Refactor "string truncate" (#32984)

models/actions/run.go

@@ -275,7 +275,7 @@ func InsertRun(ctx context.Context, run *ActionRun, jobs []*jobparser.SingleWork
 		return err
 	}
 	run.Index = index
-	run.Title, _ = util.SplitStringAtByteN(run.Title, 255)
+	run.Title = util.EllipsisDisplayString(run.Title, 255)
 
 	if err := db.Insert(ctx, run); err != nil {
 		return err
@@ -308,7 +308,7 @@ func InsertRun(ctx context.Context, run *ActionRun, jobs []*jobparser.SingleWork
 		} else {
 			hasWaiting = true
 		}
-		job.Name, _ = util.SplitStringAtByteN(job.Name, 255)
+		job.Name = util.EllipsisDisplayString(job.Name, 255)
 		runJobs = append(runJobs, &ActionRunJob{
 			RunID:             run.ID,
 			RepoID:            run.RepoID,
@@ -402,7 +402,7 @@ func UpdateRun(ctx context.Context, run *ActionRun, cols ...string) error {
 	if len(cols) > 0 {
 		sess.Cols(cols...)
 	}
-	run.Title, _ = util.SplitStringAtByteN(run.Title, 255)
+	run.Title = util.EllipsisDisplayString(run.Title, 255)
 	affected, err := sess.Update(run)
 	if err != nil {
 		return err

models/actions/runner.go

@@ -252,7 +252,7 @@ func GetRunnerByID(ctx context.Context, id int64) (*ActionRunner, error) {
 // UpdateRunner updates runner's information.
 func UpdateRunner(ctx context.Context, r *ActionRunner, cols ...string) error {
 	e := db.GetEngine(ctx)
-	r.Name, _ = util.SplitStringAtByteN(r.Name, 255)
+	r.Name = util.EllipsisDisplayString(r.Name, 255)
 	var err error
 	if len(cols) == 0 {
 		_, err = e.ID(r.ID).AllCols().Update(r)
@@ -279,7 +279,7 @@ func CreateRunner(ctx context.Context, t *ActionRunner) error {
 		// Remove OwnerID to avoid confusion; it's not worth returning an error here.
 		t.OwnerID = 0
 	}
-	t.Name, _ = util.SplitStringAtByteN(t.Name, 255)
+	t.Name = util.EllipsisDisplayString(t.Name, 255)
 	return db.Insert(ctx, t)
 }
 

models/actions/runner_token.go

@@ -10,7 +10,6 @@ import (
 	"code.gitea.io/gitea/models/db"
 	repo_model "code.gitea.io/gitea/models/repo"
 	user_model "code.gitea.io/gitea/models/user"
-	"code.gitea.io/gitea/modules/base"
 	"code.gitea.io/gitea/modules/timeutil"
 	"code.gitea.io/gitea/modules/util"
 )
@@ -52,7 +51,7 @@ func GetRunnerToken(ctx context.Context, token string) (*ActionRunnerToken, erro
 	if err != nil {
 		return nil, err
 	} else if !has {
-		return nil, fmt.Errorf(`runner token "%s...": %w`, base.TruncateString(token, 3), util.ErrNotExist)
+		return nil, fmt.Errorf(`runner token "%s...": %w`, util.TruncateRunes(token, 3), util.ErrNotExist)
 	}
 	return &runnerToken, nil
 }

models/actions/schedule.go

@@ -68,7 +68,7 @@ func CreateScheduleTask(ctx context.Context, rows []*ActionSchedule) error {
 
