gpodder/src/gpodder/model.py

# -*- coding: utf-8 -*-
#
# gPodder - A media aggregator and podcast client
# Copyright (c) 2005-2011 Thomas Perl and the gPodder Team
# Copyright (c) 2011 Neal H. Walfield
#
# gPodder is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 3 of the License, or
# (at your option) any later version.
#
# gPodder is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program.  If not, see <http://www.gnu.org/licenses/>.
#


#
#  gpodder.model - Core model classes for gPodder (2009-08-13)
#  Based on libpodcasts.py (thp, 2005-10-29)
#

import gpodder
from gpodder import util
from gpodder import feedcore
from gpodder import youtube
from gpodder import schema

import logging
logger = logging.getLogger(__name__)

import os
import re
import glob
import shutil
import time
import datetime
import rfc822
import hashlib
import feedparser
import collections
import string

_ = gpodder.gettext


class CustomFeed(feedcore.ExceptionWithData): pass

class gPodderFetcher(feedcore.Fetcher):
    """
    This class extends the feedcore Fetcher with the gPodder User-Agent and the
    Proxy handler based on the current settings in gPodder.
    """
    custom_handlers = []

    def __init__(self):
        feedcore.Fetcher.__init__(self, gpodder.user_agent)

    def fetch_channel(self, channel):
        etag = channel.http_etag
        modified = feedparser._parse_date(channel.http_last_modified)
        # If we have a username or password, rebuild the url with them included
        # Note: using a HTTPBasicAuthHandler would be pain because we need to
        # know the realm. It can be done, but I think this method works, too
        url = channel.authenticate_url(channel.url)
        for handler in self.custom_handlers:
            custom_feed = handler.handle_url(url)
            if custom_feed is not None:
                raise CustomFeed(custom_feed)
        self.fetch(url, etag, modified)

    def _resolve_url(self, url):
        return youtube.get_real_channel_url(url)

    @classmethod
    def register(cls, handler):
        cls.custom_handlers.append(handler)

# The "register" method is exposed here for external usage
register_custom_handler = gPodderFetcher.register

# Our podcast model:
#
# database -> podcast -> episode -> download/playback
#  podcast.parent == db
#  podcast.children == [episode, ...]
#  episode.parent == podcast
#
# - normally: episode.children = (None, None)
# - downloading: episode.children = (DownloadTask(), None)
# - playback: episode.children = (None, PlaybackTask())


class PodcastModelObject(object):
    """
    A generic base class for our podcast model providing common helper
    and utility functions.
    """
    __slots__ = ('id', 'parent', 'children', 'changed')

    @classmethod
    def create_from_dict(cls, d, *args):
        """
        Create a new object, passing "args" to the constructor
        and then updating the object with the values from "d".
        """
        o = cls(*args)

        o.changed = None

        # XXX: all(map(lambda k: hasattr(o, k), d))?
        for k, v in d.iteritems():
            setattr(o, k, v)

        o.changed = {}

        return o

    def __setattr__(self, name, value):
        """Track changes once "self.changed" is a dictionary

        The changed values will be stored in self.changed until
        _clear_changes is called.
        """
        if getattr(self, 'changed', None) is not None and self.id is not None:
            old_value = getattr(self, name, None)

            if old_value is not None and value != old_value:
                # Value changed (and it is not an initialization)
                if name not in self.changed:
                    self.changed[name] = old_value
                # logger.debug("%s: %s.%s changed: %s -> %s"
                #              % (self.__class__.__name__, self.id, name,
                #                 old_value, value))

        super(PodcastModelObject, self).__setattr__(name, value)

    def _clear_changes(self):
        # logger.debug("Changes: %s: %s"
        #              % ([getattr (self, a) for a in self.__slots__],
        #                 str(self.changed),))
        self.changed = {}

class PodcastEpisode(PodcastModelObject):
    """holds data for one object in a channel"""
    MAX_FILENAME_LENGTH = 200

    __slots__ = schema.EpisodeColumns

    def _deprecated(self):
        raise Exception('Property is deprecated!')

    is_played = property(fget=_deprecated, fset=_deprecated)
    is_locked = property(fget=_deprecated, fset=_deprecated)

    def _get_podcast_id(self):
        return self.channel.id

    def _set_podcast_id(self, podcast_id):
        assert self.channel.id == podcast_id

    # Accessor for the "podcast_id" DB column
    podcast_id = property(fget=_get_podcast_id, fset=_set_podcast_id)

    def has_website_link(self):
        return bool(self.link) and (self.link != self.url or \
                youtube.is_video_link(self.link))

    @classmethod
    def from_feedparser_entry(cls, entry, channel, mimetype_prefs=''):
        episode = cls(channel)

