[Feature] add precise search type for Elastic Search (#12869)

* feat: add type query parameters for specifying precise search

* feat: add select dropdown in search box

Co-authored-by: Lauris BH <lauris@nix.lv>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>

6 files changed, 35 insertions, 15 deletions

diff --git a/modules/indexer/code/bleve.go b/modules/indexer/code/bleve.go
index b0822ad222..826efde4c1 100644
--- a/modules/indexer/code/bleve.go
+++ b/modules/indexer/code/bleve.go
@@ -280,12 +280,23 @@ func (b *BleveIndexer) Delete(repoID int64) error {
 
 // Search searches for files in the specified repo.
 // Returns the matching file-paths
-func (b *BleveIndexer) Search(repoIDs []int64, language, keyword string, page, pageSize int) (int64, []*SearchResult, []*SearchResultLanguages, error) {
-	phraseQuery := bleve.NewMatchPhraseQuery(keyword)
-	phraseQuery.FieldVal = "Content"
-	phraseQuery.Analyzer = repoIndexerAnalyzer
+func (b *BleveIndexer) Search(repoIDs []int64, language, keyword string, page, pageSize int, isMatch bool) (int64, []*SearchResult, []*SearchResultLanguages, error) {
+	var (
+		indexerQuery query.Query
+		keywordQuery query.Query
+	)
+
+	if isMatch {
+		prefixQuery := bleve.NewPrefixQuery(keyword)
+		prefixQuery.FieldVal = "Content"
+		keywordQuery = prefixQuery
+	} else {
+		phraseQuery := bleve.NewMatchPhraseQuery(keyword)
+		phraseQuery.FieldVal = "Content"
+		phraseQuery.Analyzer = repoIndexerAnalyzer
+		keywordQuery = phraseQuery
+	}
 
-	var indexerQuery query.Query
 	if len(repoIDs) > 0 {
 		var repoQueries = make([]query.Query, 0, len(repoIDs))
 		for _, repoID := range repoIDs {
@@ -294,10 +305,10 @@ func (b *BleveIndexer) Search(repoIDs []int64, language, keyword string, page, p
 
 		indexerQuery = bleve.NewConjunctionQuery(
 			bleve.NewDisjunctionQuery(repoQueries...),
-			phraseQuery,
+			keywordQuery,
 		)
 	} else {
-		indexerQuery = phraseQuery
+		indexerQuery = keywordQuery
 	}
 
 	// Save for reuse without language filter
diff --git a/modules/indexer/code/elastic_search.go b/modules/indexer/code/elastic_search.go
index 0f61c4e592..f81dbb34d4 100644
--- a/modules/indexer/code/elastic_search.go
+++ b/modules/indexer/code/elastic_search.go
@@ -27,6 +27,10 @@ import (
 
 const (
 	esRepoIndexerLatestVersion = 1
+	// multi-match-types, currently only 2 types are used
+	// Reference: https://www.elastic.co/guide/en/elasticsearch/reference/7.0/query-dsl-multi-match-query.html#multi-match-types
+	esMultiMatchTypeBestFields   = "best_fields"
+	esMultiMatchTypePhrasePrefix = "phrase_prefix"
 )
 
 var (
@@ -330,8 +334,13 @@ func extractAggs(searchResult *elastic.SearchResult) []*SearchResultLanguages {
 }
 
