remove();
sc.Index = i;
break;
}
else
{
sc.Index = i + endName.ToString().Length + 3;
i = sc.Index - 1;
}
}
else
{
sc.Index = i + endName.ToString().Length + 2;
i = sc.Index - 1;
}
}
}
}
else if (fullName == "script")
{
//Script Text
//innerText += text[i];
}
else
{
throw new Exception("Invalid End Name.");
}
}
else if (fullName == "script")
{
//Script Text
//innerText += text[i];
}
else
{
//Start Tag
XContainer childNode = this.ParseNode(sc.NewIndex(i), node);
i = sc.Index - 1;
}
}
else if (!(sc.Value[i] == ' ' && innerText == string.Empty))
{
//Inner Text
if (fullName != "script")
{
innerText += sc.Value[i];
}
}
}
}
}
sc.TrimStart();
return(node);
}
else
{
throw new Exception("Invalid Start Tag. A node has to start with [<].");
}
}
private XContainer ParseNode(StringCounter sc, XContainer parent)
{
sc.TrimStart();
if (sc.StartValue == '<')
{
sc.Index = sc.Index + 1;
XContainer node = null;
bool isComment = false;
bool isDeclaration = false;
int breakOff = 0;
XName name = null;
if (sc.StartsWith("?xml"))
{
//Declaration
isDeclaration = true;
sc.Index += 4;
}
else if (sc.StartValue == '!')
{
//Comment
isComment = true;
if (sc.Value[sc.Index + 1] == '-') breakOff = 1;
if (sc.Value[sc.Index + 2] == '-') breakOff = 2;
sc.Index += breakOff + 1;
sc.TrimStart();
}
else if (IsValidTagChar(sc.StartValue))
{
//Name
name = ParseName(sc);
if (name != null)
{
node = new XElement(name);
}
else
{
throw new Exception("Invalid Node Name.");
}
}
else
{
throw new Exception("Invalid Node Name.");
}
if (node != null || isComment || isDeclaration)
{
//Attributes
bool elementAtEnd = name != null ? (name.LocalName == "br") : false;
string comment = string.Empty;
string declVer = string.Empty;
string declEnc = string.Empty;
string declSta = string.Empty;
for (int i = sc.Index; i < sc.Value.Length; i++)
{
if (!isComment && !isDeclaration)
{
//Node Attributes
if (sc.Value[i] != ' ')
{
if (sc.Value[i] == '>')
{
sc.Index = i + 1;
break;
}
else if (sc.Value[i] == '/' && sc.Value[i + 1] == '>')
{
elementAtEnd = true;
sc.Index += 2;
break;
}
else if (IsValidTagChar(sc.Value[i]))
{
XAttribute att = ParseAttribute(sc.NewIndex(i));
i = sc.Index - 1;
if (att != null)
{
node.Add(att);
}
}
}
}
else
{
if (isComment)
{
//Comment
if ((breakOff == 2 && sc.Value[i] == '-' && sc.Value[i + 1] == '-' && sc.Value[i + 2] == '>') || (breakOff == 0 && sc.Value[i] == '>') || (breakOff == 1 && sc.Value[i] == '-' && sc.Value[i + 1] == '>'))
{
if (parent != null) parent.Add(new XComment(comment));
sc.Index = i + breakOff + 1;
break;
}
else
{
comment += sc.Value[i];
}
}
else if (isDeclaration)
{
//Declaration
if (sc.Value[i] == '?' && sc.Value[i + 1] == '>')
{
if (parent != null && parent is XDocument) ((XDocument)parent).Declaration = new XDeclaration(declVer, declEnc, declSta);
sc.Index = i + 2;
break;
}
else if (IsValidTagChar(sc.Value[i]))
{
XAttribute att = ParseAttribute(sc.NewIndex(i));
i = sc.Index - 1;
if (att != null)
{
if (att.Name.LocalName.ToLower() == "version")
{
declVer = att.Value;
}
else if (att.Name.LocalName.ToLower() == "encoding")
{
declEnc = att.Value;
}
else if (att.Name.LocalName.ToLower() == "standalone")
{
declSta = att.Value;
}
}
}
}
}
}
if (node != null) parent.Add(node);
//Content
if (node != null && !elementAtEnd)
{
string innerText = string.Empty;
string fullName = name.ToString();
for (int i = sc.Index; i < sc.Value.Length; i++)
{
if (sc.Value[i] == '<')
{
if (innerText != string.Empty)
{
node.Add(XmlParser.DecodeXml(innerText));
innerText = string.Empty;
}
if (sc.Value[i + 1] == '/')
{
//End Tag
XName endName = ParseName(sc.NewIndex(i + 2));
if (endName != null)
{
if (CompareXName(endName, name))
{
//Actual End Name
sc.Index = i + 3 + fullName.Length;
break;
}
else
{
if (name.LocalName != "script")
{
//Other End Name
XContainer par = ((parent is XElement && XmlParser.CompareXName(((XElement)parent).Name, endName)) ? parent : (XContainer)this.FindParent(parent, endName));
if (par != null)
{
if (name.LocalName != "form")
{
XElement[] enm = MyHelper.EnumToArray(node.Elements());
if (enm.Length > 0)
{
foreach (XElement elem in enm)
{
parent.Add(elem);
}
}
node.Elements().Remove();
sc.Index = i;
break;
}
else
{
sc.Index = i + endName.ToString().Length + 3;
i = sc.Index - 1;
}
}
else
{
sc.Index = i + endName.ToString().Length + 2;
i = sc.Index - 1;
}
}
}
}
else if (fullName == "script")
{
//Script Text
//innerText += text[i];
}
else
{
throw new Exception("Invalid End Name.");
}
}
else if (fullName == "script")
{
//Script Text
//innerText += text[i];
}
else
{
//Start Tag
XContainer childNode = this.ParseNode(sc.NewIndex(i), node);
i = sc.Index - 1;
}
}
else if (!(sc.Value[i] == ' ' && innerText == string.Empty))
{
//Inner Text
if (fullName != "script") innerText += sc.Value[i];
}
}
}
}
sc.TrimStart();
return node;
}
else
{
throw new Exception("Invalid Start Tag. A node has to start with [<].");
}
}
