Add replace colons support for zwj joined emojis with skintones (#7)

zero-width-joined emojis would 'just work' if you appended the skin-tone
after them, i.e 👨:skin-tone-3:. However, this breaks down for
multibyte emojis with skintones, i.e :male-technologist::skin-tone-3:,
which is actually 👨:skin-tone-3:💻. This PR handles these
cases explicitly, by utilising data from the emoji source while
retaining previous functionality.

Author: Fruglemonkey

emoji_data_python/replacement.py

@@ -1,7 +1,8 @@
 import re
+from typing import cast
 
 
-def replace_colons(text: str, strip: bool=False) -> str:
+def replace_colons(text: str, strip: bool = False) -> str:
     """Parses a string with colon encoded emoji and renders found emoji.
     Unknown emoji are left as is unless `strip` is set to `True`
 
@@ -11,23 +12,29 @@ def replace_colons(text: str, strip: bool=False) -> str:
     >>> emoji_data_python.replace_colons('Hello world ! :wave::skin-tone-3: :earth_africa: :exclamation:')
     'Hello world ! 👋🏼 🌍 ❗'
     """
-    from emoji_data_python import emoji_short_names
+    from emoji_data_python import emoji_short_names, EmojiChar
 
     def emoji_repl(matchobj) -> str:
-        match = matchobj.group(0)
-        codes = match.split(':')
-        res = ''
-        for code in codes:
-            if len(code) > 0:
-                try:
-                    res += emoji_short_names.get(code.replace('-', '_')).char
-                except AttributeError:
-                    if not strip:
-                        res += f':{code}:'
-
-        return res
-
-    return re.sub(r'\:[a-zA-Z0-9-_+]+\:(\:skin-tone-[2-6]\:)?', emoji_repl, text)
+        emoji_match = matchobj.group(1)
+        base_emoji = emoji_short_names.get(emoji_match.strip(':').replace('-', '_'))
+
+        if matchobj.lastindex == 2:
+            skin_tone_match = matchobj.group(2)
+            skin_tone = cast(EmojiChar, emoji_short_names.get(skin_tone_match.strip(':')))
+
+            if base_emoji is None:
+                return f'{emoji_match if strip is False else ""}{skin_tone.char}'
+
+            emoji_with_skin_tone = base_emoji.skin_variations.get(skin_tone.unified)
+            if emoji_with_skin_tone is None:
+                return f'{base_emoji.char}{skin_tone.char}'
+            return emoji_with_skin_tone.char
+
+        if base_emoji is None:
+            return f'{emoji_match if strip is False else ""}'
+        return base_emoji.char
+
+    return re.sub(r'(\:[a-zA-Z0-9-_+]+\:)(\:skin-tone-[2-6]\:)?', emoji_repl, text)
 
 
 def get_emoji_regex():

tests/test_conversion.py

@@ -12,6 +12,7 @@ def test_longer_unified(self):
 
     def test_multiple_unified_to_char(self):
         self.assertEqual('👨‍🌾', unified_to_char('1F468-200D-1F33E'))
+        self.assertEqual('👩🏼‍🌾', unified_to_char('1F469-1F3FC-200D-1F33E'))
         self.assertEqual('🇳🇬', unified_to_char('1F1F3-1F1EC'))
         self.assertEqual('\U0001F1F3\U0001F1EC', unified_to_char('1F1F3-1F1EC'))
         self.assertEqual('4⃣', unified_to_char('0034-20E3'))
@@ -24,6 +25,7 @@ def test_longer_char_to_unified(self):
         self.assertEqual('1F1E6', char_to_unified('\U0001F1E6'))
 
     def test_multiple_char_to_unified(self):
+        self.assertEqual('1F469-1F3FC-200D-1F33E', char_to_unified('👩🏼‍🌾'))
         self.assertEqual('1F468-200D-1F33E', char_to_unified('👨‍🌾'))
         self.assertEqual('1F1F3-1F1EC', char_to_unified('🇳🇬'))
         self.assertEqual('1F1F3-1F1EC', char_to_unified('\U0001F1F3\U0001F1EC'))

tests/test_replacement.py

@@ -11,6 +11,9 @@ def test_replace_colons(self):
     def test_skin_tone(self):
         self.assertEqual('👋🏼', replace_colons(':wave::skin-tone-3:'))
 
+    def test_skin_tone_appended_to_emoji_with_no_skin_tone(self):
+        self.assertEqual('💩🏼', replace_colons(':poop::skin-tone-3:'))
+
     def test_underscore_hyphenated_codes(self):
         self.assertEqual('😙', replace_colons(':kissing_smiling_eyes:'))
         self.assertEqual('😘', replace_colons(':kissing-heart:'))
@@ -24,11 +27,17 @@ def test_zwj_emoji(self):
         self.assertEqual('👨‍👩‍👦', replace_colons(':man-woman-boy:'))
         self.assertEqual('👨‍🌾', replace_colons(':male-farmer:'))
 
+    def test_zwj_emoji_skin_tone(self):
+        """This tests zwj emojis that also have a skin tone"""
+        self.assertEqual('👨🏼‍🌾', replace_colons(':male-farmer::skin-tone-3:'))
+
     def test_unknown_code(self):
         self.assertEqual('💩💩 :poo:🏼', replace_colons(':hankey::poop: :poo::skin-tone-3:'))
+        self.assertEqual('💩:poo: 🐶 :poo:', replace_colons(':poop::poo: :dog: :poo:'))
 
     def test_strip_unknown_code(self):
         self.assertEqual('💩💩 🏼', replace_colons(':hankey::poop: :poo::skin-tone-3:', strip=True))
+        self.assertEqual('💩 🐶 ', replace_colons(':poop::poo: :dog: :poo:', strip=True))
 
     def test_multiline_sentence(self):
         self.assertEqual("""