 	// Loop through each schedule row
 	for _, row := range rows {
-		row.Title, _ = util.SplitStringAtByteN(row.Title, 255)
+		row.Title = util.EllipsisDisplayString(row.Title, 255)
 		// Create new schedule row
 		if err = db.Insert(ctx, row); err != nil {
 			return err

models/actions/task.go

@@ -298,7 +298,7 @@ func CreateTaskForRunner(ctx context.Context, runner *ActionRunner) (*ActionTask
 	if len(workflowJob.Steps) > 0 {
 		steps := make([]*ActionTaskStep, len(workflowJob.Steps))
 		for i, v := range workflowJob.Steps {
-			name, _ := util.SplitStringAtByteN(v.String(), 255)
+			name := util.EllipsisDisplayString(v.String(), 255)
 			steps[i] = &ActionTaskStep{
 				Name:   name,
 				TaskID: task.ID,

models/activities/action.go

@@ -20,12 +20,12 @@ import (
 	repo_model "code.gitea.io/gitea/models/repo"
 	"code.gitea.io/gitea/models/unit"
 	user_model "code.gitea.io/gitea/models/user"
-	"code.gitea.io/gitea/modules/base"
 	"code.gitea.io/gitea/modules/git"
 	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/setting"
 	"code.gitea.io/gitea/modules/structs"
 	"code.gitea.io/gitea/modules/timeutil"
+	"code.gitea.io/gitea/modules/util"
 
 	"xorm.io/builder"
 	"xorm.io/xorm/schemas"
@@ -226,7 +226,7 @@ func (a *Action) GetActUserName(ctx context.Context) string {
 // ShortActUserName gets the action's user name trimmed to max 20
 // chars.
 func (a *Action) ShortActUserName(ctx context.Context) string {
-	return base.EllipsisString(a.GetActUserName(ctx), 20)
+	return util.EllipsisDisplayString(a.GetActUserName(ctx), 20)
 }
 
 // GetActDisplayName gets the action's display name based on DEFAULT_SHOW_FULL_NAME, or falls back to the username if it is blank.
@@ -260,7 +260,7 @@ func (a *Action) GetRepoUserName(ctx context.Context) string {
 // ShortRepoUserName returns the name of the action repository owner
 // trimmed to max 20 chars.
 func (a *Action) ShortRepoUserName(ctx context.Context) string {
-	return base.EllipsisString(a.GetRepoUserName(ctx), 20)
+	return util.EllipsisDisplayString(a.GetRepoUserName(ctx), 20)
 }
 
 // GetRepoName returns the name of the action repository.
@@ -275,7 +275,7 @@ func (a *Action) GetRepoName(ctx context.Context) string {
 // ShortRepoName returns the name of the action repository
 // trimmed to max 33 chars.
 func (a *Action) ShortRepoName(ctx context.Context) string {
-	return base.EllipsisString(a.GetRepoName(ctx), 33)
+	return util.EllipsisDisplayString(a.GetRepoName(ctx), 33)
 }
 
 // GetRepoPath returns the virtual path to the action repository.

models/issues/issue_update.go

@@ -177,7 +177,7 @@ func ChangeIssueTitle(ctx context.Context, issue *Issue, doer *user_model.User,
 	}
 	defer committer.Close()
 
-	issue.Title, _ = util.SplitStringAtByteN(issue.Title, 255)
+	issue.Title = util.EllipsisDisplayString(issue.Title, 255)
 	if err = UpdateIssueCols(ctx, issue, "name"); err != nil {
 		return fmt.Errorf("updateIssueCols: %w", err)
 	}
@@ -440,7 +440,7 @@ func NewIssue(ctx context.Context, repo *repo_model.Repository, issue *Issue, la
 	}
 
 	issue.Index = idx
-	issue.Title, _ = util.SplitStringAtByteN(issue.Title, 255)
+	issue.Title = util.EllipsisDisplayString(issue.Title, 255)
 
 	if err = NewIssueWithIndex(ctx, issue.Poster, NewIssueOptions{
 		Repo:        repo,

models/issues/pull.go

@@ -572,7 +572,7 @@ func NewPullRequest(ctx context.Context, repo *repo_model.Repository, issue *Iss
 	}
 
 	issue.Index = idx
-	issue.Title, _ = util.SplitStringAtByteN(issue.Title, 255)
+	issue.Title = util.EllipsisDisplayString(issue.Title, 255)
 
 	if err = NewIssueWithIndex(ctx, issue.Poster, NewIssueOptions{
 		Repo:        repo,

models/project/project.go

@@ -256,7 +256,7 @@ func NewProject(ctx context.Context, p *Project) error {
 		return util.NewInvalidArgumentErrorf("project type is not valid")
 	}
 