        # Replace multi-space and newlines with single space (Maemo bug 11173)
        episode.title = re.sub('\s+', ' ', entry.get('title', ''))
        episode.link = entry.get('link', '')
        if 'content' in entry and len(entry['content']) and \
                entry['content'][0].get('type', '') == 'text/html':
            episode.description = entry['content'][0].value
        else:
            episode.description = entry.get('summary', '')

        try:
            # Parse iTunes-specific podcast duration metadata
            total_time = util.parse_time(entry.get('itunes_duration', ''))
            episode.total_time = total_time
        except:
            pass

        # Fallback to subtitle if summary is not available0
        if not episode.description:
            episode.description = entry.get('subtitle', '')

        episode.guid = entry.get('id', '')
        if entry.get('updated_parsed', None):
            episode.published = rfc822.mktime_tz(entry.updated_parsed+(0,))

        enclosures = entry.get('enclosures', [])
        media_rss_content = entry.get('media_content', [])
        audio_available = any(e.get('type', '').startswith('audio/') \
                for e in enclosures + media_rss_content)
        video_available = any(e.get('type', '').startswith('video/') \
                for e in enclosures + media_rss_content)

        # Create the list of preferred mime types
        mimetype_prefs = mimetype_prefs.split(',')

        def calculate_preference_value(enclosure):
            """Calculate preference value of an enclosure

            This is based on mime types and allows users to prefer
            certain mime types over others (e.g. MP3 over AAC, ...)
            """
            mimetype = enclosure.get('type', None)
            try:
                # If the mime type is found, return its (zero-based) index
                return mimetype_prefs.index(mimetype)
            except ValueError:
                # If it is not found, assume it comes after all listed items
                return len(mimetype_prefs)

        # Enclosures
        for e in sorted(enclosures, key=calculate_preference_value):
            episode.mime_type = e.get('type', 'application/octet-stream')
            if episode.mime_type == '':
                # See Maemo bug 10036
                logger.warn('Fixing empty mimetype in ugly feed')
                episode.mime_type = 'application/octet-stream'

            if '/' not in episode.mime_type:
                continue

            # Skip images in feeds if audio or video is available (bug 979)
            # This must (and does) also look in Media RSS enclosures (bug 1430)
            if episode.mime_type.startswith('image/') and \
                    (audio_available or video_available):
                continue

            # If we have audio or video available later on, skip
            # 'application/octet-stream' data types (fixes Linux Outlaws)
            if episode.mime_type == 'application/octet-stream' and \
                    (audio_available or video_available):
                continue

            episode.url = util.normalize_feed_url(e.get('href', ''))
            if not episode.url:
                continue

            try:
                episode.file_size = int(e.length) or -1
            except:
                episode.file_size = -1

            return episode

        # Media RSS content
        for m in sorted(media_rss_content, key=calculate_preference_value):
            episode.mime_type = m.get('type', 'application/octet-stream')
            if '/' not in episode.mime_type:
                continue

            # Skip images in Media RSS if we have audio/video (bug 1444)
            if episode.mime_type.startswith('image/') and \
                    (audio_available or video_available):
                continue

            episode.url = util.normalize_feed_url(m.get('url', ''))
            if not episode.url:
                continue

            try:
                episode.file_size = int(m.fileSize) or -1
            except:
                episode.file_size = -1

            return episode

        # Brute-force detection of any links
        for l in entry.get('links', ()):
            episode.url = util.normalize_feed_url(l.get('href', ''))
            if not episode.url:
                continue

            if youtube.is_video_link(episode.url):
                return episode

            # Check if we can resolve this link to a audio/video file
            filename, extension = util.filename_from_url(episode.url)
            file_type = util.file_type_by_extension(extension)
            if file_type is None and hasattr(l, 'type'):
                extension = util.extension_from_mimetype(l.type)
                file_type = util.file_type_by_extension(extension)

            # The link points to a audio or video file - use it!
            if file_type is not None:
                return episode

        # Scan MP3 links in description text
        mp3s = re.compile(r'http://[^"]*\.mp3')
        for content in entry.get('content', ()):
            html = content.value
            for match in mp3s.finditer(html):
                episode.url = match.group(0)
                return episode

        return None

    def __init__(self, channel):
        self.parent = channel
        self.children = (None, None)

        self.id = None
        self.url = ''
        self.title = ''
        self.file_size = 0
        self.mime_type = 'application/octet-stream'
        self.guid = ''
        self.description = ''
        self.link = ''
        self.published = 0
        self.download_filename = None

        self.state = gpodder.STATE_NORMAL
        self.is_new = True
        self.archive = channel.auto_archive_episodes

        # Time attributes
        self.total_time = 0
        self.current_position = 0
        self.current_position_updated = 0

        # Timestamp of last playback time
        self.last_playback = 0

    @property
    def channel(self):
        return self.parent

    @property
    def db(self):
        return self.parent.parent.db

    def _set_download_task(self, download_task):
        self.children = (download_task, self.children[1])

    def _get_download_task(self):
        return self.children[0]

    download_task = property(_get_download_task, _set_download_task)

