From 6b5828c1c074d9824387e7c21c54a643ba5f63f8 Mon Sep 17 00:00:00 2001
From: Mavis <Mavis@local>
Date: Mon, 8 Jun 2026 15:49:03 +0800
Subject: [PATCH] =?UTF-8?q?fix(translation):=20=E8=A7=84=E8=8C=83=E5=8C=96?=
 =?UTF-8?q?=20BCP-47=20lang=5Fsrc(=E9=81=BF=E5=85=8D=20en-gb/zh-cn=20?=
 =?UTF-8?q?=E7=AD=89=E8=A2=AB=20TMT=20=E6=8B=92)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 backend/app/services/translation/tencent.py | 20 +++++++++++++++++++-
 1 file changed, 19 insertions(+), 1 deletion(-)

diff --git a/backend/app/services/translation/tencent.py b/backend/app/services/translation/tencent.py
index 41d48f6..58b53e4 100644
--- a/backend/app/services/translation/tencent.py
+++ b/backend/app/services/translation/tencent.py
@@ -17,7 +17,22 @@ from app.services.translation.base import BaseTranslator, TranslationResult
 
 logger = logging.getLogger("news.translate.tencent")
 
-# 常见语种映射
+
+def _normalize_lang(code: str) -> str:
+    """把 BCP-47(可能带地区后缀)简化成主语言,TMT 用的 2 字母 code。
+
+    例:
+      en-gb / en-us / en-au -> en
+      zh-cn / zh-tw        -> zh
+      ja-jp                -> ja
+      "" / auto            -> 原样
+    """
+    if not code or code == "auto":
+        return code
+    return code.split("-")[0].lower()
+
+
+# 常见语种映射(归一化后再过这层;保留覆盖是兜底)
 _LANG_MAP = {
     "en": "en",
     "zh": "zh",
@@ -48,6 +63,9 @@ class TencentTranslator(BaseTranslator):
         if not text.strip():
             return TranslationResult(text=text, engine=self.name, chars=0)
 
+        # 关键:BCP-47 归一化(避免 en-gb / en-us / zh-cn / ja-jp 等被 TMT 拒)
+        source = _normalize_lang(source)
+        target = _normalize_lang(target)
         source = _LANG_MAP.get(source, source if source != "auto" else "auto")
         target = _LANG_MAP.get(target, target)