int acl_xml_decode(const char *in, ACL_VSTRING *out)
{
int n = 0, len;
const char *ptr = in, *pre;
const ACL_TOKEN *token;
const XML_SPEC *spec;
acl_pthread_once(&__token_once, xml_decode_init);
if (__token_tree == NULL)
acl_msg_fatal("__token_tree null");
while (*ptr != 0) {
pre = ptr;
token = acl_token_tree_match(__token_tree, &ptr, NULL, NULL);
if (token == NULL) {
pre = markup_unescape(pre, out);
len = (int) (ptr - pre);
if (len > 0)
acl_vstring_memcat(out, pre, len);
break;
}
spec = (const XML_SPEC*) token->ctx;
acl_assert(spec != NULL);
len = (int) (ptr - pre - spec->len);
if (len > 0)
acl_vstring_memcat(out, pre, len);
acl_vstring_strcat(out, spec->str);
n++;
}
ACL_VSTRING_TERMINATE(out);
return (n);
}
bool charset_conv::update(const char* in, size_t len, acl::string* out)
{
#ifdef HAVE_H_ICONV
if (in == NULL)
logger_fatal("in null");
if (out == NULL)
logger_fatal("out null");
if (EQ(m_fromCharset, m_toCharset))
{
out->append(in, len);
return (true);
}
if (m_iconv == (iconv_t) -1)
{
logger_error("m_iconv invalid");
m_errmsg = "m_iconv invalid";
return (false);
}
// 去掉有些 UTF-8 文档中开始的 UTF-8 引导符
if (*m_pUtf8Pre)
{
while (len > 0)
{
if (*m_pUtf8Pre == 0x00)
break;
else if (*m_pUtf8Pre != *in)
{
// 必须使 UTF-8 前缀失效
m_pUtf8Pre = &UTF8_HEADER[3];
break;
}
m_pUtf8Pre++;
in++;
len--;
}
}
if (len == 0)
return (true);
if (m_pInBuf == NULL)
m_pInBuf = acl_vstring_alloc(len);
if (m_pOutBuf == NULL)
m_pOutBuf = acl_vstring_alloc(len);
else
ACL_VSTRING_SPACE(m_pOutBuf, len);
// 先将输入数据进行缓冲
if (*m_pUtf8Pre && m_pUtf8Pre - UTF8_HEADER > 0)
acl_vstring_memcpy(m_pInBuf, UTF8_HEADER,
m_pUtf8Pre - UTF8_HEADER);
acl_vstring_memcat(m_pInBuf, in, len);
ACL_VSTRING_TERMINATE(m_pInBuf);
char *pIn, *pOut;
size_t ret, nIn, nOut;
while (true)
{
nIn = LEN(m_pInBuf);
if (nIn == 0)
break;
pIn = STR(m_pInBuf);
pOut = STR(m_pOutBuf);
nOut = SIZE(m_pOutBuf);
#ifdef WIN32
# ifdef USE_WIN_ICONV
ret = __iconv(m_iconv, (const char**) &pIn, &nIn,
&pOut, &nOut);
# else
int err;
ret = __iconv(m_iconv, (const char**) &pIn, &nIn,
&pOut, &nOut, &err);
errno = err;
# endif // USE_WIN_ICONV
#elif defined(ACL_SUNOS5) || defined(ACL_FREEBSD)
ret = __iconv(m_iconv, (const char**) &pIn, &nIn, &pOut, &nOut);
#else
ret = __iconv(m_iconv, &pIn, &nIn, &pOut, &nOut);
#endif
if (ret != (size_t) -1)
{
if ((ret = SIZE(m_pOutBuf) - nOut) > 0)
out->append(STR(m_pOutBuf), ret);
else // xxx
out->append(in, len);
ACL_VSTRING_RESET(m_pInBuf);
break;
}
else if (errno == E2BIG)
{
if ((ret = SIZE(m_pOutBuf) - nOut) > 0)
out->append(STR(m_pOutBuf), ret);
if (pIn > STR(m_pInBuf) && nIn < LEN(m_pInBuf))
acl_vstring_memmove(m_pInBuf, pIn, nIn);
// 扩大内存空间
ACL_VSTRING_SPACE(m_pOutBuf, SIZE(m_pOutBuf) * 2);
continue;
}
else if (errno == EILSEQ)
{
char *pNil = NULL;
size_t zero = 0;
// 重置状态, 似乎也没啥用处
#ifdef WIN32
# ifdef USE_WIN_ICONV
__iconv(m_iconv, (const char**) &pNil,
&zero, &pNil, &zero);
# else
__iconv(m_iconv, (const char**) &pNil,
&zero, &pNil, &zero, NULL);
# endif
#elif defined(ACL_SUNOS5) || defined(ACL_FREEBSD)
__iconv(m_iconv, (const char**) &pNil,
&zero, &pNil, &zero);
#else
__iconv(m_iconv, &pNil, &zero, &pNil, &zero);
#endif
// 遇到无效的多字节序列，pIn 指向第一个无效的位置
// 先拷贝已经转换的数据
if ((ret = SIZE(m_pOutBuf) - nOut) > 0)
out->append(STR(m_pOutBuf), ret);
if (nIn == 0)
{
ACL_VSTRING_RESET(m_pInBuf);
break;
}
acl_assert(pIn >= STR(m_pInBuf));
// 跳过无效字节
(*out) += (char)(*pIn); // 直接拷贝无效字节
nIn--;
pIn++;
if (nIn > 0)
acl_vstring_memmove(m_pInBuf, pIn, nIn);
else
ACL_VSTRING_RESET(m_pInBuf);
}
else if (errno == EINVAL)
{
char *pNil = NULL;
size_t zero = 0;
// 重置状态, 似乎也没啥用处
#ifdef WIN32
# ifdef USE_WIN_ICONV
__iconv(m_iconv, (const char**) &pNil,
&zero, &pNil, &zero);
# else
__iconv(m_iconv, (const char**) &pNil,
&zero, &pNil, &zero, NULL);
# endif // USE_WIN_ICONV
#elif defined(ACL_SUNOS5) || defined(ACL_FREEBSD)
__iconv(m_iconv, (const char**) &pNil, &zero, &pNil, &zero);
#else
__iconv(m_iconv, &pNil, &zero, &pNil, &zero);
#endif
// 输入的多字节序列不完整，pIn 指向该不完整的位置
// 先拷贝已经转换的数据
if ((ret = SIZE(m_pOutBuf) - nOut) > 0)
out->append(STR(m_pOutBuf), ret);
// 移动数据，将未转换的数据移至缓冲区起始位置
if (nIn > 0)
acl_vstring_memmove(m_pInBuf, pIn, nIn);
else
ACL_VSTRING_RESET(m_pInBuf);
break;
}
else if (LEN(m_pInBuf) > 0)
{
// 如果遇到了无效的字符集，根据设置的标志位
// 决定是否直接拷贝
if (m_addInvalid)
{
out->append(STR(m_pInBuf), LEN(m_pInBuf));
ACL_VSTRING_RESET(m_pInBuf);
}
break;
}
else
break;
}
return (true);
#else
(void) in;
(void) len;
(void) out;
logger_error("no iconv lib");
m_errmsg = "no iconv lib";
return (false);
#endif
}