    @property
    def downloading(self):
        task = self.download_task
        if task is None:
            return False

        return task.status in (task.DOWNLOADING, task.QUEUED, task.PAUSED)

    def check_is_new(self):
        return (self.state == gpodder.STATE_NORMAL and self.is_new and
                not self.downloading)

    def save(self):
        gpodder.user_hooks.on_episode_save(self)

        self._clear_changes()

        self.db.save_episode(self)

    def on_downloaded(self, filename):
        self.state = gpodder.STATE_DOWNLOADED
        self.is_new = True
        self.file_size = os.path.getsize(filename)
        self.save()

    def set_state(self, state):
        self.state = state
        self.save()

    def playback_mark(self):
        self.is_new = False
        self.last_playback = int(time.time())
        self.save()

    def mark(self, state=None, is_played=None, is_locked=None):
        if state is not None:
            self.state = state
        if is_played is not None:
            self.is_new = not is_played
        if is_locked is not None:
            self.archive = is_locked
        self.save()

    def age_in_days(self):
        return util.file_age_in_days(self.local_filename(create=False, \
                check_only=True))

    age_int_prop = property(fget=age_in_days)

    def get_age_string(self):
        return util.file_age_to_string(self.age_in_days())

    age_prop = property(fget=get_age_string)

    @property
    def description_html(self):
        # XXX: That's not a very well-informed heuristic to check
        # if the description already contains HTML. Better ideas?
        if '<' in self.description:
            return self.description

        return self.description.replace('\n', '<br>')

    def one_line_description(self):
        MAX_LINE_LENGTH = 120
        desc = util.remove_html_tags(self.description or '')
        desc = re.sub('\s+', ' ', desc).strip()
        if not desc:
            return _('No description available')
        else:
            # Decode the description to avoid gPodder bug 1277
            if isinstance(desc, str):
                desc = desc.decode('utf-8', 'ignore')

            desc = desc.strip()

            if len(desc) > MAX_LINE_LENGTH:
                return desc[:MAX_LINE_LENGTH] + '...'
            else:
                return desc

    def delete_from_disk(self):
        filename = self.local_filename(create=False, check_only=True)
        if filename is not None:
            gpodder.user_hooks.on_episode_delete(self, filename)
            util.delete_file(filename)

        self.set_state(gpodder.STATE_DELETED)

    def get_playback_url(self, fmt_id=None, allow_partial=False):
        """Local (or remote) playback/streaming filename/URL

        Returns either the local filename or a streaming URL that
        can be used to playback this episode.

        Also returns the filename of a partially downloaded file
        in case partial (preview) playback is desired.
        """
        url = self.local_filename(create=False)

        if (allow_partial and url is not None and
                os.path.exists(url + '.partial')):
            return url + '.partial'

        if url is None or not os.path.exists(url):
            url = self.url
            if youtube.is_video_link(url):
                url = youtube.get_real_download_url(url, fmt_id)

        return url

    def find_unique_file_name(self, filename, extension):
        # Remove leading and trailing whitespace + dots (to avoid hidden files)
        filename = filename.strip('.' + string.whitespace) + extension

        for name in util.generate_names(filename):
            if (not self.db.episode_filename_exists(self.podcast_id, name) or
                    self.download_filename == name):
                return name

    def local_filename(self, create, force_update=False, check_only=False,
            template=None, return_wanted_filename=False):
        """Get (and possibly generate) the local saving filename

        Pass create=True if you want this function to generate a
        new filename if none exists. You only want to do this when
        planning to create/download the file after calling this function.

        Normally, you should pass create=False. This will only
        create a filename when the file already exists from a previous
        version of gPodder (where we used md5 filenames). If the file
        does not exist (and the filename also does not exist), this
        function will return None.

        If you pass force_update=True to this function, it will try to
        find a new (better) filename and move the current file if this
        is the case. This is useful if (during the download) you get
        more information about the file, e.g. the mimetype and you want
        to include this information in the file name generation process.

        If check_only=True is passed to this function, it will never try
        to rename the file, even if would be a good idea. Use this if you
        only want to check if a file exists.

        If "template" is specified, it should be a filename that is to
        be used as a template for generating the "real" filename.

        The generated filename is stored in the database for future access.

        If return_wanted_filename is True, the filename will not be written to
        the database, but simply returned by this function (for use by the
        "import external downloads" feature).
        """
        if self.download_filename is None and (check_only or not create):
            return None

        ext = self.extension(may_call_local_filename=False).encode('utf-8', 'ignore')

        if not check_only and (force_update or not self.download_filename):
            # Avoid and catch gPodder bug 1440 and similar situations
            if template == '':
                logger.warn('Empty template. Report this podcast URL %s',
                        self.channel.url)
                template = None

