bool CXBMCTinyXML::TryParse(const std::string& data, const std::string& tryDataCharset)
{
if (tryDataCharset == "UTF-8")
InternalParse(data, TIXML_ENCODING_UTF8); // process data without conversion
else if (!tryDataCharset.empty())
{
std::string converted;
/* some wrong conversions can leave US-ASCII XML header and structure untouched but break non-English data
* so conversion must fail on wrong character and then other encodings will be tried */
if (!g_charsetConverter.ToUtf8(tryDataCharset, data, converted, true) || converted.empty())
return false; // can't convert data
InternalParse(converted, TIXML_ENCODING_UTF8);
}
else
InternalParse(data, TIXML_ENCODING_LEGACY);
// 'Error()' contains result of last run of 'TiXmlDocument::Parse()'
if (Error())
{
Clear();
location.Clear();
return false;
}
m_UsedCharset = tryDataCharset;
return true;
}
XmlNode::smart_ptr_t XmlParser::Parse( const tchar* _pBuffer, int _len, bool _isFinal )
{
if ( InternalParse( _pBuffer, _len, _isFinal ) )
return m_Root;
return XmlNode::smart_ptr_t(NULL);
}
bool CXBMCTinyXML::Parse(const std::string& data, TiXmlEncoding encoding /*= TIXML_DEFAULT_ENCODING */)
{
m_UsedCharset.clear();
if (encoding != TIXML_ENCODING_UNKNOWN)
{ // encoding != TIXML_ENCODING_UNKNOWN means "do not use m_SuggestedCharset and charset detection"
m_SuggestedCharset.clear();
if (encoding == TIXML_ENCODING_UTF8)
m_UsedCharset = "UTF-8";
return InternalParse(data, encoding);
}
if (!m_SuggestedCharset.empty() && TryParse(data, m_SuggestedCharset))
return true;
std::string detectedCharset;
if (CCharsetDetection::DetectXmlEncoding(data, detectedCharset) && TryParse(data, detectedCharset))
{
if (!m_SuggestedCharset.empty())
CLog::Log(LOGWARNING, "%s: \"%s\" charset was used instead of suggested charset \"%s\" for %s", __FUNCTION__, m_UsedCharset.c_str(), m_SuggestedCharset.c_str(),
(value.empty() ? "XML data" : ("file \"" + value + "\"").c_str()));
return true;
}
// check for valid UTF-8
if (m_SuggestedCharset != "UTF-8" && detectedCharset != "UTF-8" && CUtf8Utils::isValidUtf8(data) &&
TryParse(data, "UTF-8"))
{
if (!m_SuggestedCharset.empty())
CLog::Log(LOGWARNING, "%s: \"%s\" charset was used instead of suggested charset \"%s\" for %s", __FUNCTION__, m_UsedCharset.c_str(), m_SuggestedCharset.c_str(),
(value.empty() ? "XML data" : ("file \"" + value + "\"").c_str()));
else if (!detectedCharset.empty())
CLog::Log(LOGWARNING, "%s: \"%s\" charset was used instead of detected charset \"%s\" for %s", __FUNCTION__, m_UsedCharset.c_str(), detectedCharset.c_str(),
(value.empty() ? "XML data" : ("file \"" + value + "\"").c_str()));
return true;
}
// fallback: try user GUI charset
if (TryParse(data, g_langInfo.GetGuiCharSet()))
{
if (!m_SuggestedCharset.empty())
CLog::Log(LOGWARNING, "%s: \"%s\" charset was used instead of suggested charset \"%s\" for %s", __FUNCTION__, m_UsedCharset.c_str(), m_SuggestedCharset.c_str(),
(value.empty() ? "XML data" : ("file \"" + value + "\"").c_str()));
else if (!detectedCharset.empty())
CLog::Log(LOGWARNING, "%s: \"%s\" charset was used instead of detected charset \"%s\" for %s", __FUNCTION__, m_UsedCharset.c_str(), detectedCharset.c_str(),
(value.empty() ? "XML data" : ("file \"" + value + "\"").c_str()));
return true;
}
// can't detect correct data charset, try to process data as is
if (InternalParse(data, TIXML_ENCODING_UNKNOWN))
{
if (!m_SuggestedCharset.empty())
CLog::Log(LOGWARNING, "%s: Processed %s as unknown encoding instead of suggested \"%s\"", __FUNCTION__,
(value.empty() ? "XML data" : ("file \"" + value + "\"").c_str()), m_SuggestedCharset.c_str());
else if (!detectedCharset.empty())
CLog::Log(LOGWARNING, "%s: Processed %s as unknown encoding instead of detected \"%s\"", __FUNCTION__,
(value.empty() ? "XML data" : ("file \"" + value + "\"").c_str()), detectedCharset.c_str());
return true;
}
return false;
}
