TestScorerPerfpublic class TestScorerPerf extends TestCase
| Fields Summary |
|---|
| Random | r | | boolean | validate | | BitSet[] | sets | | IndexSearcher | s |
| Methods Summary |
|---|
| java.util.BitSet | addClause(org.apache.lucene.search.BooleanQuery bq, java.util.BitSet result)
BitSet rnd = sets[r.nextInt(sets.length)];
Query q = new ConstantScoreQuery(new BitSetFilter(rnd));
bq.add(q, BooleanClause.Occur.MUST);
if (validate) {
if (result==null) result = (BitSet)rnd.clone();
else result.and(rnd);
}
return result;
| | public void | createDummySearcher()
// Create a dummy index with nothing in it.
// This could possibly fail if Lucene starts checking for docid ranges...
RAMDirectory rd = new RAMDirectory();
IndexWriter iw = new IndexWriter(rd,new WhitespaceAnalyzer(), true);
iw.close();
s = new IndexSearcher(rd);
| | public void | createRandomTerms(int nDocs, int nTerms, double power, org.apache.lucene.store.Directory dir)
int[] freq = new int[nTerms];
for (int i=0; i<nTerms; i++) {
int f = (nTerms+1)-i; // make first terms less frequent
freq[i] = (int)Math.ceil(Math.pow(f,power));
}
IndexWriter iw = new IndexWriter(dir,new WhitespaceAnalyzer(), true);
iw.setMaxBufferedDocs(123);
for (int i=0; i<nDocs; i++) {
Document d = new Document();
for (int j=0; j<nTerms; j++) {
if (r.nextInt(freq[j]) == 0) {
d.add(new Field("f", Character.toString((char)j), Field.Store.NO, Field.Index.UN_TOKENIZED));
}
}
iw.addDocument(d);
}
iw.close();
| | public int | doConjunctions(int iter, int maxClauses)
int ret=0;
for (int i=0; i<iter; i++) {
int nClauses = r.nextInt(maxClauses-1)+2; // min 2 clauses
BooleanQuery bq = new BooleanQuery();
BitSet result=null;
for (int j=0; j<nClauses; j++) {
result = addClause(bq,result);
}
CountingHitCollector hc = validate ? new MatchingHitCollector(result)
: new CountingHitCollector();
s.search(bq, hc);
ret += hc.getSum();
if (validate) assertEquals(result.cardinality(), hc.getCount());
// System.out.println(hc.getCount());
}
return ret;
| | public int | doNestedConjunctions(int iter, int maxOuterClauses, int maxClauses)
int ret=0;
for (int i=0; i<iter; i++) {
int oClauses = r.nextInt(maxOuterClauses-1)+2;
BooleanQuery oq = new BooleanQuery();
BitSet result=null;
for (int o=0; o<oClauses; o++) {
int nClauses = r.nextInt(maxClauses-1)+2; // min 2 clauses
BooleanQuery bq = new BooleanQuery();
for (int j=0; j<nClauses; j++) {
result = addClause(bq,result);
}
oq.add(bq, BooleanClause.Occur.MUST);
} // outer
CountingHitCollector hc = validate ? new MatchingHitCollector(result)
: new CountingHitCollector();
s.search(oq, hc);
ret += hc.getSum();
if (validate) assertEquals(result.cardinality(), hc.getCount());
// System.out.println(hc.getCount());
}
return ret;
| | public int | doNestedTermConjunctions(org.apache.lucene.search.IndexSearcher s, int termsInIndex, int maxOuterClauses, int maxClauses, int iter)
int ret=0;
for (int i=0; i<iter; i++) {
int oClauses = r.nextInt(maxOuterClauses-1)+2;
BooleanQuery oq = new BooleanQuery();
for (int o=0; o<oClauses; o++) {
int nClauses = r.nextInt(maxClauses-1)+2; // min 2 clauses
BooleanQuery bq = new BooleanQuery();
BitSet terms = new BitSet(termsInIndex);
for (int j=0; j<nClauses; j++) {
int tnum;
// don't pick same clause twice
do {tnum = r.nextInt(termsInIndex);} while (terms.get(tnum));
Query tq = new TermQuery(new Term("f",Character.toString((char)tnum)));
bq.add(tq, BooleanClause.Occur.MUST);
} // inner
oq.add(bq, BooleanClause.Occur.MUST);
} // outer
CountingHitCollector hc = new CountingHitCollector();
s.search(oq, hc);
ret += hc.getSum();
}
return ret;
| | public int | doSloppyPhrase(org.apache.lucene.search.IndexSearcher s, int termsInIndex, int maxClauses, int iter)
int ret=0;
for (int i=0; i<iter; i++) {
int nClauses = r.nextInt(maxClauses-1)+2; // min 2 clauses
PhraseQuery q = new PhraseQuery();
for (int j=0; j<nClauses; j++) {
int tnum = r.nextInt(termsInIndex);
q.add(new Term("f",Character.toString((char)tnum)), j);
}
q.setSlop(termsInIndex); // this could be random too
CountingHitCollector hc = new CountingHitCollector();
s.search(q, hc);
ret += hc.getSum();
}
return ret;
| | public int | doTermConjunctions(org.apache.lucene.search.IndexSearcher s, int termsInIndex, int maxClauses, int iter)
int ret=0;
for (int i=0; i<iter; i++) {
int nClauses = r.nextInt(maxClauses-1)+2; // min 2 clauses
BooleanQuery bq = new BooleanQuery();
BitSet terms = new BitSet(termsInIndex);
for (int j=0; j<nClauses; j++) {
int tnum;
// don't pick same clause twice
do {tnum = r.nextInt(termsInIndex);} while (terms.get(tnum));
Query tq = new TermQuery(new Term("f",Character.toString((char)tnum)));
bq.add(tq, BooleanClause.Occur.MUST);
}
CountingHitCollector hc = new CountingHitCollector();
s.search(bq, hc);
ret += hc.getSum();
}
return ret;
| | public java.util.BitSet | randBitSet(int sz, int numBitsToSet)
BitSet set = new BitSet(sz);
for (int i=0; i<numBitsToSet; i++) {
set.set(r.nextInt(sz));
}
return set;
| | public java.util.BitSet[] | randBitSets(int numSets, int setSize)
BitSet[] sets = new BitSet[numSets];
for (int i=0; i<sets.length; i++) {
sets[i] = randBitSet(setSize, r.nextInt(setSize));
}
return sets;
| | public void | testConjunctions()
// test many small sets... the bugs will be found on boundary conditions
createDummySearcher();
validate=true;
sets=randBitSets(1000,10);
doConjunctions(10000,5);
doNestedConjunctions(10000,3,3);
s.close();
|
|