-	p.Title, _ = util.SplitStringAtByteN(p.Title, 255)
+	p.Title = util.EllipsisDisplayString(p.Title, 255)
 
 	return db.WithTx(ctx, func(ctx context.Context) error {
 		if err := db.Insert(ctx, p); err != nil {
@@ -311,7 +311,7 @@ func UpdateProject(ctx context.Context, p *Project) error {
 		p.CardType = CardTypeTextOnly
 	}
 
-	p.Title, _ = util.SplitStringAtByteN(p.Title, 255)
+	p.Title = util.EllipsisDisplayString(p.Title, 255)
 	_, err := db.GetEngine(ctx).ID(p.ID).Cols(
 		"title",
 		"description",

models/repo/release.go

@@ -156,7 +156,7 @@ func IsReleaseExist(ctx context.Context, repoID int64, tagName string) (bool, er
 
 // UpdateRelease updates all columns of a release
 func UpdateRelease(ctx context.Context, rel *Release) error {
-	rel.Title, _ = util.SplitStringAtByteN(rel.Title, 255)
+	rel.Title = util.EllipsisDisplayString(rel.Title, 255)
 	_, err := db.GetEngine(ctx).ID(rel.ID).AllCols().Update(rel)
 	return err
 }

models/user/user.go

@@ -190,9 +190,9 @@ func (u *User) BeforeUpdate() {
 	}
 
 	u.LowerName = strings.ToLower(u.Name)
-	u.Location = base.TruncateString(u.Location, 255)
+	u.Location = util.TruncateRunes(u.Location, 255)
-	u.Website = base.TruncateString(u.Website, 255)
+	u.Website = util.TruncateRunes(u.Website, 255)
-	u.Description = base.TruncateString(u.Description, 255)
+	u.Description = util.TruncateRunes(u.Description, 255)
 }
 
 // AfterLoad is invoked from XORM after filling all the fields of this object.
@@ -501,9 +501,9 @@ func (u *User) GitName() string {
 // ShortName ellipses username to length
 func (u *User) ShortName(length int) string {
 	if setting.UI.DefaultShowFullName && len(u.FullName) > 0 {
-		return base.EllipsisString(u.FullName, length)
+		return util.EllipsisDisplayString(u.FullName, length)
 	}
-	return base.EllipsisString(u.Name, length)
+	return util.EllipsisDisplayString(u.Name, length)
 }
 
 // IsMailable checks if a user is eligible

modules/base/tool.go

@@ -16,11 +16,11 @@ import (
 	"strconv"
 	"strings"
 	"time"
-	"unicode/utf8"
 
 	"code.gitea.io/gitea/modules/git"
 	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/setting"
+	"code.gitea.io/gitea/modules/util"
 
 	"github.com/dustin/go-humanize"
 )
@@ -35,7 +35,7 @@ func EncodeSha256(str string) string {
 // ShortSha is basically just truncating.
 // It is DEPRECATED and will be removed in the future.
 func ShortSha(sha1 string) string {
-	return TruncateString(sha1, 10)
+	return util.TruncateRunes(sha1, 10)
 }
 
 // BasicAuthDecode decode basic auth string
@@ -116,27 +116,6 @@ func FileSize(s int64) string {
 	return humanize.IBytes(uint64(s))
 }
 
-// EllipsisString returns a truncated short string,
-// it appends '...' in the end of the length of string is too large.
-func EllipsisString(str string, length int) string {
-	if length <= 3 {
-		return "..."
-	}
-	if utf8.RuneCountInString(str) <= length {
-		return str
-	}
-	return string([]rune(str)[:length-3]) + "..."
-}
-
-// TruncateString returns a truncated string with given limit,
-// it returns input string if length is not reached limit.
-func TruncateString(str string, limit int) string {
-	if utf8.RuneCountInString(str) < limit {
-		return str
-	}
-	return string([]rune(str)[:limit])
-}
-
 // StringsToInt64s converts a slice of string to a slice of int64.
 func StringsToInt64s(strs []string) ([]int64, error) {
 	if strs == nil {

modules/base/tool_test.go

@@ -113,36 +113,6 @@ func TestFileSize(t *testing.T) {
 	assert.Equal(t, "2.0 EiB", FileSize(size))
 }
 
