xmlparse.c

来自「ARM上的如果你对底层感兴趣」· C语言 代码 · 共 2,269 行 · 第 1/5 页

C
2,269
字号
  XML_Parser parser = oldParser;
  DTD *oldDtd = &dtd;
  XML_StartElementHandler oldStartElementHandler = startElementHandler;
  XML_EndElementHandler oldEndElementHandler = endElementHandler;
  XML_CharacterDataHandler oldCharacterDataHandler = characterDataHandler;
  XML_ProcessingInstructionHandler oldProcessingInstructionHandler = processingInstructionHandler;
  XML_CommentHandler oldCommentHandler = commentHandler;
  XML_DefaultHandler oldDefaultHandler = defaultHandler;
  XML_ExternalEntityRefHandler oldExternalEntityRefHandler = externalEntityRefHandler;
  XML_UnknownEncodingHandler oldUnknownEncodingHandler = unknownEncodingHandler;
  void *oldUserData = userData;
  void *oldHandlerArg = handlerArg;
  int oldDefaultExpandInternalEntities = defaultExpandInternalEntities;
 
  parser = (ns
            ? XML_ParserCreateNS(encodingName, namespaceSeparator)
	    : XML_ParserCreate(encodingName));
  if (!parser)
    return 0;
  startElementHandler = oldStartElementHandler;
  endElementHandler = oldEndElementHandler;
  characterDataHandler = oldCharacterDataHandler;
  processingInstructionHandler = oldProcessingInstructionHandler;
  commentHandler = oldCommentHandler;
  defaultHandler = oldDefaultHandler;
  externalEntityRefHandler = oldExternalEntityRefHandler;
  unknownEncodingHandler = oldUnknownEncodingHandler;
  userData = oldUserData;
  if (oldUserData == oldHandlerArg)
    handlerArg = userData;
  else
    handlerArg = parser;
  defaultExpandInternalEntities = oldDefaultExpandInternalEntities;
  if (!dtdCopy(&dtd, oldDtd) || !setContext(parser, context)) {
    XML_ParserFree(parser);
    return 0;
  }
  processor = externalEntityInitProcessor;
  return parser;
}

static
void destroyBindings(BINDING *bindings)
{
  for (;;) {
    BINDING *b = bindings;
    if (!b)
      break;
    bindings = b->nextTagBinding;
    free(b->uri);
    free(b);
  }
}

void XML_ParserFree(XML_Parser parser)
{
  for (;;) {
    TAG *p;
    if (tagStack == 0) {
      if (freeTagList == 0)
	break;
      tagStack = freeTagList;
      freeTagList = 0;
    }
    p = tagStack;
    tagStack = tagStack->parent;
    free(p->buf);
    destroyBindings(p->bindings);
    free(p);
  }
  destroyBindings(freeBindingList);
  destroyBindings(inheritedBindings);
  poolDestroy(&tempPool);
  poolDestroy(&temp2Pool);
  dtdDestroy(&dtd);
  free((void *)atts);
  free(groupConnector);
  free(buffer);
  free(dataBuf);
  free(unknownEncodingMem);
  if (unknownEncodingRelease)
    unknownEncodingRelease(unknownEncodingData);
  free(parser);
}

void XML_UseParserAsHandlerArg(XML_Parser parser)
{
  handlerArg = parser;
}

void XML_SetUserData(XML_Parser parser, void *p)
{
  if (handlerArg == userData)
    handlerArg = userData = p;
  else
    userData = p;
}

int XML_SetBase(XML_Parser parser, const XML_Char *p)
{
  if (p) {
    p = poolCopyString(&dtd.pool, p);
    if (!p)
      return 0;
    dtd.base = p;
  }
  else
    dtd.base = 0;
  return 1;
}

const XML_Char *XML_GetBase(XML_Parser parser)
{
  return dtd.base;
}

void XML_SetElementHandler(XML_Parser parser,
			   XML_StartElementHandler start,
			   XML_EndElementHandler end)
{
  startElementHandler = start;
  endElementHandler = end;
}

void XML_SetCharacterDataHandler(XML_Parser parser,
				 XML_CharacterDataHandler handler)
{
  characterDataHandler = handler;
}

void XML_SetProcessingInstructionHandler(XML_Parser parser,
					 XML_ProcessingInstructionHandler handler)
{
  processingInstructionHandler = handler;
}

void XML_SetCommentHandler(XML_Parser parser,
			   XML_CommentHandler handler)
{
  commentHandler = handler;
}

void XML_SetDefaultHandler(XML_Parser parser,
			   XML_DefaultHandler handler)
{
  defaultHandler = handler;
  defaultExpandInternalEntities = 0;
}

void XML_SetDefaultHandlerExpand(XML_Parser parser,
				 XML_DefaultHandler handler)
{
  defaultHandler = handler;
  defaultExpandInternalEntities = 1;
}

