void CopyWordCounts( WordCount& targetWordCount, WordCount& sourceWordCount, int weight )
{
for( wciter iwc=sourceWordCount.begin(); iwc!=sourceWordCount.end(); iwc++ )
{
if( targetWordCount.find( iwc->first ) == targetWordCount.end() )
targetWordCount[ iwc->first ] = 0;
targetWordCount[ iwc->first ] += weight * iwc->second;
}
}
void DumpPositionTable( PositionText& ptext )
{
for( ptiter ip=ptext.begin(); ip!=ptext.end(); ip++ )
{
cout << "DumpPositionTable for key value ='" << ip->first << "'" << endl;
WordCount* wc = ip->second;
for( wciter iwc=wc->begin(); iwc!=wc->end(); iwc++ )
{
cout << iwc->first << ": " << iwc->second << endl;
}
}
}
string GetRandWord( WordCount& wordCount, int accumulation )
{
WordCount wordTable;
int totalCount = 0;
// cout << "-----[ RANDOM TABLE ]-----" << endl;
int terminalWeight = 0;
if( accumulation > 5 )
{
int distance = accumulation - 5;
terminalWeight = distance * distance * distance;
}
for( wciter iwc=wordCount.begin(); iwc!=wordCount.end(); iwc++ )
{
totalCount += iwc->second * iwc->second;
if( iwc->first == BEGIN_KEY || iwc->first == END_KEY )
{
totalCount += terminalWeight;
}
wordTable[ iwc->first ] = totalCount;
// cout << totalCount << " " << iwc->first << endl;
}
int roll = rand() % totalCount;
// cout << " --> roll: " << roll << endl;
totalCount = 0;
string word;
// cout << "-----[ SAME TABLE? ]-----" << endl;
for( wciter iwc=wordTable.begin(); iwc!=wordTable.end(); iwc++ )
{
// cout << iwc->second << " " << iwc->first << endl;
if( roll <= iwc->second )
{
word = iwc->first;
break;
}
}
// cout << " --> word: " << word << endl;
return word;
}
string GetMaxWord( WordCount& wordCount )
{
string word;
int maxCount = 0;
for( wciter iwc=wordCount.begin(); iwc!=wordCount.end(); iwc++ )
{
// cout << iwc->first << ": " << iwc->second << endl;
if( iwc->second > maxCount )
{
word = iwc->first;
maxCount = iwc->second;
}
}
return word;
}
// impl by chenshuo
void sort_words_by_frequencies(const WordCount& counts)
{
typedef std::vector<std::pair<int, WordCount::const_iterator> > FreqList;
FreqList freq;
freq.reserve(counts.size());
for (WordCount::const_iterator it = counts.begin(); it != counts.end(); ++it)
{
freq.push_back(make_pair(it->second, it));
}
std::sort(freq.begin(), freq.end(), Greater());
for (FreqList::iterator itr = freq.begin(); itr!=freq.end(); ++itr)
{
std::cout << itr->first << '\t' << itr->second->first << '\n';
}
}