在不激活虚拟环境的情况下运行Python3 [英] Run Python3 without activating the virtual environment

问题描述

我的目标是在AWS Lambda服务上运行Python 3代码，该服务目前仅支持Python 2.7.这些是我已经完成的步骤.

My objective is to run Python 3 code on the AWS Lambda Service, which currently only supports Python 2.7. These are the steps I have done.

• 由于我在Mac上工作，因此请设置与相似的 AWS Lambda Linux实例.

• Since I work on a Mac, setup a docker image similar to the AWS Lambda Linux instance.

在Docker映像上从源代码构建 Python3.

Build Python3 from source on the docker image.

在docker映像中创建一个虚拟环境并将其复制到我的项目中.

In the docker image create a virtual environment and copy it to my project.

AWS Lambda要求您创建代码的zip并将其上传到他们的服务.对于这个原型，我有一个拉链，其根部有三个工件

AWS Lambda requires you to create a zip of the code and upload it to their service. For this prototype, I have a zip with three artifacts at the root

1. handler.py:这是一个Python 2.7文件.发生事件时(例如，在S3存储桶中创建新文件时)，AWS Lambda服务将执行此文件中的handler函数.

1. handler.py: This is a Python 2.7 file. The handler function in this file will be executed by the AWS Lambda Service when an event occurs (e.g. When a new file is created in a S3 bucket).

def handler(event, context):
    execution_uuid = uuid.uuid4()
    commands = '''
    source venv/bin/activate && venv/bin/python3.6 ./handler_python3.py --execution_uuid {ex_uuid}
    '''.format(ex_uuid=str(execution_uuid))
    p = Popen('/bin/bash', shell=True, stdin=PIPE, stdout=PIPE, stderr=PIPE, close_fds=True)
    stdout, stderr = p.communicate(commands)
    pprint(stdout)
    pprint(stderr)

handler_python3.py.这是先前的handler.py文件调用的Python3文件.注意正在读取execution_uuid.为了简洁起见，我已经删除了使用它的代码，但是我确实需要它，并且我正在使用argparse提取它.

handler_python3.py. This is the Python3 file called by the earlier handler.py file. Note the execution_uuid being read. I have taken out the code that uses it for brevity but I do need it and I am using argparse to extract it.

def read_execution_uuid():
    import argparse
    parser = argparse.ArgumentParser()
    parser.add_argument("--execution_uuid", required=True)
    args = parser.parse_args()

    return args.execution_uuid

def handler(event, context):
    import sys
    print(sys.path)

if __name__ == '__main__':
    execution_uuid = read_execution_uuid()
    handler(event, context)

1. venv文件夹.这是我从docker映像复制的虚拟环境文件夹.

1. venv folder. This is the virtual environment folder that I copied from the docker image.

运行AWS Lambda服务时，出现以下错误

When I run the AWS Lambda Service, I get the following error

Traceback (most recent call last):
  File "./handler_python3.py", line 38, in <module>
    execution_uuid = read_execution_uuid()
  File "./handler_python3.py", line 7, in read_execution_uuid
    import argparse
ModuleNotFoundError: No module named \'argparse\'

注意:

• 如果我删除了argparse代码并执行了handler_python3.py中的handler函数，它将显示sys.path

• If I remove the argparse code and the handler function in handler_python3.py executes, it shows the following values for sys.path

['/var/task', '/var/runtime', '/var/task/venv/lib/python36.zip', '/var/task/venv/lib/python3.6', '/var/task/venv/lib/python3.6/lib-dynload', '/usr/local/lib/python3.6', '/var/task/venv/lib/python3.6/site-packages']

注意:

• 我可以显式安装argparse.但是我不想.
• 请注意python 2.7文件handler.py中的source venv/bin/activate命令.尽管它在本地工作，但不适用于Lambda实例.

• I can install argparse explicitly. But I'd like not to.
• Note the source venv/bin/activate command in the python 2.7 file handler.py. That doesn't work on the Lambda instance though it works locally.

推荐答案

创建虚拟环境不会复制/usr/local/lib/python3.6目录中的所有模块.我必须将所有文件复制到那里.

Creating a virtual environment does not copy all the modules from the /usr/local/lib/python3.6 directory. I had to copy all the files there.

这篇关于在不激活虚拟环境的情况下运行Python3的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！