fix go-gitea#4479: add fuzzy keyword search to title

coldrye-collaboration · Jun 27, 2024 · a390d75 · a390d75
1 parent d655ff1
commit a390d75
Show file tree

Hide file tree

Showing 4 changed files with 16 additions and 5 deletions.
diff --git a/modules/indexer/internal/bleve/query.go b/modules/indexer/internal/bleve/query.go
@@ -28,6 +28,14 @@ func MatchPhraseQuery(matchPhrase, field, analyzer string, fuzziness int) *query
 	return q
 }
 
+// FuzzyQuery generates a fuzzy query for the given phrase, field, and fuzziness
+func FuzzyQuery(matchPhrase, field string, fuzziness int) *query.FuzzyQuery {
+	q := bleve.NewFuzzyQuery(matchPhrase)
+	q.FieldVal = field
+	q.Fuzziness = fuzziness
+	return q
+}
+
 // BoolFieldQuery generates a bool field query for the given value and field
 func BoolFieldQuery(value bool, field string) *query.BoolFieldQuery {
 	q := bleve.NewBoolFieldQuery(value)

diff --git a/modules/indexer/internal/bleve/util.go b/modules/indexer/internal/bleve/util.go
@@ -50,12 +50,14 @@ func openIndexer(path string, latestVersion int) (bleve.Index, int, error) {
 
 func GuessFuzzinessByKeyword(s string) int {
 	// according to https://github.com/blevesearch/bleve/issues/1563, the supported max fuzziness is 2
-	// magic number 4 was chosen to determine the levenshtein distance per each character of a keyword
-	// BUT, when using CJK (eg: `갃갃갃` `啊啊啊`), it mismatches a lot.
+	// BUT, when using CJK (eg: `갃갃갃` `啊啊啊`), it mismatches a lot
+	// which we need to live with, as we need to support not just ASCII
+	// in case of code points >= 128 we will increase the fuzziness to 2
+	// the standard is 1
 	for _, r := range s {
 		if r >= 128 {
-			return 0
+			return 2
 		}
 	}
-	return min(2, len(s)/4)
+	return 1
 }
diff --git a/modules/indexer/issues/bleve/bleve.go b/modules/indexer/issues/bleve/bleve.go
@@ -162,7 +162,7 @@ func (b *Indexer) Search(ctx context.Context, options *internal.SearchOptions) (
 		}
 
 		queries = append(queries, bleve.NewDisjunctionQuery([]query.Query{
-			inner_bleve.MatchPhraseQuery(options.Keyword, "title", issueIndexerAnalyzer, fuzziness),
+			inner_bleve.FuzzyQuery(options.Keyword, "title", fuzziness),
 			inner_bleve.MatchPhraseQuery(options.Keyword, "content", issueIndexerAnalyzer, fuzziness),
 			inner_bleve.MatchPhraseQuery(options.Keyword, "comments", issueIndexerAnalyzer, fuzziness),
 		}...))

diff --git a/routers/web/repo/issue.go b/routers/web/repo/issue.go
@@ -2677,6 +2677,7 @@ func SearchIssues(ctx *context.Context) {
 		MilestoneIDs:        includedMilestones,
 		ProjectID:           projectID,
 		SortBy:              issue_indexer.SortByCreatedDesc,
+		IsFuzzyKeyword:      true,
 	}
 
 	if since != 0 {