unsigned CShuangpinSegmentor::_clear (unsigned from)
{
unsigned i, j;
_locateSegment (from, i, j);
std::string new_pystr = m_pystr.substr (i, from-i);
m_pystr.resize (i);
m_nAlpha = _getNumberOfNonAlpha();
m_segs.erase (m_segs.begin()+j, m_segs.end());
if (m_nLastValidPos + 1 >= from) {
m_hasInvalid = false;
}
m_updatedFrom = from;
for (std::string::const_iterator it = new_pystr.begin();
it!= new_pystr.end(); ++it) {
unsigned tmp = _push ((*it) & 0x7f);
if (tmp < m_updatedFrom) m_updatedFrom = tmp;
}
return m_updatedFrom;
}
unsigned CShuangpinSegmentor::deleteAt (unsigned idx, bool backward)
{
unsigned pyIdx, segIdx;
if (!backward) idx += 1;
_locateSegment (idx, pyIdx, segIdx);
m_inputBuf.erase (idx, 1);
m_pystr.erase (idx, 1);
std::string new_pystr = m_pystr.substr (pyIdx);
m_pystr.resize (pyIdx);
TSegmentVec tmp_segs (m_segs.begin()+segIdx+1, m_segs.end());
m_segs.erase (m_segs.begin()+segIdx, m_segs.end());
if (m_nLastValidPos + 1 < idx) {
//del invalid ch, and do not effect current status.
m_pystr.insert(idx, new_pystr);
m_segs.insert (m_segs.end(), tmp_segs.begin(), tmp_segs.end());
return m_inputBuf.size() -1;
} else {
m_hasInvalid = false;
m_nAlpha = _getNumberOfNonAlpha();
}
m_updatedFrom = UINT_MAX;
std::string::const_iterator it = new_pystr.begin();
for (; it!= new_pystr.end(); ++it) {
unsigned tmp = _push ((*it) & 0x7f);
if (tmp < m_updatedFrom) m_updatedFrom = tmp;
}
return m_updatedFrom;
}
unsigned CHunpinSegmentor::_clear (unsigned from)
{
unsigned i, j;
_locateSegment (from, i, j);
std::string new_pystr = m_pystr.substr (i, from-i);
m_pystr.resize (i);
m_segs.erase (m_segs.begin()+j, m_segs.end());
m_updatedFrom = _updateWith (new_pystr, from);
return m_updatedFrom;
}
unsigned CHunpinSegmentor::insertAt (unsigned idx, unsigned ch)
{
unsigned i, j;
_locateSegment (idx, i, j);
m_inputBuf.insert (idx, 1, ch);
m_pystr.insert (idx, 1, ch);
std::string new_pystr = m_pystr.substr (i);
m_pystr.resize (i);
m_segs.erase (m_segs.begin()+j, m_segs.end());
m_updatedFrom = _updateWith (new_pystr);
return m_updatedFrom;
}
unsigned CHunpinSegmentor::deleteAt (unsigned idx, bool backward)
{
unsigned i, j;
if (!backward) idx += 1;
_locateSegment (idx, i, j);
m_inputBuf.erase (idx, 1);
m_pystr.erase (idx, 1);
std::string new_pystr = m_pystr.substr (i);
m_pystr.resize (i);
m_segs.erase (m_segs.begin()+j, m_segs.end());
m_updatedFrom = _updateWith (new_pystr);
return m_updatedFrom;
}
unsigned CShuangpinSegmentor::insertAt (unsigned idx, unsigned ch)
{
unsigned pyIdx, segIdx;
_locateSegment (idx, pyIdx, segIdx);
m_inputBuf.insert (idx, 1, ch);
m_pystr.insert (idx, 1, ch);
std::string new_pystr = m_pystr.substr (pyIdx);
m_pystr.resize (pyIdx);
m_segs.erase (m_segs.begin()+segIdx, m_segs.end());
if (m_nLastValidPos == idx) {
m_hasInvalid = false;
} else if (m_nLastValidPos + 1 == idx) {
m_hasInvalid = false;
int nSize = m_pystr.size();
if (islower(m_pystr[nSize-1])) {
m_nLastValidPos = idx - 1;
new_pystr.insert((size_t)0, 1, m_pystr[nSize-1]);
m_pystr.erase(nSize-1, 1);
m_segs.erase (m_segs.begin()+segIdx-1);
}
} else if (m_nLastValidPos + 1 > idx) {
m_hasInvalid = false;
m_nLastValidPos = idx;
}
m_nAlpha = _getNumberOfNonAlpha();
m_updatedFrom = UINT_MAX;
std::string::const_iterator it = new_pystr.begin();
for (; it!= new_pystr.end(); ++it) {
unsigned tmp = _push ((*it) & 0x7f);
if (tmp < m_updatedFrom) m_updatedFrom = tmp;
}
return m_updatedFrom;
}
