/// <summary>
/// Advance a PhrasePosition and update 'end', return false if exhausted </summary>
private bool AdvancePP(PhrasePositions pp)
{
if (!pp.NextPosition())
{
return(false);
}
if (pp.position > end)
{
end = pp.position;
}
return(true);
}
/// <summary> Score a candidate doc for all slop-valid position-combinations (matches)
/// encountered while traversing/hopping the PhrasePositions.
/// <br/> The score contribution of a match depends on the distance:
/// <br/> - highest score for distance=0 (exact match).
/// <br/> - score gets lower as distance gets higher.
/// <br/>Example: for query "a b"~2, a document "x a b a y" can be scored twice:
/// once for "a b" (distance=0), and once for "b a" (distance=2).
/// <br/>Possibly not all valid combinations are encountered, because for efficiency
/// we always propagate the least PhrasePosition. This allows to base on
/// PriorityQueue and move forward faster.
/// As result, for example, document "a b c b a"
/// would score differently for queries "a b c"~4 and "c b a"~4, although
/// they really are equivalent.
/// Similarly, for doc "a b c b a f g", query "c b"~2
/// would get same score as "g f"~2, although "c b"~2 could be matched twice.
/// We may want to fix this in the future (currently not, for performance reasons).
/// </summary>
protected internal override float PhraseFreq()
{
int end = InitPhrasePositions();
float freq = 0.0f;
bool done = (end < 0);
while (!done)
{
PhrasePositions pp = (PhrasePositions)pq.Pop();
int start = pp.position;
int next = ((PhrasePositions)pq.Top()).position;
bool tpsDiffer = true;
for (int pos = start; pos <= next || !tpsDiffer; pos = pp.position)
{
if (pos <= next && tpsDiffer)
{
start = pos; // advance pp to min window
}
if (!pp.NextPosition())
{
done = true; // ran out of a term -- done
break;
}
PhrasePositions pp2 = null;
tpsDiffer = !pp.repeats || (pp2 = TermPositionsDiffer(pp)) == null;
if (pp2 != null && pp2 != pp)
{
pp = Flip(pp, pp2); // flip pp to pp2
}
}
int matchLength = end - start;
if (matchLength <= slop)
{
freq += GetSimilarity().SloppyFreq(matchLength); // score match
}
if (pp.position > end)
{
end = pp.position;
}
pq.Put(pp); // restore pq
}
return(freq);
}
protected internal override float PhraseFreq()
{
pq.Clear();
int end = 0;
for (PhrasePositions pp = first; pp != null; pp = pp.next)
{
pp.FirstPosition();
if (pp.position > end)
{
end = pp.position;
}
pq.Put(pp); // build pq from list
}
float freq = 0.0f;
bool done = false;
do
{
PhrasePositions pp = (PhrasePositions)pq.Pop();
int start = pp.position;
int next = ((PhrasePositions)pq.Top()).position;
for (int pos = start; pos <= next; pos = pp.position)
{
start = pos; // advance pp to min window
if (!pp.NextPosition())
{
done = true; // ran out of a term -- done
break;
}
}
int matchLength = end - start;
if (matchLength <= slop)
{
freq += GetSimilarity().SloppyFreq(matchLength); // score match
}
if (pp.position > end)
{
end = pp.position;
}
pq.Put(pp); // restore pq
}while (!done);
return(freq);
}
/// <summary> Init PhrasePositions in place.
/// There is a one time initializatin for this scorer:
/// <br>- Put in repeats[] each pp that has another pp with same position in the doc.
/// <br>- Also mark each such pp by pp.repeats = true.
/// <br>Later can consult with repeats[] in termPositionsDiffer(pp), making that check efficient.
/// In particular, this allows to score queries with no repetiotions with no overhead due to this computation.
/// <br>- Example 1 - query with no repetitions: "ho my"~2
/// <br>- Example 2 - query with repetitions: "ho my my"~2
/// <br>- Example 3 - query with repetitions: "my ho my"~2
/// <br>Init per doc w/repeats in query, includes propagating some repeating pp's to avoid false phrase detection.
/// </summary>
/// <returns> end (max position), or -1 if any term ran out (i.e. done)
/// </returns>
/// <throws> IOException </throws>
private int InitPhrasePositions()
{
int end = 0;
// no repeats at all (most common case is also the simplest one)
if (checkedRepeats && repeats == null)
{
// build queue from list
pq.Clear();
for (PhrasePositions pp = first; pp != null; pp = pp.next)
{
pp.FirstPosition();
if (pp.position > end)
{
end = pp.position;
}
pq.Put(pp); // build pq from list
}
return(end);
}
// position the pp's
for (PhrasePositions pp = first; pp != null; pp = pp.next)
{
pp.FirstPosition();
}
// one time initializatin for this scorer
if (!checkedRepeats)
{
checkedRepeats = true;
// check for repeats
System.Collections.Hashtable m = null;
for (PhrasePositions pp = first; pp != null; pp = pp.next)
{
int tpPos = pp.position + pp.offset;
for (PhrasePositions pp2 = pp.next; pp2 != null; pp2 = pp2.next)
{
int tpPos2 = pp2.position + pp2.offset;
if (tpPos2 == tpPos)
{
if (m == null)
{
m = new System.Collections.Hashtable();
}
pp.repeats = true;
pp2.repeats = true;
m[pp] = null;
m[pp2] = null;
}
}
}
if (m != null)
{
repeats = (PhrasePositions[])(new System.Collections.ArrayList(m.Keys).ToArray(typeof(PhrasePositions)));
}
}
// with repeats must advance some repeating pp's so they all start with differing tp's
if (repeats != null)
{
// must propagate higher offsets first (otherwise might miss matches).
System.Array.Sort(repeats, new AnonymousClassComparator(this));
// now advance them
for (int i = 0; i < repeats.Length; i++)
{
PhrasePositions pp = repeats[i];
while (!TermPositionsDiffer(pp))
{
if (!pp.NextPosition())
{
return(-1); // ran out of a term -- done
}
}
}
}
// build queue from list
pq.Clear();
for (PhrasePositions pp = first; pp != null; pp = pp.next)
{
if (pp.position > end)
{
end = pp.position;
}
pq.Put(pp); // build pq from list
}
return(end);
}
