Remove duplicated words and extra spaces from full text searchable fields

persistence/persistence_suite_test.go

@@ -32,7 +32,7 @@ func TestPersistence(t *testing.T) {
 
 var (
 	artistKraftwerk = model.Artist{ID: "2", Name: "Kraftwerk", AlbumCount: 1, FullText: "kraftwerk"}
-	artistBeatles   = model.Artist{ID: "3", Name: "The Beatles", AlbumCount: 2, FullText: "the beatles"}
+	artistBeatles   = model.Artist{ID: "3", Name: "The Beatles", AlbumCount: 2, FullText: "beatles the"}
 	testArtists     = model.Artists{
 		artistKraftwerk,
 		artistBeatles,
@@ -40,9 +40,9 @@ var (
 )
 
 var (
-	albumSgtPeppers    = model.Album{ID: "1", Name: "Sgt Peppers", Artist: "The Beatles", AlbumArtistID: "3", Genre: "Rock", CoverArtId: "1", CoverArtPath: P("/beatles/1/sgt/a day.mp3"), SongCount: 1, MaxYear: 1967, FullText: "sgt peppers the beatles"}
+	albumSgtPeppers    = model.Album{ID: "1", Name: "Sgt Peppers", Artist: "The Beatles", AlbumArtistID: "3", Genre: "Rock", CoverArtId: "1", CoverArtPath: P("/beatles/1/sgt/a day.mp3"), SongCount: 1, MaxYear: 1967, FullText: "beatles peppers sgt the"}
-	albumAbbeyRoad     = model.Album{ID: "2", Name: "Abbey Road", Artist: "The Beatles", AlbumArtistID: "3", Genre: "Rock", CoverArtId: "2", CoverArtPath: P("/beatles/1/come together.mp3"), SongCount: 1, MaxYear: 1969, FullText: "abbey road the beatles"}
+	albumAbbeyRoad     = model.Album{ID: "2", Name: "Abbey Road", Artist: "The Beatles", AlbumArtistID: "3", Genre: "Rock", CoverArtId: "2", CoverArtPath: P("/beatles/1/come together.mp3"), SongCount: 1, MaxYear: 1969, FullText: "abbey beatles road the"}
-	albumRadioactivity = model.Album{ID: "3", Name: "Radioactivity", Artist: "Kraftwerk", AlbumArtistID: "2", Genre: "Electronic", CoverArtId: "3", CoverArtPath: P("/kraft/radio/radio.mp3"), SongCount: 2, FullText: "radioactivity kraftwerk"}
+	albumRadioactivity = model.Album{ID: "3", Name: "Radioactivity", Artist: "Kraftwerk", AlbumArtistID: "2", Genre: "Electronic", CoverArtId: "3", CoverArtPath: P("/kraft/radio/radio.mp3"), SongCount: 2, FullText: "kraftwerk radioactivity"}
 	testAlbums         = model.Albums{
 		albumSgtPeppers,
 		albumAbbeyRoad,
@@ -51,9 +51,9 @@ var (
 )
 
 var (
-	songDayInALife    = model.MediaFile{ID: "1", Title: "A Day In A Life", ArtistID: "3", Artist: "The Beatles", AlbumID: "1", Album: "Sgt Peppers", Genre: "Rock", Path: P("/beatles/1/sgt/a day.mp3"), FullText: "a day in a life sgt peppers the beatles"}
+	songDayInALife    = model.MediaFile{ID: "1", Title: "A Day In A Life", ArtistID: "3", Artist: "The Beatles", AlbumID: "1", Album: "Sgt Peppers", Genre: "Rock", Path: P("/beatles/1/sgt/a day.mp3"), FullText: "a beatles day in life peppers sgt the"}
-	songComeTogether  = model.MediaFile{ID: "2", Title: "Come Together", ArtistID: "3", Artist: "The Beatles", AlbumID: "2", Album: "Abbey Road", Genre: "Rock", Path: P("/beatles/1/come together.mp3"), FullText: "come together abbey road the beatles"}
+	songComeTogether  = model.MediaFile{ID: "2", Title: "Come Together", ArtistID: "3", Artist: "The Beatles", AlbumID: "2", Album: "Abbey Road", Genre: "Rock", Path: P("/beatles/1/come together.mp3"), FullText: "abbey beatles come road the together"}
-	songRadioactivity = model.MediaFile{ID: "3", Title: "Radioactivity", ArtistID: "2", Artist: "Kraftwerk", AlbumID: "3", Album: "Radioactivity", Genre: "Electronic", Path: P("/kraft/radio/radio.mp3"), FullText: "radioactivity radioactivity kraftwerk"}
+	songRadioactivity = model.MediaFile{ID: "3", Title: "Radioactivity", ArtistID: "2", Artist: "Kraftwerk", AlbumID: "3", Album: "Radioactivity", Genre: "Electronic", Path: P("/kraft/radio/radio.mp3"), FullText: "kraftwerk radioactivity"}
 	songAntenna       = model.MediaFile{ID: "4", Title: "Antenna", ArtistID: "2", Artist: "Kraftwerk", AlbumID: "3", Genre: "Electronic", Path: P("/kraft/radio/antenna.mp3"), FullText: "antenna kraftwerk"}
 	testSongs         = model.MediaFiles{
 		songDayInALife,

persistence/sql_search.go

@@ -1,6 +1,7 @@
 package persistence
 
 import (
+	"sort"
 	"strings"
 
 	. "github.com/Masterminds/squirrel"
@@ -12,7 +13,16 @@ func (r sqlRepository) getFullText(text ...string) string {
 	for _, txt := range text {
 		sanitizedText.WriteString(strings.TrimSpace(sanitize.Accents(strings.ToLower(txt))) + " ")
 	}
-	return strings.TrimSpace(sanitizedText.String())
+	words := make(map[string]struct{})
+	for _, w := range strings.Fields(sanitizedText.String()) {
+		words[w] = struct{}{}
+	}
+	var fullText []string
+	for w := range words {
+		fullText = append(fullText, w)
+	}
+	sort.Strings(fullText)
+	return strings.Join(fullText, " ")
 }
 
 func (r sqlRepository) doSearch(q string, offset, size int, results interface{}, orderBys ...string) error {

persistence/sql_search_test.go

@@ -0,0 +1,28 @@
+package persistence
+
+import (
+	. "github.com/onsi/ginkgo"
+	. "github.com/onsi/gomega"
+)
+
+var _ = Describe("sqlRepository", func() {
+	var sqlRepository = &sqlRepository{}
+
+	Describe("getFullText", func() {
+		It("returns all lowercase chars", func() {
+			Expect(sqlRepository.getFullText("Some Text")).To(Equal("some text"))
+		})
+
+		It("removes accents", func() {
+			Expect(sqlRepository.getFullText("Quintão")).To(Equal("quintao"))
+		})
+
+		It("remove extra spaces", func() {
+			Expect(sqlRepository.getFullText(" some  text  ")).To(Equal("some text"))
+		})
+
+		It("remove duplicated words", func() {
+			Expect(sqlRepository.getFullText("legião urbana urbana legiÃo")).To(Equal("legiao urbana"))
+		})
+	})
+})