python3 mayavi in docker not installing

Question

我试图让mayavi在docker容器中工作，最初我是从continuumio/anaconda3启动我的Dockerfile的。我做了一个"conda install mayavi"，它似乎会安装，但一旦我尝试导入它或 vtk，我就会得到：

"ModuleNotFoundError： No module named 'vtkRenderingOpenGL2Python'">

当我尝试从 pip3 安装它时，它无法安装"ModuleNotFoundError：没有名为'vtkOpenGLKitPython'的模块">

我从 centos：7 开始尝试并遇到同样的问题。我想值得一提的是，这些模块的 conda 搜索或 pip 搜索是空白的。但是我可以在 docker 之外安装它，一切正常。

如果有帮助，我当前的 Dockerfile 如下所示：

FROM centos:7
RUN yum install vim -y
RUN yum install python3 -y
RUN yum install python3-pip -y
RUN yum install python3-devel -y
RUN yum install gcc -y
#RUN pip3 install mayavi
#RUN pip3 install PyQt5
RUN mkdir /home/working
WORKDIR /home/working

我已经在这里工作了一段时间，任何帮助将不胜感激。

Answer 1

你可以看看我的binder repo fork，你可以在Jupyter notebooks中加载内联Mayavi。

将 Dockerfile 粘贴到此处以供后人参考：

FROM jupyter/minimal-notebook:65761486d5d3
MAINTAINER Jean-Remi King <[e-mail]>
# Install core debian packages
USER root
ENV DEBIAN_FRONTEND noninteractive
RUN apt-get update && apt-get -yq dist-upgrade 
&& apt-get install -yq --no-install-recommends 
openssh-client 
vim 
curl 
gcc 
&& apt-get clean
# Xvfb
RUN apt-get install -yq --no-install-recommends 
xvfb 
x11-utils 
libx11-dev 
qt5-default 
&& apt-get clean
ENV DISPLAY=:99
# Switch to notebook user
USER $NB_UID
# Upgrade the package managers
RUN pip install --upgrade pip
RUN npm i npm@latest -g
# Install Python packages
RUN pip install vtk && 
pip install boto && 
pip install h5py && 
pip install nose && 
pip install ipyevents && 
pip install ipywidgets && 
pip install mayavi && 
pip install nibabel && 
pip install numpy && 
pip install pillow && 
pip install pyqt5 && 
pip install scikit-learn && 
pip install scipy && 
pip install xvfbwrapper && 
pip install https://github.com/nipy/PySurfer/archive/master.zip
# Install Jupyter notebook extensions
RUN pip install RISE && 
jupyter nbextension install rise --py --sys-prefix && 
jupyter nbextension enable rise --py --sys-prefix && 
jupyter nbextension install mayavi --py --sys-prefix && 
jupyter nbextension enable mayavi --py --sys-prefix && 
npm cache clean --force
# Try to decrease initial IPython kernel load times
RUN ipython -c "import matplotlib.pyplot as plt; print(plt)"
# Add an x-server to the entrypoint. This is needed by Mayavi
ENTRYPOINT ["tini", "-g", "--", "xvfb-run"]