/**
* Get all the terms in the index with low edit distance from the supplied term
*
* @param reader the index
* @param luceneName the field to search in
* @param searchTerms search terms
* @param maxEdits maximum edit distance (Levenshtein algorithm) for matches (i.e. lower is more
* similar)
* @return the set of terms in the index that are close to our search term
* @throws BooleanQuery.TooManyClauses if the expansion resulted in too many terms
*/
public static Set<String> getMatchingTermsFromIndex(
IndexReader reader, String luceneName, Collection<String> searchTerms, int maxEdits) {
boolean doFuzzy = true;
if (maxEdits == 0) {
// Exact match; don't use fuzzy query (slow)
Set<String> result = new HashSet<>();
try {
for (String term : searchTerms) {
if (reader.docFreq(new Term(luceneName, term)) > 0) result.add(term);
}
} catch (IOException e) {
throw new RuntimeException(e);
}
return result;
}
BooleanQuery q = new BooleanQuery();
for (String s : searchTerms) {
FuzzyQuery fq = new FuzzyQuery(new Term(luceneName, s), maxEdits);
q.add(fq, Occur.SHOULD);
}
try {
Query rewritten = q.rewrite(reader);
WeightedTerm[] wts = QueryTermExtractor.getTerms(rewritten);
Set<String> terms = new HashSet<>();
for (WeightedTerm wt : wts) {
if (doFuzzy || searchTerms.contains(wt.getTerm())) {
terms.add(wt.getTerm());
}
}
return terms;
} catch (IOException e) {
throw new RuntimeException(e);
}
}
