From e858a34f3e50bc977e16ce943db0ec5484d02ce4 Mon Sep 17 00:00:00 2001
From: Michael Mayer <michael@photoprism.app>
Date: Mon, 3 Jan 2022 13:30:34 +0100
Subject: [PATCH] Search: Return empty result if keywords contain stopwords
 only #1859

---
 internal/search/geojson.go      |  4 ++++
 internal/search/photos.go       |  4 ++++
 pkg/txt/resources/stopwords.txt |  5 -----
 pkg/txt/stopwords.go            |  5 -----
 pkg/txt/words.go                | 23 +++++++++++++++++++++++
 pkg/txt/words_test.go           | 33 +++++++++++++++++++++++++++++++++
 6 files changed, 64 insertions(+), 10 deletions(-)

diff --git a/internal/search/geojson.go b/internal/search/geojson.go
index f60f77d9b..c5c7b1d23 100644
--- a/internal/search/geojson.go
+++ b/internal/search/geojson.go
@@ -132,6 +132,10 @@ func Geo(f form.SearchGeo) (results GeoResults, err error) {
 
 	// Search for one or more keywords?
 	if f.Keywords != "" {
+		if txt.StopwordsOnly(f.Keywords) {
+			return GeoResults{}, fmt.Errorf("keywords contain stopwords only")
+		}
+
 		for _, where := range LikeAnyKeyword("k.keyword", f.Keywords) {
 			s = s.Where("photos.id IN (SELECT pk.photo_id FROM keywords k JOIN photos_keywords pk ON k.id = pk.keyword_id WHERE (?))", gorm.Expr(where))
 		}
diff --git a/internal/search/photos.go b/internal/search/photos.go
index 4de1fc525..c454266b1 100644
--- a/internal/search/photos.go
+++ b/internal/search/photos.go
@@ -226,6 +226,10 @@ func Photos(f form.SearchPhotos) (results PhotoResults, count int, err error) {
 
 	// Search for one or more keywords?
 	if f.Keywords != "" {
+		if txt.StopwordsOnly(f.Keywords) {
+			return PhotoResults{}, 0, fmt.Errorf("keywords contain stopwords only")
+		}
+
 		for _, where := range LikeAnyKeyword("k.keyword", f.Keywords) {
 			s = s.Where("photos.id IN (SELECT pk.photo_id FROM keywords k JOIN photos_keywords pk ON k.id = pk.keyword_id WHERE (?))", gorm.Expr(where))
 		}
diff --git a/pkg/txt/resources/stopwords.txt b/pkg/txt/resources/stopwords.txt
index 6d860a105..20f87f7d9 100644
--- a/pkg/txt/resources/stopwords.txt
+++ b/pkg/txt/resources/stopwords.txt
@@ -13,8 +13,6 @@ handy
 tumblr
 bilder
 bild
-films
-filme
 foto
 fotos
 flickr
@@ -1958,9 +1956,6 @@ mais
 malgre
 malgré
 maximale
-meme
-memes
-merci
 mes
 mien
 mienne
diff --git a/pkg/txt/stopwords.go b/pkg/txt/stopwords.go
index 11ecd7197..083d12a34 100644
--- a/pkg/txt/stopwords.go
+++ b/pkg/txt/stopwords.go
@@ -18,8 +18,6 @@ var StopWords = map[string]bool{
 	"tumblr":           true,
 	"bilder":           true,
 	"bild":             true,
-	"films":            true,
-	"filme":            true,
 	"foto":             true,
 	"fotos":            true,
 	"flickr":           true,
@@ -1963,9 +1961,6 @@ var StopWords = map[string]bool{
 	"malgre":           true,
 	"malgré":           true,
 	"maximale":         true,
-	"meme":             true,
-	"memes":            true,
-	"merci":            true,
 	"mes":              true,
 	"mien":             true,
 	"mienne":           true,
diff --git a/pkg/txt/words.go b/pkg/txt/words.go
index 11988b047..544a72099 100644
--- a/pkg/txt/words.go
+++ b/pkg/txt/words.go
@@ -202,3 +202,26 @@ func UniqueKeywords(s string) (results []string) {
 func SortCaseInsensitive(words []string) {
 	sort.Slice(words, func(i, j int) bool { return strings.ToLower(words[i]) < strings.ToLower(words[j]) })
 }
+
+// StopwordsOnly tests if the string contains stopwords only.
+func StopwordsOnly(s string) bool {
+	s = strings.TrimSpace(s)
+
+	if s == "" {
+		return false
+	}
+
+	for _, w := range Words(s) {
+		w = strings.ToLower(w)
+
+		if UnknownWord(w) {
+			continue
+		}
+
+		if _, ok := StopWords[w]; ok == false {
+			return false
+		}
+	}
+
+	return true
+}
diff --git a/pkg/txt/words_test.go b/pkg/txt/words_test.go
index 1a3539381..96ba52b58 100644
--- a/pkg/txt/words_test.go
+++ b/pkg/txt/words_test.go
@@ -239,3 +239,36 @@ func TestRemoveFromWords(t *testing.T) {
 		assert.Equal(t, []string{"apple", "brown", "jpg", "lazy"}, result)
 	})
 }
+
+func TestStopwordsOnly(t *testing.T) {
+	t.Run("Empty", func(t *testing.T) {
+		assert.False(t, StopwordsOnly(""))
+	})
+	t.Run("FoldersDateienFile", func(t *testing.T) {
+		assert.True(t, StopwordsOnly("Folders Dateien File"))
+	})
+	t.Run("FoldersDateienFile", func(t *testing.T) {
+		assert.False(t, StopwordsOnly("Folders Dateien Meme File"))
+	})
+	t.Run("qx", func(t *testing.T) {
+		assert.True(t, StopwordsOnly("qx"))
+	})
+	t.Run("atz", func(t *testing.T) {
+		assert.True(t, StopwordsOnly("atz"))
+	})
+	t.Run("xqx", func(t *testing.T) {
+		assert.True(t, StopwordsOnly("xqx"))
+	})
+	t.Run("kuh", func(t *testing.T) {
+		assert.False(t, StopwordsOnly("kuh"))
+	})
+	t.Run("muh", func(t *testing.T) {
+		assert.False(t, StopwordsOnly("muh"))
+	})
+	t.Run("桥", func(t *testing.T) {
+		assert.False(t, StopwordsOnly("桥"))
+	})
+	t.Run("桥船", func(t *testing.T) {
+		assert.False(t, StopwordsOnly("桥船"))
+	})
+}