xmlrole.c

一个很好用的解析

    return XML_ROLE_NONE;
  case XML_TOK_LITERAL:
    state->handler = notation4;
    return XML_ROLE_NOTATION_PUBLIC_ID;
  }
  return common(state, tok);
}

static
int notation3(PROLOG_STATE *state,
	      int tok,
	      const char *ptr ATTR_UNUSED,
	      const char *end ATTR_UNUSED,
	      const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_LITERAL:
    state->handler = declClose;
    return XML_ROLE_NOTATION_SYSTEM_ID;
  }
  return common(state, tok);
}

static
int notation4(PROLOG_STATE *state,
	      int tok,
	      const char *ptr ATTR_UNUSED,
	      const char *end ATTR_UNUSED,
	      const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_LITERAL:
    state->handler = declClose;
    return XML_ROLE_NOTATION_SYSTEM_ID;
  case XML_TOK_DECL_CLOSE:
    setTopLevel(state);
    return XML_ROLE_NOTATION_NO_SYSTEM_ID;
  }
  return common(state, tok);
}

static
int attlist0(PROLOG_STATE *state,
	     int tok,
	     const char *ptr ATTR_UNUSED,
	     const char *end ATTR_UNUSED,
	     const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_NAME:
  case XML_TOK_PREFIXED_NAME:
    state->handler = attlist1;
    return XML_ROLE_ATTLIST_ELEMENT_NAME;
  }
  return common(state, tok);
}

static
int attlist1(PROLOG_STATE *state,
	     int tok,
	     const char *ptr ATTR_UNUSED,
	     const char *end ATTR_UNUSED,
	     const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_DECL_CLOSE:
    setTopLevel(state);
    return XML_ROLE_NONE;
  case XML_TOK_NAME:
  case XML_TOK_PREFIXED_NAME:
    state->handler = attlist2;
    return XML_ROLE_ATTRIBUTE_NAME;
  }
  return common(state, tok);
}

static
int attlist2(PROLOG_STATE *state,
	     int tok,
	     const char *ptr,
	     const char *end,
	     const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_NAME:
    {
      static const char *types[] = {
	KW_CDATA,
        KW_ID,
        KW_IDREF,
        KW_IDREFS,
        KW_ENTITY,
        KW_ENTITIES,
        KW_NMTOKEN,
        KW_NMTOKENS,
      };
      int i;
      for (i = 0; i < (int)(sizeof(types)/sizeof(types[0])); i++)
	if (XmlNameMatchesAscii(enc, ptr, end, types[i])) {
	  state->handler = attlist8;
	  return XML_ROLE_ATTRIBUTE_TYPE_CDATA + i;
	}
    }
    if (XmlNameMatchesAscii(enc, ptr, end, KW_NOTATION)) {
      state->handler = attlist5;
      return XML_ROLE_NONE;
    }
    break;
  case XML_TOK_OPEN_PAREN:
    state->handler = attlist3;
    return XML_ROLE_NONE;
  }
  return common(state, tok);
}

static
int attlist3(PROLOG_STATE *state,
	     int tok,
	     const char *ptr ATTR_UNUSED,
	     const char *end ATTR_UNUSED,
	     const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_NMTOKEN:
  case XML_TOK_NAME:
  case XML_TOK_PREFIXED_NAME:
    state->handler = attlist4;
    return XML_ROLE_ATTRIBUTE_ENUM_VALUE;
  }
  return common(state, tok);
}

static
int attlist4(PROLOG_STATE *state,
	     int tok,
	     const char *ptr ATTR_UNUSED,
	     const char *end ATTR_UNUSED,
	     const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_CLOSE_PAREN:
    state->handler = attlist8;
    return XML_ROLE_NONE;
  case XML_TOK_OR:
    state->handler = attlist3;
    return XML_ROLE_NONE;
  }
  return common(state, tok);
}

static
int attlist5(PROLOG_STATE *state,
	     int tok,
	     const char *ptr ATTR_UNUSED,
	     const char *end ATTR_UNUSED,
	     const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_OPEN_PAREN:
    state->handler = attlist6;
    return XML_ROLE_NONE;
  }
  return common(state, tok);
}


static
int attlist6(PROLOG_STATE *state,
	     int tok,
	     const char *ptr ATTR_UNUSED,
	     const char *end ATTR_UNUSED,
	     const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_NAME:
    state->handler = attlist7;
    return XML_ROLE_ATTRIBUTE_NOTATION_VALUE;
  }
  return common(state, tok);
}

static
int attlist7(PROLOG_STATE *state,
	     int tok,
	     const char *ptr ATTR_UNUSED,
	     const char *end ATTR_UNUSED,
	     const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_CLOSE_PAREN:
    state->handler = attlist8;
    return XML_ROLE_NONE;
  case XML_TOK_OR:
    state->handler = attlist6;
    return XML_ROLE_NONE;
  }
  return common(state, tok);
}

