xmltok_impl.c

最新apache的源代码

/*
Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
See the file COPYING for copying permission.
*/

#ifndef IS_INVALID_CHAR
#define IS_INVALID_CHAR(enc, ptr, n) (0)
#endif

#define INVALID_LEAD_CASE(n, ptr, nextTokPtr) \
    case BT_LEAD ## n: \
      if (end - ptr < n) \
	return XML_TOK_PARTIAL_CHAR; \
      if (IS_INVALID_CHAR(enc, ptr, n)) { \
        *(nextTokPtr) = (ptr); \
        return XML_TOK_INVALID; \
      } \
      ptr += n; \
      break;

#define INVALID_CASES(ptr, nextTokPtr) \
  INVALID_LEAD_CASE(2, ptr, nextTokPtr) \
  INVALID_LEAD_CASE(3, ptr, nextTokPtr) \
  INVALID_LEAD_CASE(4, ptr, nextTokPtr) \
  case BT_NONXML: \
  case BT_MALFORM: \
  case BT_TRAIL: \
    *(nextTokPtr) = (ptr); \
    return XML_TOK_INVALID;

#define CHECK_NAME_CASE(n, enc, ptr, end, nextTokPtr) \
   case BT_LEAD ## n: \
     if (end - ptr < n) \
       return XML_TOK_PARTIAL_CHAR; \
     if (!IS_NAME_CHAR(enc, ptr, n)) { \
       *nextTokPtr = ptr; \
       return XML_TOK_INVALID; \
     } \
     ptr += n; \
     break;

#define CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) \
  case BT_NONASCII: \
    if (!IS_NAME_CHAR_MINBPC(enc, ptr)) { \
      *nextTokPtr = ptr; \
      return XML_TOK_INVALID; \
    } \
  case BT_NMSTRT: \
  case BT_HEX: \
  case BT_DIGIT: \
  case BT_NAME: \
  case BT_MINUS: \
    ptr += MINBPC(enc); \
    break; \
  CHECK_NAME_CASE(2, enc, ptr, end, nextTokPtr) \
  CHECK_NAME_CASE(3, enc, ptr, end, nextTokPtr) \
  CHECK_NAME_CASE(4, enc, ptr, end, nextTokPtr)

#define CHECK_NMSTRT_CASE(n, enc, ptr, end, nextTokPtr) \
   case BT_LEAD ## n: \
     if (end - ptr < n) \
       return XML_TOK_PARTIAL_CHAR; \
     if (!IS_NMSTRT_CHAR(enc, ptr, n)) { \
       *nextTokPtr = ptr; \
       return XML_TOK_INVALID; \
     } \
     ptr += n; \
     break;

#define CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) \
  case BT_NONASCII: \
    if (!IS_NMSTRT_CHAR_MINBPC(enc, ptr)) { \
      *nextTokPtr = ptr; \
      return XML_TOK_INVALID; \
    } \
  case BT_NMSTRT: \
  case BT_HEX: \
    ptr += MINBPC(enc); \
    break; \
  CHECK_NMSTRT_CASE(2, enc, ptr, end, nextTokPtr) \
  CHECK_NMSTRT_CASE(3, enc, ptr, end, nextTokPtr) \
  CHECK_NMSTRT_CASE(4, enc, ptr, end, nextTokPtr)

#ifndef PREFIX
#define PREFIX(ident) ident
#endif

/* ptr points to character following "<!-" */

static
int PREFIX(scanComment)(const ENCODING *enc, const char *ptr, const char *end,
			const char **nextTokPtr)
{
  if (ptr != end) {
    if (!CHAR_MATCHES(enc, ptr, ASCII_MINUS)) {
      *nextTokPtr = ptr;
      return XML_TOK_INVALID;
    }
    ptr += MINBPC(enc);
    while (ptr != end) {
      switch (BYTE_TYPE(enc, ptr)) {
      INVALID_CASES(ptr, nextTokPtr)
      case BT_MINUS:
	if ((ptr += MINBPC(enc)) == end)
	  return XML_TOK_PARTIAL;
	if (CHAR_MATCHES(enc, ptr, ASCII_MINUS)) {
	  if ((ptr += MINBPC(enc)) == end)
	    return XML_TOK_PARTIAL;
	  if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
	    *nextTokPtr = ptr;
	    return XML_TOK_INVALID;
	  }
	  *nextTokPtr = ptr + MINBPC(enc);
	  return XML_TOK_COMMENT;
	}
	break;
      default:
	ptr += MINBPC(enc);
	break;
      }
    }
  }
  return XML_TOK_PARTIAL;
}

