Trigger mentions for keyword including multibyte characters in sentences (#10262)

app/notification.go

@@ -8,6 +8,7 @@ import (
 	"sort"
 	"strings"
 	"unicode"
+	"unicode/utf8"
 
 	"github.com/mattermost/mattermost-server/mlog"
 	"github.com/mattermost/mattermost-server/model"
@@ -468,6 +469,14 @@ func GetExplicitMentions(post *model.Post, keywords map[string][]string) *Explic
 
 		return isMention
 	}
+
+	var multibyteKeywords []string
+	for keyword := range keywords {
+		if len(keyword) != utf8.RuneCountInString(keyword) {
+			multibyteKeywords = append(multibyteKeywords, keyword)
+		}
+	}
+
 	processText := func(text string) {
 		for _, word := range strings.FieldsFunc(text, func(c rune) bool {
 			// Split on any whitespace or punctuation that can't be part of an at mention or emoji pattern
@@ -516,6 +525,17 @@ func GetExplicitMentions(post *model.Post, keywords map[string][]string) *Explic
 					}
 				}
 			}
+
+			// If word contains a multibyte character, check if it contains a multibyte keyword
+			if len(word) != utf8.RuneCountInString(word) {
+				for _, key := range multibyteKeywords {
+					if strings.Contains(word, key) {
+						if ids, match := keywords[key]; match {
+							addMentionedUsers(ids)
+						}
+					}
+				}
+			}
 		}
 	}
 

app/notification_test.go

@@ -474,6 +474,51 @@ func TestGetExplicitMentions(t *testing.T) {
 				ChannelMentioned: true,
 			},
 		},
+		"MultibyteCharacter": {
+			Message:  "My name is 萌",
+			Keywords: map[string][]string{"萌": {id1}},
+			Expected: &ExplicitMentions{
+				MentionedUserIds: map[string]bool{
+					id1: true,
+				},
+			},
+		},
+		"MultibyteCharacterAtBeginningOfSentence": {
+			Message:  "이메일을 보내다.",
+			Keywords: map[string][]string{"이메일": {id1}},
+			Expected: &ExplicitMentions{
+				MentionedUserIds: map[string]bool{
+					id1: true,
+				},
+			},
+		},
+		"MultibyteCharacterInPartOfSentence": {
+			Message:  "我爱吃番茄炒饭",
+			Keywords: map[string][]string{"番茄": {id1}},
+			Expected: &ExplicitMentions{
+				MentionedUserIds: map[string]bool{
+					id1: true,
+				},
+			},
+		},
+		"MultibyteCharacterAtEndOfSentence": {
+			Message:  "こんにちは、世界",
+			Keywords: map[string][]string{"世界": {id1}},
+			Expected: &ExplicitMentions{
+				MentionedUserIds: map[string]bool{
+					id1: true,
+				},
+			},
+		},
+		"MultibyteCharacterTwiceInSentence": {
+			Message:  "石橋さんが石橋を渡る",
+			Keywords: map[string][]string{"石橋": {id1}},
+			Expected: &ExplicitMentions{
+				MentionedUserIds: map[string]bool{
+					id1: true,
+				},
+			},
+		},
 
 		// The following tests cover cases where the message mentions @user.name, so we shouldn't assume that
 		// the user might be intending to mention some @user that isn't in the channel.