qreal ExactPhraseScorer::phraseFreq(){
//Func - Returns the freqency of the phrase
//Pre - first != NULL
// last != NULL
// pq != NULL
// size of the PhraseQueue pq is 0
//Post - The frequency of the phrase has been returned
CND_PRECONDITION(first != NULL,"first is NULL");
CND_PRECONDITION(last != NULL,"last is NULL");
CND_PRECONDITION(pq != NULL,"pq is NULL");
CND_PRECONDITION(pq->size()==0,"pq is not empty");
//build pq from list
//Add the nodes of the list of PhrasePositions and store them
//into the PhraseQueue pq so it can used to build
//a list of sorted nodes
for (PhrasePositions* pp = first; pp != NULL; pp = pp->_next) {
//Read the first TermPosition of the current PhrasePositions pp
pp->firstPosition();
//Store the current PhrasePositions pp into the PhraseQueue pq
pq->put(pp);
}
//pqToList requires that first and last be NULL when it's called.
//This is done at the beginning of pqToList()
//In this case, the nodes of the linked list are referenced by pq (see
//above loop), so we can clear our pointers to the head and tail of the
//linked list without fear of leaking the nodes.
//rebuild list from pq
pqToList();
//Initialize freq at 0
int32_t freq = 0;
//find position with all terms
do {
//scan forward in first
while (first->position < last->position){
do{
if (!first->nextPosition()){
return (qreal)freq;
}
} while (first->position < last->position);
//Make the current first node the last node in the list
firstToLast();
}
//all equal: a match has been found
freq++;
} while (last->nextPosition());
return (qreal)freq;
}
float_t SloppyPhraseScorer::phraseFreq() {
//Func - Returns the freqency of the phrase
//Pre - first != NULL
// last != NULL
// pq != NULL
//Post - The frequency of the phrase has been returned
CND_PRECONDITION(first != NULL,"first is NULL");
CND_PRECONDITION(last != NULL,"last is NULL");
CND_PRECONDITION(pq != NULL,"pq is NULL");
//Clear the PhraseQueue pq;
pq->clear();
int32_t end = 0;
//declare iterator
PhrasePositions* pp = NULL;
// build pq from list
//Sort the list of PhrasePositions using pq
for (pp = first; pp != NULL; pp = pp->_next) {
//Read the first TermPosition of the current PhrasePositions pp
pp->firstPosition();
//Check if the position of the pp is bigger than end
if (pp->position > end){
end = pp->position;
}
//Store the current PhrasePositions pp into the PhraseQueue pp
pq->put(pp);
}
float_t freq = 0.0f;
bool done = false;
do {
//Pop a PhrasePositions pp from the PhraseQueue pp
pp = pq->pop();
//Get start position
int32_t start = pp->position;
//Get next position
int32_t next = pq->top()->position;
for (int32_t pos = start; pos <= next; pos = pp->position) {
//advance pp to min window
start = pos;
if (!pp->nextPosition()) {
//ran out of a term -- done
done = true;
break;
}
}
//Calculate matchLength
int32_t matchLength = end - start;
//Check if matchLength is smaller than slop
if (matchLength <= slop){
// penalize longer matches
freq += 1.0 / (matchLength + 1);
}
if (pp->position > end){
end = pp->position;
}
//restore pq
pq->put(pp);
}while (!done);
return freq;
}