void XML_SetUnparsedEntityDeclHandler(XML_Parser parser,
				      XML_UnparsedEntityDeclHandler handler)
{
  unparsedEntityDeclHandler = handler;
}

void XML_SetNotationDeclHandler(XML_Parser parser,
				XML_NotationDeclHandler handler)
{
  notationDeclHandler = handler;
}

void XML_SetExternalEntityRefHandler(XML_Parser parser,
				     XML_ExternalEntityRefHandler handler)
{
  externalEntityRefHandler = handler;
}

void XML_SetUnknownEncodingHandler(XML_Parser parser,
				   XML_UnknownEncodingHandler handler,
				   void *data)
{
  unknownEncodingHandler = handler;
  unknownEncodingHandlerData = data;
}

int XML_Parse(XML_Parser parser, const char *s, int len, int isFinal)
{
  if (len == 0) {
    if (!isFinal)
      return 1;
    errorCode = processor(parser, bufferPtr, parseEndPtr = bufferEnd, 0);
    if (errorCode == XML_ERROR_NONE)
      return 1;
    eventEndPtr = eventPtr;
    return 0;
  }
  else if (bufferPtr == bufferEnd) {
    const char *end;
    int nLeftOver;
    parseEndByteIndex += len;
    positionPtr = s;
    if (isFinal) {
      errorCode = processor(parser, s, parseEndPtr = s + len, 0);
      if (errorCode == XML_ERROR_NONE)
	return 1;
      eventEndPtr = eventPtr;
      return 0;
    }
    errorCode = processor(parser, s, parseEndPtr = s + len, &end);
    if (errorCode != XML_ERROR_NONE) {
      eventEndPtr = eventPtr;
      return 0;
    }
    XmlUpdatePosition(encoding, positionPtr, end, &position);
    nLeftOver = s + len - end;
    if (nLeftOver) {
      if (buffer == 0 || nLeftOver > bufferLim - buffer) {
	/* FIXME avoid integer overflow */
	buffer = buffer == 0 ? malloc(len * 2) : realloc(buffer, len * 2);
	if (!buffer) {
	  errorCode = XML_ERROR_NO_MEMORY;
	  eventPtr = eventEndPtr = 0;
	  return 0;
	}
	bufferLim = buffer + len * 2;
      }
      memcpy(buffer, end, nLeftOver);
      bufferPtr = buffer;
      bufferEnd = buffer + nLeftOver;
    }
    return 1;
  }
  else {
    memcpy(XML_GetBuffer(parser, len), s, len);
    return XML_ParseBuffer(parser, len, isFinal);
  }
}

int XML_ParseBuffer(XML_Parser parser, int len, int isFinal)
{
  const char *start = bufferPtr;
  positionPtr = start;
  bufferEnd += len;
  parseEndByteIndex += len;
  errorCode = processor(parser, start, parseEndPtr = bufferEnd,
			isFinal ? (const char **)0 : &bufferPtr);
  if (errorCode == XML_ERROR_NONE) {
    if (!isFinal)
      XmlUpdatePosition(encoding, positionPtr, bufferPtr, &position);
    return 1;
  }
  else {
    eventEndPtr = eventPtr;
    return 0;
  }
}

void *XML_GetBuffer(XML_Parser parser, int len)
{
  if (len > bufferLim - bufferEnd) {
    /* FIXME avoid integer overflow */
    int neededSize = len + (bufferEnd - bufferPtr);
    if (neededSize  <= bufferLim - buffer) {
      memmove(buffer, bufferPtr, bufferEnd - bufferPtr);
      bufferEnd = buffer + (bufferEnd - bufferPtr);
      bufferPtr = buffer;
    }
    else {
      char *newBuf;
      int bufferSize = bufferLim - bufferPtr;
      if (bufferSize == 0)
	bufferSize = INIT_BUFFER_SIZE;
      do {
	bufferSize *= 2;
      } while (bufferSize < neededSize);
      newBuf = malloc(bufferSize);
      if (newBuf == 0) {
	errorCode = XML_ERROR_NO_MEMORY;
	return 0;
      }
      bufferLim = newBuf + bufferSize;
      if (bufferPtr) {
	memcpy(newBuf, bufferPtr, bufferEnd - bufferPtr);
	free(buffer);
      }
      bufferEnd = newBuf + (bufferEnd - bufferPtr);
      bufferPtr = buffer = newBuf;
    }
  }
  return bufferEnd;
}

enum XML_Error XML_GetErrorCode(XML_Parser parser)
{
  return errorCode;
}

long XML_GetCurrentByteIndex(XML_Parser parser)
{
  if (eventPtr)
    return parseEndByteIndex - (parseEndPtr - eventPtr);
  return -1;
}

