mew/core/indexer.py

import time
from collections import deque
from queue import Queue
from threading import Thread, Lock

import vlc
from path import Path

from core import db
from core.exceptions import NotSupportedFile
from core.file_utilities import is_media_file_ext, hash_file
from core.logging_ import Logger
from core.models import Track
from core.repositories import MusicFolderRepository, TrackRepository

logger = Logger.get()


class CyclicThread(Thread):
    DELAY = 0

    def __init__(self):
        Thread.__init__(self)
        self.interrupted = False
        self.last_exec = 0

    def act(self):
        raise NotImplementedError()

    def run(self):
        t = None
        while 1:
            if self.DELAY:
                t = time.time()

            if not self.DELAY or not self.last_exec or (t - self.last_exec) > self.DELAY:
                self.act()
                self.last_exec = t

            if self.interrupted:
                break

            time.sleep(0.1)

    def trigger(self):
        self.last_exec = 0

    def stop(self):
        self.interrupted = True


class Discoverer(CyclicThread):
    DELAY = 5

    def __init__(self, indexer):
        CyclicThread.__init__(self)
        self.indexer = indexer

    def act(self):
        session = db.Session()
        music_folder_repo = MusicFolderRepository(session)
        music_folders = music_folder_repo.get_all()

        track_repo = TrackRepository(session)
        tracks = track_repo.get_all()

        index = {t.path: t for t in tracks}

        for music_folder in music_folders:
            music_folder_path = Path(music_folder.path)

            for filename in music_folder_path.walkfiles():
                if self.indexer.in_deque(filename):
                    continue
                if filename not in index and is_media_file_ext(filename.ext):
                    self.indexer.put(filename)
                elif filename in index:
                    track = index[filename]
                    if track.status == Track.STATUS_UNAVAILABLE:
                        self.indexer.put(track.id)
                    del index[filename]

        for filename, track in index.items():
            if self.indexer.in_deque(track.id):
                continue

            filename = Path(filename)
            if not filename.exists() and track.status != Track.STATUS_UNAVAILABLE:
                self.indexer.put(track.id)


class Indexer(CyclicThread):
    DELAY = 2
    BUFFER_SIZE = 100

    def __init__(self):
        CyclicThread.__init__(self)
        self.deque = deque()
        self.interrupted = False
        self.discoverer = Discoverer(self)
        self.last_commit = None
        self.tracks = []

    def start(self):
        logger.info('** indexation thread started **')
        self.discoverer.start()
        super().start()

    def act(self):
        buffer = []
        session = db.Session()
        track_repo = TrackRepository(session)

        for _ in range(self.BUFFER_SIZE):
            try:
                track = self.index(track_repo, self.deque.pop())
                buffer.append(track)
            except (FileNotFoundError, NotSupportedFile) as e:
                logger.warning("Error during indexation: %s" % e)
                continue
            except IndexError:
                break

        if buffer:
            for track in buffer:
                if track.id is None:
                    track_repo.create(track)

            logger.info(f"{len(buffer)} tracks indexed")
            track_repo.commit()

    def put(self, filename_or_track_id):
        self.deque.appendleft(filename_or_track_id)

    def in_deque(self, filename_or_track_id):
        return filename_or_track_id in self.deque

    @staticmethod
    def index(track_repo, filename_or_track_id):
        """ index a media file from the filesystem or a track id """

        if type(filename_or_track_id) is int:
            track = track_repo.get_by_id(filename_or_track_id)
            filename = Path(track.path)
            track_hash = track.hash

            if not filename.exists() and track.status != Track.STATUS_UNAVAILABLE:
                logger.debug('Index - missing: %s' % filename)
                track.status = Track.STATUS_UNAVAILABLE
                return track
        else:
            filename = Path(filename_or_track_id)
            if not filename.exists():
                raise FileNotFoundError(f"File not found: {filename}")
            if not is_media_file_ext(filename.ext):
                raise NotSupportedFile(f"File's extension {filename.ext} is not supported")

            track_hash = hash_file(filename)

            track = track_repo.get_by_hash(track_hash)
            if not track:
                track = Track()

        vlc_media = vlc.Media(filename)
        vlc_media.parse()
        track_infos = vlc_media.get_tracks_info()

        title = vlc_media.get_meta(vlc.Meta.Title)
        if not title or title == '(null)':
            title = filename.stripext().name
        track.title = title
        track.format = filename.ext
        track.artist = vlc_media.get_meta(vlc.Meta.AlbumArtist) or vlc_media.get_meta(vlc.Meta.Artist)
        track.album = vlc_media.get_meta(vlc.Meta.Album)
        track.track_num = vlc_media.get_meta(vlc.Meta.TrackNumber)
        # track.year = vlc_media.get_meta(vlc.Meta.Date)
        # track.duration = vlc_media.get_meta(vlc.Meta.Date)
        # track.size = 0
        track.note = ""
        track.status = Track.STATUS_FOUND
        track.path = filename
        track.hash = track_hash

        return track

    def stop(self):
        self.discoverer.stop()
        super().stop()
        logger.info('** indexation thread stopped **')


if __name__ == '__main__':
    indexer = Indexer()
    indexer.start()
    try:
        indexer.join()
    except KeyboardInterrupt:
        indexer.stop()