xmlparse.c

《unix/Linux下的Curses库开发指南》综合示例程序

    case XML_TOK_PARTIAL_CHAR:
      if (nextPtr) {
	*nextPtr = s;
	return XML_ERROR_NONE;
      }
      return XML_ERROR_PARTIAL_CHAR;
    case XML_TOK_PARTIAL:
    case XML_TOK_NONE:
      if (nextPtr) {
	*nextPtr = s;
	return XML_ERROR_NONE;
      }
      return XML_ERROR_UNCLOSED_CDATA_SECTION;
    default:
      abort();
    }
    *eventPP = s = next;
  }
  /* not reached */
}

static enum XML_Error
initializeEncoding(XML_Parser parser)
{
  const char *s;
#ifdef XML_UNICODE
  char encodingBuf[128];
  if (!protocolEncodingName)
    s = 0;
  else {
    int i;
    for (i = 0; protocolEncodingName[i]; i++) {
      if (i == sizeof(encodingBuf) - 1
	  || protocolEncodingName[i] >= 0x80
	  || protocolEncodingName[i] < 0) {
	encodingBuf[0] = '\0';
	break;
      }
      encodingBuf[i] = (char)protocolEncodingName[i];
    }
    encodingBuf[i] = '\0';
    s = encodingBuf;
  }
#else
  s = protocolEncodingName;
#endif
  if ((ns ? XmlInitEncodingNS : XmlInitEncoding)(&initEncoding, &encoding, s))
    return XML_ERROR_NONE;
  return handleUnknownEncoding(parser, protocolEncodingName);
}

static enum XML_Error
processXmlDecl(XML_Parser parser, int isGeneralTextEntity,
	       const char *s, const char *next)
{
  const char *encodingName = 0;
  const ENCODING *newEncoding = 0;
  const char *version;
  int standalone = -1;
  if (!(ns
        ? XmlParseXmlDeclNS
	: XmlParseXmlDecl)(isGeneralTextEntity,
		           encoding,
		           s,
		           next,
		           &eventPtr,
		           &version,
		           &encodingName,
		           &newEncoding,
		           &standalone))
    return XML_ERROR_SYNTAX;
  if (!isGeneralTextEntity && standalone == 1)
    dtd.standalone = 1;
  if (defaultHandler)
    reportDefault(parser, encoding, s, next);
  if (!protocolEncodingName) {
    if (newEncoding) {
      if (newEncoding->minBytesPerChar != encoding->minBytesPerChar) {
	eventPtr = encodingName;
	return XML_ERROR_INCORRECT_ENCODING;
      }
      encoding = newEncoding;
    }
    else if (encodingName) {
      enum XML_Error result;
      const XML_Char *s = poolStoreString(&tempPool,
					  encoding,
					  encodingName,
					  encodingName
					  + XmlNameLength(encoding, encodingName));
      if (!s)
	return XML_ERROR_NO_MEMORY;
      result = handleUnknownEncoding(parser, s);
      poolDiscard(&tempPool);
      if (result == XML_ERROR_UNKNOWN_ENCODING)
	eventPtr = encodingName;
      return result;
    }
  }
  return XML_ERROR_NONE;
}

static enum XML_Error
handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName)
{
  if (unknownEncodingHandler) {
    XML_Encoding info;
    int i;
    for (i = 0; i < 256; i++)
      info.map[i] = -1;
    info.convert = 0;
    info.data = 0;
    info.release = 0;
    if (unknownEncodingHandler(unknownEncodingHandlerData, encodingName, &info)) {
      ENCODING *enc;
      unknownEncodingMem = malloc(XmlSizeOfUnknownEncoding());
      if (!unknownEncodingMem) {
	if (info.release)
	  info.release(info.data);
	return XML_ERROR_NO_MEMORY;
      }
      enc = (ns
	     ? XmlInitUnknownEncodingNS
	     : XmlInitUnknownEncoding)(unknownEncodingMem,
				       info.map,
				       info.convert,
				       info.data);
      if (enc) {
	unknownEncodingData = info.data;
	unknownEncodingRelease = info.release;
	encoding = enc;
	return XML_ERROR_NONE;
      }
    }
    if (info.release)
      info.release(info.data);
  }
  return XML_ERROR_UNKNOWN_ENCODING;
}