-func TestEllipsisString(t *testing.T) {
-	assert.Equal(t, "...", EllipsisString("foobar", 0))
-	assert.Equal(t, "...", EllipsisString("foobar", 1))
-	assert.Equal(t, "...", EllipsisString("foobar", 2))
-	assert.Equal(t, "...", EllipsisString("foobar", 3))
-	assert.Equal(t, "f...", EllipsisString("foobar", 4))
-	assert.Equal(t, "fo...", EllipsisString("foobar", 5))
-	assert.Equal(t, "foobar", EllipsisString("foobar", 6))
-	assert.Equal(t, "foobar", EllipsisString("foobar", 10))
-	assert.Equal(t, "测...", EllipsisString("测试文本一二三四", 4))
-	assert.Equal(t, "测试...", EllipsisString("测试文本一二三四", 5))
-	assert.Equal(t, "测试文...", EllipsisString("测试文本一二三四", 6))
-	assert.Equal(t, "测试文本一二三四", EllipsisString("测试文本一二三四", 10))
-}
-
-func TestTruncateString(t *testing.T) {
-	assert.Equal(t, "", TruncateString("foobar", 0))
-	assert.Equal(t, "f", TruncateString("foobar", 1))
-	assert.Equal(t, "fo", TruncateString("foobar", 2))
-	assert.Equal(t, "foo", TruncateString("foobar", 3))
-	assert.Equal(t, "foob", TruncateString("foobar", 4))
-	assert.Equal(t, "fooba", TruncateString("foobar", 5))
-	assert.Equal(t, "foobar", TruncateString("foobar", 6))
-	assert.Equal(t, "foobar", TruncateString("foobar", 7))
-	assert.Equal(t, "测试文本", TruncateString("测试文本一二三四", 4))
-	assert.Equal(t, "测试文本一", TruncateString("测试文本一二三四", 5))
-	assert.Equal(t, "测试文本一二", TruncateString("测试文本一二三四", 6))
-	assert.Equal(t, "测试文本一二三", TruncateString("测试文本一二三四", 7))
-}
-
 func TestStringsToInt64s(t *testing.T) {
 	testSuccess := func(input []string, expected []int64) {
 		result, err := StringsToInt64s(input)

modules/issue/template/unmarshal.go

@@ -109,7 +109,7 @@ func unmarshal(filename string, content []byte) (*api.IssueTemplate, error) {
 
 			it.Content = string(content)
 			it.Name = path.Base(it.FileName) // paths in Git are always '/' separated - do not use filepath!
-			it.About, _ = util.SplitStringAtByteN(it.Content, 80)
+			it.About = util.EllipsisDisplayString(it.Content, 80)
 		} else {
 			it.Content = templateBody
 			if it.About == "" {

modules/markup/html_link.go

@@ -173,7 +173,7 @@ func linkProcessor(ctx *RenderContext, node *html.Node) {
 
 		uri := node.Data[m[0]:m[1]]
 		remaining := node.Data[m[1]:]
-		if util.IsLikelySplitLeftPart(remaining) {
+		if util.IsLikelyEllipsisLeftPart(remaining) {
 			return
 		}
 		replaceContent(node, m[0], m[1], createLink(ctx, uri, uri, "" /*link*/))

modules/markup/html_test.go

@@ -207,12 +207,12 @@ func TestRender_links(t *testing.T) {
 		"ftps://gitea.com",
 		`<p>ftps://gitea.com</p>`)
 
-	t.Run("LinkSplit", func(t *testing.T) {
+	t.Run("LinkEllipsis", func(t *testing.T) {
-		input, _ := util.SplitStringAtByteN("http://10.1.2.3", 12)
+		input := util.EllipsisDisplayString("http://10.1.2.3", 12)
 		assert.Equal(t, "http://10…", input)
 		test(input, "<p>http://10…</p>")
 
