inline void
distinguish::segment(int width, int height, BYTE*& data)
{
int i = 0;
int j = 0;
int ct = 0;
int s_horizon[CHARACTER_NUM];
int e_horizon[CHARACTER_NUM];
vector<int> rows;
#ifdef DEBUG_ERROR
printf("width=%d,height=%d\n", width, height);
printf("c_BLACK=%d\n", c_BLACK);
#endif
for(i = 0; i < width; i++)
{
for(j = 0; j < height && data[j * width + i] < p_value; j++);
if(j == height)
{
rows.push_back(i);
}
#ifdef DEBUG_ERROR
printf("%d,rows.size(%d)=%d\n",data[j * width + i], i, j);
#endif
}
if(rows[rows.size() - 1] != width - 1)
{
rows.push_back(width - 1);
}
#ifdef DEBUG_ERROR
printf("rows.size()=%d\n", rows.size());
#endif
for(i = 1; i < int(rows.size()); i++)
{
if(rows[i] - rows[i - 1] >= MIN_WIDTH)
{
s_horizon[ct] = rows[i - 1];
e_horizon[ct] = rows[i];
subsegment(width, height, rows[i - 1], rows[i], data);
ct++;
}
#ifdef DEBUG_ERROR
printf("%d,%d\n", rows[i - 1], rows[i]);
#endif
}
#ifdef DEBUG_ERROR
printf("%d\n", ct);
#endif
}
bool CommonLanguageAnalyzer::analyze_synonym_impl(const izenelib::util::UString& inputString, SynonymOutputType& synonymOutput)
{
// cout << "[CommonLanguageAnalyzer::analyze_synonym_impl] ";
// inputString.displayStringValue(izenelib::util::UString::UTF_8);
// cout << endl;
bool retFoundSynonym = false;
const UString::CharT* pInput = inputString.c_str();
size_t length = inputString.length();
// ensure length
size_t ustring_buffer_size_ = term_ustring_buffer_limit_;
if (ustring_buffer_size_ < length+1)
{
ustring_buffer_size_ = length+1;
delete lowercase_ustring_buffer_;
lowercase_ustring_buffer_ = new UString::CharT[ustring_buffer_size_];
}
size_t string_buffer_size = term_string_buffer_limit_;
if (string_buffer_size < length*4)
{
string_buffer_size = length*4;
delete lowercase_string_buffer_;
lowercase_string_buffer_ = new char[string_buffer_size];
}
// to low case
UString::CharT* lowercaseTermUstr = lowercase_ustring_buffer_;
bool lowercaseIsDifferent = UString::toLowerString(pInput, length,
lowercase_ustring_buffer_, term_ustring_buffer_limit_);
if (lowercaseIsDifferent)
pInput = lowercaseTermUstr;
// convert input string to utf8 characters
char* chars = lowercase_string_buffer_;
vector<size_t> charOffs; charOffs.reserve(length);
size_t preCharLen = -1;
size_t curOff = 0;
for (size_t i = 0; i < length; i++)
{
curOff += (preCharLen+1);
charOffs[i] = curOff;
preCharLen = UString::convertString(UString::UTF_8, pInput+i, 1, chars+curOff, 4);
//cout << chars+curOff <<" "<<curOff<<" " << preCharLen <<endl;
}
// search synonym dict for input
izenelib::am::StrBasedVTrie strTrie(pSynonymContainer_->getData());
UString::CharT * synonymResultUstr = NULL;
size_t synonymResultUstrLen = 0;
size_t curIdx = 0;
size_t startIdx = 0;
size_t wordEndIdx;
VTrieNode endNode;
while (curIdx < length)
{
wordEndIdx = size_t(-1);
char* pch = chars+charOffs[curIdx];
strTrie.firstSearch(pch);
if (strTrie.completeSearch == false || strTrie.node->moreLong == false)
{
curIdx ++;
continue;
}
if (strTrie.exists())
{
wordEndIdx = curIdx;
strTrie.getCurrentNode(endNode);
}
//else // uncomment if minimum match
{
for (size_t j = curIdx + 1; j < length; j ++)
{
char* pnch = chars+charOffs[j];
strTrie.search(pnch);
if (strTrie.completeSearch == false)
break;
// matched a word
if (strTrie.exists())
{
wordEndIdx = j;
strTrie.getCurrentNode(endNode);
//break; // uncomment if minimum match
}
}
}
// if matched a synonym
if (wordEndIdx != size_t(-1))
{
retFoundSynonym = true;
// segment with out synonym
if (startIdx < curIdx)
{
std::vector<UString> segment;
UString subsegment(pInput+startIdx, curIdx-startIdx);
//subsegment.displayStringValue(izenelib::util::UString::UTF_8); cout << endl;///
segment.push_back(subsegment);
synonymOutput.push_back(segment);
}
// segment with synonyms
std::vector<UString> segment;
//UString(pInput+curIdx,wordEndIdx+1-curIdx).displayStringValue(izenelib::util::UString::UTF_8); cout <<" [has synonym] ";
pSynonymContainer_->setSynonym(pSynonymResult_, &endNode);
size_t cnt, idx = 0;
set<UString> synonymSet; // avoid duplication
do
{
cnt = pSynonymResult_->getSynonymCount(idx);
for (size_t off = 0; off < cnt; off++)
{
char * synonymResult = pSynonymResult_->getWord(idx, off);
if (synonymResult)
{
size_t synonymResultLen = strlen(synonymResult);
if (synonymResultLen <= term_ustring_buffer_limit_)
{
synonymResultUstr = synonym_ustring_buffer_;
synonymResultUstrLen = UString::toUcs2(synonymEncode_,
synonymResult, synonymResultLen, synonym_ustring_buffer_, term_ustring_buffer_limit_);
}
UString synonym(synonymResultUstr, synonymResultUstrLen);
if (synonymSet.find(synonym) != synonymSet.end())
continue;
synonymSet.insert(synonym);
segment.push_back(synonym);
}
}
idx++;
}
while (cnt > 0);
synonymOutput.push_back(segment);
curIdx = wordEndIdx+1;
startIdx = curIdx;
}
else
{
curIdx++;
}
}
if (!retFoundSynonym)
return false;
if (startIdx < curIdx)
{
std::vector<UString> segment;
UString subsegment(pInput+startIdx, curIdx-startIdx);
//subsegment.displayStringValue(izenelib::util::UString::UTF_8); cout << endl; ///
segment.push_back(subsegment);
synonymOutput.push_back(segment);
}
return true;
}
