diora-web/podcasts/models.py

from django.contrib.auth.models import User
from django.db import models


class PodcastFeed(models.Model):
    # What happens to a new episode of this feed (one that passed the filter).
    NEW_INBOX = 'inbox'
    NEW_QUEUE = 'queue'
    NEW_IGNORE = 'ignore'
    NEW_EPISODE_ACTIONS = [
        (NEW_INBOX, 'In „Neu“'),
        (NEW_QUEUE, 'In die Warteschlange'),
        (NEW_IGNORE, 'Nicht anzeigen'),
    ]
    # Volume factors: below 1 quieter, above 1 louder (the apps boost those).
    VOLUMES = (0.2, 0.5, 1.0, 1.5, 2.0, 2.5)

    user = models.ForeignKey(User, on_delete=models.CASCADE, related_name='podcast_feeds')
    rss_url = models.URLField(max_length=1000)
    title = models.CharField(max_length=300)
    description = models.TextField(blank=True)
    artwork_url = models.URLField(max_length=1000, blank=True)
    author = models.CharField(max_length=300, blank=True)
    link = models.URLField(max_length=1000, blank=True)
    last_refreshed_at = models.DateTimeField(null=True, blank=True)
    added_at = models.DateTimeField(auto_now_add=True)
    # Always the same as new_episodes == 'queue' (see save()); kept for the
    # web UI's toggle and older clients.
    auto_queue = models.BooleanField(default=False)
    # HTTP validators of the last full (200) response, sent back as
    # If-None-Match / If-Modified-Since so an unchanged feed costs a 304.
    http_etag = models.CharField(max_length=500, blank=True)
    http_last_modified = models.CharField(max_length=100, blank=True)
    # The last failed refresh (network, HTTP status, not a feed); cleared on
    # success. error_count counts failures in a row and drives the backoff.
    last_error = models.CharField(max_length=500, blank=True)
    last_error_at = models.DateTimeField(null=True, blank=True)
    error_count = models.PositiveIntegerField(default=0)

    # Playback, for every device: null = the device's own speed.
    playback_speed = models.FloatField(null=True, blank=True)
    skip_intro = models.PositiveIntegerField(default=0)   # seconds
    skip_outro = models.PositiveIntegerField(default=0)   # seconds
    volume = models.FloatField(default=1.0)               # one of VOLUMES
    # New episodes: what happens to them, and which ones count at all.
    new_episodes = models.CharField(max_length=10, choices=NEW_EPISODE_ACTIONS, default=NEW_INBOX)
    # Comma-separated words, case-insensitive, matched against the title.
    include_words = models.CharField(max_length=500, blank=True)
    exclude_words = models.CharField(max_length=500, blank=True)
    min_duration = models.PositiveIntegerField(default=0)  # minutes, 0 = any
    # Devices download this feed's new episodes (those in the inbox) on
    # their own, for listening offline.
    auto_download = models.BooleanField(default=False)

    class Meta:
        unique_together = ('user', 'rss_url')
        ordering = ['title']

    def __str__(self):
        return self.title

    @classmethod
    def from_db(cls, *args, **kwargs):
        obj = super().from_db(*args, **kwargs)
        obj._remember_sync()
        return obj

    def refresh_from_db(self, *args, **kwargs):
        super().refresh_from_db(*args, **kwargs)
        self._remember_sync()

    def _remember_sync(self):
        self._synced = (self.__dict__.get('auto_queue'), self.__dict__.get('new_episodes'))

    def _sync_auto_queue(self):
        """Keep auto_queue == (new_episodes == 'queue'). Whichever of the two
        was changed since loading (or, for a new feed, set away from its
        default) wins; new_episodes if both were."""
        auto_before, new_before = getattr(self, '_synced', (False, self.NEW_INBOX))
        if self.new_episodes != new_before or self.auto_queue == auto_before:
            self.auto_queue = self.new_episodes == self.NEW_QUEUE
        elif self.auto_queue:
            self.new_episodes = self.NEW_QUEUE
        elif self.new_episodes == self.NEW_QUEUE:
            self.new_episodes = self.NEW_INBOX

    def save(self, *args, **kwargs):
        update_fields = kwargs.get('update_fields')
        if update_fields is not None and not {'auto_queue', 'new_episodes'} & set(update_fields):
            return super().save(*args, **kwargs)
        self._sync_auto_queue()
        if update_fields is not None:
            kwargs['update_fields'] = set(update_fields) | {'auto_queue', 'new_episodes'}
        super().save(*args, **kwargs)
        self._remember_sync()

    def matches_filter(self, title, duration_seconds):
        """Whether a new episode passes this feed's filter: an include word
        (if any are given), no exclude word, and long enough (if known)."""
        title = (title or '').casefold()
        include = _words(self.include_words)
        if include and not any(w in title for w in include):
            return False
        if any(w in title for w in _words(self.exclude_words)):
            return False
        if self.min_duration and duration_seconds and duration_seconds < self.min_duration * 60:
            return False
        return True


def _words(raw):
    return [w.strip().casefold() for w in (raw or '').split(',') if w.strip()]


