Python包管理工具(pip)

Posted 2020-09-16

 

1 前言

pip 是一个Python包管理工具，主要是用于安装 PyPI 上的软件包，可以替代 easy_install 工具。

• GitHub: https://github.com/pypa/pip
• Doc: https://pip.pypa.io/en/latest/

2 获取pip

 

2.1 脚本安装pip

$ curl -O https://raw.github.com/pypa/pip/master/contrib/get-pip.py

$ python get-pip.py

2.2 使用包管理软件安装

$ sudo yum install python-pip

$ sudo apt-get install python-pip

2.3 更新pip

$ pip install -U pip

3 pip基本使用

 

3.1 安装PyPI软件

$ pip install SomePackage

  [...]
  Successfully installed SomePackage

3.2 查看具体安装文件

$ pip show --files SomePackage

  Name: SomePackage
  Version: 1.0
  Location: /my/env/lib/pythonx.x/site-packages
  Files:
   ../somepackage/__init__.py
   [...]

3.3 查看哪些软件需要更新

$ pip list --outdated

  SomePackage (Current: 1.0 Latest: 2.0)

3.4 升级软件包

$ pip install --upgrade SomePackage

  [...]
  Found existing installation: SomePackage 1.0
  Uninstalling SomePackage:
    Successfully uninstalled SomePackage
  Running setup.py install for SomePackage
  Successfully installed SomePackage

3.5 卸载软件包

$ pip uninstall SomePackage

  Uninstalling SomePackage:
    /my/env/lib/pythonx.x/site-packages/somepackage
  Proceed (y/n)? y
  Successfully uninstalled SomePackage

4 pip简明手册

 

4.1 安装具体版本软件

$ pip install SomePackage            # latest version
$ pip install SomePackage==1.0.4     # specific version
$ pip install ‘SomePackage>=1.0.4‘     # minimum version

4.2 Requirements文件安装依赖软件

Requirements文件 一般记录的是依赖软件列表，通过pip可以一次性安装依赖软件包:

$ pip freeze > requirements.txt

$ pip install -r requirements.txt

4.3 列出软件包清单

$ pip list

$ pip list --outdated

ipython (Current: 1.2.0 Latest: 2.3.0)

4.4 查看软件包信息

$ pip show pip
---
Name: pip
Version: 1.4.1
Location: /Library/Python/2.7/site-packages/pip-1.4.1-py2.7.egg
Requires:

$ pip show pyopencl
---
Name: pyopencl
Version: 2014.1
Location: /Library/Python/2.7/site-packages
Requires: pytools, pytest, decorator

4.5 搜寻

$ pip search pycuda

pycuda                    - Python wrapper for Nvidia CUDA
pyfft                     - FFT library for PyCuda and PyOpenCL
cudatree                  - Random Forests for the GPU using PyCUDA
reikna                    - GPGPU algorithms for PyCUDA and PyOpenCL
compyte                   - A common set of compute primitives for PyCUDA and PyOpenCL (to be created)

4.6 配置文件

配置文件: $HOME/.pip/pip.conf, 举例:

[global]
timeout = 60
index-url = http://download.zope.org/ppix

[install]
ignore-installed = true
no-dependencies = yes

4.7 命令行自动补全

对于bash:

$ pip completion --bash >> ~/.profile

对于zsh:

$ pip completion --zsh >> ~/.zprofile

加载此配置文件后，则pip命令支持自动补全功能.

5 后记

应该尽量使用pip，不要继续使用easy_install.