            # Try to find a new filename for the current file
            if template is not None:
                # If template is specified, trust the template's extension
                episode_filename, ext = os.path.splitext(template)
            else:
                episode_filename, _ = util.filename_from_url(self.url)
            fn_template = util.sanitize_filename(episode_filename, self.MAX_FILENAME_LENGTH)

            if 'redirect' in fn_template and template is None:
                # This looks like a redirection URL - force URL resolving!
                logger.warn('Looks like a redirection to me: %s', self.url)
                url = util.get_real_url(self.channel.authenticate_url(self.url))
                logger.info('Redirection resolved to: %s', url)
                episode_filename, _ = util.filename_from_url(url)
                fn_template = util.sanitize_filename(episode_filename, self.MAX_FILENAME_LENGTH)

            # Use title for YouTube downloads and Soundcloud streams
            if youtube.is_video_link(self.url) or fn_template == 'stream':
                sanitized = util.sanitize_filename(self.title, self.MAX_FILENAME_LENGTH)
                if sanitized:
                    fn_template = sanitized

            # If the basename is empty, use the md5 hexdigest of the URL
            if not fn_template or fn_template.startswith('redirect.'):
                logger.error('Report this feed: Podcast %s, episode %s',
                        self.channel.url, self.url)
                fn_template = hashlib.md5(self.url).hexdigest()

            # Find a unique filename for this episode
            wanted_filename = self.find_unique_file_name(fn_template, ext)

            if return_wanted_filename:
                # return the calculated filename without updating the database
                return wanted_filename

            # The old file exists, but we have decided to want a different filename
            if self.download_filename and wanted_filename != self.download_filename:
                # there might be an old download folder crawling around - move it!
                new_file_name = os.path.join(self.channel.save_dir, wanted_filename)
                old_file_name = os.path.join(self.channel.save_dir, self.download_filename)
                if os.path.exists(old_file_name) and not os.path.exists(new_file_name):
                    logger.info('Renaming %s => %s', old_file_name, new_file_name)
                    os.rename(old_file_name, new_file_name)
                elif force_update and not os.path.exists(old_file_name):
                    # When we call force_update, the file might not yet exist when we
                    # call it from the downloading code before saving the file
                    logger.info('Choosing new filename: %s', new_file_name)
                else:
                    logger.warn('%s exists or %s does not', new_file_name, old_file_name)
                logger.info('Updating filename of %s to "%s".', self.url, wanted_filename)
            elif self.download_filename is None:
                logger.info('Setting download filename: %s', wanted_filename)
            self.download_filename = wanted_filename
            self.save()

        return os.path.join(self.channel.save_dir, self.download_filename)

    def set_mimetype(self, mimetype, commit=False):
        """Sets the mimetype for this episode"""
        self.mime_type = mimetype
        if commit:
            self.db.commit()

    def extension(self, may_call_local_filename=True):
        filename, ext = util.filename_from_url(self.url)
        if may_call_local_filename:
            filename = self.local_filename(create=False)
            if filename is not None:
                filename, ext = os.path.splitext(filename)
        # if we can't detect the extension from the url fallback on the mimetype
        if ext == '' or util.file_type_by_extension(ext) is None:
            ext = util.extension_from_mimetype(self.mime_type)
        return ext

    def mark_new(self):
        self.is_new = True
        self.save()

    def mark_old(self):
        self.is_new = False
        self.save()

    def file_exists(self):
        filename = self.local_filename(create=False, check_only=True)
        if filename is None:
            return False
        else:
            return os.path.exists(filename)

    def was_downloaded(self, and_exists=False):
        if self.state != gpodder.STATE_DOWNLOADED:
            return False
        if and_exists and not self.file_exists():
            return False
        return True

    def sync_filename(self):
        return self.title

    def file_type(self):
        # Assume all YouTube links are video files
        if youtube.is_video_link(self.url):
            return 'video'

        return util.file_type_by_extension(self.extension())

    @property
    def basename( self):
        return os.path.splitext( os.path.basename( self.url))[0]

    @property
    def pubtime(self):
        """
        Returns published time as HHMM (or 0000 if not available)
        """
        try:
            return datetime.datetime.fromtimestamp(self.published).strftime('%H%M')
        except:
            logger.warn('Cannot format pubtime: %s', self.title, exc_info=True)
            return '0000'

    def playlist_title(self):
        """Return a title for this episode in a playlist

        The title will be composed of the podcast name, the
        episode name and the publication date. The return
        value is the canonical representation of this episode
        in playlists (for example, M3U playlists).
        """
        return '%s - %s (%s)' % (self.channel.title, \
                self.title, \
                self.cute_pubdate())

    def cute_pubdate(self):
        result = util.format_date(self.published)
        if result is None:
            return '(%s)' % _('unknown')
        else:
            return result
    
    pubdate_prop = property(fget=cute_pubdate)

    def calculate_filesize(self):
        filename = self.local_filename(create=False)
        if filename is None:
            return

        try:
            self.file_size = os.path.getsize(filename)
        except:
            logger.error('Could not get file size: %s', filename, exc_info=True)

    def is_finished(self):
        """Return True if this episode is considered "finished playing"

