0ct0pu5/fess
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Do not add fuzzy query for terms with double-quotes

Browse source 
By default, for single-phrase Term Queries, Fess adds fuzzy query
(with a low boost value) to the default query builder, causing results with
similar phrases to appear as well.
(For example, searching "plawright" without quotes will also display results
 with "playwright")

However, there are no mechanisms for the user to indicates that they only want
results with their specified term (as in, without fuzzy query results)

This patch allows users to ignore fuzzy results by surrounding the exact-search
terms with double quotes, or by using phrase search in Advanced search page
This commit is contained in:
hieu1.hoangtrung 2023-10-20 17:11:27 +07:00
parent 2bc92694bc
commit de2809286b
2 changed files with 14 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

7
src/main/java/org/codelibs/fess/query/TermQueryCommand.java
View file

@ -164,7 +164,12 @@ public class TermQueryCommand extends QueryCommand {
final BoolQueryBuilder boolQuery =
buildDefaultQueryBuilder(fessConfig, context, (f, b) -> buildMatchPhraseQuery(f, text).boost(b * boost));
final Integer fuzzyMinLength = fessConfig.getQueryBoostFuzzyMinLengthAsInteger();
if (fuzzyMinLength >= 0 && text.length() >= fuzzyMinLength) {
// Do not add fuzzy queries if the single-word query is surrounded by double quotes
final String exactWordMatchLookupRegex = ".*\"" + text + "\".*";
final boolean queryIsExactMatch = context.getQueryString().matches(exactWordMatchLookupRegex);
if (!queryIsExactMatch && fuzzyMinLength >= 0 && text.length() >= fuzzyMinLength) {
boolQuery.should(QueryBuilders.fuzzyQuery(fessConfig.getIndexFieldTitle(), text)
.boost(fessConfig.getQueryBoostFuzzyTitleAsDecimal().floatValue())
.prefixLength(fessConfig.getQueryBoostFuzzyTitlePrefixLengthAsInteger())

8
src/test/java/org/codelibs/fess/query/TermQueryCommandTest.java
View file

@ -83,6 +83,14 @@ public class TermQueryCommandTest extends UnitFessTestCase {
"{\"prefix\":{\"site\":{\"value\":\"aaa\",\"boost\":1.0}}}", //
"site:aaa");
// assertion for fuzzy search
assertQueryBuilder(BoolQueryBuilder.class,
"{\"bool\":{\"should\":[{\"match_phrase\":{\"title\":{\"query\":\"helloworld\",\"slop\":0,\"zero_terms_query\":\"NONE\",\"boost\":0.5}}},{\"match_phrase\":{\"content\":{\"query\":\"helloworld\",\"slop\":0,\"zero_terms_query\":\"NONE\",\"boost\":0.05}}},{\"fuzzy\":{\"title\":{\"value\":\"helloworld\",\"fuzziness\":\"AUTO\",\"prefix_length\":0,\"max_expansions\":10,\"transpositions\":true,\"boost\":0.01}}},{\"fuzzy\":{\"content\":{\"value\":\"helloworld\",\"fuzziness\":\"AUTO\",\"prefix_length\":0,\"max_expansions\":10,\"transpositions\":true,\"boost\":0.005}}}],\"adjust_pure_negative\":true,\"boost\":1.0}}",
"helloworld");
assertQueryBuilder(BoolQueryBuilder.class,
"{\"bool\":{\"should\":[{\"match_phrase\":{\"title\":{\"query\":\"helloworld\",\"slop\":0,\"zero_terms_query\":\"NONE\",\"boost\":0.5}}},{\"match_phrase\":{\"content\":{\"query\":\"helloworld\",\"slop\":0,\"zero_terms_query\":\"NONE\",\"boost\":0.05}}}],\"adjust_pure_negative\":true,\"boost\":1.0}}",
"\"helloworld\"");
assertQueryBuilder("{\"timestamp\":{\"order\":\"asc\"}}", "sort:timestamp");
assertQueryBuilder("{\"timestamp\":{\"order\":\"asc\"}}", "sort:timestamp.asc");
assertQueryBuilder("{\"timestamp\":{\"order\":\"desc\"}}", "sort:timestamp.desc");