xmltok_impl.c

很牛的GUI源码wxWidgets-2.8.0.zip 可在多种平台下运行.

        /* in attribute value */
        for (;;) {
          int t;
          if (ptr == end)
            return XML_TOK_PARTIAL;
          t = BYTE_TYPE(enc, ptr);
          if (t == open)
            break;
          switch (t) {
          INVALID_CASES(ptr, nextTokPtr)
          case BT_AMP:
            {
              int tok = PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, &ptr);
              if (tok <= 0) {
                if (tok == XML_TOK_INVALID)
                  *nextTokPtr = ptr;
                return tok;
              }
              break;
            }
          case BT_LT:
            *nextTokPtr = ptr;
            return XML_TOK_INVALID;
          default:
            ptr += MINBPC(enc);
            break;
          }
        }
        ptr += MINBPC(enc);
        if (ptr == end)
          return XML_TOK_PARTIAL;
        switch (BYTE_TYPE(enc, ptr)) {
        case BT_S:
        case BT_CR:
        case BT_LF:
          break;
        case BT_SOL:
          goto sol;
        case BT_GT:
          goto gt;
        default:
          *nextTokPtr = ptr;
          return XML_TOK_INVALID;
        }
        /* ptr points to closing quote */
        for (;;) {
          ptr += MINBPC(enc);
          if (ptr == end)
            return XML_TOK_PARTIAL;
          switch (BYTE_TYPE(enc, ptr)) {
          CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
          case BT_S: case BT_CR: case BT_LF:
            continue;
          case BT_GT:
          gt:
            *nextTokPtr = ptr + MINBPC(enc);
            return XML_TOK_START_TAG_WITH_ATTS;
          case BT_SOL:
          sol:
            ptr += MINBPC(enc);
            if (ptr == 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_EMPTY_ELEMENT_WITH_ATTS;
          default:
            *nextTokPtr = ptr;
            return XML_TOK_INVALID;
          }
          break;
        }
        break;
      }
    default:
      *nextTokPtr = ptr;
      return XML_TOK_INVALID;
    }
  }
  return XML_TOK_PARTIAL;
}

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

static int PTRCALL
PREFIX(scanLt)(const ENCODING *enc, const char *ptr, const char *end,
               const char **nextTokPtr)
{
#ifdef XML_NS
  int hadColon;
#endif
  if (ptr == end)
    return XML_TOK_PARTIAL;
  switch (BYTE_TYPE(enc, ptr)) {
  CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
  case BT_EXCL:
    if ((ptr += MINBPC(enc)) == 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:
      return PREFIX(scanCdataSection)(enc, ptr + MINBPC(enc),
                                      end, nextTokPtr);
    }
    *nextTokPtr = ptr;
    return XML_TOK_INVALID;
  case BT_QUEST:
    return PREFIX(scanPi)(enc, ptr + MINBPC(enc), end, nextTokPtr);
  case BT_SOL:
    return PREFIX(scanEndTag)(enc, ptr + MINBPC(enc), end, nextTokPtr);
  default:
    *nextTokPtr = ptr;
    return XML_TOK_INVALID;
  }
#ifdef XML_NS
  hadColon = 0;
#endif
  /* we have a start-tag */
  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:
      {
        ptr += MINBPC(enc);
        while (ptr != end) {
          switch (BYTE_TYPE(enc, ptr)) {
          CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
          case BT_GT:
            goto gt;
          case BT_SOL:
            goto sol;
          case BT_S: case BT_CR: case BT_LF:
            ptr += MINBPC(enc);
            continue;
          default:
            *nextTokPtr = ptr;
            return XML_TOK_INVALID;
          }
          return PREFIX(scanAtts)(enc, ptr, end, nextTokPtr);
        }
        return XML_TOK_PARTIAL;
      }
    case BT_GT:
    gt:
      *nextTokPtr = ptr + MINBPC(enc);
      return XML_TOK_START_TAG_NO_ATTS;
    case BT_SOL:
    sol:
      ptr += MINBPC(enc);
      if (ptr == 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_EMPTY_ELEMENT_NO_ATTS;
    default:
      *nextTokPtr = ptr;
      return XML_TOK_INVALID;
    }
  }
  return XML_TOK_PARTIAL;
}

