html.cpp

一个完整的编辑器的代码（很值得参考

///////////////////////////////////////////////////////////////////////////
//  File:    html.cpp
//  Version: 1.1.0.4
//  Updated: 19-Jul-1998
//
//  Copyright:  Ferdinand Prantl, portions by Stcherbatchenko Andrei
//  E-mail:     [e-mail]
//
//  HTML syntax highlighing definition
//
//  You are free to use or modify this code to the following restrictions:
//  - Acknowledge me somewhere in your about box, simple "Parts of code by.."
//  will be enough. If you can't (or don't want to), contact me personally.
//  - LEAVE THIS HEADER INTACT
////////////////////////////////////////////////////////////////////////////

#include "stdafx.h"
#include "ccrystaltextview.h"

#ifdef _DEBUG
#define new DEBUG_NEW
#undef THIS_FILE
static char THIS_FILE[] = __FILE__;
#endif

//  C++ keywords (MSVC5.0 + POET5.0)
static LPTSTR s_apszHtmlKeywordList[] =
  {
    // HTML section
    _T ("DOCTYPE"),
    _T ("PUBLIC"),
    _T ("FRAME"),
    _T ("FRAMESET"),
    _T ("NOFRAMES"),
    // HEAD section
    _T ("HEAD"),
    _T ("TITLE"),
    _T ("ISINDEX"),
    _T ("META"),
    _T ("LINK"),
    _T ("BASE"),
    _T ("SCRIPT"),
    _T ("STYLE"),
    // BODY section
    _T ("BODY"),
    // headings
    _T ("H1"),
    _T ("H2"),
    _T ("H3"),
    _T ("H4"),
    _T ("H5"),
    _T ("H6"),
    // lists
    _T ("UL"),
    _T ("OL"),
    _T ("DIR"),
    _T ("MENU"),
    _T ("LI"),
    _T ("DL"),
    _T ("DT"),
    _T ("DD"),
    // text containers
    _T ("P"),
    _T ("PRE"),
    _T ("BLOCKQUOTE"),
    _T ("ADDRESS"),
    // others
    _T ("DIV"),
    _T ("SPAN"),
    _T ("CENTER"),
    _T ("HR"),
    _T ("FORM"),
    _T ("TABLE"),
    _T ("LEFT"),
    _T ("RIGHT"),
    _T ("TOP"),
    // logical markup
    _T ("EM"),
    _T ("STRONG"),
    _T ("DFN"),
    _T ("CODE"),
    _T ("SAMP"),
    _T ("KBD"),
    _T ("VAR"),
    _T ("CITE"),
    // physical markup
    _T ("TT"),
    _T ("I"),
    _T ("B"),
    _T ("U"),
    _T ("STRIKE"),
    _T ("BIG"),
    _T ("SMALL"),
    _T ("SUB"),
    _T ("SUP"),
    // special markup
    _T ("A"),
    _T ("BASEFONT"),
    _T ("IMG"),
    _T ("APPLET"),
    _T ("PARAM"),
    _T ("FONT"),
    _T ("BR"),
    _T ("MAP"),
    _T ("AREA"),
    // forms
    _T ("INPUT"),
    _T ("SELECT"),
    _T ("OPTION"),
    _T ("TEXTAREA"),
    _T ("ONCLICK"),
    // tables
    _T ("CAPTION"),
    _T ("TR"),
    _T ("TH"),
    _T ("TD"),
    _T ("HTML"),
    NULL
  };

static LPTSTR s_apszUser1KeywordList[] =
  {
    _T ("ACTION"),
    _T ("ALIGN"),
    _T ("ALINK"),
    _T ("BACKGROUND"),
    _T ("BGCOLOR"),
    _T ("COLOR"),
    _T ("COMPACT"),
    _T ("CONTENT"),
    _T ("ENCTYPE"),
    _T ("FACE"),
    _T ("HEIGHT"),
    _T ("HREF"),
    _T ("HTTP-EQUIV"),
    _T ("LINK"),
    _T ("METHOD"),
    _T ("NAME"),
    _T ("PROMPT"),
    _T ("REL"),
    _T ("REV"),
    _T ("START"),
    _T ("TEXT"),
    _T ("TYPE"),
    _T ("VALUE"),
    _T ("VLINK"),
    _T ("WIDTH"),
    _T ("ADD_DATE"),
    _T ("ALT"),
    _T ("BORDER"),
    _T ("CELLPADDING"),
    _T ("CELLSPACING"),
    _T ("CHECKED"),
    _T ("CLEAR"),
    _T ("CODE"),
    _T ("CODEBASE"),
    _T ("COLS"),
    _T ("COLSPAN"),
    _T ("COORDS"),
    _T ("FOLDED"),
    _T ("HSPACE"),
    _T ("ISMAP"),
    _T ("LAST_MODIFIED"),
    _T ("LAST_VISIT"),
    _T ("MAXLENGTH"),
    _T ("MULTIPLE"),
    _T ("NORESIZE"),
    _T ("NOSHADE"),
    _T ("NOWRAP"),
    _T ("ROWS"),
    _T ("ROWSPAN"),
    _T ("SELECTED"),
    _T ("SHAPE"),
    _T ("SIZE"),
    _T ("SRC"),
    _T ("TARGET"),
    _T ("USEMAP"),
    _T ("VALIGN"),
    _T ("VSPACE"),
    NULL
  };

