Hello, I got an error ModuleNotFoundError: No module named 'numpy.typing' due to the typing hint from numpy.typing import ArrayLike.

Would you mind remove this hint to support older numpy version like 1.19.* ? It seems no performance issue after remove it.

This PR solves #30 by adding an alignment argument to show_bar_values defaulting to "vertical".

I couldn't think of a robust way of determining the alignment automatically. Checking if the width of the bar is greater or lower than its height seemed a bit dodgy in some cases... I don't know. What do you think @nschloe ?

Usage (adapted from README demo):

import matplotlib.pyplot as plt
import matplotx

labels = ["Australia", "Brazil", "China", "Germany", "Mexico", "United\nStates"]
vals = [21.65, 24.5, 6.95, 8.40, 21.00, 8.55]
ypos = range(len(vals))


with plt.style.context(matplotx.styles.dufte_bar):
    plt.barh(ypos, vals)
    plt.yticks(ypos, labels)
    matplotx.show_bar_values("{:.2f}", alignment="horizontal")
    plt.title("average temperature [°C]")
    plt.tight_layout()
    plt.show()

Produces:

matplotx.show_bar_values works perfectly with vertical bar charts but not with horizontal bar charts.

These are often used with long text labels.

import matplotlib.pyplot as plt
import matplotx

labels = ["Australia", "Brazil", "China", "Germany", "Mexico", "United\nStates"]
vals = [21.65, 24.5, 6.95, 8.40, 21.00, 8.55]
ypos = range(len(vals))

with plt.style.context(matplotx.styles.dufte_bar):
    plt.barh(ypos, vals)
    plt.yticks(ypos, labels)
    matplotx.show_bar_values("{:.2f}")
    plt.title("average temperature [°C]")
    plt.tight_layout()
    plt.show()

Produces:

I can write a PR and add a show_hbar_values() function that works with horizontal bar charts and produces:

Or it can also be an argument of matplotx.show_bar_value defaulting to "vertical" like show_bar_value(alignement="horizontal")

What do you think @nschloe ?

Great package! Thank you so much it really helps!

I will surely use this in my next paper/talk. How can I cite this package?

Do you plan on adding a Zenodo DOI?

Cheers Remi

Using the code example in the readme:

import matplotlib.pyplot as plt
import matplotx
plt.style.use(matplotx.styles.ayu)

I get this error:

File ~/.conda/envs/.../lib/python3.10/site-packages/matplotlib/style/core.py:117, in use(style)
    115 for style in styles:
    116     if not isinstance(style, (str, Path)):
--> 117         _apply_style(style)
    118     elif style == 'default':
    119         # Deprecation warnings were already handled when creating
    120         # rcParamsDefault, no need to reemit them here.
    121         with _api.suppress_matplotlib_deprecation_warning():

File ~/.conda/envs/.../lib/python3.10/site-packages/matplotlib/style/core.py:62, in _apply_style(d, warn)
     61 def _apply_style(d, warn=True):
---> 62     mpl.rcParams.update(_remove_blacklisted_style_params(d, warn=warn))

File ~/.conda/envs/.../lib/python3.10/_collections_abc.py:994, in MutableMapping.update(self, other, **kwds)
    992 if isinstance(other, Mapping):
    993     for key in other:
--> 994         self[key] = other[key]
    995 elif hasattr(other, "keys"):
    996     for key in other.keys():

File ~/.conda/envs/.../lib/python3.10/site-packages/matplotlib/__init__.py:649, in RcParams.__setitem__(self, key, val)
    647     dict.__setitem__(self, key, cval)
    648 except KeyError as err:
--> 649     raise KeyError(
    650         f"{key} is not a valid rc parameter (see rcParams.keys() for "
    651         f"a list of valid parameters)") from err

KeyError: 'dark is not a valid rc parameter (see rcParams.keys() for a list of valid parameters)'

Lib versions:

matplotlib-base           3.5.2           py310h5701ce4_1    conda-forge
matplotx                  0.3.7                    pypi_0    pypi

This happens with aura, ayu, github, gruvbox and others.

Some of the themes working are: challenger_deep, dracula, dufte, nord, tab10

Related to the issue I opened. It seems that small changes already go quite a long way towards support for subplots. This does not yet work for the style.