class PodcastEpisode(models.Model):
    feed = models.ForeignKey(PodcastFeed, on_delete=models.CASCADE, related_name='episodes')
    guid = models.CharField(max_length=1000)
    title = models.CharField(max_length=500)
    description = models.TextField(blank=True)
    audio_url = models.URLField(max_length=1000)
    duration_seconds = models.IntegerField(default=0)
    pub_date = models.DateTimeField(null=True, blank=True)
    episode_number = models.IntegerField(null=True, blank=True)
    season_number = models.IntegerField(null=True, blank=True)
    artwork_url = models.URLField(max_length=1000, blank=True)
    discovered_at = models.DateTimeField(auto_now_add=True)
    # Podcasting 2.0 <podcast:chapters url=…>: a JSON file, loaded on demand.
    chapters_url = models.URLField(max_length=1000, blank=True)
    # [{start: seconds, title, url?, img?}], from Podlove Simple Chapters in
    # the feed, or once fetched: the JSON at chapters_url, else (or if that
    # has none) the CHAP frames of the audio file's ID3v2 tag.
    chapters = models.JSONField(default=list, blank=True)
    # When chapters were fetched into `chapters` (from chapters_url or the
    # audio file); null = not yet.
    chapters_fetched_at = models.DateTimeField(null=True, blank=True)
    # [{url, type, language}] from <podcast:transcript>.
    transcripts = models.JSONField(default=list, blank=True)

    class Meta:
        unique_together = ('feed', 'guid')
        ordering = ['-pub_date']

    def __str__(self):
        return self.title


class EpisodeTranscript(models.Model):
    """An episode's transcript, fetched and parsed once (transcripts.py).

    A table of its own rather than fields on PodcastEpisode: a parsed
    transcript runs to hundreds of KB, and episode lists load whole rows.
    """
    episode = models.OneToOneField(PodcastEpisode, on_delete=models.CASCADE, related_name='transcript_cache')
    # URLs of the transcripts it was made from, best first. When the feed
    # offers different ones, the cache is stale.
    sources = models.JSONField(default=list)
    # {"segments": [...], "source": {...}}, or null with an error.
    data = models.JSONField(null=True, blank=True)
    # '' when data is there; else why there's none for good: 'unsupported'
    # (no format with timing), 'invalid' (missing, broken, too large).
    error = models.CharField(max_length=20, blank=True)
    fetched_at = models.DateTimeField(auto_now=True)

    def __str__(self):
        return f'Transcript of {self.episode}'


# An episode counts as played this close to its end (AntennaPod's default),
# the same rule as the apps' "als gehört zählen".
PLAYED_BEFORE_END_SECONDS = 30


def counts_as_played(position_seconds, duration_seconds):
    return bool(duration_seconds) and duration_seconds > 0 and \
        position_seconds >= max(duration_seconds - PLAYED_BEFORE_END_SECONDS, 1)


class EpisodeProgress(models.Model):
    user = models.ForeignKey(User, on_delete=models.CASCADE, related_name='episode_progress')
    episode = models.ForeignKey(PodcastEpisode, on_delete=models.CASCADE, related_name='progress')
    position_seconds = models.IntegerField(default=0)
    played = models.BooleanField(default=False)
    dismissed = models.BooleanField(default=False)
    favorite = models.BooleanField(default=False)
    updated_at = models.DateTimeField(auto_now=True)

    class Meta:
        unique_together = ('user', 'episode')
        ordering = ['-updated_at']

    def __str__(self):
        return f'{self.user} - {self.episode}'


class PodcastQueue(models.Model):
    user = models.ForeignKey(User, on_delete=models.CASCADE, related_name='podcast_queue')
    episode = models.ForeignKey(PodcastEpisode, on_delete=models.CASCADE, related_name='queued_by')
    position = models.PositiveIntegerField(default=0)
    added_at = models.DateTimeField(auto_now_add=True)

    class Meta:
        unique_together = ('user', 'episode')
        ordering = ['position']

    def __str__(self):
        return f'{self.user} queue pos {self.position}: {self.episode}'

class PodcastSettings(models.Model):
    """A user's podcast preferences. Where episodes go in the queue: when added
    by hand (also when played from outside the queue), and when a feed that
    queues its new episodes brings some; and whether the queue stays sorted."""
    BACK = 'back'
    FRONT = 'front'
    AFTER_CURRENT = 'after_current'
    POSITIONS = [
        (BACK, 'Ans Ende'),
        (FRONT, 'An den Anfang'),
        (AFTER_CURRENT, 'Nach der laufenden Folge'),
    ]

    # Keep the queue sorted: after every change the whole queue is sorted
    # again by this rule ('' = off, the user's own order).
    SORT_ORDERS = [
        ('', 'Aus'),
        ('new', 'Neueste zuerst'),
        ('old', 'Älteste zuerst'),
        ('short', 'Kürzeste zuerst'),
        ('long', 'Längste zuerst'),
        ('feed', 'Nach Podcast'),
    ]

    user = models.OneToOneField(User, on_delete=models.CASCADE, related_name='podcast_settings')
    enqueue_position = models.CharField(max_length=20, choices=POSITIONS, default=BACK)
    # front = how auto-queue always worked before this setting existed
    auto_queue_position = models.CharField(max_length=20, choices=POSITIONS, default=FRONT)
    keep_sorted = models.CharField(max_length=10, choices=SORT_ORDERS, default='', blank=True)
    # What PodcastFeed.auto_download starts as for a new subscription.
    auto_download_default = models.BooleanField(default=False)
    # Episodes that were downloaded in AntennaPod when its backup was
    # imported: [{episode_id, size (bytes, 0 = unknown)}]. The app offers once
    # to download them again, then clears the list.
    import_downloads = models.JSONField(default=list, blank=True)

    @classmethod
    def for_user(cls, user):
        obj, _ = cls.objects.get_or_create(user=user)
        return obj

    @classmethod
    def new_feed_defaults(cls, user):
        """Field values a new PodcastFeed of the user starts with."""
        auto_download = cls.objects.filter(user=user).values_list('auto_download_default', flat=True).first()
        return {'auto_download': bool(auto_download)}

    def __str__(self):
        return f'Podcast settings of {self.user}'