        An episode is considered "finished" when there is a
        current position mark on the track, and when the
        current position is greater than 99 percent of the
        total time or inside the last 10 seconds of a track.
        """
        return self.current_position > 0 and \
                (self.current_position + 10 >= self.total_time or \
                 self.current_position >= self.total_time*.99)

    def get_play_info_string(self, duration_only=False):
        duration = util.format_time(self.total_time)
        if duration_only and self.total_time > 0:
            return duration
        elif self.current_position > 0 and \
                self.current_position != self.total_time:
            position = util.format_time(self.current_position)
            return '%s / %s' % (position, duration)
        elif self.total_time > 0:
            return duration
        else:
            return '-'

    def is_duplicate(self, episode):
        if self.title == episode.title and self.published == episode.published:
            logger.warn('Possible duplicate detected: %s', self.title)
            return True
        return False

    def duplicate_id(self):
        return hash((self.title, self.published))

    def update_from(self, episode):
        for k in ('title', 'url', 'description', 'link', 'published', 'guid', 'file_size'):
            setattr(self, k, getattr(episode, k))



class PodcastChannel(PodcastModelObject):
    __slots__ = schema.PodcastColumns

    UNICODE_TRANSLATE = {ord(u'ö'): u'o', ord(u'ä'): u'a', ord(u'ü'): u'u'}

    MAX_FOLDERNAME_LENGTH = 60
    SECONDS_PER_WEEK = 7*24*60*60
    EpisodeClass = PodcastEpisode

    feed_fetcher = gPodderFetcher()

    def __init__(self, model):
        self.parent = model
        self.children = None

        self.id = None
        self.url = None
        self.title = ''
        self.link = ''
        self.description = ''
        self.cover_url = None

        self.auth_username = ''
        self.auth_password = ''

        self.http_last_modified = None
        self.http_etag = None

        self.auto_archive_episodes = False
        self.download_folder = None
        self.pause_subscription = False

        self.section = _('Other')

    @property
    def model(self):
        return self.parent

    @property
    def db(self):
        return self.parent.db

    def import_external_files(self):
        """Check the download folder for externally-downloaded files

        This will try to assign downloaded files with episodes in the
        database and (failing that) will move downloaded files into
        the "Unknown" subfolder in the download directory, so that
        the user knows that gPodder doesn't know to which episode the
        file belongs (the "Unknown" folder may be used by external
        tools or future gPodder versions for better import support).
        """
        known_files = set(e.local_filename(create=False) \
                for e in self.get_downloaded_episodes())
        existing_files = set(filename for filename in \
                glob.glob(os.path.join(self.save_dir, '*')) \
                if not filename.endswith('.partial'))
        external_files = existing_files.difference(list(known_files) + \
                [os.path.join(self.save_dir, x) \
                for x in ('folder.jpg', 'Unknown')])
        if not external_files:
            return 0

        all_episodes = self.get_all_episodes()

        count = 0
        for filename in external_files:
            found = False

            basename = os.path.basename(filename)
            existing = [e for e in all_episodes if e.download_filename == basename]
            if existing:
                existing = existing[0]
                logger.info('Importing external download: %s', filename)
                existing.on_downloaded(filename)
                count += 1
                continue

            for episode in all_episodes:
                wanted_filename = episode.local_filename(create=True, \
                        return_wanted_filename=True)
                if basename == wanted_filename:
                    logger.info('Importing external download: %s', filename)
                    episode.download_filename = basename
                    episode.on_downloaded(filename)
                    count += 1
                    found = True
                    break

                wanted_base, wanted_ext = os.path.splitext(wanted_filename)
                target_base, target_ext = os.path.splitext(basename)
                if wanted_base == target_base:
                    # Filenames only differ by the extension
                    wanted_type = util.file_type_by_extension(wanted_ext)
                    target_type = util.file_type_by_extension(target_ext)

                    # If wanted type is None, assume that we don't know
                    # the right extension before the download (e.g. YouTube)
                    # if the wanted type is the same as the target type,
                    # assume that it's the correct file
                    if wanted_type is None or wanted_type == target_type:
                        logger.info('Importing external download: %s', filename)
                        episode.download_filename = basename
                        episode.on_downloaded(filename)
                        found = True
                        count += 1
                        break

            if not found:
                logger.warn('Unknown external file: %s', filename)
                target_dir = os.path.join(self.save_dir, 'Unknown')
                if util.make_directory(target_dir):
                    target_file = os.path.join(target_dir, basename)
                    logger.info('Moving %s => %s', filename, target_file)
                    try:
                        shutil.move(filename, target_file)
                    except Exception, e:
                        logger.error('Could not move file: %s', e, exc_info=True)

        return count

    @classmethod
    def sort_key(cls, podcast):
        key = podcast.title.lower()
        if not isinstance(key, unicode):
            key = key.decode('utf-8', 'ignore')
        return re.sub('^the ', '', key).translate(cls.UNICODE_TRANSLATE)

