feat: implement algorithm for better string matching (#124)

Algorithm works as follows:
- Matches full answer string to the guess
- Matches answer string up to first non-alphanumerical character to the guess
- Matches first N words in answer to guess, where N is the number of words in the guess

src/classes/osudle.py

@@ -42,10 +42,9 @@ def check_guess(self, message: str) -> bool:
         return (
             message.channel is self.interaction.channel
             and not message.author.bot
-            and humanizer.fuzzy_string_match(
+            and humanizer.song_title_match(
                 message.content,
                 self.current_beatmapset.title,
-                permit_low_match=False,
             )
         )
 

src/common/humanizer.py

@@ -5,6 +5,7 @@
 
 from datetime import timedelta
 
+from common.regex import alphanumeric_rx
 from thefuzz import fuzz
 
 FUZZY_MATCH_THRESHOLD = 80
@@ -87,3 +88,14 @@ def milliseconds_to_duration(position: float) -> str:
 def fuzzy_string_match(str1: str, str2: str, permit_low_match: bool = False) -> bool:
     THRESHOLD = FUZZY_MATCH_THRESHOLD_LOW if permit_low_match else FUZZY_MATCH_THRESHOLD
     return fuzz.ratio(str1.casefold(), str2.casefold()) >= THRESHOLD
+
+
+def song_title_match(guess: str, answer: str) -> bool:
+    alphanumeric = alphanumeric_rx.split(answer)[0]
+    guess_length = len(guess.split(" "))
+    partial_words = " ".join(answer.split(" ")[:guess_length])
+    return (
+        fuzzy_string_match(guess, answer)
+        or fuzzy_string_match(guess, alphanumeric)
+        or fuzzy_string_match(guess, partial_words)
+    )

src/common/regex.py

@@ -12,3 +12,4 @@
 user_link_rx = re.compile(
     r"(https?)://(?P<domain>osu|lazer)\.ppy\.sh/u(sers)?/(?P<userid>[0-9]+)",
 )
+alphanumeric_rx = re.compile(r"[^a-zA-Z0-9]")