Using ColorEmoji in Matplotlib

In previous post, we introduced mpl-simple-svg-parser package that lets you render svg. You can further use this to render color emoji.

You will need a color emoji font with COLR table. For example, Noto Color Emoji can be downaloaded from https://fonts.google.com/noto/specimen/Noto+Color+Emoji

You can transform the glyphs in the font to SVG, and uses mpl-simple-svg-parser, for example, to render it with matplotlib.

mpl-colr2svg is a package that does this. To convert emoji to svg, it uses nanoemoji. Altenatively, it can use blackrenderer to render emoji font to svg.

Code

import matplotlib.pyplot as plt
import pandas as pd

emoji_popularity = [["😂", 223.94],
                    ["🤣", 170.29],
                    ["❤️", 95.02],
                    ["🙏", 116.92],
                    ["😭", 95.02],
                    ["😍", 77.3],
                    ["✨", 76.34],
                    ["🔥", 71.52],
                    ["😊", 70.15],
                    ["🥰", 67.35],
                    ]

df = pd.DataFrame(emoji_popularity, columns=["Emoji", "Popularity"])

import seaborn as sns
sns.set_color_codes("muted")
sns.set(font_scale = 1.5)

fig, ax = plt.subplots(1, 1, figsize=(7, 4), clear=True, num=2, layout="constrained")

sns.barplot(x="Emoji", y="Popularity", data=df,
            label="Emoji Popularity", color="b")

from mpl_colr2svg import ColorEmoji

ftname = "NotoColorEmoji-Regular.ttf"
fontsize = 25

emoji = ColorEmoji(ftname)

for l in ax.get_xticklabels():
    c = l.get_text()
    xy = l.get_position()

    l.set_visible(False)

    emoji.annotate(ax, c, xy, xycoords=("data", "axes fraction"),
                   box_alignment=(0.5, 1), fontsize=fontsize,
                   annotation_clip=True)

ax.xaxis.labelpad = fontsize * 1.7

plt.show()

Let’s start with a simple example.

import matplotlib.pyplot as plt
from mpl_colr2svg import ColorEmoji


ftname = "NotoColorEmoji-Regular.ttf"
emoji = ColorEmoji(ftname)
c = "🤩"

fig, axs = plt.subplots(1, 2, figsize=(10, 5), clear=True, num=1)
ax1 = axs[0]
ax1.set_title("draw in data coordinate")

emoji.draw(ax1, c, size=128)

ax2 = axs[1]

emoji.annotate(ax2, c, (0.5, 0.5), fontsize=128)
ax2.set_title("drawing_area with annotate")

plt.show()

You may use blackrender. The default ColorEmoji only support a sing glyph, but the blackrenderer version supports string of multiple charaters.


import matplotlib.pyplot as plt
from mpl_colr2svg.color_emoji_blackrenderer import ColorEmojiBlackrenderer

ftname = "seguiemj.ttf"
emoji = ColorEmojiBlackrenderer(ftname)
textString = "L❤️VE"

fig, ax = plt.subplots(1, 1, clear=True, num=1)

emoji.annotate(ax, textString, (0.5, 0.5), fontsize=128)

plt.show()

Here is a more useful example. We start with a seabron plot of emoji popularity.

import matplotlib.pyplot as plt
import pandas as pd

emoji_popularity = [["😂", 223.94],
                    ["🤣", 170.29],
                    ["❤️", 95.02],
                    ["🙏", 116.92],
                    ["😭", 95.02],
                    ["😍", 77.3],
                    ["✨", 76.34],
                    ["🔥", 71.52],
                    ["😊", 70.15],
                    ["🥰", 67.35],
                    ]

df = pd.DataFrame(emoji_popularity, columns=["Emoji", "Popularity"])

import seaborn as sns
sns.set_color_codes("muted")
sns.set(font_scale = 1.5)

fig, ax = plt.subplots(1, 1, figsize=(7, 4), clear=True, num=2, layout="constrained")

sns.barplot(x="Emoji", y="Popularity", data=df,
            label="Emoji Popularity", color="b")

