html.cpp

一个完整的编辑器的代码（很值得参考

        nActualItems ++;}\
  }

#define COOKIE_COMMENT          0x0001
#define COOKIE_PREPROCESSOR     0x0002
#define COOKIE_EXT_COMMENT      0x0004
#define COOKIE_STRING           0x0008
#define COOKIE_CHAR             0x0010
#define COOKIE_USER1            0x0020
#define COOKIE_EXT_USER1        0x0040

DWORD CCrystalTextView::
ParseLineHtml (DWORD dwCookie, int nLineIndex, TEXTBLOCK * pBuf, int &nActualItems)
{
  int nLength = GetLineLength (nLineIndex);
  if (nLength <= 1)
    return dwCookie & (COOKIE_EXT_COMMENT|COOKIE_EXT_USER1);

  LPCTSTR pszChars = GetLineChars (nLineIndex);
  BOOL bFirstChar = (dwCookie & ~(COOKIE_EXT_COMMENT|COOKIE_EXT_USER1)) == 0;
  BOOL bRedefineBlock = TRUE;
  BOOL bDecIndex = FALSE;
  int nIdentBegin = -1;
  for (int I = 0;; I++)
    {
      if (bRedefineBlock)
        {
          int nPos = I;
          if (bDecIndex)
            nPos--;
          if (dwCookie & (COOKIE_COMMENT | COOKIE_EXT_COMMENT))
            {
              DEFINE_BLOCK (nPos, COLORINDEX_COMMENT);
            }
          else if (dwCookie & (COOKIE_CHAR | COOKIE_STRING))
            {
              DEFINE_BLOCK (nPos, COLORINDEX_STRING);
            }
          else if (dwCookie & COOKIE_PREPROCESSOR)
            {
              DEFINE_BLOCK (nPos, COLORINDEX_PREPROCESSOR);
            }
          else if (dwCookie & COOKIE_EXT_USER1)
            {
              DEFINE_BLOCK (nPos, COLORINDEX_FUNCNAME);
            }
          else
            {
              if (xisalnum (pszChars[nPos]) || pszChars[nPos] == '.')
                {
                  DEFINE_BLOCK (nPos, COLORINDEX_NORMALTEXT);
                }
              else
                {
                  DEFINE_BLOCK (nPos, COLORINDEX_OPERATOR);
                  bRedefineBlock = TRUE;
                  bDecIndex = TRUE;
                  goto out;
                }
            }
          bRedefineBlock = FALSE;
          bDecIndex = FALSE;
        }
out:

      if (I == nLength)
        break;

      if (dwCookie & COOKIE_COMMENT)
        {
          DEFINE_BLOCK (I, COLORINDEX_COMMENT);
          dwCookie |= COOKIE_COMMENT;
          break;
        }

      //  String constant "...."
      if (dwCookie & COOKIE_STRING)
        {
          if (pszChars[I] == '"' && (I == 0 || I == 1 && pszChars[I - 1] != '\\' || I >= 2 && (pszChars[I - 1] != '\\' || pszChars[I - 1] == '\\' && pszChars[I - 2] == '\\')))
            {
              dwCookie &= ~COOKIE_STRING;
              bRedefineBlock = TRUE;
            }
          continue;
        }

      //  Char constant '..'
      if (dwCookie & COOKIE_CHAR)
        {
          if (pszChars[I] == '\'' && (I == 0 || I == 1 && pszChars[I - 1] != '\\' || I >= 2 && (pszChars[I - 1] != '\\' || pszChars[I - 1] == '\\' && pszChars[I - 2] == '\\')))
            {
              dwCookie &= ~COOKIE_CHAR;
              bRedefineBlock = TRUE;
            }
          continue;
        }

      //  Extended comment <!--....-->
      if (dwCookie & COOKIE_EXT_COMMENT)
        {
          if (I > 1 && pszChars[I] == '>' && pszChars[I - 1] == '-' && pszChars[I - 2] == '-')
            {
              dwCookie &= ~COOKIE_EXT_COMMENT;
              bRedefineBlock = TRUE;
            }
          continue;
        }

      //  Extended comment <?....?>
      if (dwCookie & COOKIE_EXT_USER1)
        {
          if (I > 0 && pszChars[I] == '>' && (pszChars[I - 1] == '?' || pszChars[I - 1] == '%'))
            {
              dwCookie &= ~COOKIE_EXT_USER1;
              bRedefineBlock = TRUE;
            }
          continue;
        }