static enum XML_Error
prologInitProcessor(XML_Parser parser,
		    const char *s,
		    const char *end,
		    const char **nextPtr)
{
  enum XML_Error result = initializeEncoding(parser);
  if (result != XML_ERROR_NONE)
    return result;
  processor = prologProcessor;
  return prologProcessor(parser, s, end, nextPtr);
}

static enum XML_Error
prologProcessor(XML_Parser parser,
		const char *s,
		const char *end,
		const char **nextPtr)
{
  for (;;) {
    const char *next;
    int tok = XmlPrologTok(encoding, s, end, &next);
    if (tok <= 0) {
      if (nextPtr != 0 && tok != XML_TOK_INVALID) {
	*nextPtr = s;
	return XML_ERROR_NONE;
      }
      switch (tok) {
      case XML_TOK_INVALID:
	eventPtr = next;
	return XML_ERROR_INVALID_TOKEN;
      case XML_TOK_NONE:
	return XML_ERROR_NO_ELEMENTS;
      case XML_TOK_PARTIAL:
	return XML_ERROR_UNCLOSED_TOKEN;
      case XML_TOK_PARTIAL_CHAR:
	return XML_ERROR_PARTIAL_CHAR;
      case XML_TOK_TRAILING_CR:
	eventPtr = s + encoding->minBytesPerChar;
	return XML_ERROR_NO_ELEMENTS;
      default:
	abort();
      }
    }
    switch (XmlTokenRole(&prologState, tok, s, next, encoding)) {
    case XML_ROLE_XML_DECL:
      {
	enum XML_Error result = processXmlDecl(parser, 0, s, next);
	if (result != XML_ERROR_NONE)
	  return result;
      }
      break;
    case XML_ROLE_DOCTYPE_SYSTEM_ID:
      if (!dtd.standalone
	  && notStandaloneHandler
	  && !notStandaloneHandler(handlerArg))
	return XML_ERROR_NOT_STANDALONE;
      hadExternalDoctype = 1;
      break;
    case XML_ROLE_DOCTYPE_PUBLIC_ID:
    case XML_ROLE_ENTITY_PUBLIC_ID:
      if (!XmlIsPublicId(encoding, s, next, &eventPtr))
	return XML_ERROR_SYNTAX;
      if (declEntity) {
	XML_Char *tem = poolStoreString(&dtd.pool,
	                                encoding,
					s + encoding->minBytesPerChar,
	  				next - encoding->minBytesPerChar);
	if (!tem)
	  return XML_ERROR_NO_MEMORY;
	normalizePublicId(tem);
	declEntity->publicId = tem;
	poolFinish(&dtd.pool);
      }
      break;
    case XML_ROLE_INSTANCE_START:
      processor = contentProcessor;
      if (hadExternalDoctype)
	dtd.complete = 0;
      return contentProcessor(parser, s, end, nextPtr);
    case XML_ROLE_ATTLIST_ELEMENT_NAME:
      {
	const XML_Char *name = poolStoreString(&dtd.pool, encoding, s, next);
	if (!name)
	  return XML_ERROR_NO_MEMORY;
	declElementType = (ELEMENT_TYPE *)lookup(&dtd.elementTypes, name, sizeof(ELEMENT_TYPE));
