Jupyter容器搭建`开发环境`过程记录

黑麋鹿

2023-02-08 2023-02-08 约 2291 字预计阅读 5 分钟

概述

为了搭建一个开箱即用的Python开发环境，结合个人平时用到的场景，将常用的Python库封装进Docker容器中，实现换电脑后可迅速重建开发环境，同时也是用于内网（无法连接互联网）环境的开发环境快速准确搭建。遂有此文：

`Jupyter`官方容器关系图，供选择时参考

https://jupyter-docker-stacks.readthedocs.io/en/latest/using/selecting.html#image-relationships 经测试，base、minimal等容器开启后的默认状态（不启动Python等内核）内存消耗都差不多，都是 85M 左右。

常用启动参数

重启jupyter内核不会重启容器

-e RESTARTABLE=yes 所有的容器默认启动JupyterLab，要改回传统的Notebook，需要指定变量DOCKER_STACKS_JUPYTER_CMD参数来启动：

DOCKER_STACKS_JUPYTER_CMD	后端	前端
lab (default)	Jupyter Server	JupyterLab
notebook	Jupyter Notebook	Jupyter Notebook
nbclassic	Jupyter Server	Jupyter Notebook
server	Jupyter Server	None
retro*	Jupyter Server	RetroLab

最高权限运行

默认情况下jovyan用户不能使用sudu权限，也不能通过为该用户设置密码来获得root权限，这是为了暴露在公网上的jupyterlab容器的安全。但是，有一个名为GRANT_SUDO的环境变量改变这一点。以下启动将赋予jovyan用户无需密码的root权限: docker run -it --rm -p 8888:8888 -e GRANT_SUDO=yes --user root gcr.io/kubeflow-images-staging/tensorflow-notebook-cpu

选择`JupyterLab`

JupyterLab具有更方便的布局、更完善的插件支持等等，改用JupyterLab。自行构建可参考：https://www.cnblogs.com/ulysessweb/p/14317515.html 关键是apt-get安装库文件之前使用USER root来使用root权限进行安装，否则安装不了，库文件安全之后，改回官方提供的默认用户USER $NB_UID进行Python包安装

自定义密码

https://jupyter-docker-stacks.readthedocs.io/en/latest/using/common.html#jupyter-server-options

中文包

pip install jupyterlab-language-pack-zh-CN

自定义构建

成品镜像

以下Dockerfile对应镜像地址：https://hub.docker.com/r/blackelks/jupyterlab-datascience

`Dockerfile`

构建命令docker build -t docker.io/blackelks/jupyterlab-datascience .

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62


FROM jupyter/datascience-notebook

# 安装lsp自动补全插件、excel查看插件,中文语言包、 onedarkpro主题、 创建硬链接（lsp官方推荐）
RUN pip install --no-cache-dir jupyterlab-language-pack-zh-CN jupyterlab-lsp 'python-lsp-server[all]' jupyterlab-drawio \
    && jupyterlab_execute_time pyinstall openpyxl xlrd XlsxWriter xlutils xlwings xlwt pdfminer python-docx docx2pdf \
    && pyexecjs selenium aiohttp httpx feapder[all] beautifulsoup4 requests lxml html5lib playwright matplotlib_inline \
    && opencv-python matplotlib pyecharts scipy plotly wordcloud flask django fastapi psycopg2-binary pymongo sqlalchemy \
    && pymysql pandas geopandas jieba tqdm unzip ddddocr loguru lars -i http://pypi.douban.com/simple/ --trusted-host pypi.douban.com \
    && playwright install && ln -s / .lsp_symlink 

RUN jupyter labextension install jupyterlab_onedarkpro 
RUN jupyter labextension install jupyterlab-spreadsheet

# 设置显示语言为中文
RUN mkdir -p /home/jovyan/.jupyter/lab/user-settings/@jupyterlab/translation-extension/ \
    && echo '{"locale":"zh_CN"}'>/home/jovyan/.jupyter/lab/user-settings/@jupyterlab/translation-extension/plugin.jupyterlab-settings \
    # 不接收新闻
    && mkdir -p /home/jovyan/.jupyter/lab/user-settings/@jupyterlab/apputils-extension/ \
    && echo '{"fetchNews": "false"}'>/home/jovyan/.jupyter/lab/user-settings/@jupyterlab/apputils-extension/notification.jupyterlab-settings \
    # 启用插件
    && mkdir -p /home/jovyan/.jupyter/lab/user-settings/@jupyterlab/extensionmanager-extension/ \
    && echo '{"disclaimed": true}'>/home/jovyan/.jupyter/lab/user-settings/@jupyterlab/extensionmanager-extension/plugin.jupyterlab-settings \
    