      //  Normal text
      if ((dwCookie & COOKIE_PREPROCESSOR) && pszChars[I] == '"')
        {
          DEFINE_BLOCK (I, COLORINDEX_STRING);
          dwCookie |= COOKIE_STRING;
          continue;
        }

      if ((dwCookie & COOKIE_PREPROCESSOR) && pszChars[I] == '\'')
        {
          // if (I + 1 < nLength && pszChars[I + 1] == '\'' || I + 2 < nLength && pszChars[I + 1] != '\\' && pszChars[I + 2] == '\'' || I + 3 < nLength && pszChars[I + 1] == '\\' && pszChars[I + 3] == '\'')
          if (!I || !xisalnum (pszChars[I - 1]))
            {
              DEFINE_BLOCK (I, COLORINDEX_STRING);
              dwCookie |= COOKIE_CHAR;
              continue;
            }
        }

      if (!(dwCookie & COOKIE_EXT_USER1) && I < nLength - 3 && pszChars[I] == '<' && pszChars[I + 1] == '!' && pszChars[I + 2] == '-' && pszChars[I + 3] == '-')
        {
          DEFINE_BLOCK (I, COLORINDEX_COMMENT);
          I += 3;
          dwCookie |= COOKIE_EXT_COMMENT;
          dwCookie &= ~COOKIE_PREPROCESSOR;
          continue;
        }

      if (bFirstChar)
        {
          if (!isspace (pszChars[I]))
            bFirstChar = FALSE;
        }

      if (pBuf == NULL)
        continue;               //  We don't need to extract keywords,
      //  for faster parsing skip the rest of loop

      if (xisalnum (pszChars[I]) || pszChars[I] == '.')
        {
          if (nIdentBegin == -1)
            nIdentBegin = I;
        }
      else
        {
          if (nIdentBegin >= 0)
            {
              if (dwCookie & COOKIE_PREPROCESSOR)
                {
                  if (IsHtmlKeyword (pszChars + nIdentBegin, I - nIdentBegin) && (pszChars[nIdentBegin - 1] == _T ('<') || pszChars[nIdentBegin - 1] == _T ('/')))
                    {
                      DEFINE_BLOCK (nIdentBegin, COLORINDEX_KEYWORD);
                    }
                  else if (IsUser1Keyword (pszChars + nIdentBegin, I - nIdentBegin))
                    {
                      DEFINE_BLOCK (nIdentBegin, COLORINDEX_USER1);
                    }
                  else if (IsHtmlNumber (pszChars + nIdentBegin, I - nIdentBegin))
                    {
                      DEFINE_BLOCK (nIdentBegin, COLORINDEX_NUMBER);
                    }
                  else
                    {
                      goto next;
                    }
                }
              else if (dwCookie & COOKIE_USER1)
                {
                  if (IsUser2Keyword (pszChars + nIdentBegin, I - nIdentBegin))
                    {
                      DEFINE_BLOCK (nIdentBegin, COLORINDEX_USER2);
                    }
                  else
                    {
                      goto next;
                    }
                }
              bRedefineBlock = TRUE;
              bDecIndex = TRUE;
              nIdentBegin = -1;
next:
              ;
            }

          //  User1 start: <?
          if (I < nLength && pszChars[I] == '<' && I < nLength - 1 && (pszChars[I + 1] == '?' || pszChars[I + 1] == '%'))
            {
              DEFINE_BLOCK (I, COLORINDEX_FUNCNAME);
              dwCookie |= COOKIE_EXT_USER1;
              nIdentBegin = -1;
              continue;
            }

          //  Preprocessor start: < or bracket
          if (!(dwCookie & COOKIE_EXT_USER1) && I < nLength && (pszChars[I] == '<' && !(I < nLength - 3 && pszChars[I + 1] == '!' && pszChars[I + 2] == '-' && pszChars[I + 3] == '-')/* || pszChars[I] == '{'*/))
            {
              DEFINE_BLOCK (I + 1, COLORINDEX_PREPROCESSOR);
              dwCookie |= COOKIE_PREPROCESSOR;
              nIdentBegin = -1;
              continue;
            }

