music_kraken/utils/string_processing.py

@@ -11,6 +11,9 @@ from pathvalidate import sanitize_filename
 COMMON_TITLE_APPENDIX_LIST: Tuple[str, ...] = (
     "(official video)",
 )
+OPEN_BRACKETS = "(["
+CLOSE_BRACKETS = ")]"
+DISALLOWED_SUBSTRING_IN_BRACKETS = ("official", "video", "audio", "lyrics", "prod", "remix", "ft", "feat", "ft.", "feat.")
 
 @lru_cache
 def unify(string: str) -> str:
@@ -71,6 +74,26 @@ def clean_song_title(raw_song_title: str, artist_name: Optional[str] = None) ->
         if raw_song_title.lower().endswith(dirty_appendix):
             raw_song_title = raw_song_title[:-len(dirty_appendix)].strip()
 
+    # remove brackets and their content if they contain disallowed substrings
+    for open_bracket, close_bracket in zip(OPEN_BRACKETS, CLOSE_BRACKETS):
+        start = 0
+
+        while True:
+            try:
+                open_bracket_index = raw_song_title.index(open_bracket, start)
+            except ValueError:
+                break
+            try:
+                close_bracket_index = raw_song_title.index(close_bracket, open_bracket_index + 1)
+            except ValueError:
+                break
+
+            substring = raw_song_title[open_bracket_index + 1:close_bracket_index]
+            if any(disallowed_substring in substring for disallowed_substring in DISALLOWED_SUBSTRING_IN_BRACKETS):
+                raw_song_title = raw_song_title[:open_bracket_index] + raw_song_title[close_bracket_index + 1:]
+            else:
+                start = close_bracket_index + 1
+
     # everything that requires the artist name
     if artist_name is not None:
         artist_name = artist_name.strip()