-		input, _ = util.SplitStringAtByteN("http://10.1.2.3", 13)
+		input = util.EllipsisDisplayString("http://10.1.2.3", 13)
 		assert.Equal(t, "http://10.…", input)
 		test(input, "<p>http://10.…</p>")
 	})

modules/templates/mailer.go

@@ -11,9 +11,9 @@ import (
 	"strings"
 	texttmpl "text/template"
 
-	"code.gitea.io/gitea/modules/base"
 	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/setting"
+	"code.gitea.io/gitea/modules/util"
 )
 
 var mailSubjectSplit = regexp.MustCompile(`(?m)^-{3,}\s*$`)
@@ -24,7 +24,7 @@ func mailSubjectTextFuncMap() texttmpl.FuncMap {
 		"dict": dict,
 		"Eval": evalTokens,
 
-		"EllipsisString": base.EllipsisString,
+		"EllipsisString": util.EllipsisDisplayString,
 		"AppName": func() string {
 			return setting.AppName
 		},

modules/templates/util_string.go

@@ -8,7 +8,7 @@ import (
 	"html/template"
 	"strings"
 
-	"code.gitea.io/gitea/modules/base"
+	"code.gitea.io/gitea/modules/util"
 )
 
 type StringUtils struct{}
@@ -54,7 +54,7 @@ func (su *StringUtils) Cut(s, sep string) []any {
 }
 
 func (su *StringUtils) EllipsisString(s string, maxLength int) string {
-	return base.EllipsisString(s, maxLength)
+	return util.EllipsisDisplayString(s, maxLength)
 }
 
 func (su *StringUtils) ToUpper(s string) string {

modules/util/truncate.go

@@ -14,31 +14,92 @@ const (
 	asciiEllipsis = "..."
 )
 
-func IsLikelySplitLeftPart(s string) bool {
+func IsLikelyEllipsisLeftPart(s string) bool {
 	return strings.HasSuffix(s, utf8Ellipsis) || strings.HasSuffix(s, asciiEllipsis)
 }
 
-// SplitStringAtByteN splits a string at byte n accounting for rune boundaries. (Combining characters are not accounted for.)
+// EllipsisDisplayString returns a truncated short string for display purpose.
-func SplitStringAtByteN(input string, n int) (left, right string) {
+// The length is the approximate number of ASCII-width in the string (CJK/emoji are 2-ASCII width)
-	if len(input) <= n {
+// It appends "…" or "..." at the end of truncated string.
-		return input, ""
+// It guarantees the length of the returned runes doesn't exceed the limit.
-	}
+func EllipsisDisplayString(str string, limit int) string {
+	s, _, _, _ := ellipsisDisplayString(str, limit)
+	return s
+}
 
-	if !utf8.ValidString(input) {
+// EllipsisDisplayStringX works like EllipsisDisplayString while it also returns the right part
-		if n-3 < 0 {
+func EllipsisDisplayStringX(str string, limit int) (left, right string) {
-			return input, ""
+	left, offset, truncated, encounterInvalid := ellipsisDisplayString(str, limit)
+	if truncated {
+		right = str[offset:]
+		r, _ := utf8.DecodeRune(UnsafeStringToBytes(right))
+		encounterInvalid = encounterInvalid || r == utf8.RuneError
+		ellipsis := utf8Ellipsis
+		if encounterInvalid {
+			ellipsis = asciiEllipsis
 		}
-		return input[:n-3] + asciiEllipsis, asciiEllipsis + input[n-3:]
+		right = ellipsis + right
 	}
+	return left, right
+}
 