static LPTSTR s_apszUser2KeywordList[] =
  {
    _T ("nbsp"),
    _T ("quot"),
    _T ("amp"),
    _T ("lt"),
    _T ("lt"),
    _T ("gt"),
    _T ("copy"),
    _T ("reg"),
    _T ("acute"),
    _T ("laquo"),
    _T ("raquo"),
    _T ("iexcl"),
    _T ("iquest"),
    _T ("Agrave"),
    _T ("agrave"),
    _T ("Aacute"),
    _T ("aacute"),
    _T ("Acirc"),
    _T ("acirc"),
    _T ("Atilde"),
    _T ("atilde"),
    _T ("Auml"),
    _T ("auml"),
    _T ("Aring"),
    _T ("aring"),
    _T ("AElig"),
    _T ("aelig"),
    _T ("Ccedil"),
    _T ("ccedil"),
    _T ("ETH"),
    _T ("eth"),
    _T ("Egrave"),
    _T ("egrave"),
    _T ("Eacute"),
    _T ("eacute"),
    _T ("Ecirc"),
    _T ("ecirc"),
    _T ("Euml"),
    _T ("euml"),
    _T ("Igrave"),
    _T ("igrave"),
    _T ("Iacute"),
    _T ("iacute"),
    _T ("Icirc"),
    _T ("icirc"),
    _T ("Iuml"),
    _T ("iuml"),
    _T ("Ntilde"),
    _T ("ntilde"),
    _T ("Ograve"),
    _T ("ograve"),
    _T ("Oacute"),
    _T ("oacute"),
    _T ("Ocirc"),
    _T ("ocirc"),
    _T ("Otilde"),
    _T ("otilde"),
    _T ("Ouml"),
    _T ("ouml"),
    _T ("Oslash"),
    _T ("oslash"),
    _T ("Ugrave"),
    _T ("ugrave"),
    _T ("Uacute"),
    _T ("uacute"),
    _T ("Ucirc"),
    _T ("ucirc"),
    _T ("Uuml"),
    _T ("uuml"),
    _T ("Yacute"),
    _T ("yacute"),
    _T ("yuml"),
    _T ("THORN"),
    _T ("thorn"),
    _T ("szlig"),
    _T ("sect"),
    _T ("para"),
    _T ("micro"),
    _T ("brvbar"),
    _T ("plusmn"),
    _T ("middot"),
    _T ("uml"),
    _T ("cedil"),
    _T ("ordf"),
    _T ("ordm"),
    _T ("not"),
    _T ("shy"),
    _T ("macr"),
    _T ("deg"),
    _T ("sup1"),
    _T ("sup2"),
    _T ("sup3"),
    _T ("frac14"),
    _T ("frac12"),
    _T ("frac34"),
    _T ("times"),
    _T ("divide"),
    _T ("cent"),
    _T ("pound"),
    _T ("curren"),
    _T ("yen"),
    NULL
  };

static BOOL
IsXKeyword (LPTSTR apszKeywords[], LPCTSTR pszChars, int nLength)
{
  for (int L = 0; apszKeywords[L] != NULL; L++)
    {
      if (_tcsnicmp (apszKeywords[L], pszChars, nLength) == 0
            && apszKeywords[L][nLength] == 0)
        return TRUE;
    }
  return FALSE;
}

static BOOL
IsHtmlKeyword (LPCTSTR pszChars, int nLength)
{
  return IsXKeyword (s_apszHtmlKeywordList, pszChars, nLength);
}

static BOOL
IsUser1Keyword (LPCTSTR pszChars, int nLength)
{
  return IsXKeyword (s_apszUser1KeywordList, pszChars, nLength);
}

static BOOL
IsUser2Keyword (LPCTSTR pszChars, int nLength)
{
  return IsXKeyword (s_apszUser2KeywordList, pszChars, nLength);
}

static BOOL
IsHtmlNumber (LPCTSTR pszChars, int nLength)
{
  if (nLength > 2 && pszChars[0] == '0' && pszChars[1] == 'x')
    {
      for (int I = 2; I < nLength; I++)
        {
          if (_istdigit (pszChars[I]) || (pszChars[I] >= 'A' && pszChars[I] <= 'F') ||
                (pszChars[I] >= 'a' && pszChars[I] <= 'f'))
            continue;
          return FALSE;
        }
      return TRUE;
    }
  if (!_istdigit (pszChars[0]))
    return FALSE;
  for (int I = 1; I < nLength; I++)
    {
      if (!_istdigit (pszChars[I]) && pszChars[I] != '+' &&
            pszChars[I] != '-' && pszChars[I] != '.' && pszChars[I] != 'e' &&
            pszChars[I] != 'E')
        return FALSE;
    }
  return TRUE;
}

#define DEFINE_BLOCK(pos, colorindex)   \
ASSERT((pos) >= 0 && (pos) <= nLength);\
if (pBuf != NULL)\
  {\
    if (nActualItems == 0 || pBuf[nActualItems - 1].m_nCharPos <= (pos)){\
        pBuf[nActualItems].m_nCharPos = (pos);\
        pBuf[nActualItems].m_nColorIndex = (colorindex);\