LuaXml解析xml报文时，unicode编码无法被解析

最新推荐文章于 2024-08-20 14:21:59 发布

冷曦_sole

最新推荐文章于 2024-08-20 14:21:59 发布

阅读量785

点赞数

分类专栏： luaxml 文章标签： lua xml unicode

本文链接：https://blog.csdn.net/surongle123/article/details/79984558

版权

luaxml 专栏收录该内容

1 篇文章 0 订阅

订阅专栏

在使用LuaXml解析xml报文时，如果xml报文中带有unicode编码，解码后依然是unicode编码，无法被解析成实际的内容，

这种情况需要修改Lua_XML_lib.c,修改如下

将251行到257行替换成如下内容：

    while(pos < s_size)
   {
       if(found&&(pos+5<s_size)&&(*(found+5)==';')&&isdigit(*(found+2))&&isdigit(*(found+3))&&isdigit(*(found+4)) )
       {
           if(pos>start) luaL_addlstring(&b,s+start, pos-start);
           luaL_addchar(&b, 100*(s[pos+2]-48)+10*(s[pos+3]-48)+(s[pos+4]-48));
           start=pos+6;
           found = strstr(found+6, "&#");
           if(!found) pos = s_size;
           else pos = found-s;
       }
       else if(found&&(pos+7<s_size)&&(*(found+2)=='x')&&(*(found+7)==';')&&isxdigit(*(found+3))&&isxdigit(*(found+4))&&isxdigit(*(found+5))&&isxdigit(*(found+6)))
       {
           if(pos>start) luaL_addlstring(&b,s+start, pos-start);
           char buf[32];
           memset(buf,0,sizeof(buf));
           strncpy(buf,found+3,4);
           unsigned int value_int;
           sscanf(buf,"%04X",&value_int);
           char result[32];
           memset(result,0,sizeof(result));
           if(1)
           {
               if (value_int <= 0x7f)
               {
                   result[0] = value_int;
               }
               else if (value_int <= 0x7FF)
               {
                   result[1] = (0x80 | (0x3f & value_int));
                   result[0] = (0xC0 | (0x1f & (value_int >> 6)));
               }
               else if (value_int <= 0xFFFF)
               {
                   result[2] = (0x80 | (0x3f & value_int));
                   result[1] = 0x80 | ((0x3f & (value_int >> 6)));
                   result[0] = 0xE0 | ((0xf & (value_int >> 12)));
               }
               else if (value_int <= 0x10FFFF)
               {
                   result[3] = (0x80 | (0x3f & value_int));
                   result[2] = (0x80 | (0x3f & (value_int >> 6)));
                   result[1] = (0x80 | (0x3f & (value_int >> 12)));
                   result[0] = (0xF0 | (0x7 & (value_int >> 18)));
               }
           }

           luaL_addstring(&b,result);

           start=pos+8;
           found = strstr(found+8, "&#");
           if(!found) pos = s_size;
           else pos = found-s;
       }
       else
       {
           luaL_addlstring(&b,s+start, pos-start);

}

差分如下：

if(!found) pos = s_size;		if(!found) pos = s_size;
else pos = found-s;		else pos = found-s;
while(pos < s_size)	<>
{
if(found&&(pos+5<s_size)&&((found+5)==';')&&isdigit((found+2))&&isdigit((found+3))&&isdigit((found+4)) )		while(found&&(pos+5<s_size)&&((found+5)==';')&&isdigit((found+2))&&isdigit((found+3))&&isdigit((found+4)) ) {
{
if(pos>start) luaL_addlstring(&b,s+start, pos-start);	=	if(pos>start) luaL_addlstring(&b,s+start, pos-start);
luaL_addchar(&b, 100(s[pos+2]-48)+10(s[pos+3]-48)+(s[pos+4]-48));		luaL_addchar(&b, 100(s[pos+2]-48)+10(s[pos+3]-48)+(s[pos+4]-48));
start=pos+6;		start=pos+6;
found = strstr(found+6, "&#");		found = strstr(found+6, "&#");
if(!found) pos = s_size;		if(!found) pos = s_size;
else pos = found-s;		else pos = found-s;
}	+-
else if(found&&(pos+7<s_size)&&((found+2)=='x')&&((found+7)==';')&&isxdigit((found+3))&&isxdigit((found+4))&&isxdigit((found+5))&&isxdigit((found+6)))
{
if(pos>start) luaL_addlstring(&b,s+start, pos-start);
char buf[32];
memset(buf,0,sizeof(buf));
strncpy(buf,found+3,4);
unsigned int value_int;
sscanf(buf,"%04X",&value_int);
char result[32];
memset(result,0,sizeof(result));
if(1)
{
if (value_int <= 0x7f)
{
result[0] = value_int;
}
else if (value_int <= 0x7FF)
{
result[1] = (0x80 \| (0x3f & value_int));
result[0] = (0xC0 \| (0x1f & (value_int >> 6)));
}
else if (value_int <= 0xFFFF)
{
result[2] = (0x80 \| (0x3f & value_int));
result[1] = 0x80 \| ((0x3f & (value_int >> 6)));
result[0] = 0xE0 \| ((0xf & (value_int >> 12)));
}
else if (value_int <= 0x10FFFF)
{
result[3] = (0x80 \| (0x3f & value_int));
result[2] = (0x80 \| (0x3f & (value_int >> 6)));
result[1] = (0x80 \| (0x3f & (value_int >> 12)));
result[0] = (0xF0 \| (0x7 & (value_int >> 18)));
}
}
	=
luaL_addstring(&b,result);	+-
	=
start=pos+8;	+-
found = strstr(found+8, "&#");
if(!found) pos = s_size;
else pos = found-s;
}
else
{
luaL_addlstring(&b,s+start, pos-start);
}
}	=	}
if(pos>start) luaL_addlstring(&b,s+start, pos-start);		if(pos>start) luaL_addlstring(&b,s+start, pos-start);
luaL_pushresult(&b);		luaL_pushresult(&b);
size_t i;		size_t i;
for(i=sv_code_size-1; i<sv_code_size; i-=2) {		for(i=sv_code_size-1; i<sv_code_size; i-=2) {
luaL_gsub(L, lua_tostring(L,-1), sv_code[i], sv_code[i-1]);		luaL_gsub(L, lua_tostring(L,-1), sv_code[i], sv_code[i-1]);
lua_remove(L,-2);		lua_remove(L,-2);
}		}
}		}