/// <summary>
/// Merging constructor. Note that this just grabs seqnum from the first info.
/// </summary>
public WeightedPhraseInfo(ICollection <WeightedPhraseInfo> toMerge)
{
IEnumerator <Toffs>[] allToffs = new IEnumerator <Toffs> [toMerge.Count];
try
{
// Pretty much the same idea as merging FieldPhraseLists:
// Step 1. Sort by startOffset, endOffset
// While we are here merge the boosts and termInfos
using (IEnumerator <WeightedPhraseInfo> toMergeItr = toMerge.GetEnumerator())
{
if (!toMergeItr.MoveNext())
{
throw new ArgumentException("toMerge must contain at least one WeightedPhraseInfo.");
}
WeightedPhraseInfo first = toMergeItr.Current;
termsInfos = new List <TermInfo>();
seqnum = first.seqnum;
boost = first.boost;
allToffs[0] = first.termsOffsets.GetEnumerator();
int index = 1;
while (toMergeItr.MoveNext())
{
WeightedPhraseInfo info = toMergeItr.Current;
boost += info.boost;
termsInfos.AddRange(info.termsInfos);
allToffs[index++] = info.termsOffsets.GetEnumerator();
}
// Step 2. Walk the sorted list merging overlaps
using (MergedEnumerator <Toffs> itr = new MergedEnumerator <Toffs>(false, allToffs))
{
termsOffsets = new List <Toffs>();
if (!itr.MoveNext())
{
return;
}
Toffs work = itr.Current;
while (itr.MoveNext())
{
Toffs current = itr.Current;
if (current.StartOffset <= work.EndOffset)
{
work.EndOffset = Math.Max(work.EndOffset, current.EndOffset);
}
else
{
termsOffsets.Add(work);
work = current;
}
}
termsOffsets.Add(work);
}
}
}
finally
{
IOUtils.Dispose(allToffs);
}
}
public virtual void TestMergeOne()
{
Term t1 = new Term("foo", "a");
PrefixCodedTerms.Builder b1 = new PrefixCodedTerms.Builder();
b1.Add(t1);
PrefixCodedTerms pb1 = b1.Finish();
Term t2 = new Term("foo", "b");
PrefixCodedTerms.Builder b2 = new PrefixCodedTerms.Builder();
b2.Add(t2);
PrefixCodedTerms pb2 = b2.Finish();
IEnumerator <Term> merged = new MergedEnumerator <Term>(pb1.GetEnumerator(), pb2.GetEnumerator());
Assert.IsTrue(merged.MoveNext());
Assert.AreEqual(t1, merged.Current);
Assert.IsTrue(merged.MoveNext());
Assert.AreEqual(t2, merged.Current);
}
public virtual void TestMergeRandom()
{
PrefixCodedTerms[] pb = new PrefixCodedTerms[TestUtil.NextInt32(Random, 2, 10)];
JCG.SortedSet <Term> superSet = new JCG.SortedSet <Term>();
for (int i = 0; i < pb.Length; i++)
{
JCG.SortedSet <Term> terms = new JCG.SortedSet <Term>();
int nterms = TestUtil.NextInt32(Random, 0, 10000);
for (int j = 0; j < nterms; j++)
{
Term term = new Term(TestUtil.RandomUnicodeString(Random, 2), TestUtil.RandomUnicodeString(Random, 4));
terms.Add(term);
}
superSet.UnionWith(terms);
PrefixCodedTerms.Builder b = new PrefixCodedTerms.Builder();
foreach (Term @ref in terms)
{
b.Add(@ref);
}
pb[i] = b.Finish();
}
List <IEnumerator <Term> > subs = new List <IEnumerator <Term> >();
for (int i = 0; i < pb.Length; i++)
{
subs.Add(pb[i].GetEnumerator());
}
IEnumerator <Term> expected = superSet.GetEnumerator();
IEnumerator <Term> actual = new MergedEnumerator <Term>(subs.ToArray());
while (actual.MoveNext())
{
Assert.IsTrue(expected.MoveNext());
Assert.AreEqual(expected.Current, actual.Current);
}
Assert.IsFalse(expected.MoveNext());
}
/// <summary>
/// Merging constructor.
/// </summary>
/// <param name="toMerge"><see cref="FieldPhraseList"/>s to merge to build this one</param>
public FieldPhraseList(FieldPhraseList[] toMerge)
{
// Merge all overlapping WeightedPhraseInfos
// Step 1. Sort by startOffset, endOffset, and boost, in that order.
IEnumerator <WeightedPhraseInfo>[] allInfos = new IEnumerator <WeightedPhraseInfo> [toMerge.Length];
try
{
int index = 0;
foreach (FieldPhraseList fplToMerge in toMerge)
{
allInfos[index++] = fplToMerge.phraseList.GetEnumerator();
}
using MergedEnumerator <WeightedPhraseInfo> itr = new MergedEnumerator <WeightedPhraseInfo>(false, allInfos);
// Step 2. Walk the sorted list merging infos that overlap
phraseList = new List <WeightedPhraseInfo>();
if (!itr.MoveNext())
{
return;
}
List <WeightedPhraseInfo> work = new List <WeightedPhraseInfo>();
WeightedPhraseInfo first = itr.Current;
work.Add(first);
int workEndOffset = first.EndOffset;
while (itr.MoveNext())
{
WeightedPhraseInfo current = itr.Current;
if (current.StartOffset <= workEndOffset)
{
workEndOffset = Math.Max(workEndOffset, current.EndOffset);
work.Add(current);
}
else
{
if (work.Count == 1)
{
phraseList.Add(work[0]);
work[0] = current;
}
else
{
phraseList.Add(new WeightedPhraseInfo(work));
work.Clear();
work.Add(current);
}
workEndOffset = current.EndOffset;
}
}
if (work.Count == 1)
{
phraseList.Add(work[0]);
}
else
{
phraseList.Add(new WeightedPhraseInfo(work));
work.Clear();
}
}
finally
{
IOUtils.Dispose(allInfos);
}
}