virtual bool cut(Unicode::const_iterator begin, Unicode::const_iterator end, vector<Unicode>& res) const
{
vector<Unicode> words;
words.reserve(end - begin);
if(!_mpSeg.cut(begin, end, words))
{
Rcout<<"mpSeg cutDAG failed."<<std::endl;
return false;
}
vector<Unicode> hmmRes;
hmmRes.reserve(end - begin);
Unicode piece;
piece.reserve(end - begin);
for (size_t i = 0, j = 0; i < words.size(); i++)
{
//if mp get a word, it's ok, put it into result
if (1 != words[i].size() || (words[i].size() == 1 && _mpSeg.isUserDictSingleChineseWord(words[i][0])))
{
res.push_back(words[i]);
continue;
}
// if mp get a single one and it is not in userdict, collect it in sequence
j = i;
while (j < words.size() && 1 == words[j].size() && !_mpSeg.isUserDictSingleChineseWord(words[j][0]))
{
piece.push_back(words[j][0]);
j++;
}
// cut the sequence with hmm
if (!_hmmSeg.cut(piece.begin(), piece.end(), hmmRes))
{
Rcout<<"_hmmSeg cut failed."<<std::endl;
return false;
}
//put hmm result to result
for (size_t k = 0; k < hmmRes.size(); k++)
{
res.push_back(hmmRes[k]);
}
//clear tmp vars
piece.clear();
hmmRes.clear();
//let i jump over this piece
i = j - 1;
}
return true;
}
inline bool DecodeRunesInString(const char* s, size_t len, Unicode& unicode) {
unicode.clear();
RuneStrArray runes;
if (!DecodeRunesInString(s, len, runes)) {
return false;
}
unicode.reserve(runes.size());
for (size_t i = 0; i < runes.size(); i++) {
unicode.push_back(runes[i].rune);
}
return true;
}
bool MixSegment::cut(Unicode::const_iterator begin, Unicode::const_iterator end, vector<string>& res)const
{
if(!_getInitFlag())
{
LogError("not inited.");
return false;
}
if(begin == end)
{
return false;
}
vector<TrieNodeInfo> infos;
if(!_mpSeg.cut(begin, end, infos))
{
LogError("mpSeg cutDAG failed.");
return false;
}
Unicode unico;
vector<Unicode> hmmRes;
string tmp;
for(uint i= 0; i < infos.size(); i++)
{
TransCode::encode(infos[i].word,tmp);
if(1 == infos[i].word.size())
{
unico.push_back(infos[i].word[0]);
}
else
{
if(!unico.empty())
{
hmmRes.clear();
if(!_hmmSeg.cut(unico.begin(), unico.end(), hmmRes))
{
LogError("_hmmSeg cut failed.");
return false;
}
for(uint j = 0; j < hmmRes.size(); j++)
{
TransCode::encode(hmmRes[j], tmp);
res.push_back(tmp);
}
}
unico.clear();
TransCode::encode(infos[i].word, tmp);
res.push_back(tmp);
}
}
if(!unico.empty())
{
hmmRes.clear();
if(!_hmmSeg.cut(unico.begin(), unico.end(), hmmRes))
{
LogError("_hmmSeg cut failed.");
return false;
}
for(uint j = 0; j < hmmRes.size(); j++)
{
TransCode::encode(hmmRes[j], tmp);
res.push_back(tmp);
}
}
return true;
}