static int PTRCALL
PREFIX(contentTok)(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_LT:
    return PREFIX(scanLt)(enc, ptr + MINBPC(enc), end, nextTokPtr);
  case BT_AMP:
    return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
  case BT_CR:
    ptr += MINBPC(enc);
    if (ptr == end)
      return XML_TOK_TRAILING_CR;
    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;
  case BT_RSQB:
    ptr += MINBPC(enc);
    if (ptr == end)
      return XML_TOK_TRAILING_RSQB;
    if (!CHAR_MATCHES(enc, ptr, ASCII_RSQB))
      break;
    ptr += MINBPC(enc);
    if (ptr == end)
      return XML_TOK_TRAILING_RSQB;
    if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
      ptr -= MINBPC(enc);
      break;
    }
    *nextTokPtr = ptr;
    return XML_TOK_INVALID;
  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_RSQB:
      if (ptr + MINBPC(enc) != end) {
         if (!CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_RSQB)) {
           ptr += MINBPC(enc);
           break;
         }
         if (ptr + 2*MINBPC(enc) != end) {
           if (!CHAR_MATCHES(enc, ptr + 2*MINBPC(enc), ASCII_GT)) {
             ptr += MINBPC(enc);
             break;
           }
           *nextTokPtr = ptr + 2*MINBPC(enc);
           return XML_TOK_INVALID;
         }
      }
      /* fall through */
    case BT_AMP:
    case BT_LT:
    case BT_NONXML:
    case BT_MALFORM:
    case BT_TRAIL:
    case BT_CR:
    case BT_LF:
      *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 PTRCALL
PREFIX(scanPercent)(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_S: case BT_LF: case BT_CR: case BT_PERCNT:
    *nextTokPtr = ptr;
    return XML_TOK_PERCENT;
  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_PARAM_ENTITY_REF;
    default:
      *nextTokPtr = ptr;
      return XML_TOK_INVALID;
    }
  }
  return XML_TOK_PARTIAL;
}

static int PTRCALL
PREFIX(scanPoundName)(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_CR: case BT_LF: case BT_S:
    case BT_RPAR: case BT_GT: case BT_PERCNT: case BT_VERBAR:
      *nextTokPtr = ptr;
      return XML_TOK_POUND_NAME;
    default:
      *nextTokPtr = ptr;
      return XML_TOK_INVALID;
    }
  }
  return -XML_TOK_POUND_NAME;
}

static int PTRCALL
PREFIX(scanLit)(int open, const ENCODING *enc,
                const char *ptr, const char *end,
                const char **nextTokPtr)
{
  while (ptr != end) {
    int t = BYTE_TYPE(enc, ptr);
    switch (t) {
    INVALID_CASES(ptr, nextTokPtr)
    case BT_QUOT:
    case BT_APOS:
      ptr += MINBPC(enc);
      if (t != open)
        break;
      if (ptr == end)
        return -XML_TOK_LITERAL;
      *nextTokPtr = ptr;
      switch (BYTE_TYPE(enc, ptr)) {
      case BT_S: case BT_CR: case BT_LF:
      case BT_GT: case BT_PERCNT: case BT_LSQB:
        return XML_TOK_LITERAL;
      default:
        return XML_TOK_INVALID;
      }
    default:
      ptr += MINBPC(enc);
      break;
    }
  }
  return XML_TOK_PARTIAL;
}

