获取用户在 conda 环境中安装的软件包列表无依赖

如何解决获取用户在 conda 环境中安装的软件包列表无依赖

此线程与to this one（可能还有also to this one）相关。比如说，我克隆了默认的 base conda 环境，随着时间的推移，我在克隆的环境中使用 conda install 和 pip install 安装了许多软件包。过了一会儿，我想获得这些包的列表（最好的情况是没有与这些包一起安装的依赖项）。有没有办法做到这一点？我想出了这个简单的脚本，它只读取用 .json 生成的两个 conda env export --json 文件，并返回只能在克隆环境中但不能在基本环境中找到的唯一包。但最终的列表仍然包含很多我没有明确安装但作为依赖项安装的包。

import json
  
# both base and 'base + x' environment where exported using 
# conda env export --json -n env_name > env_name.json
# get json as dicts
with open('./base.json') as json_file:
    base_env = json_file.read()
    base_env = json.loads(base_env)
  
with open('./base_clone.json') as json_file:
    base_clone_env = json_file.read()
    base_clone_env = json.loads(base_clone_env)

# get packages for both environments as lists
base_list = base_env['dependencies']
base_clone_list = base_clone_env['dependencies']

# the 'base + x' environment also contains pip installed packages that we get
# as separate list
base_clone_pip = base_clone_list.pop()['pip']

# we only want to find out which packages are installed in the 'base + x' 
# environment that are not installed in the base environment (not if packages
# differ in the versions so we strip out the version part)
base_list = [s.split('=',1)[0] for s in base_list]
base_clone_list = [s.split('=',1)[0] for s in base_clone_list]

# these are the final lists
base_clone_pip = [s.split('=',1)[0] for s in base_clone_pip]
envs_diff = list(set(base_clone_list) - set(base_list))