feat: Improved emoji matching (#712)

* tests: implement target tests

* feat: improved emoji matching

naff/models/discord/emoji.py

@@ -1,4 +1,5 @@
 import re
+import string
 from typing import TYPE_CHECKING, Any, Dict, List, Optional, Union
 
 import attrs
@@ -21,6 +22,7 @@
 __all__ = ("PartialEmoji", "CustomEmoji", "process_emoji_req_format", "process_emoji")
 
 emoji_regex = re.compile(r"<?(a)?:(\w*):(\d*)>?")
+unicode_emoji_reg = re.compile(r"[^\w\s,]")
 
 
 @attrs.define(eq=False, order=False, hash=False, kw_only=False)
@@ -77,9 +79,15 @@ def from_str(cls, emoji_str: str, *, language: str = "alias") -> Optional["Parti
             _emoji_list = emoji.distinct_emoji_list(emoji_str)
             if _emoji_list:
                 return cls(name=_emoji_list[0])
-            if len(emoji_str) == 1:
-                # likely a regional indicator
-                return cls(name=emoji_str)
+
+            # the emoji lib handles *most* emoji, however there are certain ones that it misses
+            # this acts as a fallback check
+            if matches := unicode_emoji_reg.search(emoji_str):
+                match = matches.group()
+
+                # the regex will match certain special characters, so this acts as a final failsafe
+                if match not in string.printable:
+                    return cls(name=match)
         return None
 
     def __str__(self) -> str:

tests/test_emoji.py

@@ -1,3 +1,7 @@
+import string
+
+import emoji
+
 from naff.models.discord.emoji import PartialEmoji, process_emoji, process_emoji_req_format
 
 __all__ = ()
@@ -54,3 +58,52 @@ def test_emoji_processing() -> None:
     assert str(from_str) == raw_sample
 
     assert PartialEmoji.from_str("<a:sparklesnek:910496037708374016>").animated is True
+
+
+def test_unicode_recognition() -> None:
+    for _e in emoji.EMOJI_DATA:
+        assert PartialEmoji.from_str(_e) is not None
+
+
+def test_regional_indicators() -> None:
+    regional_indicators = [
+        "🇦",
+        "🇧",
+        "🇨",
+        "🇩",
+        "🇪",
+        "🇫",
+        "🇬",
+        "🇭",
+        "🇮",
+        "🇯",
+        "🇰",
+        "🇱",
+        "🇲",
+        "🇳",
+        "🇴",
+        "🇵",
+        "🇶",
+        "🇷",
+        "🇸",
+        "🇹",
+        "🇺",
+        "🇻",
+        "🇼",
+        "🇽",
+        "🇾",
+        "🇿",
+    ]
+    for _e in regional_indicators:
+        assert PartialEmoji.from_str(_e) is not None
+
+
+def test_numerical_emoji() -> None:
+    numerical_emoji = ["0️⃣", "1️⃣", "2️⃣", "3️⃣", "4️⃣", "5️⃣", "6️⃣", "7️⃣", "8️⃣", "9️⃣"]
+    for _e in numerical_emoji:
+        assert PartialEmoji.from_str(_e) is not None
+
+
+def test_false_positives() -> None:
+    for _e in string.printable:
+        assert PartialEmoji.from_str(_e) is None