Fix UTF-8-safe truncation for Telegram and CLI summaries

Author: emperorhan

cmd/ralphctl/main.go

@@ -2,6 +2,7 @@ package main
 
 import (
 	"bufio"
+	"bytes"
 	"context"
 	"errors"
 	"flag"
@@ -15,6 +16,7 @@ import (
 	"strings"
 	"syscall"
 	"time"
+	"unicode/utf8"
 
 	"codex-ralph/internal/ralph"
 )
@@ -1550,13 +1552,20 @@ func compactSingleLine(raw string, maxLen int) string {
 	v = strings.ReplaceAll(v, "\n", " ")
 	v = strings.ReplaceAll(v, "\r", " ")
 	v = strings.Join(strings.Fields(v), " ")
-	if maxLen <= 0 || len(v) <= maxLen {
+	if !utf8.ValidString(v) {
+		v = string(bytes.ToValidUTF8([]byte(v), []byte("?")))
+	}
+	if maxLen <= 0 {
+		return v
+	}
+	runes := []rune(v)
+	if len(runes) <= maxLen {
 		return v
 	}
 	if maxLen <= 3 {
-		return v[:maxLen]
+		return string(runes[:maxLen])
 	}
-	return v[:maxLen-3] + "..."
+	return string(runes[:maxLen-3]) + "..."
 }
 
 func valueOrDash(raw string) string {

cmd/ralphctl/main_test.go

@@ -4,6 +4,7 @@ import (
 	"os"
 	"path/filepath"
 	"testing"
+	"unicode/utf8"
 
 	"codex-ralph/internal/ralph"
 )
@@ -53,6 +54,32 @@ func TestCompactSingleLine(t *testing.T) {
 	}
 }
 
+func TestCompactSingleLineUnicodeSafe(t *testing.T) {
+	t.Parallel()
+
+	// 4-byte emoji + Korean should not be cut mid-rune.
+	got := compactSingleLine("🔥비트코인 자동화", 5)
+	if !utf8.ValidString(got) {
+		t.Fatalf("output must be valid UTF-8: %q", got)
+	}
+	if got == "" {
+		t.Fatalf("output should not be empty")
+	}
+}
+
+func TestCompactSingleLineInvalidUTF8Sanitized(t *testing.T) {
+	t.Parallel()
+
+	raw := string([]byte{0xff, 0xfe, 'a', 'b', 'c'})
+	got := compactSingleLine(raw, 10)
+	if !utf8.ValidString(got) {
+		t.Fatalf("output must be valid UTF-8: %q", got)
+	}
+	if got == "" {
+		t.Fatalf("output should not be empty")
+	}
+}
+
 func TestResolveReloadTargetsFleetOnlyWhenCurrentUnmanaged(t *testing.T) {
 	t.Parallel()
 

internal/ralph/telegram.go

@@ -15,6 +15,7 @@ import (
 	"strings"
 	"sync"
 	"time"
+	"unicode/utf8"
 )
 
 const defaultTelegramBaseURL = "https://api.telegram.org"
@@ -628,8 +629,12 @@ func compactTelegramError(raw string) string {
 	}
 	raw = strings.ReplaceAll(raw, "\n", " ")
 	raw = strings.Join(strings.Fields(raw), " ")
-	if len(raw) > 300 {
-		return raw[:297] + "..."
+	if !utf8.ValidString(raw) {
+		raw = string(bytes.ToValidUTF8([]byte(raw), []byte("?")))
+	}
+	runes := []rune(raw)
+	if len(runes) > 300 {
+		return string(runes[:297]) + "..."
 	}
 	return raw
 }

internal/ralph/telegram_test.go

@@ -9,6 +9,7 @@ import (
 	"strings"
 	"testing"
 	"time"
+	"unicode/utf8"
 )
 
 func TestParseTelegramChatIDs(t *testing.T) {
@@ -81,6 +82,32 @@ func TestSplitTelegramMessage(t *testing.T) {
 	}
 }
 
+func TestCompactTelegramErrorUnicodeSafe(t *testing.T) {
+	t.Parallel()
+
+	raw := strings.Repeat("🔥비트코인 자동화 ", 40)
+	got := compactTelegramError(raw)
+	if !utf8.ValidString(got) {
+		t.Fatalf("output must be valid UTF-8: %q", got)
+	}
+	if len([]rune(got)) > 300 {
+		t.Fatalf("output should be capped at 300 runes: %d", len([]rune(got)))
+	}
+}
+
+func TestCompactTelegramErrorInvalidUTF8Sanitized(t *testing.T) {
+	t.Parallel()
+
+	raw := string([]byte{0xff, 0xfe, 'a', 'b', 'c'})
+	got := compactTelegramError(raw)
+	if !utf8.ValidString(got) {
+		t.Fatalf("output must be valid UTF-8: %q", got)
+	}
+	if !strings.Contains(got, "abc") {
+		t.Fatalf("sanitized output should preserve readable content: %q", got)
+	}
+}
+
 func TestIsTelegramChatAllowed(t *testing.T) {
 	t.Parallel()