-	end := 0
+func ellipsisDisplayString(str string, limit int) (res string, offset int, truncated, encounterInvalid bool) {
-	for end <= n-3 {
+	if len(str) <= limit {
-		_, size := utf8.DecodeRuneInString(input[end:])
+		return str, len(str), false, false
-		if end+size > n-3 {
+	}
+
+	// To future maintainers: this logic must guarantee that the length of the returned runes doesn't exceed the limit,
+	// because the returned string will also be used as database value. UTF-8 VARCHAR(10) could store 10 rune characters,
+	// So each rune must be countered as at least 1 width.
+	// Even if there are some special Unicode characters (zero-width, combining, etc.), they should NEVER be counted as zero.
+	pos, used := 0, 0
+	for i, r := range str {
+		encounterInvalid = encounterInvalid || r == utf8.RuneError
+		pos = i
+		runeWidth := 1
+		if r >= 128 {
+			runeWidth = 2 // CJK/emoji chars are considered as 2-ASCII width
+		}
+		if used+runeWidth+3 > limit {
 			break
 		}
-		end += size
+		used += runeWidth
+		offset += utf8.RuneLen(r)
+	}
+
+	// if the remaining are fewer than 3 runes, then maybe we could add them, no need to ellipse
+	if len(str)-pos <= 12 {
+		var nextCnt, nextWidth int
+		for _, r := range str[pos:] {
+			if nextCnt >= 4 {
+				break
+			}
+			nextWidth++
+			if r >= 128 {
+				nextWidth++ // CJK/emoji chars are considered as 2-ASCII width
+			}
+			nextCnt++
+		}
+		if nextCnt <= 3 && used+nextWidth <= limit {
+			return str, len(str), false, false
+		}
+	}
+	if limit < 3 {
+		// if the limit is so small, do not add ellipsis
+		return str[:offset], offset, true, false
 	}
+	ellipsis := utf8Ellipsis
+	if encounterInvalid {
+		ellipsis = asciiEllipsis
+	}
+	return str[:offset] + ellipsis, offset, true, encounterInvalid
+}
 
-	return input[:end] + utf8Ellipsis, utf8Ellipsis + input[end:]
+// TruncateRunes returns a truncated string with given rune limit,
+// it returns input string if its rune length doesn't exceed the limit.
+func TruncateRunes(str string, limit int) string {
+	if utf8.RuneCountInString(str) < limit {
+		return str
+	}
+	return string([]rune(str)[:limit])
 }

modules/util/truncate_test.go

@@ -4,43 +4,94 @@
 package util
 
 import (
+	"fmt"
+	"strings"
 	"testing"
 
 	"github.com/stretchr/testify/assert"
 )
 
-func TestSplitString(t *testing.T) {
+func TestEllipsisString(t *testing.T) {
-	type testCase struct {
+	cases := []struct {
-		input    string
+		limit int
-		n        int
+
-		leftSub  string
+		input, left, right string
-		ellipsis string
+	}{
+		{limit: 0, input: "abcde", left: "", right: "…abcde"},
+		{limit: 1, input: "abcde", left: "", right: "…abcde"},
+		{limit: 2, input: "abcde", left: "", right: "…abcde"},
+		{limit: 3, input: "abcde", left: "…", right: "…abcde"},
+		{limit: 4, input: "abcde", left: "a…", right: "…bcde"},
+		{limit: 5, input: "abcde", left: "abcde", right: ""},
+		{limit: 6, input: "abcde", left: "abcde", right: ""},
+		{limit: 7, input: "abcde", left: "abcde", right: ""},
+
+		// a CJK char or emoji is considered as 2-ASCII width, the ellipsis is 3-ASCII width
+		{limit: 0, input: "测试文本", left: "", right: "…测试文本"},
+		{limit: 1, input: "测试文本", left: "", right: "…测试文本"},
+		{limit: 2, input: "测试文本", left: "", right: "…测试文本"},
+		{limit: 3, input: "测试文本", left: "…", right: "…测试文本"},
+		{limit: 4, input: "测试文本", left: "…", right: "…测试文本"},
+		{limit: 5, input: "测试文本", left: "测…", right: "…试文本"},
+		{limit: 6, input: "测试文本", left: "测…", right: "…试文本"},
+		{limit: 7, input: "测试文本", left: "测试…", right: "…文本"},
+		{limit: 8, input: "测试文本", left: "测试文本", right: ""},
+		{limit: 9, input: "测试文本", left: "测试文本", right: ""},
+	}
+	for _, c := range cases {
+		t.Run(fmt.Sprintf("%s(%d)", c.input, c.limit), func(t *testing.T) {
+			left, right := EllipsisDisplayStringX(c.input, c.limit)
+			assert.Equal(t, c.left, left, "left")
+			assert.Equal(t, c.right, right, "right")
+		})
 	}
 
