Merge pull request #14907 from kevin9327/fix/bounded-segment-rune

fix(evidence): cut quoted mutation-scan segments on a rune boundary

internal/safety/evidence/bounded_segment_test.go

@@ -0,0 +1,29 @@
+package evidence
+
+import (
+	"strings"
+	"testing"
+	"unicode/utf8"
+)
+
+func TestBoundedSegmentCutsOnRuneBoundary(t *testing.T) {
+	// 120 bytes into this segment falls inside a three-byte character.
+	segment := `python3 -c "print('` + strings.Repeat("测试", 30) + `')"`
+	got := boundedSegment(segment)
+	if !utf8.ValidString(got) {
+		t.Fatalf("boundedSegment split a character: %q", got)
+	}
+	kept := strings.TrimSuffix(got, "…")
+	if kept == got || !strings.HasPrefix(segment, kept) {
+		t.Fatalf("boundedSegment = %q, want a prefix of the segment ending in …", got)
+	}
+	if len(kept) > unprovenSegmentLimit || len(kept) < unprovenSegmentLimit-3 {
+		t.Fatalf("boundedSegment kept %d bytes, want just under %d", len(kept), unprovenSegmentLimit)
+	}
+}
+
+func TestBoundedSegmentKeepsShortSegment(t *testing.T) {
+	if got := boundedSegment("  python -c 'print(1)'  "); got != "python -c 'print(1)'" {
+		t.Fatalf("boundedSegment = %q", got)
+	}
+}

internal/safety/evidence/mutation_scan.go

@@ -3,6 +3,7 @@ package evidence
 import (
 	"encoding/json"
 	"strings"
+	"unicode/utf8"
 
 	"reasonix/internal/base/shellparse"
 )
@@ -113,7 +114,11 @@ func namedInlineNonTerminalPair(args json.RawMessage) (string, string) {
 func boundedSegment(segment string) string {
 	segment = strings.TrimSpace(segment)
 	if len(segment) > unprovenSegmentLimit {
-		segment = segment[:unprovenSegmentLimit] + "…"
+		cut := unprovenSegmentLimit
+		for cut > 0 && !utf8.RuneStart(segment[cut]) {
+			cut--
+		}
+		segment = segment[:cut] + "…"
 	}
 	return segment
 }