/* ptr points to character following "<!" */

static
int PREFIX(scanDecl)(const ENCODING *enc, const char *ptr, const char *end,
		     const char **nextTokPtr)
{
  if (ptr == end)
    return XML_TOK_PARTIAL;
  switch (BYTE_TYPE(enc, ptr)) {
  case BT_MINUS:
    return PREFIX(scanComment)(enc, ptr + MINBPC(enc), end, nextTokPtr);
  case BT_LSQB:
    *nextTokPtr = ptr + MINBPC(enc);
    return XML_TOK_COND_SECT_OPEN;
  case BT_NMSTRT:
  case BT_HEX:
    ptr += MINBPC(enc);
    break;
  default:
    *nextTokPtr = ptr;
    return XML_TOK_INVALID;
  }
  while (ptr != end) {
    switch (BYTE_TYPE(enc, ptr)) {
    case BT_PERCNT:
      if (ptr + MINBPC(enc) == end)
	return XML_TOK_PARTIAL;
      /* don't allow <!ENTITY% foo "whatever"> */
      switch (BYTE_TYPE(enc, ptr + MINBPC(enc))) {
      case BT_S: case BT_CR: case BT_LF: case BT_PERCNT:
	*nextTokPtr = ptr;
	return XML_TOK_INVALID;
      }
      /* fall through */
    case BT_S: case BT_CR: case BT_LF:
      *nextTokPtr = ptr;
      return XML_TOK_DECL_OPEN;
    case BT_NMSTRT:
    case BT_HEX:
      ptr += MINBPC(enc);
      break;
    default:
      *nextTokPtr = ptr;
      return XML_TOK_INVALID;
    }
  }
  return XML_TOK_PARTIAL;
}

static
int PREFIX(checkPiTarget)(const ENCODING *enc, const char *ptr, const char *end, int *tokPtr)
{
  int upper = 0;
  *tokPtr = XML_TOK_PI;
  if (end - ptr != MINBPC(enc)*3)
    return 1;
  switch (BYTE_TO_ASCII(enc, ptr)) {
  case ASCII_x:
    break;
  case ASCII_X:
    upper = 1;
    break;
  default:
    return 1;
  }
  ptr += MINBPC(enc);
  switch (BYTE_TO_ASCII(enc, ptr)) {
  case ASCII_m:
    break;
  case ASCII_M:
    upper = 1;
    break;
  default:
    return 1;
  }
  ptr += MINBPC(enc);
  switch (BYTE_TO_ASCII(enc, ptr)) {
  case ASCII_l:
    break;
  case ASCII_L:
    upper = 1;
    break;
  default:
    return 1;
  }
  if (upper)
    return 0;
  *tokPtr = XML_TOK_XML_DECL;
  return 1;
}

/* ptr points to character following "<?" */

static
int PREFIX(scanPi)(const ENCODING *enc, const char *ptr, const char *end,
		   const char **nextTokPtr)
{
  int tok;
  const char *target = ptr;
  if (ptr == end)
    return XML_TOK_PARTIAL;
  switch (BYTE_TYPE(enc, ptr)) {
  CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
  default:
    *nextTokPtr = ptr;
    return XML_TOK_INVALID;
  }
  while (ptr != end) {
    switch (BYTE_TYPE(enc, ptr)) {
    CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
    case BT_S: case BT_CR: case BT_LF:
      if (!PREFIX(checkPiTarget)(enc, target, ptr, &tok)) {
	*nextTokPtr = ptr;
	return XML_TOK_INVALID;
      }
      ptr += MINBPC(enc);
      while (ptr != end) {
        switch (BYTE_TYPE(enc, ptr)) {
        INVALID_CASES(ptr, nextTokPtr)
	case BT_QUEST:
	  ptr += MINBPC(enc);
	  if (ptr == end)
	    return XML_TOK_PARTIAL;
	  if (CHAR_MATCHES(enc, ptr, ASCII_GT)) {
	    *nextTokPtr = ptr + MINBPC(enc);
	    return tok;
	  }
	  break;
	default:
	  ptr += MINBPC(enc);
	  break;
	}
      }
      return XML_TOK_PARTIAL;
    case BT_QUEST:
      if (!PREFIX(checkPiTarget)(enc, target, ptr, &tok)) {
	*nextTokPtr = ptr;
	return XML_TOK_INVALID;
      }
      ptr += MINBPC(enc);
      if (ptr == end)
	return XML_TOK_PARTIAL;
      if (CHAR_MATCHES(enc, ptr, ASCII_GT)) {
	*nextTokPtr = ptr + MINBPC(enc);
	return tok;
      }
      /* fall through */
    default:
      *nextTokPtr = ptr;
      return XML_TOK_INVALID;
    }
  }
  return XML_TOK_PARTIAL;
}