    # 设置默认主题为JupyterLab Dark----the one of two default themes.
    && echo '{"theme": "JupyterLab Dark"}'>/home/jovyan/.jupyter/lab/user-settings/@jupyterlab/apputils-extension/themes.jupyterlab-settings \
    # 开启自动补全
    && mkdir -p /home/jovyan/.jupyter/lab/user-settings/@krassowski/jupyterlab-lsp \
    && echo '{"continuousHinting": true}'>/home/jovyan/.jupyter/lab/user-settings/@krassowski/jupyterlab-lsp/completion.jupyterlab-settings \
    # 设置默认开启自动框号补全等
    && mkdir -p /home/jovyan/.jupyter/lab/user-settings/@jupyterlab/notebook-extension/ \
    && echo '{"codeCellConfig":{"autoClosingBrackets":true,"lineNumbers":true,"codeFolding":true,},"experimentalDisableDocumentWideUndoRedo":true,"kernelShutdown":true,"kernelStatus":{"showOnStatusBar":true,"showProgress":true},"markdownCellConfig":{"autoClosingBrackets":true,"lineNumbers":true,"lineWrap":"on","matchBrackets":true,"codeFolding":true,},"rawCellConfig":{"autoClosingBrackets":true,"lineNumbers":true,"lineWrap":"on","matchBrackets":true,"codeFolding":true,},"recordTiming":true}'>/home/jovyan/.jupyter/lab/user-settings/@jupyterlab/notebook-extension/tracker.jupyterlab-settings

# `jupyterlab-variableInspector`帮助我们在`jupyter lab`中查看当前环境中存在的变量相关信息，
# 以美观的界面形式对多种类型的对象予以呈现。
# https://cloud.tencent.com/developer/article/1971947
RUN pip install --no-cache-dir lckr-jupyterlab-variableinspector -i http://pypi.douban.com/simple/ --trusted-host pypi.douban.com

# jupyterlab-system-monitor通过在jupyter lab界面中添加资源监视器部件，
# 能帮助我们在工作过程中方便的看到CPU、内存的实时占用情况
RUN pip install --no-cache-dir nbresuse -i http://pypi.douban.com/simple/ --trusted-host pypi.douban.com \
    && jupyter labextension install jupyterlab-topbar-extension jupyterlab-system-monitor
# docker run -itd --name=lab -p 8888:8888 -e CHOWN_HOME=yes -e NB_UID=1000 -e NB_GID=1000 -e RESTARTABLE=yes -e GRANT_SUDO=yes --user root docker.io/blackelks/jupyterlab-datascience:latest

# 补充库:faker 代码格式化、black isort
RUN pip install --no-cache-dir faker jupyterlab-code-formatter black isort -i http://pypi.douban.com/simple/ --trusted-host pypi.douban.com

# 代码格式化插件设置
RUN mkdir -p /home/jovyan/.jupyter/lab/user-settings/@ryantam626/jupyterlab_code_formatter/ \
    && echo '{"preferences":{"default_formatter":{"python":["isort","black"],}},"black":{"line_length":88,"string_normalization":true},"yapf":{"style_config":"google"},"autopep8":{"max_line_length":120,"ignore":["E226","E302","E41"]},"isort":{"multi_line_output":3,"include_trailing_comma":true,"force_grid_wrap":0,"use_parentheses":true,"line_length":88},"formatOnSave":true,"astyle":{"args":[]},"suppressFormatterErrors":false}'>/home/jovyan/.jupyter/lab/user-settings/@ryantam626/jupyterlab_code_formatter/settings.jupyterlab-settings

# 新增插件`ipydatagrid`：https://github.com/bloomberg/ipydatagrid
RUN pip --no-cache-dir install ipydatagrid -i http://pypi.douban.com/simple/ --trusted-host pypi.douban.com

# 下载ipydatagrid的示例文件到工作目录
RUN git clone https://github.com/bloomberg/ipydatagrid.git \
    && mkdir -p ipyDataGrid示例 \
    && mv ipydatagrid/examples/* ipyDataGrid示例 \
    && rm ipydatagrid -rf

# 安装darcula主题并激活，为了不费时的重复构建，使用此层覆盖前面的主题设置
RUN pip install --no-cache-dir theme-darcula -i http://pypi.douban.com/simple/ --trusted-host pypi.douban.com \
    && echo '{"theme":"Darcula","theme-scrollbars":true}'>/home/jovyan/.jupyter/lab/user-settings/@jupyterlab/apputils-extension/themes.jupyterlab-settings

Notebook相关(弃用，需要时可参考)

安装主题

pip install jupyterthemes

更新主题

pip install --upgrade jupyterthemes

查看可用主题

jt -l

我个人喜欢亮一点的背景主题，于是选择了grade3，它还支持语法高亮。下面是我的背景主题设置：

1

jt -t grade3 -f fira -fs 13 -cellw 90% -ofs 11 -dfs 11 -T -N

-f(字体) -fs(字体大小) -cellw(占屏比或宽度) -ofs(输出段的字号) -T(显示工具栏) -N(显示自己主机名)

2.代码自动补全

首先安装 nbextensions：

1
2


pip install --no-cache-dir jupyter_contrib_nbextensions
jupyter contrib nbextension install --user

然后安装 nbextensions_configurator：

1
2


pip install --no-cache-dir jupyter_nbextensions_configurator
jupyter nbextensions_configurator enable --user

如果提示缺少依赖，就使用pip安装对应依赖即可。

最后重启jupyter

在弹出的Home面里，能看到增加了一个Nbextensions标签页，在这个页面里，勾选Hinterland即启用了代码自动补全，如图所示：

效果图： https://www.cnblogs.com/shanger/p/12006161.html

其他备注信息

`Notebook`配置文件路径:

/home/jovyan/.jupyter/jupyter_notebook_config.py

`Notebook`语言文件目录

/opt/conda/lib/python3.10/site-packages/notebook/i18n

容器内工作目录

默认为/home/jovyan/work，可通过自定义环境变量来变更。

目录