From 1c7a64bc5881385b632b7beb4ce241462eb6148d Mon Sep 17 00:00:00 2001
From: codey <codey@code.ee>
Date: Mon, 20 Jan 2025 05:47:09 -0500
Subject: [PATCH] stoof

---
 endpoints/lyric_search.py               | 12 +++++---
 lyric_search_new/sources/aggregate.py   |  3 ++
 lyric_search_new/sources/redis_cache.py | 38 +++++++++++++++++++++----
 lyric_search_new/utils.py               |  2 +-
 4 files changed, 45 insertions(+), 10 deletions(-)

diff --git a/endpoints/lyric_search.py b/endpoints/lyric_search.py
index 1dd6dcb..75399b9 100644
--- a/endpoints/lyric_search.py
+++ b/endpoints/lyric_search.py
@@ -112,7 +112,8 @@ class LyricSearch(FastAPI):
             "IRC-KALI",
             "DISC-ACES",
             "DISC-HAVOC",
-            "IRC-SHARED"
+            "IRC-SHARED",
+            "LIMNORIA-SHARED",
         ]
 
         self.lrc_regex = regex.compile(r'\[([0-9]{2}:[0-9]{2})\.[0-9]{1,3}\](\s(.*)){0,}')
@@ -143,8 +144,8 @@ class LyricSearch(FastAPI):
     
     async def song_typeahead_handler(self, data: ValidTypeAheadRequest):
         """Song Type Ahead Handler"""
-        if not isinstance(data.pre_query, str) or len(data.pre_query) < 2\
-            or not isinstance(data.query, str) or len(data.query) < 2:
+        if not isinstance(data.pre_query, str)\
+            or not isinstance(data.query, str|None):
             return {
                 'err': True,
                 'errorText': 'Invalid request',
@@ -183,7 +184,10 @@ class LyricSearch(FastAPI):
             raise HTTPException(detail="Invalid request", status_code=500)
         
         if data.src.upper() not in self.acceptable_request_sources:
-            raise HTTPException(detail="Invalid request", status_code=500)
+            return {
+                'err': True,
+                'errorText': f'Unknown request source: {data.src.upper()}',
+            }
         
         search_artist: Optional[str] = data.a
         search_song: Optional[str] = data.s
diff --git a/lyric_search_new/sources/aggregate.py b/lyric_search_new/sources/aggregate.py
index 57b9799..c70e878 100644
--- a/lyric_search_new/sources/aggregate.py
+++ b/lyric_search_new/sources/aggregate.py
@@ -43,6 +43,8 @@ class Aggregate:
         sources: list = [cache_search,
                    lrclib_search,
                    genius_search]
+        if not plain:
+            sources = [lrclib_search] # Only LRCLib supported for synced lyrics 
         search_result: Optional[LyricsResult] = None
         for source in sources:
             if source.label.lower() in self.exclude_methods:
@@ -53,6 +55,7 @@ class Aggregate:
                     if plain:
                         logging.info("Skipping source: %s, excluded.", source.label)
                         continue
+                
             search_result = await source.search(artist=artist, song=song,
                                                  plain=plain)
             if search_result:
diff --git a/lyric_search_new/sources/redis_cache.py b/lyric_search_new/sources/redis_cache.py
index 2e0a3bc..449662a 100644
--- a/lyric_search_new/sources/redis_cache.py
+++ b/lyric_search_new/sources/redis_cache.py
@@ -8,6 +8,7 @@ import traceback
 import json
 import time
 import sys
+import regex
 sys.path.insert(1,'..')
 from lyric_search_new import notifier
 from lyric_search_new.constructors import LyricsResult
@@ -21,6 +22,7 @@ from . import private
 
 
 
+
 logger = logging.getLogger()
 log_level = logging.getLevelName(logger.level)
 
@@ -38,6 +40,10 @@ class RedisCache:
         self.redis_client = redis.Redis(password=private.REDIS_PW)
         self.notifier = notifier.DiscordNotifier()
         self.notify_warnings = True
+        self.regexes = [
+            regex.compile(r'\-'),
+            regex.compile(r'[^a-zA-Z0-9\s]'),
+        ]
     
     async def create_index(self) -> None:
         """Create Index"""
@@ -54,6 +60,24 @@ class RedisCache:
                 raise RedisException(f"Redis: Failed to create index: {result}")
         except Exception as e:
             await self.notifier.send(f"ERROR @ {__file__}", f"Failed to create idx: {str(e)}")
+            
+    def sanitize_input(self, artist: str, song: str) -> tuple[str, str]:
+        """
+        Sanitize artist/song input (convert to redis matchable fuzzy query)
+        Args:
+            artist: Input artist
+            song: Input song
+        Returns:
+            tuple[str, str]: Tuple containing the 2 output strings (artist, song)
+        """
+        artist = self.regexes[0].sub(" ", artist)
+        artist = self.regexes[1].sub("", artist).strip()
+        song = self.regexes[0].sub(" ", song)
+        song = self.regexes[1].sub("", song).strip()
+        artist = " ".join([f"(%{artist_word}%)" for artist_word in artist.split(" ")])
+        song = "".join([f"(%{song_word}%)" for song_word in song.split(" ")])
+        return (artist, song)        
+        
     
     async def search(self, **kwargs) -> list[tuple]:
         """
@@ -77,11 +101,15 @@ class RedisCache:
                 raise RedisException("Lyric search not yet implemented")
             
             if not is_random_search:
-                artist = artist.replace("-", " ")
-                song = song.replace("-", " ")
-                search_res = await self.redis_client.ft().search(
-                    Query(f"@artist:{artist} @song:{song}"
-                          )) 
+                (artist, song) = self.sanitize_input(artist=artist,
+                                                    song=song)
+                search_res = await self.redis_client.ft().search(Query(
+                    f"@artist:{artist} @song:{song}"
+                    ))
+                
+
+
+
                 search_res_out = [(result['id'].split(":",
                                                       maxsplit=1)[1], dict(json.loads(result['json'])))
                                   for result in search_res.docs]
diff --git a/lyric_search_new/utils.py b/lyric_search_new/utils.py
index 37279d3..ec7e1f3 100644
--- a/lyric_search_new/utils.py
+++ b/lyric_search_new/utils.py
@@ -140,7 +140,7 @@ class DataUtils:
             if not reg_helper[1].strip():
                 _words = "♪"
             else:
-                _words = reg_helper[1]
+                _words = reg_helper[1].strip()
             lrc_out.append({
                 "timeTag": _timetag,
                 "words": _words,