docs-i18n: avoid ambiguous body-only wrapper unwrap (openclaw#63808)

hxy91819 · web-flow · commit 164287f0569a · 2026-04-10T00:01:17.000+08:00
* docs-i18n: avoid ambiguous body-only wrapper unwrap

* docs: clarify targeted testing tip

* changelog: include docs-i18n follow-up thanks
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -7,7 +7,7 @@ Docs: https://docs.openclaw.ai
 ### Changes
 
 - macOS/Talk: add an experimental local MLX speech provider for Talk Mode, with explicit provider selection, local utterance playback, interruption handling, and system-voice fallback. (#63539) Thanks @ImLukeF.
-- Docs i18n: chunk raw doc translation, reject truncated tagged outputs, and recover from terminated Pi translation sessions without changing the default `openai/gpt-5.4` path. (#62969) Thanks @hxy91819.
+- Docs i18n: chunk raw doc translation, reject truncated tagged outputs, avoid ambiguous body-only wrapper unwrapping, and recover from terminated Pi translation sessions without changing the default `openai/gpt-5.4` path. (#62969, #63808) Thanks @hxy91819.
 
 ### Fixes
 
diff --git a/docs/help/testing.md b/docs/help/testing.md
@@ -26,6 +26,7 @@ Most days:
 - Faster local full-suite run on a roomy machine: `pnpm test:max`
 - Direct Vitest watch loop: `pnpm test:watch`
 - Direct file targeting now routes extension/channel paths too: `pnpm test extensions/discord/src/monitor/message-handler.preflight.test.ts`
+- Prefer targeted runs first when you are iterating on a single failure.
 - Docker-backed QA site: `pnpm qa:lab:up`
 
 When you touch tests or want extra confidence:
diff --git a/scripts/docs-i18n/doc_chunked_raw.go b/scripts/docs-i18n/doc_chunked_raw.go
@@ -231,14 +231,21 @@ func sanitizeDocChunkProtocolWrappers(source, translated string) string {
 			return body
 		}
 	}
-	body, ok := stripBodyOnlyWrapper(trimmedTranslated)
+	body, ok := stripBodyOnlyWrapper(source, trimmedTranslated)
 	if !ok || strings.TrimSpace(body) == "" {
 		return translated
 	}
 	return body
 }
 
-func stripBodyOnlyWrapper(text string) (string, bool) {
+func stripBodyOnlyWrapper(source, text string) (string, bool) {
+	sourceLower := strings.ToLower(source)
+	// When the source itself documents <body> tokens, a bare body-only payload is
+	// ambiguous: the trailing </body> can be literal translated content instead of
+	// a real wrapper close. Keep it for validation/retry instead of truncating.
+	if strings.Contains(sourceLower, strings.ToLower(bodyTagStart)) || strings.Contains(sourceLower, strings.ToLower(bodyTagEnd)) {
+		return "", false
+	}
 	lower := strings.ToLower(text)
 	bodyStartLower := strings.ToLower(bodyTagStart)
 	bodyEndLower := strings.ToLower(bodyTagEnd)
diff --git a/scripts/docs-i18n/doc_mode_test.go b/scripts/docs-i18n/doc_mode_test.go
@@ -512,18 +512,15 @@ func TestTranslateDocBodyChunkedStripsUppercaseBodyWrapper(t *testing.T) {
 	}
 }
 
-func TestSanitizeDocChunkProtocolWrappersStripsTopLevelWrapperEvenWhenSourceMentionsBodyTag(t *testing.T) {
+func TestSanitizeDocChunkProtocolWrappersKeepsBodyOnlyWrapperWhenSourceMentionsBodyTag(t *testing.T) {
 	t.Parallel()
 
 	source := "Use `<body>` and `</body>` in examples, but keep the paragraph text plain.\n"
 	translated := "<body>\nTranslated paragraph.\n</body>\n"
 
 	got := sanitizeDocChunkProtocolWrappers(source, translated)
-	if strings.Contains(got, "<body>") || strings.Contains(got, "</body>") {
-		t.Fatalf("expected top-level wrapper to be stripped, got %q", got)
-	}
-	if strings.TrimSpace(got) != "Translated paragraph." {
-		t.Fatalf("unexpected sanitized body %q", got)
+	if got != translated {
+		t.Fatalf("expected ambiguous body-only wrapper to remain unchanged for retry\nwant:\n%s\ngot:\n%s", translated, got)
 	}
 }
 
@@ -539,6 +536,21 @@ func TestSanitizeDocChunkProtocolWrappersKeepsLegitimateTopLevelBodyBlock(t *tes
 	}
 }
 
+func TestSanitizeDocChunkProtocolWrappersStripsBodyOnlyWrapperWhenSourceHasNoBodyTokens(t *testing.T) {
+	t.Parallel()
+
+	source := "Regular paragraph.\n"
+	translated := "<body>\nTranslated paragraph.\n</body>\n"
+
+	got := sanitizeDocChunkProtocolWrappers(source, translated)
+	if strings.Contains(got, "<body>") || strings.Contains(got, "</body>") {
+		t.Fatalf("expected body-only wrapper to be stripped, got %q", got)
+	}
+	if strings.TrimSpace(got) != "Translated paragraph." {
+		t.Fatalf("unexpected sanitized body %q", got)
+	}
+}
+
 func TestSanitizeDocChunkProtocolWrappersKeepsAmbiguousTaggedWrapperForRetry(t *testing.T) {
 	t.Parallel()
 

Original file line number	Diff line number	Diff line change
`@@ -231,14 +231,21 @@ func sanitizeDocChunkProtocolWrappers(source, translated string) string {`
`231`	`231`	`return body`
`232`	`232`	`}`
`233`	`233`	`}`
`234`		`- body, ok := stripBodyOnlyWrapper(trimmedTranslated)`
	`234`	`+ body, ok := stripBodyOnlyWrapper(source, trimmedTranslated)`
`235`	`235`	`if !ok \|\| strings.TrimSpace(body) == "" {`
`236`	`236`	`return translated`
`237`	`237`	`}`
`238`	`238`	`return body`
`239`	`239`	`}`
`240`	`240`
`241`		`-func stripBodyOnlyWrapper(text string) (string, bool) {`
	`241`	`+func stripBodyOnlyWrapper(source, text string) (string, bool) {`
	`242`	`+ sourceLower := strings.ToLower(source)`
	`243`	`+ // When the source itself documents <body> tokens, a bare body-only payload is`
	`244`	`+ // ambiguous: the trailing </body> can be literal translated content instead of`
	`245`	`+ // a real wrapper close. Keep it for validation/retry instead of truncating.`
	`246`	`+ if strings.Contains(sourceLower, strings.ToLower(bodyTagStart)) \|\| strings.Contains(sourceLower, strings.ToLower(bodyTagEnd)) {`
	`247`	`+ return "", false`
	`248`	`+ }`
`242`	`249`	`lower := strings.ToLower(text)`
`243`	`250`	`bodyStartLower := strings.ToLower(bodyTagStart)`
`244`	`251`	`bodyEndLower := strings.ToLower(bodyTagEnd)`