如何在Robot Framework中使Trace32的self.t32api保持全局作用域？

我开发了一个与Lauterbach Trace32通信的Robot Framework测试套件，希望通过循环调用函数并在每次循环中设置断点。已编写Python脚本作为Robot库，但仅第一次循环的断点生效，后续循环失效。现需让库中的self.t32api保持全局作用域，直至测试完全执行完毕。

test.robot 文件 #

import os

*** Settings ***
Documentation  simple test script to control Trace32
Library        Collections
Library        can.Trace32


Suite Setup
Suite Teardown

*** Variables ***
${temp}                1
    
*** Test Cases ***
Check Input and Output
    [Documentation]  test script
    [Setup]
    
    #Retrive Data . This list has 5 values
    ${MainList}  Create List 
    
    #start debugger
    start Debugger
    #connect debugger
    Connect Debugger
    
    #Iterate 5 times
    FOR  ${UserAttribute}  IN  @{MainList}
        
        #sleep 1 sec
        Sleep  1 seconds
        
        #call for breakpoint
        break Debugger
        
        ${temp} +=1
    END
    
   Disconnect Debugger

    [Teardown]

Trace32 Python库脚本 #

import time
import ctypes
from ctypes import c_void_p
import enum
T32_DEV = 1

class Trace32:
  
  def start_Debugger(self):
    self.t32api = ctypes.cdll.LoadLibrary('D:/test/api/python/t32api64.dll')
    self.t32api.T32_Config(b"NODE=",b"localhost")
    self.t32api.T32_Config(b"PORT=",b"20000")
    self.t32api.T32_Config(b"PACKLEN=",b"1024")
    rc = self.t32api.T32_GetChannelSize()
    ch1 = ctypes.create_string_buffer(rc)
    self.t32api.T32_GetChannelDefaults(ctypes.cast(ch1,ctypes.c_void_p))
    ch2 = ctypes.create_string_buffer(rc)
    self.t32api.T32_GetChannelDefaults(ctypes.cast(ch2,ctypes.c_void_p))
    self.t32api.T32_SetChannel(ctypes.cast(ch2,c_void_p))
    
    
  def Connect_Debugger(self):
    rc = self.t32api.T32_Init()
    rc = self.t32api.T32_Attach(T32_DEV)
    
    
  def breakpoint_Debugger(self):
    rc = self.t32api.T32_Ping()
    time.sleep(2)
    rc = self.t32api.T32_Cmd(b"InterCom M7_0 Break")
    time.sleep(3)
    rc = self.t32api.T32_Cmd(b"InterCom M7_0 Go")
    time.sleep(2)
    rc = self.t32api.T32_Cmd(b"InterCom M7_0 break.Set My_func")
    time.sleep(2)
    
  
  def Disconnect_Debugger(self):
    rc = self.t32api.T32_Exit()

问题根源是Robot Framework默认会为关键字/测试用例创建新的库实例，导致每次循环调用时t32api被重新初始化，丢失了之前的连接状态。以下两种方式可实现t32api的全局作用域：

方式1：使用类静态属性 #

将t32api定义为类的静态变量，确保整个测试过程中只初始化一次连接：

import time
import ctypes
from ctypes import c_void_p
import enum
T32_DEV = 1

class Trace32:
    # 类级静态属性，全局共享
    t32api = None
  
    def start_Debugger(self):
        if Trace32.t32api is None:
            Trace32.t32api = ctypes.cdll.LoadLibrary('D:/test/api/python/t32api64.dll')
            Trace32.t32api.T32_Config(b"NODE=",b"localhost")
            Trace32.t32api.T32_Config(b"PORT=",b"20000")
            Trace32.t32api.T32_Config(b"PACKLEN=",b"1024")
            rc = Trace32.t32api.T32_GetChannelSize()
            ch1 = ctypes.create_string_buffer(rc)
            Trace32.t32api.T32_GetChannelDefaults(ctypes.cast(ch1,ctypes.c_void_p))
            ch2 = ctypes.create_string_buffer(rc)
            Trace32.t32api.T32_GetChannelDefaults(ctypes.cast(ch2,ctypes.c_void_p))
            Trace32.t32api.T32_SetChannel(ctypes.cast(ch2,c_void_p))
    
    
    def Connect_Debugger(self):
        rc = Trace32.t32api.T32_Init()
        rc = Trace32.t32api.T32_Attach(T32_DEV)
    
    
    def breakpoint_Debugger(self):
        rc = Trace32.t32api.T32_Ping()
        time.sleep(2)
        rc = Trace32.t32api.T32_Cmd(b"InterCom M7_0 Break")
        time.sleep(3)
        rc = Trace32.t32api.T32_Cmd(b"InterCom M7_0 Go")
        time.sleep(2)
        rc = Trace32.t32api.T32_Cmd(b"InterCom M7_0 break.Set My_func")
        time.sleep(2)
    
  
    def Disconnect_Debugger(self):
        if Trace32.t32api is not None:
            rc = Trace32.t32api.T32_Exit()
            Trace32.t32api = None  # 测试结束后清理资源

方式2：配置Robot Framework库为单例 #

在Robot Framework的库导入语句中添加SINGLETON参数，强制整个测试套件只创建一个库实例：

*** Settings ***
Documentation  simple test script to control Trace32
Library        Collections
Library        can.Trace32    SINGLETON

额外修正点 #

• 测试用例中break Debugger关键字与库中breakpoint_Debugger方法名不匹配，需统一为breakpoint Debugger或修改方法名为break_Debugger
• ${MainList}需添加实际元素（如Create List 1 2 3 4 5），否则循环不会执行
• 建议将start Debugger/Connect Debugger移至Suite Setup，Disconnect Debugger移至Suite Teardown，避免重复初始化

内容的提问来源于stack exchange，提问作者user2986042