For the original code, everything was correctly calculated with the axes in mind, but then it was applied to plt instead of ax, even if an ax parameter was supplied for line_labels, it was still applied to plt.

The code changes should have no effect when there are no subplots. When there are subplots, the code now offers better support.

import matplotlib.pyplot as plt
import matplotx
import numpy as np

# create data
rng = np.random.default_rng(0)
offsets = [1.0, 1.50, 1.60]
labels = ["no balancing", "CRV-27", "CRV-27*"]
names = ["Plot left", "Plot right"]
x0 = np.linspace(0.0, 3.0, 100)
y = [offset * x0 / (x0 + 1) + 0.1 * rng.random(len(x0)) for offset in offsets]

fig, axes = plt.subplots(2,1)                                           

for ax, name in zip(axes, names):                                                         
    with plt.style.context(matplotx.styles.dufte):
        for yy, label in zip(y, labels):
            ax.plot(x0, yy, label=label)                                
        ax.set_xlabel("distance [m]")                                   
    matplotx.ylabel_top(name)    
    matplotx.line_labels(ax=ax)

Original code

New code

To draw the legend dufte uses plt.text() https://github.com/nschloe/dufte/blob/main/src/dufte/main.py#L196

plt.text() allows for additional kwargs to customize the text https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.text.html

If possible, could you loop through the additional text kwargs to allow for a higher customizable legend?

I've been using matplotx.ylabel_top and just noticed an issue with the label placement after setting the y tick labels explicitly. A working example is below.

import numpy as np
from seaborn import scatterplot
import matplotx

rng = np.random.default_rng(42)
x = rng.random(100)
y = -2*x + rng.normal(0, 0.5, 100)
ax = scatterplot(
    x=x,
    y=y
)
ax.set_yticks([0, -1, -2])
matplotx.ylabel_top('Example\nLabel')

i'm using

numpy==1.23.4
seaborn==0.12.1
matplotx==0.3.10

I just came across this project, looks really neat. Especially the smooth contourf got me curious.

I've noticed in the readme that (at least on firefox) if I click any of the three images, the link that opens (even with the "open image in new tab" context menu option) is https://github.com/nschloe/matplotx/blob/main/tests/dufte_comparison.py. In contrast, the contourf images open just fine, for instance.

I assume the reason for this is the enclosing a tag for the first example: https://github.com/nschloe/matplotx/blob/c767b08ea91492b1db9626b8b2c8786b4bc99458/README.md?plain=1#L39

In case this is not just a firefox thing, I would recommend trying to make the first three images clickable on their own right.

I'm keen on making a PR to adapt line_labels to make it work with fill_between objects (PolyCollection)

This would be the usage and output:

import matplotlib.pyplot as plt
import matplotx
import numpy as np

x = np.linspace(0, 1)
y1 = np.linspace(1, 2)
y2 = np.linspace(2, 4)

plt.fill_between(x, y1, label="label1")
plt.fill_between(x, y1, y2, label="label1")

matplotx.label_fillbetween()
plt.show()

@nschloe would you be interested in this feature?

Perhaps this is already implemented and I'm just unable to find it. I think this package in general is great; very easy to use and very beautiful. Thank you for your time making it.

I'm unable to get matplotx working properly when using subplots. Adapting the Clean line plots (dufte) example to include two subplots (side-by-side, or one-below-the-other) appears not to work.

import matplotlib.pyplot as plt
import matplotx
import numpy as np

# create data
rng = np.random.default_rng(0)
offsets = [1.0, 1.50, 1.60]
labels = ["no balancing", "CRV-27", "CRV-27*"]
x0 = np.linspace(0.0, 3.0, 100)
y = [offset * x0 / (x0 + 1) + 0.1 * rng.random(len(x0)) for offset in offsets]

fig, axes = plt.subplots(2,1)                                           # add subplots

for ax in axes:                                                         # Let's make two identical subplots
    with plt.style.context(matplotx.styles.dufte):
        for yy, label in zip(y, labels):
            ax.plot(x0, yy, label=label)                                # changed plt. to ax.
        ax.set_xlabel("distance [m]")                                   # changed plt. to ax.
        matplotx.ylabel_top("voltage [V]")                              # move ylabel to the top, rotate
        matplotx.line_labels()                                          # line labels to the right
        #plt.show()                                                     # Including this adds the 'pretty axis' below the subplots.