static int PTRCALL
PREFIX(prologTok)(const ENCODING *enc, const char *ptr, const char *end,
                  const char **nextTokPtr)
{
  int tok;
  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_QUOT:
    return PREFIX(scanLit)(BT_QUOT, enc, ptr + MINBPC(enc), end, nextTokPtr);
  case BT_APOS:
    return PREFIX(scanLit)(BT_APOS, enc, ptr + MINBPC(enc), end, nextTokPtr);
  case BT_LT:
    {
      ptr += MINBPC(enc);
      if (ptr == end)
        return XML_TOK_PARTIAL;
      switch (BYTE_TYPE(enc, ptr)) {
      case BT_EXCL:
        return PREFIX(scanDecl)(enc, ptr + MINBPC(enc), end, nextTokPtr);
      case BT_QUEST:
        return PREFIX(scanPi)(enc, ptr + MINBPC(enc), end, nextTokPtr);
      case BT_NMSTRT:
      case BT_HEX:
      case BT_NONASCII:
      case BT_LEAD2:
      case BT_LEAD3:
      case BT_LEAD4:
        *nextTokPtr = ptr - MINBPC(enc);
        return XML_TOK_INSTANCE_START;
      }
      *nextTokPtr = ptr;
      return XML_TOK_INVALID;
    }
  case BT_CR:
    if (ptr + MINBPC(enc) == end) {
      *nextTokPtr = end;
      /* indicate that this might be part of a CR/LF pair */
      return -XML_TOK_PROLOG_S;
    }
    /* fall through */
  case BT_S: case BT_LF:
    for (;;) {
      ptr += MINBPC(enc);
      if (ptr == end)
        break;
      switch (BYTE_TYPE(enc, ptr)) {
      case BT_S: case BT_LF:
        break;
      case BT_CR:
        /* don't split CR/LF pair */
        if (ptr + MINBPC(enc) != end)
          break;
        /* fall through */
      default:
        *nextTokPtr = ptr;
        return XML_TOK_PROLOG_S;
      }
    }
    *nextTokPtr = ptr;
    return XML_TOK_PROLOG_S;
  case BT_PERCNT:
    return PREFIX(scanPercent)(enc, ptr + MINBPC(enc), end, nextTokPtr);
  case BT_COMMA:
    *nextTokPtr = ptr + MINBPC(enc);
    return XML_TOK_COMMA;
  case BT_LSQB:
    *nextTokPtr = ptr + MINBPC(enc);
    return XML_TOK_OPEN_BRACKET;
  case BT_RSQB:
    ptr += MINBPC(enc);
    if (ptr == end)
      return -XML_TOK_CLOSE_BRACKET;
    if (CHAR_MATCHES(enc, ptr, ASCII_RSQB)) {
      if (ptr + MINBPC(enc) == end)
        return XML_TOK_PARTIAL;
      if (CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_GT)) {
        *nextTokPtr = ptr + 2*MINBPC(enc);
        return XML_TOK_COND_SECT_CLOSE;
      }
    }
    *nextTokPtr = ptr;
    return XML_TOK_CLOSE_BRACKET;
  case BT_LPAR:
    *nextTokPtr = ptr + MINBPC(enc);
    return XML_TOK_OPEN_PAREN;
  case BT_RPAR:
    ptr += MINBPC(enc);
    if (ptr == end)
      return -XML_TOK_CLOSE_PAREN;
    switch (BYTE_TYPE(enc, ptr)) {
    case BT_AST:
      *nextTokPtr = ptr + MINBPC(enc);
      return XML_TOK_CLOSE_PAREN_ASTERISK;
    case BT_QUEST:
      *nextTokPtr = ptr + MINBPC(enc);
      return XML_TOK_CLOSE_PAREN_QUESTION;
    case BT_PLUS:
      *nextTokPtr = ptr + MINBPC(enc);
      return XML_TOK_CLOSE_PAREN_PLUS;
    case BT_CR: case BT_LF: case BT_S:
    case BT_GT: case BT_COMMA: case BT_VERBAR:
    case BT_RPAR:
      *nextTokPtr = ptr;
      return XML_TOK_CLOSE_PAREN;
    }
    *nextTokPtr = ptr;
    return XML_TOK_INVALID;
  case BT_VERBAR:
    *nextTokPtr = ptr + MINBPC(enc);
    return XML_TOK_OR;
  case BT_GT:
    *nextTokPtr = ptr + MINBPC(enc);
    return XML_TOK_DECL_CLOSE;
  case BT_NUM:
    return PREFIX(scanPoundName)(enc, ptr + MINBPC(enc), end, nextTokPtr);
#define LEAD_CASE(n) \
  case BT_LEAD ## n: \
    if (end - ptr < n) \
      return XML_TOK_PARTIAL_CHAR; \
    if (IS_NMSTRT_CHAR(enc, ptr, n)) { \
      ptr += n; \
      tok = XML_TOK_NAME; \
      break; \
    } \
    if (IS_NAME_CHAR(enc, ptr, n)) { \
      ptr += n; \
      tok = XML_TOK_NMTOKEN; \
      break; \
    } \
    *nextTokPtr = ptr; \
    return XML_TOK_INVALID;
    LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
#undef LEAD_CASE
  case BT_NMSTRT:
  case BT_HEX:
    tok = XML_TOK_NAME;
    ptr += MINBPC(enc);
    break;
  case BT_DIGIT:
  case BT_NAME:
  case BT_MINUS:
#ifdef XML_NS
  case BT_COLON:
#endif
    tok = XML_TOK_NMTOKEN;
    ptr += MINBPC(enc);
    break;
  case BT_NONASCII:
    if (IS_NMSTRT_CHAR_MINBPC(enc, ptr)) {
      ptr += MINBPC(enc);
      tok = XML_TOK_NAME;
      break;
    }
    if (IS_NAME_CHAR_MINBPC(enc, ptr)) {
      ptr += MINBPC(enc);
      tok = XML_TOK_NMTOKEN;
      break;
    }
    /* fall through */
  default:
    *nextTokPtr = ptr;
    return XML_TOK_INVALID;
  }
  while (ptr != end) {
    switch (BYTE_TYPE(enc, ptr)) {
    CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
    case BT_GT: case BT_RPAR: case BT_COMMA:
    case BT_VERBAR: case BT_LSQB: case BT_PERCNT:
    case BT_S: case BT_CR: case BT_LF:
      *nextTokPtr = ptr;
      return tok;
#ifdef XML_NS
    case BT_COLON:
      ptr += MINBPC(enc);
      switch (tok) {
      case XML_TOK_NAME:
        if (ptr == end)
          return XML_TOK_PARTIAL;
        tok = XML_TOK_PREFIXED_NAME;
        switch (BYTE_TYPE(enc, ptr)) {
        CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
        default:
          tok = XML_TOK_NMTOKEN;
          break;
        }
        break;
      case XML_TOK_PREFIXED_NAME:
        tok = XML_TOK_NMTOKEN;
        break;
      }
      break;
#endif
    case BT_PLUS:
      if (tok == XML_TOK_NMTOKEN)  {
        *nextTokPtr = ptr;
        return XML_TOK_INVALID;
      }
      *nextTokPtr = ptr + MINBPC(enc);
      return XML_TOK_NAME_PLUS;
    case BT_AST:
      if (tok == XML_TOK_NMTOKEN)  {
        *nextTokPtr = ptr;
        return XML_TOK_INVALID;
      }
      *nextTokPtr = ptr + MINBPC(enc);
      return XML_TOK_NAME_ASTERISK;
    case BT_QUEST:
      if (tok == XML_TOK_NMTOKEN)  {
        *nextTokPtr = ptr;
        return XML_TOK_INVALID;
      }