    @classmethod
    def load(cls, model, url, create=True, authentication_tokens=None,\
            max_episodes=0, \
            mimetype_prefs=''):
        if isinstance(url, unicode):
            url = url.encode('utf-8')

        existing = filter(lambda p: p.url == url, model.get_podcasts())

        if existing:
            return existing[0]

        if create:
            tmp = cls(model)
            tmp.url = url
            if authentication_tokens is not None:
                tmp.auth_username = authentication_tokens[0]
                tmp.auth_password = authentication_tokens[1]

            # Save podcast, so it gets an ID assigned before
            # updating the feed and adding saving episodes
            tmp.save()

            try:
                tmp.update(max_episodes, mimetype_prefs)
            except Exception, e:
                logger.debug('Fetch failed. Removing buggy feed.')
                tmp.remove_downloaded()
                tmp.delete()
                raise

            # Determine the section in which this podcast should appear
            tmp.section = tmp._get_content_type()

            # Determine a new download folder now that we have the title
            tmp.get_save_dir(force_new=True)

            # Mark episodes as downloaded if files already exist (bug 902)
            tmp.import_external_files()

            tmp.save()

            return tmp

    def episode_factory(self, d):
        """
        This function takes a dictionary containing key-value pairs for
        episodes and returns a new PodcastEpisode object that is connected
        to this object.

        Returns: A new PodcastEpisode object
        """
        return self.EpisodeClass.create_from_dict(d, self)

    def _consume_updated_title(self, new_title):
        # Replace multi-space and newlines with single space (Maemo bug 11173)
        new_title = re.sub('\s+', ' ', new_title).strip()

        # Only update the podcast-supplied title when we
        # don't yet have a title, or if the title is the
        # feed URL (e.g. we didn't find a title before).
        if not self.title or self.title == self.url:
            self.title = new_title

    def _consume_custom_feed(self, custom_feed, max_episodes=0):
        self._consume_updated_title(custom_feed.get_title())
        self.link = custom_feed.get_link()
        self.description = custom_feed.get_description()
        self.cover_url = custom_feed.get_image()
        self.save()

        guids = [episode.guid for episode in self.get_all_episodes()]

        assert self.children is not None

        # Insert newly-found episodes into the database + local cache
        self.children.extend(custom_feed.get_new_episodes(self, guids))

        # Sort episodes by pubdate, descending
        self.children.sort(key=lambda e: e.published, reverse=True)

        self.save()

        self.db.purge(max_episodes, self.id)

    def _consume_updated_feed(self, feed, max_episodes=0, mimetype_prefs=''):
        #self.parse_error = feed.get('bozo_exception', None)

        self._consume_updated_title(feed.feed.get('title', self.url))

        self.link = feed.feed.get('link', self.link)
        self.description = feed.feed.get('subtitle', self.description)
        # Start YouTube-specific title FIX
        YOUTUBE_PREFIX = 'Uploads by '
        if self.title.startswith(YOUTUBE_PREFIX):
            self.title = self.title[len(YOUTUBE_PREFIX):] + ' on YouTube'
        # End YouTube-specific title FIX

        if hasattr(feed.feed, 'image'):
            for attribute in ('href', 'url'):
                new_value = getattr(feed.feed.image, attribute, None)
                if new_value is not None:
                    self.cover_url = new_value

        if hasattr(feed.feed, 'icon'):
            self.cover_url = feed.feed.icon

        self.save()

        # Load all episodes to update them properly.
        existing = self.get_all_episodes()

        # We can limit the maximum number of entries that gPodder will parse
        if max_episodes > 0 and len(feed.entries) > max_episodes:
            # We have to sort the entries in descending chronological order,
            # because if the feed lists items in ascending order and has >
            # max_episodes old episodes, new episodes will not be shown.
            # See also: gPodder Bug 1186
            try:
                entries = sorted(feed.entries, \
                        key=lambda x: x.get('updated_parsed', (0,)*9), \
                        reverse=True)[:max_episodes]
            except Exception, e:
                logger.warn('Could not sort episodes: %s', e, exc_info=True)
                entries = feed.entries[:max_episodes]
        else:
            entries = feed.entries

        # Title + PubDate hashes for existing episodes
        existing_dupes = dict((e.duplicate_id(), e) for e in existing)

        # GUID-based existing episode list
        existing_guids = dict((e.guid, e) for e in existing)

        # Get most recent published of all episodes
        last_published = self.db.get_last_published(self) or 0

        # Keep track of episode GUIDs currently seen in the feed
        seen_guids = set()

        # Search all entries for new episodes
        for entry in entries:
            try:
                episode = self.EpisodeClass.from_feedparser_entry(entry, self, mimetype_prefs)
                if episode is not None:
                    if not episode.title:
                        logger.warn('Using filename as title for %s',
                                episode.url)
                        basename = os.path.basename(episode.url)
                        episode.title, ext = os.path.splitext(basename)