<Axes: xlabel='Emoji', ylabel='Popularity'>

/home/jjlee/miniconda3/envs/mpl-dev/lib/python3.10/site-packages/IPython/core/events.py:82: UserWarning: Glyph 129315 (\N{ROLLING ON THE FLOOR LAUGHING}) missing from font(s) DejaVu Sans.
  func(*args, **kwargs)
/home/jjlee/miniconda3/envs/mpl-dev/lib/python3.10/site-packages/IPython/core/events.py:82: UserWarning: Glyph 128591 (\N{PERSON WITH FOLDED HANDS}) missing from font(s) DejaVu Sans.
  func(*args, **kwargs)
/home/jjlee/miniconda3/envs/mpl-dev/lib/python3.10/site-packages/IPython/core/events.py:82: UserWarning: Glyph 10024 (\N{SPARKLES}) missing from font(s) DejaVu Sans.
  func(*args, **kwargs)
/home/jjlee/miniconda3/envs/mpl-dev/lib/python3.10/site-packages/IPython/core/events.py:82: UserWarning: Glyph 128293 (\N{FIRE}) missing from font(s) DejaVu Sans.
  func(*args, **kwargs)
/home/jjlee/miniconda3/envs/mpl-dev/lib/python3.10/site-packages/IPython/core/events.py:82: UserWarning: Glyph 129392 (\N{SMILING FACE WITH SMILING EYES AND THREE HEARTS}) missing from font(s) DejaVu Sans.
  func(*args, **kwargs)
/home/jjlee/miniconda3/envs/mpl-dev/lib/python3.10/site-packages/IPython/core/pylabtools.py:152: UserWarning: Glyph 129315 (\N{ROLLING ON THE FLOOR LAUGHING}) missing from font(s) DejaVu Sans.
  fig.canvas.print_figure(bytes_io, **kw)
/home/jjlee/miniconda3/envs/mpl-dev/lib/python3.10/site-packages/IPython/core/pylabtools.py:152: UserWarning: Glyph 128591 (\N{PERSON WITH FOLDED HANDS}) missing from font(s) DejaVu Sans.
  fig.canvas.print_figure(bytes_io, **kw)
/home/jjlee/miniconda3/envs/mpl-dev/lib/python3.10/site-packages/IPython/core/pylabtools.py:152: UserWarning: Glyph 10024 (\N{SPARKLES}) missing from font(s) DejaVu Sans.
  fig.canvas.print_figure(bytes_io, **kw)
/home/jjlee/miniconda3/envs/mpl-dev/lib/python3.10/site-packages/IPython/core/pylabtools.py:152: UserWarning: Glyph 128293 (\N{FIRE}) missing from font(s) DejaVu Sans.
  fig.canvas.print_figure(bytes_io, **kw)
/home/jjlee/miniconda3/envs/mpl-dev/lib/python3.10/site-packages/IPython/core/pylabtools.py:152: UserWarning: Glyph 129392 (\N{SMILING FACE WITH SMILING EYES AND THREE HEARTS}) missing from font(s) DejaVu Sans.
  fig.canvas.print_figure(bytes_io, **kw)

Don’t worry if you see tofus. We will replace the x-tick labels with color emoji.

from mpl_colr2svg import ColorEmoji

ftname = "NotoColorEmoji-Regular.ttf"
fontsize = 25

emoji = ColorEmoji(ftname)

for l in ax.get_xticklabels():
    c = l.get_text()
    xy = l.get_position()

    l.set_visible(False)

    emoji.annotate(ax, c, xy, xycoords=("data", "axes fraction"),
                   box_alignment=(0.5, 1), fontsize=fontsize,
                   annotation_clip=True)

ax.xaxis.labelpad = fontsize * 1.7