Skip to content

My custom helper to tune Matplotlib experience in Atom/Hydrogen, Pandoctools/Knitty or Jupyter

License

Notifications You must be signed in to change notification settings

kiwi0fruit/matplotlibhelper

Repository files navigation

Matplotlib Helper

Matplotlib Helper is my custom helper to tune Matplotlib experience for Pandoctools/Knitty (but it can be used by itself). I tuned fonts (that are shipped with this python package), please see default fonts and other options in default keyword arguments of ready(...), img(...) and img_path(...). I made some tweaks to use it with SugarTeX, some tweaks to automatically use interactive Qt5 plots in Atom/Hydrogen or non-jupyter Python. It can also export plots to SVG or PNG.

Works in Jupyter as well.

Contents

Install

Via conda:

conda install -c defaults -c conda-forge matplotlibhelper

Via pip:

pip install matplotlibhelper

Additionally you may need to install to use in Jupyter Lab:

conda install jupyterlab jupyterlab_server tk nodejs
pip install ipympl
jupyter labextension install @jupyter-widgets/jupyterlab-manager jupyter-matplotlib

Instead of pip install ipympl you can also try:

conda update --all
conda install -c defaults -c conda-forge ipympl
conda update --all

Usage example

Usage example that works both in Atom+Hydrogen and in Pandoctools+Knitty:

import IPython.display as ds
import matplotlibhelper as mh
mh.ready(font_size=14)  # should be run before import matplotlib.pyplot
import matplotlib.pyplot as plt


plt.figure(figsize=mh.figsize(w=6))  # height is automatic via the golden ration
plt.plot([1, 2, 3, 4])
plt.ylabel(mh.stex('ˎ∇ ⋅ [⠘E]ˎ, V/m'))  # using SugarTeX

# this code in Knitty would be parsed by Pandoc,
# in Atom/Hydrogen or Jupyter it would be displayed:
ds.display(ds.Markdown(
    f'![My beautiful figure]({mh.img()}){{#fig:1}}'
))
# or if you have image size issues in Hydrogen try:
ds.display(ds.HTML(
    f'<img src="{mh.img()}" width="900">'
))

Qt backend gives interactive plots in Atom/Hydrogen.

Hints

  1. Delete fontList.cache, fontList.py3k.cache or fontList.json from %USERPROFILE%\.matplotlib folder after installing a new font.
  2. If font becomes bold without a reason try (source):
from matplotlib import font_manager
if 'roman' in font_manager.weight_dict:
    del font_manager.weight_dict['roman']
    # noinspection PyProtectedMember
    font_manager._rebuild()
  1. Install Computer Modern Unicode for bold-italic unicode support: "mathtext.sf": "CMU Serif:bold:italic". Sans-serif command \mathsf{} is reassigned because sans-serif font is rarely used in serif docs.

About

My custom helper to tune Matplotlib experience in Atom/Hydrogen, Pandoctools/Knitty or Jupyter

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published