fix(verification): 验证码提取保留原文大小写

CHANGELOG.md

@@ -6,6 +6,7 @@ All notable changes to OutlookMail Plus are documented in this file.
 
 ### 修复 / Bug Fixes
 
+- **验证码提取大小写被篡改**：验证码提取收口处（规则分支 `_smart_extract_code_by_keywords` / `_fallback_extract_code` 及历史函数 `smart_extract_verification_code` / `fallback_extract_verification_code`，以及 AI 回退 `enhance_verification_with_ai_fallback`）不再强制 `.upper()`，`verification_code` 与 `formatted` 保持邮件原文大小写（如 `ab12cd` 不再返回 `AB12CD`）。补充回归测试。
 - **Issue #65 Watchtower 容器镜像过时**：`docker-compose.yml` 中固定 Watchtower 版本为 `containrrr/watchtower:1.7.1`，避免本地缓存的旧版镜像（内嵌 Docker 客户端 API 1.25）连接新版本 Docker Engine（要求 API 1.44+）时失败。README 新增故障排查指引。
 
 ## [v2.7.0] - 2026-05-29

outlook_web/services/verification_extractor.py

@@ -154,7 +154,7 @@ def smart_extract_verification_code(email_content: str) -> Optional[str]:
                 # 过滤掉纯字母的匹配（验证码通常包含数字）
                 for match in matches:
                     if any(c.isdigit() for c in match):
-                        return match.upper()
+                        return match
 
     return None
 
@@ -183,8 +183,6 @@ def fallback_extract_verification_code(email_content: str) -> Optional[str]:
     # 过滤规则
     filtered = []
     for match in matches:
-        match_upper = match.upper()
-
         # 必须包含至少一个数字
         if not any(c.isdigit() for c in match):
             continue
@@ -209,7 +207,7 @@ def fallback_extract_verification_code(email_content: str) -> Optional[str]:
             if 2020 <= num <= 2030:
                 continue
 
-        filtered.append(match_upper)
+        filtered.append(match)
 
     return filtered[0] if filtered else None
 
@@ -460,7 +458,7 @@ def _smart_extract_code_by_keywords(email_content: str, code_re: re.Pattern) ->
         for m in code_re.finditer(context):
             value = m.group(0)
             if value and any(c.isdigit() for c in value):
-                return value.upper()
+                return value
 
     return None
 
@@ -492,7 +490,7 @@ def _fallback_extract_code(email_content: str, code_re: re.Pattern) -> Optional[
             if 2020 <= num <= 2030:
                 continue
 
-        candidates.append(value.upper())
+        candidates.append(value)
 
     return candidates[0] if candidates else None
 
@@ -1055,7 +1053,7 @@ def _apply_output_policy(payload: Dict[str, Any]) -> Dict[str, Any]:
 
     updated = False
     if ai_code:
-        result["verification_code"] = ai_code.upper()
+        result["verification_code"] = ai_code
         result["code_confidence"] = "high" if ai_confidence == "high" else "low"
         updated = True
     if ai_link:

tests/test_ai_fallback_trigger_condition.py

@@ -193,6 +193,31 @@ def test_code_high_link_low_enforces_mutual_exclusion(self, mock_config, mock_ai
         self.assertEqual(result["link_confidence"], "low")
 
 
+class AiFallbackCasePreservationTests(unittest.TestCase):
+    """验证码大小写：AI fallback 返回的验证码必须保持原大小写"""
+
+    @patch("outlook_web.services.verification_extractor._call_verification_ai")
+    @patch("outlook_web.services.verification_extractor.get_verification_ai_runtime_config")
+    def test_ai_code_case_preserved(self, mock_config, mock_ai_call):
+        """AI 返回小写 code → 结果保持小写，不转大写"""
+        mock_config.return_value = _AI_CONFIG
+        mock_ai_call.return_value = {
+            "schema_version": "verification_ai_v1",
+            "verification_code": "cbdfdb",
+            "verification_link": "",
+            "confidence": "high",
+            "reason": "test",
+        }
+        extracted = _make_extracted("low", "low")
+
+        result = extractor.enhance_verification_with_ai_fallback(email=_EMAIL_OBJ, extracted=extracted)
+
+        mock_ai_call.assert_called_once()
+        self.assertTrue(result.get("ai_used"))
+        self.assertEqual(result["verification_code"], "cbdfdb")
+        self.assertEqual(result["formatted"], "cbdfdb")
+
+
 class AiFallbackEdgeCaseTests(unittest.TestCase):
     """边界情况：缺失 confidence 字段时默认为 low"""
 

tests/test_verification_extractor.py

@@ -413,6 +413,51 @@ def test_extract_email_text_body_content_type(self):
         text = extract_email_text(email)
         self.assertIn("555666", text)
 
+    # ==================== 大小写保持测试（回归） ====================
+
+    def test_smart_extract_preserves_lowercase(self):
+        """
+        测试用例：智能识别 - 保持小写验证码原样
+
+        测试目的：验证智能识别不再私自转换大小写（含数字的小写字母验证码）
+        """
+        content = "Your verification code is ab12cd. Please verify."
+        result = smart_extract_verification_code(content)
+        self.assertEqual(result, "ab12cd")
+
+    def test_smart_extract_preserves_mixed_case(self):
+        """
+        测试用例：智能识别 - 保持大小写混合验证码原样
+        """
+        content = "您的验证码是 Ab12Cd，请尽快使用。"
+        result = smart_extract_verification_code(content)
+        self.assertEqual(result, "Ab12Cd")
+
+    def test_fallback_extract_preserves_lowercase(self):
+        """
+        测试用例：保底提取 - 保持小写验证码原样
+        """
+        content = "Please use ab12cd to complete your registration."
+        result = fallback_extract_verification_code(content)
+        self.assertEqual(result, "ab12cd")
+
+    def test_fallback_extract_preserves_mixed_case(self):
+        """
+        测试用例：保底提取 - 保持大小写混合验证码原样
+        """
+        content = "Please use Ab12Cd to complete your registration."
+        result = fallback_extract_verification_code(content)
+        self.assertEqual(result, "Ab12Cd")
+
+    def test_extract_verification_info_preserves_case(self):
+        """
+        测试用例：完整流程 - verification_code 与 formatted 均保持原大小写
+        """
+        email = {"body": "Your verification code is ab12cd."}
+        result = extract_verification_info_from_text(email["body"])
+        self.assertEqual(result["verification_code"], "ab12cd")
+        self.assertEqual(result["formatted"], "ab12cd")
+
 
 if __name__ == "__main__":
     unittest.main()