 // Search searches for codes and language stats by given conditions.
-func (b *ElasticSearchIndexer) Search(repoIDs []int64, language, keyword string, page, pageSize int) (int64, []*SearchResult, []*SearchResultLanguages, error) {
-	kwQuery := elastic.NewMultiMatchQuery(keyword, "content")
+func (b *ElasticSearchIndexer) Search(repoIDs []int64, language, keyword string, page, pageSize int, isMatch bool) (int64, []*SearchResult, []*SearchResultLanguages, error) {
+	searchType := esMultiMatchTypeBestFields
+	if isMatch {
+		searchType = esMultiMatchTypePhrasePrefix
+	}
+
+	kwQuery := elastic.NewMultiMatchQuery(keyword, "content").Type(searchType)
 	query := elastic.NewBoolQuery()
 	query = query.Must(kwQuery)
 	if len(repoIDs) > 0 {
diff --git a/modules/indexer/code/indexer.go b/modules/indexer/code/indexer.go
index 35c298a548..a7d78e9fdc 100644
--- a/modules/indexer/code/indexer.go
+++ b/modules/indexer/code/indexer.go
@@ -43,7 +43,7 @@ type SearchResultLanguages struct {
 type Indexer interface {
 	Index(repo *models.Repository, sha string, changes *repoChanges) error
 	Delete(repoID int64) error
-	Search(repoIDs []int64, language, keyword string, page, pageSize int) (int64, []*SearchResult, []*SearchResultLanguages, error)
+	Search(repoIDs []int64, language, keyword string, page, pageSize int, isMatch bool) (int64, []*SearchResult, []*SearchResultLanguages, error)
 	Close()
 }
 
diff --git a/modules/indexer/code/indexer_test.go b/modules/indexer/code/indexer_test.go
index 0b4851a48a..8fcb7a0e8a 100644
--- a/modules/indexer/code/indexer_test.go
+++ b/modules/indexer/code/indexer_test.go
@@ -64,7 +64,7 @@ func testIndexer(name string, t *testing.T, indexer Indexer) {
 
 		for _, kw := range keywords {
 			t.Run(kw.Keyword, func(t *testing.T) {
-				total, res, langs, err := indexer.Search(kw.RepoIDs, "", kw.Keyword, 1, 10)
+				total, res, langs, err := indexer.Search(kw.RepoIDs, "", kw.Keyword, 1, 10, false)
 				assert.NoError(t, err)
 				assert.EqualValues(t, len(kw.IDs), total)
 				assert.EqualValues(t, kw.Langs, len(langs))
diff --git a/modules/indexer/code/search.go b/modules/indexer/code/search.go
index 29ed416541..51b7c9427d 100644
--- a/modules/indexer/code/search.go
+++ b/modules/indexer/code/search.go
@@ -106,12 +106,12 @@ func searchResult(result *SearchResult, startIndex, endIndex int) (*Result, erro
 }
 
 // PerformSearch perform a search on a repository
-func PerformSearch(repoIDs []int64, language, keyword string, page, pageSize int) (int, []*Result, []*SearchResultLanguages, error) {
+func PerformSearch(repoIDs []int64, language, keyword string, page, pageSize int, isMatch bool) (int, []*Result, []*SearchResultLanguages, error) {
 	if len(keyword) == 0 {
 		return 0, nil, nil, nil
 	}
 
-	total, results, resultLanguages, err := indexer.Search(repoIDs, language, keyword, page, pageSize)
+	total, results, resultLanguages, err := indexer.Search(repoIDs, language, keyword, page, pageSize, isMatch)
 	if err != nil {
 		return 0, nil, nil, err
 	}
diff --git a/modules/indexer/code/wrapped.go b/modules/indexer/code/wrapped.go
index d839544874..5b19f9c625 100644
--- a/modules/indexer/code/wrapped.go
+++ b/modules/indexer/code/wrapped.go
@@ -73,12 +73,12 @@ func (w *wrappedIndexer) Delete(repoID int64) error {
 	return indexer.Delete(repoID)
 }
 
-func (w *wrappedIndexer) Search(repoIDs []int64, language, keyword string, page, pageSize int) (int64, []*SearchResult, []*SearchResultLanguages, error) {
+func (w *wrappedIndexer) Search(repoIDs []int64, language, keyword string, page, pageSize int, isMatch bool) (int64, []*SearchResult, []*SearchResultLanguages, error) {
 	indexer, err := w.get()
 	if err != nil {
 		return 0, nil, nil, err
 	}
-	return indexer.Search(repoIDs, language, keyword, page, pageSize)
+	return indexer.Search(repoIDs, language, keyword, page, pageSize, isMatch)
 
 }