          //  User1 end: ?>
          if (dwCookie & COOKIE_EXT_USER1)
            {
              if (I > 0 && pszChars[I] == '>' && (pszChars[I - 1] == '?' || pszChars[I - 1] == '%'))
                {
                  dwCookie &= ~COOKIE_EXT_USER1;
                  nIdentBegin = -1;
                  bRedefineBlock = TRUE;
                  bDecIndex = TRUE;
                  continue;
                }
            }

          //  Preprocessor end: > or bracket
          if (dwCookie & COOKIE_PREPROCESSOR)
            {
              if (pszChars[I] == '>'/* || pszChars[I] == '}'*/)
                {
                  dwCookie &= ~COOKIE_PREPROCESSOR;
                  nIdentBegin = -1;
                  bRedefineBlock = TRUE;
                  bDecIndex = TRUE;
                  continue;
                }
            }

          //  Preprocessor start: &
          if (pszChars[I] == '&')
            {
              dwCookie |= COOKIE_USER1;
              nIdentBegin = -1;
              continue;
            }

          //  Preprocessor end: ;
          if (dwCookie & COOKIE_USER1)
            {
              if (pszChars[I] == ';')
                {
                  dwCookie &= ~COOKIE_USER1;
                  nIdentBegin = -1;
                  continue;
                }
            }
        }
    }

  if (nIdentBegin >= 0 && (dwCookie & COOKIE_PREPROCESSOR))
    {
      if (IsHtmlKeyword (pszChars + nIdentBegin, I - nIdentBegin) && (pszChars[nIdentBegin - 1] == _T ('<') || pszChars[nIdentBegin - 1] == _T ('/')))
        {
          DEFINE_BLOCK (nIdentBegin, COLORINDEX_KEYWORD);
        }
      else if (IsUser1Keyword (pszChars + nIdentBegin, I - nIdentBegin))
        {
          DEFINE_BLOCK (nIdentBegin, COLORINDEX_USER1);
        }
      else if (IsUser2Keyword (pszChars + nIdentBegin, I - nIdentBegin))
        {
          DEFINE_BLOCK (nIdentBegin, COLORINDEX_USER2);
        }
      else if (IsHtmlNumber (pszChars + nIdentBegin, I - nIdentBegin))
        {
          DEFINE_BLOCK (nIdentBegin, COLORINDEX_NUMBER);
        }
      else
        {
          bool bFunction = FALSE;

          for (int j = I; j < nLength; j++)
            {
              if (!isspace (pszChars[j]))
                {
                  if (pszChars[j] == '(')
                    {
                      bFunction = TRUE;
                    }
                  break;
                }
            }
          if (bFunction)
            {
              DEFINE_BLOCK (nIdentBegin, COLORINDEX_FUNCNAME);
            }
        }
    }

  //  User1 start: <?
  if (I < nLength && pszChars[I] == '<' && I < nLength - 1 && (pszChars[I + 1] == '?' || pszChars[I + 1] == '%'))
    {
      DEFINE_BLOCK (I, COLORINDEX_FUNCNAME);
      dwCookie |= COOKIE_EXT_USER1;
      nIdentBegin = -1;
      goto end;
    }

  //  Preprocessor start: < or {
  if (!(dwCookie & COOKIE_EXT_USER1) && I < nLength && (pszChars[I] == '<' && !(I < nLength - 3 && pszChars[I + 1] == '!' && pszChars[I + 2] == '-' && pszChars[I + 3] == '-')/* || pszChars[I] == '{'*/))
    {
      DEFINE_BLOCK (I + 1, COLORINDEX_PREPROCESSOR);
      dwCookie |= COOKIE_PREPROCESSOR;
      nIdentBegin = -1;
      goto end;
    }

  //  User1 end: ?>
  if (dwCookie & COOKIE_EXT_USER1)
    {
      if (I > 0 && pszChars[I] == '>' && (pszChars[I - 1] == '?' || pszChars[I - 1] == '%'))
        {
          dwCookie &= ~COOKIE_EXT_USER1;
          nIdentBegin = -1;
        }
    }

  //  Preprocessor end: > or }
  if (dwCookie & COOKIE_PREPROCESSOR)
    {
      if (pszChars[I] == '>'/* || pszChars[I] == '}'*/)
        {
          dwCookie &= ~COOKIE_PREPROCESSOR;
          nIdentBegin = -1;
        }
    }

end:
  dwCookie &= (COOKIE_EXT_COMMENT | COOKIE_STRING | COOKIE_PREPROCESSOR | COOKIE_EXT_USER1);
  return dwCookie;
}