                    # Maemo bug 12073
                    if not episode.guid:
                        logger.warn('Using download URL as GUID for %s',
                                episode.title)
                        episode.guid = episode.url

                    seen_guids.add(episode.guid)
            except Exception, e:
                logger.error('Skipping episode: %s', e, exc_info=True)
                continue

            if episode is None:
                continue

            # Detect (and update) existing episode based on GUIDs
            existing_episode = existing_guids.get(episode.guid, None)
            if existing_episode:
                existing_episode.update_from(episode)
                existing_episode.save()
                continue

            # Detect (and update) existing episode based on duplicate ID
            existing_episode = existing_dupes.get(episode.duplicate_id(), None)
            if existing_episode:
                if existing_episode.is_duplicate(episode):
                    existing_episode.update_from(episode)
                    existing_episode.save()
                    continue

            # Workaround for bug 340: If the episode has been
            # published earlier than one week before the most
            # recent existing episode, do not mark it as new.
            if episode.published < last_published - self.SECONDS_PER_WEEK:
                logger.debug('Episode with old date: %s', episode.title)
                episode.is_new = False

            episode.save()

            # This episode is new - if we already loaded the
            # list of "children" episodes, add it to this list
            if self.children is not None:
                self.children.append(episode)

        # Remove "unreachable" episodes - episodes that have not been
        # downloaded and that the feed does not list as downloadable anymore
        if self.id is not None:
            episodes_to_purge = (e for e in existing if \
                    e.state != gpodder.STATE_DOWNLOADED and \
                    e.guid not in seen_guids)

            for episode in episodes_to_purge:
                logger.debug('Episode removed from feed: %s (%s)',
                        episode.title, episode.guid)
                gpodder.user_hooks.on_episode_removed_from_podcast(episode)
                self.db.delete_episode_by_guid(episode.guid, self.id)

                # Remove the episode from the "children" episodes list
                if self.children is not None:
                    self.children.remove(episode)

        # This *might* cause episodes to be skipped if there were more than
        # max_episodes_per_feed items added to the feed between updates.
        # The benefit is that it prevents old episodes from apearing as new
        # in certain situations (see bug #340).
        self.db.purge(max_episodes, self.id) # TODO: Remove from self.children!

        # Sort episodes by pubdate, descending
        self.children.sort(key=lambda e: e.published, reverse=True)

    def _update_etag_modified(self, feed):
        self.http_etag = feed.headers.get('etag', self.http_etag)
        self.http_last_modified = feed.headers.get('last-modified', self.http_last_modified)

    def update(self, max_episodes=0, mimetype_prefs=''):
        try:
            self.feed_fetcher.fetch_channel(self)
        except CustomFeed, updated:
            custom_feed = updated.data
            self._consume_custom_feed(custom_feed, max_episodes)
            self.save()
        except feedcore.UpdatedFeed, updated:
            feed = updated.data
            self._consume_updated_feed(feed, max_episodes, mimetype_prefs)
            self._update_etag_modified(feed)
            self.save()
        except feedcore.NewLocation, updated:
            feed = updated.data
            logger.info('New feed location: %s => %s', self.url, feed.href)
            self.url = feed.href
            self._consume_updated_feed(feed, max_episodes, mimetype_prefs)
            self._update_etag_modified(feed)
            self.save()
        except feedcore.NotModified, updated:
            feed = updated.data
            self._update_etag_modified(feed)
            self.save()
        except Exception, e:
            # "Not really" errors
            #feedcore.AuthenticationRequired
            # Temporary errors
            #feedcore.Offline
            #feedcore.BadRequest
            #feedcore.InternalServerError
            #feedcore.WifiLogin
            # Permanent errors
            #feedcore.Unsubscribe
            #feedcore.NotFound
            #feedcore.InvalidFeed
            #feedcore.UnknownStatusCode
            gpodder.user_hooks.on_podcast_update_failed(self, e)
            raise

        gpodder.user_hooks.on_podcast_updated(self)

        self.db.commit()

    def delete(self):
        gpodder.user_hooks.on_podcast_delete(self)
        self.db.delete_podcast(self)
        self.model._remove_podcast(self)

    def save(self):
        if self.download_folder is None:
            self.get_save_dir()

        if self.id is None:
            gpodder.user_hooks.on_podcast_subscribe(self)

        gpodder.user_hooks.on_podcast_save(self)

        self._clear_changes()

        self.db.save_podcast(self)
        self.model._append_podcast(self)

    def get_statistics(self):
        if self.id is None:
            return (0, 0, 0, 0, 0)
        else:
            return self.db.get_podcast_statistics(self.id)