static
int PREFIX(scanCdataSection)(const ENCODING *enc, const char *ptr, const char *end,
			     const char **nextTokPtr)
{
  static const char CDATA_LSQB[] = { ASCII_C, ASCII_D, ASCII_A, ASCII_T, ASCII_A, ASCII_LSQB };
  int i;
  /* CDATA[ */
  if (end - ptr < 6 * MINBPC(enc))
    return XML_TOK_PARTIAL;
  for (i = 0; i < 6; i++, ptr += MINBPC(enc)) {
    if (!CHAR_MATCHES(enc, ptr, CDATA_LSQB[i])) {
      *nextTokPtr = ptr;
      return XML_TOK_INVALID;
    }
  }
  *nextTokPtr = ptr;
  return XML_TOK_CDATA_SECT_OPEN;
}

static
int PREFIX(cdataSectionTok)(const ENCODING *enc, const char *ptr, const char *end,
			    const char **nextTokPtr)
{
  if (ptr == end)
    return XML_TOK_NONE;
  if (MINBPC(enc) > 1) {
    size_t n = end - ptr;
    if (n & (MINBPC(enc) - 1)) {
      n &= ~(MINBPC(enc) - 1);
      if (n == 0)
	return XML_TOK_PARTIAL;
      end = ptr + n;
    }
  }
  switch (BYTE_TYPE(enc, ptr)) {
  case BT_RSQB:
    ptr += MINBPC(enc);
    if (ptr == end)
      return XML_TOK_PARTIAL;
    if (!CHAR_MATCHES(enc, ptr, ASCII_RSQB))
      break;
    ptr += MINBPC(enc);
    if (ptr == end)
      return XML_TOK_PARTIAL;
    if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
      ptr -= MINBPC(enc);
      break;
    }
    *nextTokPtr = ptr + MINBPC(enc);
    return XML_TOK_CDATA_SECT_CLOSE;
  case BT_CR:
    ptr += MINBPC(enc);
    if (ptr == end)
      return XML_TOK_PARTIAL;
    if (BYTE_TYPE(enc, ptr) == BT_LF)
      ptr += MINBPC(enc);
    *nextTokPtr = ptr;
    return XML_TOK_DATA_NEWLINE;
  case BT_LF:
    *nextTokPtr = ptr + MINBPC(enc);
    return XML_TOK_DATA_NEWLINE;
  INVALID_CASES(ptr, nextTokPtr)
  default:
    ptr += MINBPC(enc);
    break;
  }
  while (ptr != end) {
    switch (BYTE_TYPE(enc, ptr)) {
#define LEAD_CASE(n) \
    case BT_LEAD ## n: \
      if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \
	*nextTokPtr = ptr; \
	return XML_TOK_DATA_CHARS; \
      } \
      ptr += n; \
      break;
    LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
#undef LEAD_CASE
    case BT_NONXML:
    case BT_MALFORM:
    case BT_TRAIL:
    case BT_CR:
    case BT_LF:
    case BT_RSQB:
      *nextTokPtr = ptr;
      return XML_TOK_DATA_CHARS;
    default:
      ptr += MINBPC(enc);
      break;
    }
  }
  *nextTokPtr = ptr;
  return XML_TOK_DATA_CHARS;
}

/* ptr points to character following "</" */

static
int PREFIX(scanEndTag)(const ENCODING *enc, const char *ptr, const char *end,
		       const char **nextTokPtr)
{
  if (ptr == end)
    return XML_TOK_PARTIAL;
  switch (BYTE_TYPE(enc, ptr)) {
  CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
  default:
    *nextTokPtr = ptr;
    return XML_TOK_INVALID;
  }
  while (ptr != end) {
    switch (BYTE_TYPE(enc, ptr)) {
    CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
    case BT_S: case BT_CR: case BT_LF:
      for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) {
	switch (BYTE_TYPE(enc, ptr)) {
	case BT_S: case BT_CR: case BT_LF:
	  break;
	case BT_GT:
	  *nextTokPtr = ptr + MINBPC(enc);
          return XML_TOK_END_TAG;
	default:
	  *nextTokPtr = ptr;
	  return XML_TOK_INVALID;
	}
      }
      return XML_TOK_PARTIAL;
#ifdef XML_NS
    case BT_COLON:
      /* no need to check qname syntax here, since end-tag must match exactly */
      ptr += MINBPC(enc);
      break;
#endif
    case BT_GT:
      *nextTokPtr = ptr + MINBPC(enc);
      return XML_TOK_END_TAG;
    default:
      *nextTokPtr = ptr;
      return XML_TOK_INVALID;
    }
  }
  return XML_TOK_PARTIAL;
}

