Stroka CCommonParm::GetOutputFileName() const {
if (NULL != Config.Get() && Config->has_output() && Config->output().has_file()) {
Stroka fn = Config->output().GetFile();
fn.to_lower();
if ("stdout" == fn)
return "-";
return Config->output().GetFile();
}
return Stroka("-");
}
Stroka CCommonParm::GetPrettyOutputFileName() const {
if (NULL != Config.Get() && Config->has_prettyoutput()) {
Stroka fn = Config->GetPrettyOutput();
fn.to_lower();
if ("stdout" == fn || "stderr" == fn)
return fn;
if ("-" == fn)
return "stdout";
if (fn.size() > 0)
return Config->GetPrettyOutput();
}
return Stroka("");
}
void CCommonParm::AnalizeParameter(const TStringBuf& name, const TStringBuf& value) {
Stroka paramName = ToString(StripString(name));
Stroka paramValue = ToString(StripString(value));
paramName.to_lower();
}
bool CCommonParm::ParseConfig(const Stroka& fn) {
Config.Reset(NProtoConf::LoadFromFile<TTextMinerConfig>(fn));
if (!Config)
ythrow yexception() << "Cannot read the config from \"" << fn << "\".";
if (m_args.has(OPT_BINARY_DIR)) {
if (Config->has_binarydir() && Config->binarydir().length() > 0)
ythrow yexception() << "Both \"--" << OPT_BINARY_DIR << "\" command line argument and \"BinaryDir\" config parameter specified";
Config->set_binarydir(m_args[OPT_BINARY_DIR]);
}
if (Config->has_input()) {
TTextMinerConfig_TInputParams inputParams = Config->input();
if (inputParams.has_file() && !inputParams.file().empty()
&& inputParams.has_dir() && !inputParams.dir().empty())
ythrow yexception() << "Input\\File and Input\\Dir options are meaningless together";
Stroka fn = inputParams.file();
fn.to_lower();
if (fn.empty() || "stdin" == fn || "-" == fn)
m_strInputFileName = "";
else
m_strInputFileName = inputParams.file();
if (inputParams.has_dir()) {
if (inputParams.has_type())
ythrow yexception() << "Input\\Type field is meaningless for directory processing";
m_strSourceType = "dir";
m_strInputFileName = inputParams.dir();
m_strDocDir = inputParams.dir();
if (!PathHelper::IsDir(m_strDocDir))
ythrow yexception() << "\"" << m_strDocDir << "\" isn't a directory";
} else {
if (inputParams.has_type()) {
switch (inputParams.type()) {
case TTextMinerConfig::TInputParams::no:
m_strSourceType = "no";
break;
case TTextMinerConfig::TInputParams::dpl:
m_strSourceType = "dpl";
break;
case TTextMinerConfig::TInputParams::arcview:
m_strSourceType = "arcview";
break;
case TTextMinerConfig::TInputParams::mapreduce:
m_strSourceType = "mapreduce";
break;
case TTextMinerConfig::TInputParams::tar:
if (m_strInputFileName.empty())
ythrow yexception() << "Please specify Input\\File field in configuration file in order to use .tar archive.";
m_strSourceType = "tar";
break;
case TTextMinerConfig::TInputParams::som:
if (m_strInputFileName.empty())
ythrow yexception() << "Please specify Input\\File field in configuration file in order to read SOM data.";
m_strSourceType = "som";
break;
case TTextMinerConfig::TInputParams::yarchive:
if (m_strInputFileName.empty())
ythrow yexception() << "Please specify Input\\File field in configuration file in order to read Yandex archive.";
m_strSourceType = "yarchive";
break;
default:
ythrow yexception() << "This type of input isn't supported";
}
} else
m_strSourceType = "no";
}
}
if (NULL == ParserOptions.Get())
ParserOptions.Reset(new CParserOptions);
ParserOptions->InitFromConfigObject(*Config.Get());
return true;
}