-	test := func(tc []*testCase, f func(input string, n int) (left, right string)) {
+	t.Run("LongInput", func(t *testing.T) {
-		for _, c := range tc {
+		left, right := EllipsisDisplayStringX(strings.Repeat("abc", 240), 90)
-			l, r := f(c.input, c.n)
+		assert.Equal(t, strings.Repeat("abc", 29)+"…", left)
-			if c.ellipsis != "" {
+		assert.Equal(t, "…"+strings.Repeat("abc", 211), right)
-				assert.Equal(t, c.leftSub+c.ellipsis, l, "test split %q at %d, expected leftSub: %q", c.input, c.n, c.leftSub)
+	})
-				assert.Equal(t, c.ellipsis+c.input[len(c.leftSub):], r, "test split %s at %d, expected rightSub: %q", c.input, c.n, c.input[len(c.leftSub):])
+
-			} else {
+	t.Run("InvalidUtf8", func(t *testing.T) {
-				assert.Equal(t, c.leftSub, l, "test split %q at %d, expected leftSub: %q", c.input, c.n, c.leftSub)
+		invalidCases := []struct {
-				assert.Empty(t, r, "test split %q at %d, expected rightSub: %q", c.input, c.n, "")
+			limit       int
-			}
+			left, right string
+		}{
+			{limit: 0, left: "", right: "...\xef\x03\xfe\xef\x03\xfe"},
+			{limit: 1, left: "", right: "...\xef\x03\xfe\xef\x03\xfe"},
+			{limit: 2, left: "", right: "...\xef\x03\xfe\xef\x03\xfe"},
+			{limit: 3, left: "...", right: "...\xef\x03\xfe\xef\x03\xfe"},
+			{limit: 4, left: "...", right: "...\xef\x03\xfe\xef\x03\xfe"},
+			{limit: 5, left: "\xef\x03\xfe...", right: "...\xef\x03\xfe"},
+			{limit: 6, left: "\xef\x03\xfe\xef\x03\xfe", right: ""},
+			{limit: 7, left: "\xef\x03\xfe\xef\x03\xfe", right: ""},
 		}
-	}
+		for _, c := range invalidCases {
+			t.Run(fmt.Sprintf("%d", c.limit), func(t *testing.T) {
+				left, right := EllipsisDisplayStringX("\xef\x03\xfe\xef\x03\xfe", c.limit)
+				assert.Equal(t, c.left, left, "left")
+				assert.Equal(t, c.right, right, "right")
+			})
+		}
+	})
 
-	tc := []*testCase{
+	t.Run("IsLikelyEllipsisLeftPart", func(t *testing.T) {
-		{"abc123xyz", 0, "", utf8Ellipsis},
+		assert.True(t, IsLikelyEllipsisLeftPart("abcde…"))
-		{"abc123xyz", 1, "", utf8Ellipsis},
+		assert.True(t, IsLikelyEllipsisLeftPart("abcde..."))
-		{"abc123xyz", 4, "a", utf8Ellipsis},
+	})
-		{"啊bc123xyz", 4, "", utf8Ellipsis},
+}
-		{"啊bc123xyz", 6, "啊", utf8Ellipsis},
+
-		{"啊bc", 5, "啊bc", ""},
+func TestTruncateRunes(t *testing.T) {
-		{"啊bc", 6, "啊bc", ""},
+	assert.Equal(t, "", TruncateRunes("", 0))
-		{"abc\xef\x03\xfe", 3, "", asciiEllipsis},
+	assert.Equal(t, "", TruncateRunes("", 1))
-		{"abc\xef\x03\xfe", 4, "a", asciiEllipsis},
+
-		{"\xef\x03", 1, "\xef\x03", ""},
+	assert.Equal(t, "", TruncateRunes("ab", 0))
-	}
+	assert.Equal(t, "a", TruncateRunes("ab", 1))
-	test(tc, SplitStringAtByteN)
+	assert.Equal(t, "ab", TruncateRunes("ab", 2))
+	assert.Equal(t, "ab", TruncateRunes("ab", 3))
+
+	assert.Equal(t, "", TruncateRunes("测试", 0))
+	assert.Equal(t, "测", TruncateRunes("测试", 1))
+	assert.Equal(t, "测试", TruncateRunes("测试", 2))
+	assert.Equal(t, "测试", TruncateRunes("测试", 3))
 }

