From 93b3e896c805b1946749c1a3ab3ea4c5423bb4e1 Mon Sep 17 00:00:00 2001
From: xiaoxia <xiaoxia@xiaoxiajianji.com>
Date: Tue, 6 Oct 2026 10:15:27 +0800
Subject: [PATCH] =?UTF-8?q?fix(viral-video):=20=E7=BC=96=E5=AF=BC=E8=84=9A?=
 =?UTF-8?q?=E6=9C=AC=E7=94=9F=E6=88=90timeout=2090=E2=86=92150s=20+=20?=
 =?UTF-8?q?=E4=BF=AEai=5Fclient=20logger=E5=8D=A0=E4=BD=8D=E7=AC=A6bug?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

P0根因：用户09:57反馈『生成不了文案』，排查确认：
1. #2215的call_llm→chat_completion修复本身正确，凌晨/上午日志均验证意图解析HTTP 200成功（33s返回1505 tokens）
2. 编导脚本fast tmo=90s超时（#2183沿用至今，非#2215引入）：doubao-seed-2-1-pro输出完整分镜脚本token量大，高峰期read timeout 90.1s，重试期间被watchtower容器重启打断
3. ai_client.py:287 logger.info格式串6个占位符只传5个参数（缺_req_timeout），每次chat_completion成功都打TypeError假错误栈，干扰排查

修复：
- 编导脚本fast tmo 90→150s，pro 120→150s，支持ENV VIRAL_VIDEO_SCRIPT_FAST_TIMEOUT/VIRAL_VIDEO_SCRIPT_PRO_TIMEOUT覆盖
- 补ai_client.py logger.info缺失的_req_timeout参数

文件变更：
- apps/worker/worker_app/tasks/viral_video.py: 编导脚本超时上调
- packages/shared/ai_client.py: logger占位符补参
---
 apps/worker/worker_app/tasks/viral_video.py | 13 +++++++------
 packages/shared/ai_client.py                |  1 +
 2 files changed, 8 insertions(+), 6 deletions(-)

diff --git a/apps/worker/worker_app/tasks/viral_video.py b/apps/worker/worker_app/tasks/viral_video.py
index ed648fe0f..04bda80cf 100644
--- a/apps/worker/worker_app/tasks/viral_video.py
+++ b/apps/worker/worker_app/tasks/viral_video.py
@@ -910,18 +910,19 @@ def _step_script_generation(job: ViralVideoJob, intent: dict, image_analysis: di
     _s = get_shared_settings()
     _fast = _s.doubao_fast_model
     _pro = getattr(_s, "doubao_model", None) or _fast
+    _script_fast_tmo = int(os.environ.get("VIRAL_VIDEO_SCRIPT_FAST_TIMEOUT", "150"))
+    _script_pro_tmo = int(os.environ.get("VIRAL_VIDEO_SCRIPT_PRO_TIMEOUT", "150"))
     try:
-        # 第一次：快模型 25s
-        normalized = _try_gen(_fast, 0.8, 2500, "fast-first", tmo=90)
+        # #2217: doubao-seed-2-1-pro生成长编导脚本高峰期>90s，上调到150s，支持ENV覆盖
+        normalized = _try_gen(_fast, 0.8, 2500, "fast-first", tmo=_script_fast_tmo)
         if normalized is not None:
             return normalized
-        # #2183: 实测pro 1500tok输出需75.8s，单次timeout提到90s
-        normalized = _try_gen(_fast, 0.6, 3200, "fast-retry", tmo=90)
+        normalized = _try_gen(_fast, 0.6, 3200, "fast-retry", tmo=_script_fast_tmo)
         if normalized is not None:
             return normalized
-        # 第三次：用主力模型兜底，给 120s
+        # 第三次：用主力模型兜底
         if _pro and _pro != _fast:
-            normalized = _try_gen(_pro, 0.7, 3500, "pro-fallback", tmo=120)
+            normalized = _try_gen(_pro, 0.7, 3500, "pro-fallback", tmo=_script_pro_tmo)
             if normalized is not None:
                 return normalized
         logger.warning("[爆款视频] 编导脚本三次都未生成合格结果，使用兜底脚本")
diff --git a/packages/shared/ai_client.py b/packages/shared/ai_client.py
index 5e45ca31f..3b4dc4692 100755
--- a/packages/shared/ai_client.py
+++ b/packages/shared/ai_client.py
@@ -291,6 +291,7 @@ class DoubaoClient:
                     data.get("usage", {}).get("completion_tokens", 0),
                     _elapsed,
                     attempt + 1,
+                    _req_timeout,
                 )
                 return content.strip()
             except Exception as e: