python.cpp

一个完整的编辑器的代码（很值得参考

///////////////////////////////////////////////////////////////////////////
//  File:    python.cpp
//  Version: 1.1.0.4
//  Updated: 19-Jul-1998
//
//  Copyright:  Ferdinand Prantl, portions by Stcherbatchenko Andrei
//  E-mail:     [e-mail]
//
//  Python 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_apszPythonKeywordList[] =
  {
    _T ("access"),
    _T ("and"),
    _T ("break"),
    _T ("class"),
    _T ("continue"),
    _T ("def"),
    _T ("del"),
    _T ("elif"),
    _T ("else"),
    _T ("except"),
    _T ("exec"),
    _T ("finally"),
    _T ("for"),
    _T ("from"),
    _T ("global"),
    _T ("if"),
    _T ("import"),
    _T ("in"),
    _T ("is"),
    _T ("lambda"),
    _T ("not"),
    _T ("or"),
    _T ("pass"),
    _T ("print"),
    _T ("raise"),
    _T ("return"),
    _T ("try"),
    _T ("while"),
    NULL
  };

static LPTSTR s_apszUser1KeywordList[] =
  {
    _T ("argv"),
    _T ("builtin_module_names"),
    _T ("exc_type"),
    _T ("exc_value"),
    _T ("exc_traceback"),
    _T ("exit"),
    _T ("exitfunc"),
    _T ("last_type"),
    _T ("last_value"),
    _T ("last_traceback"),
    _T ("modules"),
    _T ("path"),
    _T ("ps1"),
    _T ("ps2"),
    _T ("settrace"),
    _T ("setprofile"),
    _T ("stdin"),
    _T ("stdout"),
    _T ("stderr"),
    _T ("tracebacklimit"),
    _T ("AttributeError"),
    _T ("EOFError"),
    _T ("IOError"),
    _T ("ImportError"),
    _T ("IndexError"),
    _T ("KeyError"),
    _T ("KeyboardInterrupt"),
    _T ("MemoryError"),
    _T ("NameError"),
    _T ("OverflowError"),
    _T ("RuntimeError"),
    _T ("SyntaxError"),
    _T ("SystemError"),
    _T ("SystemExit"),
    _T ("TypeError"),
    _T ("ValueError"),
    _T ("ZeroDivisionError"),
    NULL
  };

static LPTSTR s_apszUser2KeywordList[] =
  {
    _T ("id"),
    _T ("__init__"),
    _T ("__del__"),
    _T ("__repr__"),
    _T ("__str__"),
    _T ("__cmp__"),
    _T ("__hash__"),
    _T ("__dict__"),
    _T ("__methods__"),
    _T ("__members__"),
    _T ("__class__"),
    _T ("__bases__"),
    _T ("__add__"),
    _T ("__mul__"),
    _T ("__mod__"),
    _T ("__sub__"),
    _T ("__div__"),
    _T ("divmod"),
    _T ("__divmod__"),
    _T ("pow"),
    _T ("__pow__"),
    _T ("__and__"),
    _T ("__xor__"),
    _T ("__or__"),
    _T ("__lshift__"),
    _T ("__rshift__"),
    _T ("nonzero"),
    _T ("__nonzero__"),
    _T ("coerce"),
    _T ("__coerce__"),
    _T ("__neg__"),
    _T ("__pos__"),
    _T ("abs"),
    _T ("__abs__"),
    _T ("__invert__"),
    _T ("int"),
    _T ("__int__"),
    _T ("float"),
    _T ("__float__"),
    _T ("long"),
    _T ("__long__"),
    _T ("float"),
    _T ("__float__"),
    _T ("oct"),
    _T ("__oct__"),
    _T ("hex"),
    _T ("__hex__"),
    _T ("range"),
    _T ("round"),
    _T ("xrange"),
    _T ("len"),
    _T ("__len__"),
    _T ("__getitem__"),
    NULL
  };

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

static BOOL
IsPythonKeyword (LPCTSTR pszChars, int nLength)
{
  return IsXKeyword (s_apszPythonKeywordList, 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
IsPythonNumber (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);\
        nActualItems ++;}\
  }

#define COOKIE_COMMENT          0x0001
#define COOKIE_PREPROCESSOR     0x0002
#define COOKIE_EXT_COMMENT      0x0004
#define COOKIE_STRING           0x0008
#define COOKIE_CHAR             0x0010

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

  LPCTSTR pszChars = GetLineChars (nLineIndex);
  BOOL bFirstChar = (dwCookie & ~COOKIE_EXT_COMMENT) == 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 (xisalnum (pszChars[nPos]) || pszChars[nPos] == '.' && nPos > 0 && (!xisalpha (pszChars[nPos - 1]) && !xisalpha (pszChars[nPos + 1])))
                {
                  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;
        }

      if (pszChars[I] == '#')
        {
          DEFINE_BLOCK (I, COLORINDEX_COMMENT);
          dwCookie |= COOKIE_COMMENT;
          break;
        }

      //  Normal text
      if (pszChars[I] == '"')
        {
          DEFINE_BLOCK (I, COLORINDEX_STRING);
          dwCookie |= COOKIE_STRING;
          continue;
        }
      if (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 (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] == '.' && I > 0 && (!xisalpha (pszChars[I - 1]) && !xisalpha (pszChars[I + 1])))
        {
          if (nIdentBegin == -1)
            nIdentBegin = I;
        }
      else
        {
          if (nIdentBegin >= 0)
            {
              if (IsPythonKeyword (pszChars + nIdentBegin, I - nIdentBegin))
                {
                  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 (IsPythonNumber (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);
                    }
                }
              bRedefineBlock = TRUE;
              bDecIndex = TRUE;
              nIdentBegin = -1;
            }
        }
    }

  if (nIdentBegin >= 0)
    {
      if (IsPythonKeyword (pszChars + nIdentBegin, I - nIdentBegin))
        {
          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 (IsPythonNumber (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);
            }
        }
    }

  if (pszChars[nLength - 1] != '\\')
    dwCookie &= COOKIE_EXT_COMMENT;
  return dwCookie;
}