simplify Dockerfile numpy-pandas dependecy

[jeremydouglass]

Currently pandas fails to build on a generic python3 container. Specifically, cchardet fails with a gcc error. In commit c28702d this is worked-around by building from an available pre-built numpy-pandas image instead:

FROM jjuanda/numpy-pandas

....however it is a personal image that lacks a public Dockerfile.

In the future we should rework the Dockerfile to add pandas / flask on top of an official distribution image.

If the Dockerfile is:

FROM python:3
RUN pip3 install pandas flask

Below are the error details:

Running setup.py bdist_wheel for cchardet: started

Running setup.py bdist_wheel for cchardet: finished with status 'error'

Complete output from command /usr/local/bin/python -u -c "import setuptools, tokenize;file='/tmp/pip-install-8x1_b8z6/cchardet/setup.py';f=getattr(tokenize, 'open', open)(file);code=f.read().replace('\r\n', '\n');f.close();exec(compile(code, file, 'exec'))" bdist_wheel -d /tmp/pip-wheel-uno61gbs --python-tag cp37:
running bdist_wheel
running build
running build_py
creating build
creating build/lib.linux-x86_64-3.7
creating build/lib.linux-x86_64-3.7/cchardet
copying src/cchardet/version.py -> build/lib.linux-x86_64-3.7/cchardet
copying src/cchardet/init.py -> build/lib.linux-x86_64-3.7/cchardet
running build_ext
building 'cchardet._cchardet' extension
creating build/temp.linux-x86_64-3.7
creating build/temp.linux-x86_64-3.7/src
creating build/temp.linux-x86_64-3.7/src/cchardet
creating build/temp.linux-x86_64-3.7/src/ext
creating build/temp.linux-x86_64-3.7/src/ext/libcharsetdetect
creating build/temp.linux-x86_64-3.7/src/ext/libcharsetdetect/mozilla
creating build/temp.linux-x86_64-3.7/src/ext/libcharsetdetect/mozilla/extensions
creating build/temp.linux-x86_64-3.7/src/ext/libcharsetdetect/mozilla/extensions/universalchardet
creating build/temp.linux-x86_64-3.7/src/ext/libcharsetdetect/mozilla/extensions/universalchardet/src
creating build/temp.linux-x86_64-3.7/src/ext/libcharsetdetect/mozilla/extensions/universalchardet/src/base
gcc -pthread -Wno-unused-result -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -fPIC -Isrc/ext/libcharsetdetect/mozilla/extensions/universalchardet/src/base/ -Isrc/ext/libcharsetdetect/nspr-emu/ -Isrc/ext/libcharsetdetect/ -I/usr/local/include/python3.7m -c src/cchardet/_cchardet.cpp -o build/temp.linux-x86_64-3.7/src/cchardet/_cchardet.o
src/cchardet/_cchardet.cpp: In function ‘PyObject* __Pyx_PyCFunction_FastCall(PyObject*, PyObject**, Py_ssize_t)’:
src/cchardet/_cchardet.cpp:2351:69: error: too many arguments to function
return (*((__Pyx_PyCFunctionFast)meth)) (self, args, nargs, NULL);
^
error: command 'gcc' failed with exit status 1

[jeremydouglass]

I was able to locally work around cchardet failing to build by running pip install cython first.

Cython is listed as an optional dependency for pandas. It is required by other dependencies of our project, like tabulator.

Examples of fixing related cchardet install problems by pre-installing cython:

• Installation on OSX fails PyYoshi/cChardet#15
• fail install on mac os 10.13.3 (command 'clang' failed with exit status 1) kivy/kivy#5653

Will test to see if adding that to requirements can also be used for a generic image build on DockerHub.