int XML_GetCurrentLineNumber(XML_Parser parser)
{
  if (eventPtr) {
    XmlUpdatePosition(encoding, positionPtr, eventPtr, &position);
    positionPtr = eventPtr;
  }
  return position.lineNumber + 1;
}

int XML_GetCurrentColumnNumber(XML_Parser parser)
{
  if (eventPtr) {
    XmlUpdatePosition(encoding, positionPtr, eventPtr, &position);
    positionPtr = eventPtr;
  }
  return position.columnNumber;
}

void XML_DefaultCurrent(XML_Parser parser)
{
  if (defaultHandler) {
    if (openInternalEntities)
      reportDefault(parser,
	            ns ? XmlGetInternalEncodingNS() : XmlGetInternalEncoding(),
		    openInternalEntities->internalEventPtr,
		    openInternalEntities->internalEventEndPtr);
    else
      reportDefault(parser, encoding, eventPtr, eventEndPtr);
  }
}

const XML_LChar *XML_ErrorString(int code)
{
  static const XML_LChar *message[] = {
    0,
    XML_T("out of memory"),
    XML_T("syntax error"),
    XML_T("no element found"),
    XML_T("not well-formed"),
    XML_T("unclosed token"),
    XML_T("unclosed token"),
    XML_T("mismatched tag"),
    XML_T("duplicate attribute"),
    XML_T("junk after document element"),
    XML_T("illegal parameter entity reference"),
    XML_T("undefined entity"),
    XML_T("recursive entity reference"),
    XML_T("asynchronous entity"),
    XML_T("reference to invalid character number"),
    XML_T("reference to binary entity"),
    XML_T("reference to external entity in attribute"),
    XML_T("xml processing instruction not at start of external entity"),
    XML_T("unknown encoding"),
    XML_T("encoding specified in XML declaration is incorrect"),
    XML_T("unclosed CDATA section"),
    XML_T("error in processing external entity reference")
  };
  if (code > 0 && code < sizeof(message)/sizeof(message[0]))
    return message[code];
  return 0;
}

static
enum XML_Error contentProcessor(XML_Parser parser,
				const char *start,
				const char *end,
				const char **endPtr)
{
  return doContent(parser, 0, encoding, start, end, endPtr);
}

static
enum XML_Error externalEntityInitProcessor(XML_Parser parser,
					   const char *start,
					   const char *end,
					   const char **endPtr)
{
  enum XML_Error result = initializeEncoding(parser);
  if (result != XML_ERROR_NONE)
    return result;
  processor = externalEntityInitProcessor2;
  return externalEntityInitProcessor2(parser, start, end, endPtr);
}

static
enum XML_Error externalEntityInitProcessor2(XML_Parser parser,
					    const char *start,
					    const char *end,
					    const char **endPtr)
{
  const char *next;
  int tok = XmlContentTok(encoding, start, end, &next);
  switch (tok) {
  case XML_TOK_BOM:
    start = next;
    break;
  case XML_TOK_PARTIAL:
    if (endPtr) {
      *endPtr = start;
      return XML_ERROR_NONE;
    }
    eventPtr = start;
    return XML_ERROR_UNCLOSED_TOKEN;
  case XML_TOK_PARTIAL_CHAR:
    if (endPtr) {
      *endPtr = start;
      return XML_ERROR_NONE;
    }
    eventPtr = start;
    return XML_ERROR_PARTIAL_CHAR;
  }
  processor = externalEntityInitProcessor3;
  return externalEntityInitProcessor3(parser, start, end, endPtr);
}

static
enum XML_Error externalEntityInitProcessor3(XML_Parser parser,
					    const char *start,
					    const char *end,
					    const char **endPtr)
{
  const char *next;
  int tok = XmlContentTok(encoding, start, end, &next);
  switch (tok) {
  case XML_TOK_XML_DECL:
    {
      enum XML_Error result = processXmlDecl(parser, 1, start, next);
      if (result != XML_ERROR_NONE)
	return result;
      start = next;
    }
    break;
  case XML_TOK_PARTIAL:
    if (endPtr) {
      *endPtr = start;
      return XML_ERROR_NONE;
    }
    eventPtr = start;
    return XML_ERROR_UNCLOSED_TOKEN;
  case XML_TOK_PARTIAL_CHAR:
    if (endPtr) {
      *endPtr = start;
      return XML_ERROR_NONE;
    }
    eventPtr = start;
    return XML_ERROR_PARTIAL_CHAR;
  }
  processor = externalEntityContentProcessor;
  tagLevel = 1;
  return doContent(parser, 1, encoding, start, end, endPtr);
}

static

⌨️ 快捷键说明

复制代码Ctrl + C
搜索代码Ctrl + F
全屏模式F11
增大字号Ctrl + =
减小字号Ctrl + -
显示快捷键?