api/lyric_search_new/sources/lrclib.py

#!/usr/bin/env python3.12
# pylint: disable=bare-except, broad-exception-caught, wrong-import-position

import sys
import time
sys.path.insert(1,'..')
import traceback
import logging
from typing import Optional
from aiohttp import ClientTimeout, ClientSession
from lyric_search_new import utils
from lyric_search_new.constructors import LyricsResult
from . import common
from . import cache

logger = logging.getLogger()
log_level = logging.getLevelName(logger.level)

class InvalidResponseException(Exception):
    """
    Invalid Response Exception
    """

class LRCLib:
    """LRCLib Search Module"""
    def __init__(self):
        self.label: str = "LRCLib"
        self.lrclib_url: str = "https://lrclib.net/api/search"
        self.headers: dict = common.SCRAPE_HEADERS
        self.timeout = ClientTimeout(connect=2, sock_read=4)
        self.datautils = utils.DataUtils()
        self.matcher = utils.TrackMatcher() 
        self.cache = cache.Cache()

    async def search(self, artist: str, song: str, plain: bool = True) -> Optional[LyricsResult]:
        """
        @artist: the artist to search
        @song: the song to search
        """
        try:          
            artist: str = artist.strip().lower()
            song: str = song.strip().lower()
            time_start: float = time.time()
            lrc_obj: Optional[list[dict]] = None

            logging.info("Searching %s - %s on %s",
                         artist, song, self.label) 
                
            input_track: str = f"{artist} - {song}"         
            returned_lyrics: str = ''
            async with ClientSession() as client:
                async with client.get(self.lrclib_url,
                                      params = {
                                          'artist_name': artist,
                                          'track_name': song,
                                      },
                                      timeout=self.timeout,
                                      headers=self.headers) as request:
                    request.raise_for_status()
                    text: str|None = await request.text()
                     
                    if len(text) < 100:
                        raise InvalidResponseException("Search response text was invalid (len < 100 chars.)")
                    
                    search_data: dict|None = await request.json()

                    # logging.info("Search Data:\n%s", search_data)
                    
                    if not isinstance(search_data, list):
                        raise InvalidResponseException("Invalid JSON.")
                    
                    if plain:
                        possible_matches = [(x, f"{result.get('artistName')} - {result.get('trackName')}")
                                         for x, result in enumerate(search_data)]
                    else:
                        logging.info("Limiting possible matches to only those with non-null syncedLyrics")
                        possible_matches = [(x, f"{result.get('artistName')} - {result.get('trackName')}")
                                            for x, result in enumerate(search_data) if isinstance(result['syncedLyrics'], str)]                        

                    

                    best_match = self.matcher.find_best_match(input_track,
                                                                 possible_matches)[0]
                    if not best_match:
                        return
                    best_match_id = best_match[0]
                    
                    if not isinstance(search_data[best_match_id]['artistName'], str):
                        raise InvalidResponseException(f"Invalid JSON: Cannot find artistName key.\n{search_data}")                     
                     
                    if not isinstance(search_data[best_match_id]['trackName'], str):
                        raise InvalidResponseException(f"Invalid JSON: Cannot find trackName key.\n{search_data}")
                     
                    returned_artist: str = search_data[best_match_id]['artistName']
                    returned_song: str = search_data[best_match_id]['trackName']
                    if plain:
                        if not isinstance(search_data[best_match_id]['plainLyrics'], str):
                            raise InvalidResponseException(f"Invalid JSON: Cannot find plainLyrics key.\n{search_data}")
                        returned_lyrics: str = search_data[best_match_id]['plainLyrics']
                        returned_lyrics = self.datautils.scrub_lyrics(returned_lyrics)
                    else:
                        if not isinstance(search_data[best_match_id]['syncedLyrics'], str):
                            raise InvalidResponseException(f"Invalid JSON: Cannot find syncedLyrics key.\n{search_data}")
                        returned_lyrics: str = search_data[best_match_id]['syncedLyrics']
                        lrc_obj = self.datautils.create_lrc_object(returned_lyrics)
                    returned_track: str = f"{returned_artist} - {returned_song}"
                    (_matched, confidence) = self.matcher.find_best_match(input_track=input_track,
                                                                        candidate_tracks=[(0, returned_track)])
                    if not confidence:
                        return # No suitable match found
                    logging.info("Result found on %s", self.label)
                    time_end: float = time.time()
                    time_diff: float = time_end - time_start
                    matched = LyricsResult(artist=returned_artist,
                                        song=returned_song,
                                        src=self.label,
                                        lyrics=returned_lyrics if plain else lrc_obj,
                                        confidence=confidence,
                                        time=time_diff)
                    await self.cache.store(matched)
                    return matched
        except:
            traceback.print_exc()
            return