public static unsafe TokenList ParseTokens(string text)
{
var tokens = new TokenList(text);
long depth = 0;
fixed (char* pText = text)
{
char* p = pText;
char* pEnd = pText + text.Length;
while (p < pEnd)
{
// skip past whitespace between tags
char* pStart = p;
while (p != pEnd && (*p == ' ' || *p == '\t' || *p == '\r' || *p == '\n'))
++p;
// identify text region (if there is one)
if (p != pEnd && *p != '<')
{
while (p != pEnd && *p != '<')
++p;
tokens.AddRegion(pStart - pText, TokenType.TextNormal, p - pStart, depth);
}
//else if (p != pStart)
//{
// // remember that this is whitespace, but no more details
// tokens.AddRegion(pStart - pText, TokenType.TextNormal, 0, depth);
//}
// identify tag region
if (p != pEnd)
{
pStart = p;
++p;
if (*p == '!')
{
p = HandleExclamationPoint(pText, pStart, pEnd, depth, tokens);
}
else if (*p == '?')
{
p = HandleQuestionMark(pText, pStart, pEnd, depth, tokens);
}
else
{
// normal tags (closing, opening, self-closing)
bool isClosing = (*p == '/');
while (p != pEnd && *p != '>') ++p;
if (isClosing)
{
--depth;
}
else
{
// QName format
// [prefix:]local
char* pFirstSymbol = pStart + 1;
char* pTmp = pFirstSymbol;
long namePrefixLength = 0;
while (pTmp != p && (*pTmp != ' ' && *pTmp != '\t' && *pTmp != '\r' && *pTmp != '\n'))
{
if (*pTmp == ':' && namePrefixLength == 0)
namePrefixLength = (pTmp - pFirstSymbol);
++pTmp;
}
long length = pTmp - pFirstSymbol;
tokens.AddTag(pFirstSymbol - pText, length, namePrefixLength, depth);
// check for self-closing
if ((*(p - 1) != '/'))
++depth;
// attributes
ParseAttributes(pFirstSymbol, pTmp, p, tokens);
}
}
}
++p;
}
}
if (depth != 0)
throw new Exception("bad depth");
return tokens;
}
