Replace unicode emojis in a text. Supports Unicode 8 standard.
- Python 2.7 (wide-build), 3.3, 3.4 and 3.5 (recommended)
$ pip install emoji-unicode
emoji_unicode.replace(
u'Time to ⛽',
lambda e: u'<img src="{filename}.svg" alt="{raw}">'.format(filename=e.code_points, raw=e.unicode)
)
# Time to <img src="26fd.svg" alt="⛽">Note: the Emoji.code_points are normalized.
This function removes optional characters that may appear depending on
the input source (Android, iOS, etc). For example the emoji variation \\uFE0F
may (or may not) appear in between a emoji and a skin tone modifier,
making the code points to be different. It should be used
to rename the image files.
emoji_unicode.normalize(u'1F468-200D-2764-FE0F-200D-1F468')
# 1f468-2764-1f468PATTERN = re.compile(emoji_unicode.RE_PATTERN_TEMPLATE)
def match_handler(m):
e = emoji_unicode.Emoji(unicode=m.group('emoji'))
return u'<img src="{filename}.svg" alt="{raw}">'.format(
filename=e.code_points,
raw=e.unicode
)
re.sub(PATTERN, match_handler, u'Time to ⛽')
# Time to <img src="26fd.svg" alt="⛽">If your current emoji package supports unicode 8, which means it supports skin tones and sequences, then normalizing the file names should be enough. But to handle unsupported emojis, for example future sequences, they should be displayed as multiple glyphs.
Instead of displaying the woman-kissing-man glyph you may
display woman, heart, kiss, man glyphs.
Here is a example of how this could be handled:
EMOJI_FILES = set(['1f469', '2764', '1f48b', '1f468']) # A set containing the emoji file names
def _render(unicode, code_points):
return u'<img src="{filename}.svg" alt="{alt}">'.format(filename=code_points, alt=unicode)
def render(e):
"""
Return the rendered html for the passed Emoji.
Return the html as multiple glyphs when the
emoji is a sequence not found within the files.
Return the raw unicode when one or more glyphs
are missing.
"""
if e.code_points in EMOJI_FILES:
return _render(e.unicode, e.code_points)
if any(c not in EMOJI_FILES for u, c in e.as_map()):
return e.unicode
return u''.join(_render(u, c) for u, c in e.as_map())
# This assumes `woman-kissing-man.svg` is missing
emoji_unicode.replace(
u'\U0001f469\u200d\u2764\ufe0f\u200d\U0001f48b\u200d\U0001f468',
render
)
# <img src="1f469.svg" alt="\U0001f469"><img src="2764.svg" alt="\u2764"> ...The ./emoji_unicode/pattern.py file is generated
by parsing the ./emoji_unicode/emoji-data.txt file,
then putting the output in a in-memory copy of
./emoji_unicode/pattern_template.py, and lastly
writing the result into pattern.py.
To generate the pattern.py file, run:
$ python ./build.py
$ python ./runtests.py
This will run some silly benchmarks.
$ python ./benchmark.py
Here is the output on my machine:
emoji.replace()
text len: 10000
0.01640868396498263
re.sub() (raw match)
text len: 10000
0.005225047003477812
Text with no emojis
emoji.replace()
text len: 10000
0.0014624089817516506
Thanks to iamcal/emoji-data for maintaining an incredible source of emojis that allowed me to make a robust test suite.
MIT