    @property
    def group_by(self):
        if not self.section:
            self.section = self._get_content_type()
            self.save()

        return self.section

    def _get_content_type(self):
        if 'youtube.com' in self.url:
            return _('Video')

        audio, video, other = 0, 0, 0
        for content_type in self.db.get_content_types(self.id):
            content_type = content_type.lower()
            if content_type.startswith('audio'):
                audio += 1
            elif content_type.startswith('video'):
                video += 1
            else:
                other += 1

        if audio >= video:
            return _('Audio')
        elif video > other:
            return _('Video')

        return _('Other')

    def authenticate_url(self, url):
        return util.url_add_authentication(url, self.auth_username, self.auth_password)

    def _get_cover_url(self):
        return self.cover_url

    image = property(_get_cover_url)

    def rename(self, new_title):
        new_title = new_title.strip()
        if self.title == new_title:
            return

        new_folder_name = self.find_unique_folder_name(new_title)
        if new_folder_name and new_folder_name != self.download_folder:
            new_folder = os.path.join(gpodder.downloads, new_folder_name)
            old_folder = os.path.join(gpodder.downloads, self.download_folder)
            if os.path.exists(old_folder):
                if not os.path.exists(new_folder):
                    # Old folder exists, new folder does not -> simply rename
                    logger.info('Renaming %s => %s', old_folder, new_folder)
                    os.rename(old_folder, new_folder)
                else:
                    # Both folders exist -> move files and delete old folder
                    logger.info('Moving files from %s to %s', old_folder,
                            new_folder)
                    for file in glob.glob(os.path.join(old_folder, '*')):
                        shutil.move(file, new_folder)
                    logger.info('Removing %s', old_folder)
                    shutil.rmtree(old_folder, ignore_errors=True)
            self.download_folder = new_folder_name

        self.title = new_title
        self.save()

    def get_downloaded_episodes(self):
        return filter(lambda e: e.was_downloaded(), self.get_all_episodes())

    def get_all_episodes(self):
        if self.children is None:
            self.children = self.db.load_episodes(self, self.episode_factory)
        return self.children

    def find_unique_folder_name(self, download_folder):
        # Remove trailing dots to avoid errors on Windows (bug 600)
        # Also remove leading dots to avoid hidden folders on Linux
        download_folder = download_folder.strip('.' + string.whitespace)

        for folder_name in util.generate_names(download_folder):
            if (not self.db.podcast_download_folder_exists(folder_name) or
                    self.download_folder == folder_name):
                return folder_name

    def get_save_dir(self, force_new=False):
        if self.download_folder is None or force_new:
            # we must change the folder name, because it has not been set manually
            fn_template = util.sanitize_filename(self.title, self.MAX_FOLDERNAME_LENGTH)

            if not fn_template:
                fn_template = util.sanitize_filename(self.url, self.MAX_FOLDERNAME_LENGTH)

            # Find a unique folder name for this podcast
            download_folder = self.find_unique_folder_name(fn_template)

            # Try removing the download folder if it has been created previously
            if self.download_folder is not None:
                folder = os.path.join(gpodder.downloads, self.download_folder)
                try:
                    os.rmdir(folder)
                except OSError:
                    logger.info('Old download folder is kept for %s', self.url)

            logger.info('Updating download_folder of %s to %s', self.url,
                    download_folder)
            self.download_folder = download_folder
            self.save()

        save_dir = os.path.join(gpodder.downloads, self.download_folder)

        # Create save_dir if it does not yet exist
        if not util.make_directory(save_dir):
            logger.error('Could not create save_dir: %s', save_dir)

        return save_dir
    
    save_dir = property(fget=get_save_dir)

    def remove_downloaded(self):
        # Remove the download directory
        for episode in self.get_downloaded_episodes():
            filename = episode.local_filename(create=False, check_only=True)
            if filename is not None:
                gpodder.user_hooks.on_episode_delete(episode, filename)

        shutil.rmtree(self.save_dir, True)

    @property
    def cover_file(self):
        return os.path.join(self.save_dir, 'folder.jpg')


class Model(object):
    PodcastClass = PodcastChannel

    def __init__(self, db):
        self.db = db
        self.children = None

    def _append_podcast(self, podcast):
        if podcast not in self.children:
            self.children.append(podcast)
        # XXX: Sort?

    def _remove_podcast(self, podcast):
        self.children.remove(podcast)

    def get_podcasts(self):
        def podcast_factory(dct, db):
            return self.PodcastClass.create_from_dict(dct, self)

        if self.children is None:
            self.children = sorted(self.db.load_podcasts(podcast_factory),
                    key=self.PodcastClass.sort_key)
        return self.children

    def load_podcast(self, url, create=True, authentication_tokens=None,
                     max_episodes=0, mimetype_prefs=''):
        return self.PodcastClass.load(self, url, create,
                                      authentication_tokens,
                                      max_episodes, mimetype_prefs)

    @classmethod
    def podcast_sort_key(cls, podcast):
        return cls.PodcastClass.sort_key(podcast)

    @classmethod
    def episode_sort_key(cls, episode):
        return episode.published

    @classmethod
    def sort_episodes_by_pubdate(cls, episodes, reverse=False):
        """Sort a list of PodcastEpisode objects chronologically

        Returns a iterable, sorted sequence of the episodes
        """
        return sorted(episodes, key=cls.episode_sort_key, reverse=reverse)