@Override
public final int getDocCount(String field) throws IOException {
final Terms terms = terms(field);
if (terms == null) {
return 0;
}
return terms.getDocCount();
}
private void verifyCount(IndexReader ir) throws Exception {
Fields fields = MultiFields.getFields(ir);
for (String field : fields) {
Terms terms = fields.terms(field);
if (terms == null) {
continue;
}
int docCount = terms.getDocCount();
FixedBitSet visited = new FixedBitSet(ir.maxDoc());
TermsEnum te = terms.iterator();
while (te.next() != null) {
PostingsEnum de = TestUtil.docs(random(), te, null, PostingsEnum.NONE);
while (de.nextDoc() != DocIdSetIterator.NO_MORE_DOCS) {
visited.set(de.docID());
}
}
assertEquals(visited.cardinality(), docCount);
}
}
public static TermsEnum filter(
TermsEnum toFilter, Terms terms, AtomicReader reader, Settings settings)
throws IOException {
int docCount = terms.getDocCount();
if (docCount == -1) {
docCount = reader.maxDoc();
}
final double minFrequency = settings.getAsDouble("min", 0d);
final double maxFrequency = settings.getAsDouble("max", docCount + 1d);
final double minSegmentSize = settings.getAsInt("min_segment_size", 0);
if (minSegmentSize < docCount) {
final int minFreq =
minFrequency >= 1.0 ? (int) minFrequency : (int) (docCount * minFrequency);
final int maxFreq =
maxFrequency >= 1.0 ? (int) maxFrequency : (int) (docCount * maxFrequency);
assert minFreq < maxFreq;
return new FrequencyFilter(toFilter, minFreq, maxFreq);
}
return toFilter;
}