/* default value */
static
int attlist8(PROLOG_STATE *state,
	     int tok,
	     const char *ptr,
	     const char *end,
	     const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_POUND_NAME:
    if (XmlNameMatchesAscii(enc,
			    ptr + MIN_BYTES_PER_CHAR(enc),
			    end,
			    KW_IMPLIED)) {
      state->handler = attlist1;
      return XML_ROLE_IMPLIED_ATTRIBUTE_VALUE;
    }
    if (XmlNameMatchesAscii(enc,
			    ptr + MIN_BYTES_PER_CHAR(enc),
			    end,
			    KW_REQUIRED)) {
      state->handler = attlist1;
      return XML_ROLE_REQUIRED_ATTRIBUTE_VALUE;
    }
    if (XmlNameMatchesAscii(enc,
			    ptr + MIN_BYTES_PER_CHAR(enc),
			    end,
			    KW_FIXED)) {
      state->handler = attlist9;
      return XML_ROLE_NONE;
    }
    break;
  case XML_TOK_LITERAL:
    state->handler = attlist1;
    return XML_ROLE_DEFAULT_ATTRIBUTE_VALUE;
  }
  return common(state, tok);
}

static
int attlist9(PROLOG_STATE *state,
	     int tok,
	     const char *ptr ATTR_UNUSED,
	     const char *end ATTR_UNUSED,
	     const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_LITERAL:
    state->handler = attlist1;
    return XML_ROLE_FIXED_ATTRIBUTE_VALUE;
  }
  return common(state, tok);
}

static
int element0(PROLOG_STATE *state,
	     int tok,
	     const char *ptr ATTR_UNUSED,
	     const char *end ATTR_UNUSED,
	     const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_NAME:
  case XML_TOK_PREFIXED_NAME:
    state->handler = element1;
    return XML_ROLE_ELEMENT_NAME;
  }
  return common(state, tok);
}

static
int element1(PROLOG_STATE *state,
	     int tok,
	     const char *ptr,
	     const char *end,
	     const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_NAME:
    if (XmlNameMatchesAscii(enc, ptr, end, KW_EMPTY)) {
      state->handler = declClose;
      return XML_ROLE_CONTENT_EMPTY;
    }
    if (XmlNameMatchesAscii(enc, ptr, end, KW_ANY)) {
      state->handler = declClose;
      return XML_ROLE_CONTENT_ANY;
    }
    break;
  case XML_TOK_OPEN_PAREN:
    state->handler = element2;
    state->level = 1;
    return XML_ROLE_GROUP_OPEN;
  }
  return common(state, tok);
}

static
int element2(PROLOG_STATE *state,
	     int tok,
	     const char *ptr,
	     const char *end,
	     const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_POUND_NAME:
    if (XmlNameMatchesAscii(enc,
			    ptr + MIN_BYTES_PER_CHAR(enc),
			    end,
			    KW_PCDATA)) {
      state->handler = element3;
      return XML_ROLE_CONTENT_PCDATA;
    }
    break;
  case XML_TOK_OPEN_PAREN:
    state->level = 2;
    state->handler = element6;
    return XML_ROLE_GROUP_OPEN;
  case XML_TOK_NAME:
  case XML_TOK_PREFIXED_NAME:
    state->handler = element7;
    return XML_ROLE_CONTENT_ELEMENT;
  case XML_TOK_NAME_QUESTION:
    state->handler = element7;
    return XML_ROLE_CONTENT_ELEMENT_OPT;
  case XML_TOK_NAME_ASTERISK:
    state->handler = element7;
    return XML_ROLE_CONTENT_ELEMENT_REP;
  case XML_TOK_NAME_PLUS:
    state->handler = element7;
    return XML_ROLE_CONTENT_ELEMENT_PLUS;
  }
  return common(state, tok);
}

static
int element3(PROLOG_STATE *state,
	     int tok,
	     const char *ptr ATTR_UNUSED,
	     const char *end ATTR_UNUSED,
	     const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_CLOSE_PAREN:
  case XML_TOK_CLOSE_PAREN_ASTERISK:
    state->handler = declClose;
    return XML_ROLE_GROUP_CLOSE_REP;
  case XML_TOK_OR:
    state->handler = element4;
    return XML_ROLE_NONE;
  }
  return common(state, tok);
}

static
int element4(PROLOG_STATE *state,
	     int tok,
	     const char *ptr ATTR_UNUSED,
	     const char *end ATTR_UNUSED,
	     const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_NAME:
  case XML_TOK_PREFIXED_NAME:
    state->handler = element5;
    return XML_ROLE_CONTENT_ELEMENT;
  }
  return common(state, tok);
}

static
int element5(PROLOG_STATE *state,
	     int tok,
	     const char *ptr ATTR_UNUSED,
	     const char *end ATTR_UNUSED,
	     const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_CLOSE_PAREN_ASTERISK:
    state->handler = declClose;
    return XML_ROLE_GROUP_CLOSE_REP;
  case XML_TOK_OR:
    state->handler = element4;
    return XML_ROLE_NONE;
  }
  return common(state, tok);
}

