From 9a4f51d3781c5534805dcf857831f6f86ab29172 Mon Sep 17 00:00:00 2001
From: panda <1415243231@qq.com>
Date: Sat, 23 May 2026 09:20:55 +0800
Subject: [PATCH] fix: add retry limit to startup wait loops to prevent
 infinite hang

Each service wait loop now fails after 30 retries (~60s) instead of
spinning forever when a port is occupied by a stuck process.
Also added cleanup label that kills partially-started services on failure.
---
 start.bat     | 13 ++++++++++++-
 start_all.bat | 42 +++++++++++++++++++++++++++++++++++++++---
 2 files changed, 51 insertions(+), 4 deletions(-)

diff --git a/start.bat b/start.bat
index 53e6c58..bdc90a2 100644
--- a/start.bat
+++ b/start.bat
@@ -29,10 +29,21 @@ start "jrxml-validator" /MIN .venv\Scripts\python.exe -c "import uvicorn; uvicor
 
 :: 等待验证服务就绪 (用 PowerShell 检测)
 echo [等待] 验证服务就绪...
+set /a RETRY=0
 :wait_val
 ping -n 2 127.0.0.1 >nul
 powershell -Command "try{$r=Invoke-WebRequest -Uri http://localhost:8001/health -TimeoutSec 2 -UseBasicParsing;exit 0}catch{exit 1}" >nul 2>&1
-if errorlevel 1 goto wait_val
+if not errorlevel 1 goto val_ok
+set /a RETRY+=1
+if %RETRY% GEQ 30 (
+    echo    [失败] 验证服务启动超时，请检查端口 8001 是否被占用
+    taskkill /F /FI "WINDOWTITLE eq jrxml-validator*" >nul 2>&1
+    for /f "tokens=5" %%a in ('netstat -ano ^| findstr ":8001.*LISTENING"') do taskkill /F /PID %%a >nul 2>&1
+    pause
+    exit /b 1
+)
+goto wait_val
+:val_ok
 echo    :8001 就绪
 
 :: 启动 API 服务 (前台，Ctrl+C 退出)
diff --git a/start_all.bat b/start_all.bat
index 72309b3..45f43dd 100644
--- a/start_all.bat
+++ b/start_all.bat
@@ -23,19 +23,35 @@ echo.
 :: 1. 验证服务
 echo [1/3] 验证服务 :8001
 start "jrxml-validator" /MIN .venv\Scripts\python.exe -c "import uvicorn; uvicorn.run('validation_service.main:app',host='0.0.0.0',port=8001,reload=False)"
+set /a RETRY=0
 :wait_val
 ping -n 2 127.0.0.1 >nul
 powershell -Command "try{$r=Invoke-WebRequest -Uri http://localhost:8001/health -TimeoutSec 2 -UseBasicParsing;exit 0}catch{exit 1}" >nul 2>&1
-if errorlevel 1 goto wait_val
+if not errorlevel 1 goto val_ok
+set /a RETRY+=1
+if %RETRY% GEQ 30 (
+    echo    [失败] 验证服务启动超时，请检查端口 8001 是否被占用
+    goto cleanup
+)
+goto wait_val
+:val_ok
 echo    :8001 就绪
 
 :: 2. API 服务
 echo [2/3] API 服务 :8000
 start "jrxml-api" /MIN .venv\Scripts\python.exe -c "import uvicorn; uvicorn.run('api_server:app',host='0.0.0.0',port=8000,reload=False)"
+set /a RETRY=0
 :wait_api
 ping -n 2 127.0.0.1 >nul
 powershell -Command "try{$r=Invoke-WebRequest -Uri http://localhost:8000/api/health -TimeoutSec 2 -UseBasicParsing;exit 0}catch{exit 1}" >nul 2>&1
-if errorlevel 1 goto wait_api
+if not errorlevel 1 goto api_ok
+set /a RETRY+=1
+if %RETRY% GEQ 30 (
+    echo    [失败] API 服务启动超时，请检查端口 8000 是否被占用
+    goto cleanup
+)
+goto wait_api
+:api_ok
 echo    :8000 就绪
 
 :: 3. 前端
@@ -49,10 +65,18 @@ if not exist "%~dp0frontend\node_modules" (
 cd /d "%~dp0frontend"
 start "jrxml-frontend" /MIN cmd /c "npm run dev"
 cd /d "%~dp0"
+set /a RETRY=0
 :wait_fe
 ping -n 3 127.0.0.1 >nul
 powershell -Command "try{$r=Invoke-WebRequest -Uri http://localhost:5173 -TimeoutSec 3 -UseBasicParsing;exit 0}catch{exit 1}" >nul 2>&1
-if errorlevel 1 goto wait_fe
+if not errorlevel 1 goto fe_ok
+set /a RETRY+=1
+if %RETRY% GEQ 30 (
+    echo    [失败] 前端启动超时，请检查端口 5173 是否被占用
+    goto cleanup
+)
+goto wait_fe
+:fe_ok
 echo    :5173 就绪
 
 echo.
@@ -64,3 +88,15 @@ echo     验证:   http://localhost:8001/health
 echo   运行 stop.bat 停止所有服务
 echo ================================================
 pause
+exit /b 0
+
+:cleanup
+echo [清理] 停止已启动的服务...
+taskkill /F /FI "WINDOWTITLE eq jrxml-validator*" >nul 2>&1
+taskkill /F /FI "WINDOWTITLE eq jrxml-api*" >nul 2>&1
+taskkill /F /FI "WINDOWTITLE eq jrxml-frontend*" >nul 2>&1
+for /f "tokens=5" %%a in ('netstat -ano ^| findstr ":8001.*LISTENING"') do taskkill /F /PID %%a >nul 2>&1
+for /f "tokens=5" %%a in ('netstat -ano ^| findstr ":8000.*LISTENING"') do taskkill /F /PID %%a >nul 2>&1
+echo 已清理，请重试
+pause
+exit /b 1