	if (!declElementType)
	  return XML_ERROR_NO_MEMORY;
	if (declElementType->name != name)
	  poolDiscard(&dtd.pool);
	else {
	  poolFinish(&dtd.pool);
	  if (!setElementTypePrefix(parser, declElementType))
            return XML_ERROR_NO_MEMORY;
	}
	break;
      }
    case XML_ROLE_ATTRIBUTE_NAME:
      declAttributeId = getAttributeId(parser, encoding, s, next);
      if (!declAttributeId)
	return XML_ERROR_NO_MEMORY;
      declAttributeIsCdata = 0;
      break;
    case XML_ROLE_ATTRIBUTE_TYPE_CDATA:
      declAttributeIsCdata = 1;
      break;
    case XML_ROLE_IMPLIED_ATTRIBUTE_VALUE:
    case XML_ROLE_REQUIRED_ATTRIBUTE_VALUE:
      if (dtd.complete
	  && !defineAttribute(declElementType, declAttributeId, declAttributeIsCdata, 0))
	return XML_ERROR_NO_MEMORY;
      break;
    case XML_ROLE_DEFAULT_ATTRIBUTE_VALUE:
    case XML_ROLE_FIXED_ATTRIBUTE_VALUE:
      {
	const XML_Char *attVal;
	enum XML_Error result
	  = storeAttributeValue(parser, encoding, declAttributeIsCdata,
				s + encoding->minBytesPerChar,
			        next - encoding->minBytesPerChar,
			        &dtd.pool);
	if (result)
	  return result;
	attVal = poolStart(&dtd.pool);
	poolFinish(&dtd.pool);
	if (dtd.complete
	    && !defineAttribute(declElementType, declAttributeId, declAttributeIsCdata, attVal))
	  return XML_ERROR_NO_MEMORY;
	break;
      }
    case XML_ROLE_ENTITY_VALUE:
      {
	enum XML_Error result = storeEntityValue(parser, s, next);
	if (result != XML_ERROR_NONE)
	  return result;
      }
      break;
    case XML_ROLE_ENTITY_SYSTEM_ID:
      if (declEntity) {
	declEntity->systemId = poolStoreString(&dtd.pool, encoding,
	                                       s + encoding->minBytesPerChar,
	  				       next - encoding->minBytesPerChar);
	if (!declEntity->systemId)
	  return XML_ERROR_NO_MEMORY;
	declEntity->base = dtd.base;
	poolFinish(&dtd.pool);
      }
      break;
    case XML_ROLE_ENTITY_NOTATION_NAME:
      if (declEntity) {
	declEntity->notation = poolStoreString(&dtd.pool, encoding, s, next);
	if (!declEntity->notation)
	  return XML_ERROR_NO_MEMORY;
	poolFinish(&dtd.pool);
	if (unparsedEntityDeclHandler) {
	  eventPtr = eventEndPtr = s;
	  unparsedEntityDeclHandler(handlerArg,
				    declEntity->name,
				    declEntity->base,
				    declEntity->systemId,
				    declEntity->publicId,
				    declEntity->notation);
	}

