Description
Bug summary
I'm just getting started with the development side of matplotlib, and I'm getting a UnicodeDecodeError in some cases (not all cases) when TeX is being used in unit tests and building docs. The cause of the issue may be specific to my installation and seems to be isolated to the use of certain special and accented characters, but I'm not sure how to resolve it.
The code below is an example (from tex_demo.py in the docs) of one instance where this error is occurring, and \N{DEGREE SIGN} is causing the issue (removing it allows the code to run without error).
Code for reproduction
import numpy as np
import matplotlib.pyplot as plt
plt.rcParams['text.usetex'] = True
t = np.linspace(0.0, 1.0, 100)
s = np.cos(4 * np.pi * t) + 2
fig, ax = plt.subplots(figsize=(6, 4), tight_layout=True)
ax.plot(t, s)
ax.set_xlabel(r'\textbf{time (s)}')
ax.set_ylabel('\\textit{Velocity (\N{DEGREE SIGN}/sec)}', fontsize=16) # \N{DEGREE SIGN} appears to be the culprit
ax.set_title(r'\TeX\ is Number $\displaystyle\sum_{n=1}^\infty'
r'\frac{-e^{i\pi}}{2^n}$!', fontsize=16, color='r')Actual outcome
Traceback from warning/error in building docs:
generating gallery for gallery\text_labels_and_annotations... [ 78%] tex_demo.pydemo.py.pyxample.py
Warning, treated as error:
C:\Users\username\Documents\GitHub\matplotlib\examples\text_labels_and_annotations\tex_demo.py failed to execute correctly: Traceback (most recent call last):
File "C:\Users\username\miniconda3\envs\matplotlib-dev\lib\site-packages\sphinx_gallery\scrapers.py", line 378, in save_figures
rst = scraper(block, block_vars, gallery_conf)
File "C:\Users\username\Documents\GitHub\matplotlib\doc\conf.py", line 173, in matplotlib_reduced_latex_scraper
return matplotlib_scraper(block, block_vars, gallery_conf, **kwargs)
File "C:\Users\username\miniconda3\envs\matplotlib-dev\lib\site-packages\sphinx_gallery\scrapers.py", line 171, in matplotlib_scraper
fig.savefig(image_path, **these_kwargs)
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\figure.py", line 3099, in savefig
self.canvas.print_figure(fname, **kwargs)
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\backend_bases.py", line 2262, in print_figure
self.figure.draw(renderer)
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\artist.py", line 73, in draw_wrapper
result = draw(artist, renderer, *args, **kwargs)
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\artist.py", line 50, in draw_wrapper
return draw(artist, renderer)
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\figure.py", line 2891, in draw
mimage._draw_list_compositing_images(
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\image.py", line 131, in _draw_list_compositing_images
a.draw(renderer)
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\artist.py", line 50, in draw_wrapper
return draw(artist, renderer)
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\axes\_base
.py", line 3042, in draw
self._update_title_position(renderer)
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\axes\_base
.py", line 2982, in _update_title_position
ax.yaxis.get_tightbbox(renderer) # update offsetText
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\text.py", line 915, in get_window_extent
bbox, info, descent = self._get_layout(self._renderer)
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\text.py", line 321, in _get_layout File "c:\users\username\documents\github\matplotlib\lib\matplotlib\axis
6692
.py",
line 1244, in get_tightbbox
bb = self.label.get_window_extent(renderer)
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\text.py",
line 915, in get_window_extent
bbox, info, descent = self._get_layout(self._renderer)
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\text.py",
line 321, in _get_layout
w, h, d = _get_text_metrics_with_cache(
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\text.py", line 97, in _get_text_metrics_with_cache
return _get_text_metrics_with_cache_impl(
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\text.py", line 105, in _get_text_metrics_with_cache_impl
return renderer_ref().get_text_width_height_descent(text, fontprop, ismath)
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\backends\backend_agg.py", line 235, in get_text_width_height_descent
w, h, d = texmanager.get_text_width_height_descent(
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\texmanager.py", line 359, in get_text_width_height_descent
dvifile = cls.make_dvi(tex, fontsize)
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\texmanager.py", line 292, in make_dvi
cls._run_checked_subprocess(
File "c:\users\username\documents\github\matplotlib\lib\matplotlib\texmanager.py", line 268, in _run_checked_subprocess
exc=exc.output.decode('utf-8'))) from exc
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xb0 in position 1673: invalid start byte
The test_savefig_to_stringio() test also fails four times, each time with a different set of arguments. This is also a UnicodeDecodeError, but it originates from the title of the plot having accented characters ("Déjà vu"). Removing those characters resolves the UnicodeDecodeError, but leaves me with an AssertionError at the end of the test. Traceback is omitted here because it is over 200 lines long, and results in a similar error as above.
Expected outcome
Expected outcome is to build docs and pass all tests without failure. Referring to the failing instance in the docs from tex_demo.py, that code should produce:
Additional information
Again, this could be isolated to my installation of matplotlib, but if anyone can reproduce this error or knows of a possible fix, that would be great.
Operating system
Windows 11
Matplotlib Version
3.6.0.dev2170+g40676bb62f
Matplotlib Backend
module://matplotlib_inline.backend_inline
Python version
3.8.13
Jupyter version
3.4.0
Installation
git checkout