static
int element6(PROLOG_STATE *state,
	     int tok,
	     const char *ptr ATTR_UNUSED,
	     const char *end ATTR_UNUSED,
	     const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_OPEN_PAREN:
    state->level += 1;
    return XML_ROLE_GROUP_OPEN;
  case XML_TOK_NAME:
  case XML_TOK_PREFIXED_NAME:
    state->handler = element7;
    return XML_ROLE_CONTENT_ELEMENT;
  case XML_TOK_NAME_QUESTION:
    state->handler = element7;
    return XML_ROLE_CONTENT_ELEMENT_OPT;
  case XML_TOK_NAME_ASTERISK:
    state->handler = element7;
    return XML_ROLE_CONTENT_ELEMENT_REP;
  case XML_TOK_NAME_PLUS:
    state->handler = element7;
    return XML_ROLE_CONTENT_ELEMENT_PLUS;
  }
  return common(state, tok);
}

static
int element7(PROLOG_STATE *state,
	     int tok,
	     const char *ptr ATTR_UNUSED,
	     const char *end ATTR_UNUSED,
	     const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_CLOSE_PAREN:
    state->level -= 1;
    if (state->level == 0)
      state->handler = declClose;
    return XML_ROLE_GROUP_CLOSE;
  case XML_TOK_CLOSE_PAREN_ASTERISK:
    state->level -= 1;
    if (state->level == 0)
      state->handler = declClose;
    return XML_ROLE_GROUP_CLOSE_REP;
  case XML_TOK_CLOSE_PAREN_QUESTION:
    state->level -= 1;
    if (state->level == 0)
      state->handler = declClose;
    return XML_ROLE_GROUP_CLOSE_OPT;
  case XML_TOK_CLOSE_PAREN_PLUS:
    state->level -= 1;
    if (state->level == 0)
      state->handler = declClose;
    return XML_ROLE_GROUP_CLOSE_PLUS;
  case XML_TOK_COMMA:
    state->handler = element6;
    return XML_ROLE_GROUP_SEQUENCE;
  case XML_TOK_OR:
    state->handler = element6;
    return XML_ROLE_GROUP_CHOICE;
  }
  return common(state, tok);
}

#ifdef XML_DTD

static
int condSect0(PROLOG_STATE *state,
	      int tok,
	      const char *ptr,
	      const char *end,
	      const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_NAME:
    if (XmlNameMatchesAscii(enc, ptr, end, KW_INCLUDE)) {
      state->handler = condSect1;
      return XML_ROLE_NONE;
    }
    if (XmlNameMatchesAscii(enc, ptr, end, KW_IGNORE)) {
      state->handler = condSect2;
      return XML_ROLE_NONE;
    }
    break;
  }
  return common(state, tok);
}

static
int condSect1(PROLOG_STATE *state,
	      int tok,
	      const char *ptr,
	      const char *end,
	      const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_OPEN_BRACKET:
    state->handler = externalSubset1;
    state->includeLevel += 1;
    return XML_ROLE_NONE;
  }
  return common(state, tok);
}

static
int condSect2(PROLOG_STATE *state,
	      int tok,
	      const char *ptr,
	      const char *end,
	      const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_OPEN_BRACKET:
    state->handler = externalSubset1;
    return XML_ROLE_IGNORE_SECT;
  }
  return common(state, tok);
}

#endif /* XML_DTD */

static
int declClose(PROLOG_STATE *state,
	      int tok,
	      const char *ptr ATTR_UNUSED,
	      const char *end ATTR_UNUSED,
	      const ENCODING *enc ATTR_UNUSED)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_DECL_CLOSE:
    setTopLevel(state);
    return XML_ROLE_NONE;
  }
  return common(state, tok);
}

#if 0

static
int ignore(PROLOG_STATE *state,
	   int tok,
	   const char *ptr,
	   const char *end,
	   const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_DECL_CLOSE:
    state->handler = internalSubset;
    return 0;
  default:
    return XML_ROLE_NONE;
  }
  return common(state, tok);
}
#endif

static
int error(PROLOG_STATE *state ATTR_UNUSED,
	  int tok ATTR_UNUSED,
	  const char *ptr ATTR_UNUSED,
	  const char *end ATTR_UNUSED,
	  const ENCODING *enc ATTR_UNUSED)
{
  return XML_ROLE_NONE;
}

static
int common(PROLOG_STATE *state, int tok ATTR_UNUSED)
{
#ifdef XML_DTD
  if (!state->documentEntity && tok == XML_TOK_PARAM_ENTITY_REF)
    return XML_ROLE_INNER_PARAM_ENTITY_REF;
#endif
  state->handler = error;
  return XML_ROLE_ERROR;
}

void xmlrpc_XmlPrologStateInit(PROLOG_STATE *state)
{
  state->handler = prolog0;
#ifdef XML_DTD
  state->documentEntity = 1;
  state->includeLevel = 0;
#endif /* XML_DTD */
}

#ifdef XML_DTD

void XmlPrologStateInitExternalEntity(PROLOG_STATE *state)
{
  state->handler = externalSubset0;
  state->documentEntity = 0;
  state->includeLevel = 0;
}

#endif /* XML_DTD */