      }
      break;
    case XML_ROLE_GENERAL_ENTITY_NAME:
      {
	const XML_Char *name;
	if (XmlPredefinedEntityName(encoding, s, next)) {
	  declEntity = 0;
	  break;
	}
	name = poolStoreString(&dtd.pool, encoding, s, next);
	if (!name)
	  return XML_ERROR_NO_MEMORY;
	if (dtd.complete) {
	  declEntity = (ENTITY *)lookup(&dtd.generalEntities, name, sizeof(ENTITY));
	  if (!declEntity)
	    return XML_ERROR_NO_MEMORY;
	  if (declEntity->name != name) {
	    poolDiscard(&dtd.pool);
	    declEntity = 0;
	  }
	  else
	    poolFinish(&dtd.pool);
	}
	else {
	  poolDiscard(&dtd.pool);
	  declEntity = 0;
	}
      }
      break;
    case XML_ROLE_PARAM_ENTITY_NAME:
      declEntity = 0;
      break;
    case XML_ROLE_NOTATION_NAME:
      declNotationPublicId = 0;
      declNotationName = 0;
      if (notationDeclHandler) {
	declNotationName = poolStoreString(&tempPool, encoding, s, next);
	if (!declNotationName)
	  return XML_ERROR_NO_MEMORY;
	poolFinish(&tempPool);
      }
      break;
    case XML_ROLE_NOTATION_PUBLIC_ID:
      if (!XmlIsPublicId(encoding, s, next, &eventPtr))
	return XML_ERROR_SYNTAX;
      if (declNotationName) {
	XML_Char *tem = poolStoreString(&tempPool,
	                                encoding,
					s + encoding->minBytesPerChar,
	  				next - encoding->minBytesPerChar);
	if (!tem)
	  return XML_ERROR_NO_MEMORY;
	normalizePublicId(tem);
	declNotationPublicId = tem;
	poolFinish(&tempPool);
      }
      break;
    case XML_ROLE_NOTATION_SYSTEM_ID:
      if (declNotationName && notationDeclHandler) {
	const XML_Char *systemId
	  = poolStoreString(&tempPool, encoding,
			    s + encoding->minBytesPerChar,
	  		    next - encoding->minBytesPerChar);
	if (!systemId)
	  return XML_ERROR_NO_MEMORY;
	eventPtr = eventEndPtr = s;
	notationDeclHandler(handlerArg,
			    declNotationName,
			    dtd.base,
			    systemId,
			    declNotationPublicId);
      }
      poolClear(&tempPool);
      break;
    case XML_ROLE_NOTATION_NO_SYSTEM_ID:
      if (declNotationPublicId && notationDeclHandler) {
	eventPtr = eventEndPtr = s;
	notationDeclHandler(handlerArg,
			    declNotationName,
			    dtd.base,
			    0,
			    declNotationPublicId);
      }
      poolClear(&tempPool);
      break;
    case XML_ROLE_ERROR:
      eventPtr = s;
      switch (tok) {
      case XML_TOK_PARAM_ENTITY_REF:
	return XML_ERROR_PARAM_ENTITY_REF;
      case XML_TOK_XML_DECL:
	return XML_ERROR_MISPLACED_XML_PI;
      default:
	return XML_ERROR_SYNTAX;
      }
    case XML_ROLE_GROUP_OPEN:
      if (prologState.level >= groupSize) {
	if (groupSize)
	  groupConnector = realloc(groupConnector, groupSize *= 2);
	else
	  groupConnector = malloc(groupSize = 32);
	if (!groupConnector)
	  return XML_ERROR_NO_MEMORY;
      }
      groupConnector[prologState.level] = 0;
      break;
    case XML_ROLE_GROUP_SEQUENCE:
      if (groupConnector[prologState.level] == '|') {
	eventPtr = s;
	return XML_ERROR_SYNTAX;
      }
      groupConnector[prologState.level] = ',';
      break;
    case XML_ROLE_GROUP_CHOICE:
      if (groupConnector[prologState.level] == ',') {
	eventPtr = s;
	return XML_ERROR_SYNTAX;
      }
      groupConnector[prologState.level] = '|';
      break;
    case XML_ROLE_PARAM_ENTITY_REF:
      if (!dtd.standalone
	  && notStandaloneHandler
	  && !notStandaloneHandler(handlerArg))
	return XML_ERROR_NOT_STANDALONE;
      dtd.complete = 0;
      break;
    case XML_ROLE_NONE:
      switch (tok) {
      case XML_TOK_PI:
	eventPtr = s;
	eventEndPtr = next;
	if (!reportProcessingInstruction(parser, encoding, s, next))
	  return XML_ERROR_NO_MEMORY;
	break;
      case XML_TOK_COMMENT:
	eventPtr = s;
	eventEndPtr = next;
	if (!reportComment(parser, encoding, s, next))
	  return XML_ERROR_NO_MEMORY;
	break;
      }
      break;
    }
    if (defaultHandler) {
      switch (tok) {
      case XML_TOK_PI:
      case XML_TOK_COMMENT:
      ca