/** * @param reader * @param numTerms * @param field * @return TermStats[] ordered by terms with highest docFreq first. * @throws Exception */ public static TermStats[] getHighFreqTerms(IndexReader reader, int numTerms, String[] fieldNames) throws Exception { TermStatsQueue tiq = null; TermsEnum te = null; if (fieldNames != null) { Fields fields = MultiFields.getFields(reader); if (fields == null) { LOG.info("Index with no fields - probably empty or corrupted"); return EMPTY_STATS; } tiq = new TermStatsQueue(numTerms); for (String field : fieldNames) { Terms terms = fields.terms(field); if (terms != null) { te = terms.iterator(te); fillQueue(te, tiq, field); } } } else { Fields fields = MultiFields.getFields(reader); if (fields == null) { LOG.info("Index with no fields - probably empty or corrupted"); return EMPTY_STATS; } tiq = new TermStatsQueue(numTerms); Iterator<String> fieldIterator = fields.iterator(); while (fieldIterator.hasNext()) { String field = fieldIterator.next(); Terms terms = fields.terms(field); if (terms != null) { te = terms.iterator(te); fillQueue(te, tiq, field); } } } TermStats[] result = new TermStats[tiq.size()]; // we want highest first so we read the queue and populate the array // starting at the end and work backwards int count = tiq.size() - 1; while (tiq.size() != 0) { result[count] = tiq.pop(); count--; } return result; }
public static void fillQueue(TermsEnum termsEnum, TermStatsQueue tiq, String field) throws Exception { BytesRef term; while ((term = termsEnum.next()) != null) { BytesRef r = new BytesRef(); r.copyBytes(term); tiq.insertWithOverflow(new TermStats(field, r, termsEnum.docFreq())); } }