从Python调用时SEG故障，但通常从命令行运行

Question

我正在编写一个用于连续集成和测试的python脚本，该脚本将由bitten调用。我们的单元测试使用Google测试框架。每个软件组件都有一个BASH脚本，该脚本运行配置和其他必需的服务，并运行GTEST可执行文件。Python脚本在存储库中寻找BASH脚本，并使用OS.Popen（）命令调用每个脚本。

python脚本（unittest.py）

#!/usr/bin/python
import os
import fnmatch
import sys
import subprocess
repository_location = '/home/actuv/workspace/eclipse/iccs/'
unit_test_script_name = 'RunUnitTests.sh'
def file_locator(repo, script_name):
    # Function for determining all unit test scripts
    test_location = []
    for root, dirnames, filenames in os.walk(repo):
        for filename in fnmatch.filter(filenames, script_name):
            test_location.append(os.path.join(root))
    return test_location
def run_tests(test_locations, script_name):
    # Runs test scripts located at each test location
    for tests in test_locations:
        cmd = 'cd ' + tests + ';./' + script_name
        print 'Running Unit Test at: ' + tests
        os.popen(cmd)
################    MAIN    ################
# Find Test Locations
script_locations = file_locator(repository_location, unit_test_script_name)
# Run tests located at each location
run_tests(script_locations)
# End of tests
sys.exit(0)

bash脚本

#!/bin/sh
echo "Running unit tests..."
# update the LD_LIBRARY_PATH to include paths to our shared libraries
# start the test server
# Run the tests
# wait to allow all processes to end before terminating the server
sleep 10s

当我从终端窗口手动运行BASH脚本时，它运行良好。当我有Python脚本调用bash脚本时，我会在bash脚本的testsingleclient和testmulticlientla线上得到一个分割故障。

Answer 1

尝试更换

os.popen(cmd)

proc = subprocess.Popen('./scriptname', shell = True, 
                       cwd = tests)
proc.communicate()

Answer 2

绝对查看子进程模块 - 特别是查看 subprocess.call（）便利方法。我投入了操作系统检查，以确保您的测试目录也存在。

def run_tests(test_locations, script_name):
    # Runs test scripts located at each test location
    for tests in test_locations:
        #  Make sure tests directory exists and is a dir
        if os.path.isdir(tests):
            print 'Running Unit Test at: ' + tests
            subprocess.call(script_name, shell=True, cwd=tests)

另外 - 您对Stdout和Stderr的观察是正确的，尤其是在有很多数据时。当有大量或未知数的输出时，我将temp文件用于stdout/stderr。
例如。

def execute_some_command(cmd="arbitrary_exe"):
    """ Execute some command using subprocess.call()"""
    #  open/create temportary file for stdout    
    tmp_out=file('.tmp','w+')
    #  Run command, pushing stdout to tmp_out file handle
    retcode = subprocess.call(cmd, stdout=tmp_out, shell=True)
    if retcode != 0:
        # do something useful (like bailout) based on the OS return code
        print "FAILED"
    #  Flush any queued data
    tmp_out.flush()
    #  Jump to the beginning
    tmp_out.seek(0)
    #  Parse output
    for line in tmp_out.readlines():
        # do something useful with output
    # cleanup 
    tmp_out.close()
    os.remove(_out.name)
return     

查看python文件对象上的方法，以了解如何从 _OUT 文件句柄中处理您的Stdout数据。

良好的狩猎。