routers/api/actions/runner/runner.go

@@ -69,7 +69,7 @@ func (s *Service) Register(
 	labels := req.Msg.Labels
 
 	// create new runner
-	name, _ := util.SplitStringAtByteN(req.Msg.Name, 255)
+	name := util.EllipsisDisplayString(req.Msg.Name, 255)
 	runner := &actions_model.ActionRunner{
 		UUID:        gouuid.New().String(),
 		Name:        name,

routers/web/repo/compare.go

@@ -664,7 +664,7 @@ func PrepareCompareDiff(
 	}
 	if len(title) > 255 {
 		var trailer string
-		title, trailer = util.SplitStringAtByteN(title, 255)
+		title, trailer = util.EllipsisDisplayStringX(title, 255)
 		if len(trailer) > 0 {
 			if ctx.Data["content"] != nil {
 				ctx.Data["content"] = fmt.Sprintf("%s\n\n%s", trailer, ctx.Data["content"])

services/feed/notifier.go

@@ -109,7 +109,7 @@ func (a *actionNotifier) CreateIssueComment(ctx context.Context, doer *user_mode
 		IsPrivate: issue.Repo.IsPrivate,
 	}
 
-	truncatedContent, truncatedRight := util.SplitStringAtByteN(comment.Content, 200)
+	truncatedContent, truncatedRight := util.EllipsisDisplayStringX(comment.Content, 200)
 	if truncatedRight != "" {
 		// in case the content is in a Latin family language, we remove the last broken word.
 		lastSpaceIdx := strings.LastIndex(truncatedContent, " ")

services/mailer/sender/message.go

@@ -10,9 +10,9 @@ import (
 	"strings"
 	"time"
 
-	"code.gitea.io/gitea/modules/base"
 	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/setting"
+	"code.gitea.io/gitea/modules/util"
 
 	"github.com/jaytaylor/html2text"
 	gomail "github.com/wneessen/go-mail"
@@ -54,7 +54,7 @@ func (m *Message) ToMessage() *gomail.Msg {
 
 	plainBody, err := html2text.FromString(m.Body)
 	if err != nil || setting.MailService.SendAsPlainText {
-		if strings.Contains(base.TruncateString(m.Body, 100), "<html>") {
+		if strings.Contains(util.TruncateRunes(m.Body, 100), "<html>") {
 			log.Warn("Mail contains HTML but configured to send as plain text.")
 		}
 		msg.SetBodyString("text/plain", plainBody)

services/migrations/gitea_uploader.go

@@ -19,7 +19,6 @@ import (
 	issues_model "code.gitea.io/gitea/models/issues"
 	repo_model "code.gitea.io/gitea/models/repo"
 	user_model "code.gitea.io/gitea/models/user"
-	base_module "code.gitea.io/gitea/modules/base"
 	"code.gitea.io/gitea/modules/git"
 	"code.gitea.io/gitea/modules/gitrepo"
 	"code.gitea.io/gitea/modules/label"
@@ -409,7 +408,7 @@ func (g *GiteaLocalUploader) CreateIssues(issues ...*base.Issue) error {
 			RepoID:      g.repo.ID,
 			Repo:        g.repo,
 			Index:       issue.Number,
-			Title:       base_module.TruncateString(issue.Title, 255),
+			Title:       util.TruncateRunes(issue.Title, 255),
 			Content:     issue.Content,
 			Ref:         issue.Ref,
 			IsClosed:    issue.State == "closed",

services/release/release.go

@@ -179,7 +179,7 @@ func CreateRelease(gitRepo *git.Repository, rel *repo_model.Release, attachmentU
 		return err
 	}
 
-	rel.Title, _ = util.SplitStringAtByteN(rel.Title, 255)
+	rel.Title = util.EllipsisDisplayString(rel.Title, 255)
 	rel.LowerTagName = strings.ToLower(rel.TagName)
 	if err = db.Insert(gitRepo.Ctx, rel); err != nil {
 		return err