void TreeScorerBoundVariables::AddVariable( const lem::UCString & upper_name, const Solarix::Word_Form * wordform )
{
LEM_CHECKIT_Z( !upper_name.empty() );
LEM_CHECKIT_Z( wordform!=NULL );
bound_variables[ upper_name ] = wordform;
return;
}
const lem::CString lem::to_ascii( const lem::UCString &str, const CodeConverter *cp )
{
if( str.empty() )
return CString();
lem::CString ascii;
lem_unicode_to_ascii( ascii.ptr(), str.c_str(), cp ? cp : &lem::UI::get_UI().GetSessionCp() );
ascii.ptr()[str.length()] = 0;
ascii.calc_hash();
return ascii;
}
int Lemmatizator::MatchSuffix( const lem::UCString & suffix ) const
{
LEM_CHECKIT_Z( !suffix.empty() );
int id = -1;
std::map<lem::UCString,int>::const_iterator it = suffix2id.find(suffix);
if( it!=suffix2id.end() )
id = it->second;
return id;
}
lem::UCString Lemmatizator::GetSuffix( const lem::UCString & word ) const
{
LEM_CHECKIT_Z( !word.empty() );
lem::UCString uword(word);
uword.to_lower();
if( word.length()==7 && uword==L"~~end~~" )
return word;
else if( word.length()==9 && uword==L"~~begin~~" )
return word;
else if( word.length()<=model_suffix_len )
return uword;
else
return lem::UCString(L'~')+lem::right( uword, model_suffix_len );
}