/* ptr points to character following "&#X" */

static
int PREFIX(scanHexCharRef)(const ENCODING *enc, const char *ptr, const char *end,
			   const char **nextTokPtr)
{
  if (ptr != end) {
    switch (BYTE_TYPE(enc, ptr)) {
    case BT_DIGIT:
    case BT_HEX:
      break;
    default:
      *nextTokPtr = ptr;
      return XML_TOK_INVALID;
    }
    for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) {
      switch (BYTE_TYPE(enc, ptr)) {
      case BT_DIGIT:
      case BT_HEX:
	break;
      case BT_SEMI:
	*nextTokPtr = ptr + MINBPC(enc);
	return XML_TOK_CHAR_REF;
      default:
	*nextTokPtr = ptr;
	return XML_TOK_INVALID;
      }
    }
  }
  return XML_TOK_PARTIAL;
}

/* ptr points to character following "&#" */

static
int PREFIX(scanCharRef)(const ENCODING *enc, const char *ptr, const char *end,
			const char **nextTokPtr)
{
  if (ptr != end) {
    if (CHAR_MATCHES(enc, ptr, ASCII_x))
      return PREFIX(scanHexCharRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
    switch (BYTE_TYPE(enc, ptr)) {
    case BT_DIGIT:
      break;
    default:
      *nextTokPtr = ptr;
      return XML_TOK_INVALID;
    }
    for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) {
      switch (BYTE_TYPE(enc, ptr)) {
      case BT_DIGIT:
	break;
      case BT_SEMI:
	*nextTokPtr = ptr + MINBPC(enc);
	return XML_TOK_CHAR_REF;
      default:
	*nextTokPtr = ptr;
	return XML_TOK_INVALID;
      }
    }
  }
  return XML_TOK_PARTIAL;
}

/* ptr points to character following "&" */

static
int PREFIX(scanRef)(const ENCODING *enc, const char *ptr, const char *end,
		    const char **nextTokPtr)
{
  if (ptr == end)
    return XML_TOK_PARTIAL;
  switch (BYTE_TYPE(enc, ptr)) {
  CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
  case BT_NUM:
    return PREFIX(scanCharRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
  default:
    *nextTokPtr = ptr;
    return XML_TOK_INVALID;
  }
  while (ptr != end) {
    switch (BYTE_TYPE(enc, ptr)) {
    CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
    case BT_SEMI:
      *nextTokPtr = ptr + MINBPC(enc);
      return XML_TOK_ENTITY_REF;
    default:
      *nextTokPtr = ptr;
      return XML_TOK_INVALID;
    }
  }
  return XML_TOK_PARTIAL;
}

/* ptr points to character following first character of attribute name */

static
int PREFIX(scanAtts)(const ENCODING *enc, const char *ptr, const char *end,
		     const char **nextTokPtr)
{
#ifdef XML_NS
  int hadColon = 0;
#endif
  while (ptr != end) {
    switch (BYTE_TYPE(enc, ptr)) {
    CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
#ifdef XML_NS
    case BT_COLON:
      if (hadColon) {
	*nextTokPtr = ptr;
	return XML_TOK_INVALID;
      }
      hadColon = 1;
      ptr += MINBPC(enc);
      if (ptr == end)
	return XML_TOK_PARTIAL;
      switch (BYTE_TYPE(enc, ptr)) {
      CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
      default:
	*nextTokPtr = ptr;
	return XML_TOK_INVALID;
      }
      break;
#endif
    case BT_S: case BT_CR: case BT_LF:
      for (;;) {
	int t;

	ptr += MINBPC(enc);
	if (ptr == end)
	  return XML_TOK_PARTIAL;
	t = BYTE_TYPE(enc, ptr);
	if (t == BT_EQUALS)
	  break;
	switch (t) {
	case BT_S:
	case BT_LF:
	case BT_CR:
	  break;
	default:
	  *nextTokPtr = ptr;
	  return XML_TOK_INVALID;
	}
      }
    /* fall through */
    case BT_EQUALS:
      {
	int open;
#ifdef XML_NS
	hadColon = 0;
#endif
	for (;;) {
	  
	  ptr += MINBPC(enc);
	  if (ptr == end)
	    return XML_TOK_PARTIAL;
	  open = BYTE_TYPE(enc, ptr);
	  if (open == BT_QUOT || open == BT_APOS)
	    break;
	  switch (open) {
	  case BT_S:
	  case BT_LF:
	  case BT_CR:
	    break;
	  default:
	    *nextTokPtr = ptr;