def add_youtube_feeds(cls): channels = Channel.all().fetch(100) for channel in channels: for keyword in channel.keywords: yt_service = gdata.youtube.service.YouTubeService() gdata.alt.appengine.run_on_appengine(yt_service) uri = Programming.YOUTUBE_FEED % ('most_popular', keyword) feed = yt_service.GetRecentlyFeaturedVideoFeed() medias = Media.add_from_entry(feed.entry) for media in medias: Program.add_program(channel, media)
def add_program(self, channel, media, time): programming = memcache.get('programming') or {} program = Program.add_program(channel, media, time=time, min_time=time, max_time=(time + datetime.timedelta(media.duration + 50))) logging.info(program) if program: if not programming.get(channel.id, None): programming[channel.id] = [] programming.get(channel.id).append(program.toJson(fetch_channel=False, fetch_media=True, media_desc=False, pub_desc=False)) memcache.set('programming', programming)
def set_user_channel_programs(self, key, channel, medias, time=None, reset=False): from model import Program programs = [] programs_json = [] channel_json = {} if not reset and memcache.get(channel.id): channel_json = memcache.get(channel.id) or {} programs_json = channel_json.get('programs', []) else: channel_json['channel'] = channel.toJson() next_time = time or datetime.datetime.now() if len(programs_json): next_time = iso8601.parse_date(programs_json[-1]['time']).replace(tzinfo=None) + \ datetime.timedelta(seconds=programs_json[-1]['media']['duration']) for media in medias: program = Program.add_program(channel, media, time=next_time) if program: programs.append(program) programs_json.append(program.toJson(fetch_channel=False, fetch_media=True, media_desc=False, pub_desc=False)) next_time = next_time + datetime.timedelta(seconds=media.duration) if len(pickle.dumps(programs_json)) > 1000000: # We can only fit 1mb into memcache break deferred.defer(Programming.fetch_related_tweets, medias, _name='twitter-' + channel.id + '-' + str(uuid.uuid1()), _queue='twitter') user_obj = memcache.get(key) or {} user_channels = (user_obj.get('channels') or []) if user_obj else [] if not channel.id in user_channels: user_channels.append(channel.id) user_obj['channels'] = user_channels memcache.set(key, user_obj) channel_json['programs'] = programs_json memcache.set(channel.id, channel_json) return programs
def set_programming(cls, channel_id, duration=3600, schedule_next=False, fetch_twitter=True, queue='programming', target=None, kickoff=False): import broadcast import constants from model import Channel from model import Program # Stored programming programming = memcache.get('programming') or {} onlineUsers = memcache.get('web_channels') or {} logging.info('programming: ' + channel_id) next_programs = Programming.next_programs(programming.get(channel_id, []), duration, prelude=300) gap = Programming.gap(programming.get(channel_id, []), duration) logging.info('GAP: ' + str(gap)) if programming.get(channel_id) and len(programming[channel_id]) and \ programming[channel_id][0]['media'].get('live') == True: logging.info('live tweets') # Update tweets for live events media = Media.get_by_key_name(programming[channel_id][0]['media']['id']) deferred.defer(Programming.fetch_related_tweets, [], _name='twitter-' + channel_id + '-' + str(uuid.uuid1()), _queue='twitter') programs = [] if not programming.get(channel_id) or gap > 60: logging.info('PROGRAOMMING') channel = Channel.get_by_key_name(channel_id) #channel.update_next_time() viewers = (memcache.get('channel_viewers') or {}).get(str(channel_id), []) cols = channel.get_collections() all_medias = [] backup_medias = [] limit = 100 for col in cols: medias = [] filtered_medias = [] offset = 0 while offset <= 400: medias = col.get_medias(limit=limit, offset=offset) logging.info('fetched medias: ' + str(len(medias))) if not len(medias): break backup_medias += medias # Dont reprogram anytihng already scheduled filtered_medias = Programming.no_reprogram(programming.get(channel_id, []), medias) # Don't repeat the same program within two hour cutoff = 7200 if col.lifespan else 43200 filtered_medias = [c for c in filtered_medias if not c.last_programmed or (datetime.datetime.now() - c.last_programmed).seconds > cutoff] # At most, 30% of the audience has already "witnessed" this program # filtered_medias = [m for m in filtered_medias if not len(viewers) or # float(len(Programming.have_seen(m, viewers)))/len(viewers) < .3] all_medias += filtered_medias logging.info('all medias: ' + str(len(all_medias))) offset += limit all_medias = backup_medias if not len(all_medias) else all_medias # Don't repeat already programmed # all_medias = Programming.no_reprogram(next_programs, all_medias) # StorySort algorithm # all_medias = Programming.story_sort(all_medias) # Only one publisher per story all_medias = Programming.unique_publishers(all_medias) # Grab "duration" seconds of programming all_medias = Programming.timed_subset(all_medias, duration) if fetch_twitter: # Find related twitter posts deferred.defer(Programming.fetch_related_tweets, all_medias, _name='twitter-' + channel.name.replace(' ', '') + '-' + str(uuid.uuid1()), _queue='twitter', _countdown=30) # Truncate old programs programming[channel_id] = Programming.cutoff_programs(programming.get(channel_id), 300) for media in all_medias: program = Program.add_program(channel, media, min_time=datetime.datetime.now(), max_time=(datetime.datetime.now() + datetime.timedelta(seconds=duration))) logging.info(program) if program: if not programming.get(channel_id, None): programming[channel_id] = [] programming.get(channel_id).append(program.toJson(fetch_channel=False, fetch_media=True, media_desc=False, pub_desc=False)) programs.append(program) logging.info('ADDING: ' + media.name + ' at: ' + program.time.isoformat()) if len(pickle.dumps(programming)) > 1000000: # We can only fit 1mb into memcache break if len(programs): broadcast.broadcastNewPrograms(channel, programs) memcache.set('programming', programming) channels = memcache.get('channels') or [] updated = False for i,c in enumerate(channels): if c['id'] == channel_id: channels[i] = channel.toJson(get_programming=False) updated = True if not updated: channels.append(channel.toJson(get_programming=False)) memcache.set('channels', channels) # Schedule our next programming selection if schedule_next and (not constants.SLEEP_PROGRAMMING or (constants.SLEEP_PROGRAMMING and (kickoff or len(onlineUsers.keys())))): logging.info('NUMBER OF PROGRAMS: ' + str(len(programs))) if len(programs) > 1: next_gen = (programs[-2].time - datetime.datetime.now()).seconds / 2 elif len(programs) == 1: next_gen = programs[0].media.duration / 2 else: next_gen = 60 next_gen = min(next_gen, reduce(lambda x, y: x + y, [p.media.duration for p in programs], 0) \ if len(programs) else 10) next_gen = min(next_gen, duration / 2) logging.info('COUNTDOWN FOR ' + channel_id + ': ' + str(next_gen)) deferred.defer(Programming.set_programming, channel_id, fetch_twitter=fetch_twitter, _name=channel_id + '-' + str(uuid.uuid1()), _countdown=next_gen, _queue=queue) return programs
def schedule_youtube_channel(name, user_id=None, token=None, channel_id=None, yt_channel_id=None, yt_playlist_id=None): import broadcast import constants import datetime import programming from model import Media from model import Channel from model import User from model import common from model import Program user = None if user_id: user = User.get_by_key_name(user_id) channel = Channel(key_name=(user_id or token) + '-' + (yt_channel_id or yt_playlist_id), name=name, privacy=constants.Privacy.FRIENDS, online=True, user=user) medias = [] youtube3 = common.get_youtube3_service() next_page_token = '' search_response = {} if yt_channel_id: if yt_channel_id.startswith('HC'): channel_response = youtube3.channels().list( id=yt_channel_id, part='topicDetails', maxResults=1 ).execute() if len(channel_response.get('items', [])): topic_id = channel_response.get('items')[0]['topicDetails']['topicIds'][0] search_response = youtube3.search().list( topicId=topic_id, order='date', part='id,snippet', maxResults=10, fields='items', type='video' ).execute() else: search_response = youtube3.search().list( channelId=yt_channel_id, part='id,snippet', order='date', maxResults=10, type='video' ).execute() elif yt_playlist_id: search_response = youtube3.playlistItems().list( playlistId=yt_playlist_id, part='id,snippet', maxResults=10, fields='items' ).execute() search_ids = '' for item in search_response.get('items', [])[1:]: if item['kind'] == 'youtube#searchResult': search_ids += item['id']['videoId'] + ',' elif item['kind'] == 'youtube#playlistItem': search_ids += item['snippet']['resourceId']['videoId'] + ',' videos_response = youtube3.videos().list( id=search_ids, part="id,snippet,topicDetails,contentDetails,statistics" ).execute() for item in videos_response.get("items", []): medias = Media.add_from_snippet([item], approve=True) programs = [] if user_id: programs = programming.Programming.set_user_channel_programs(user_id, channel, medias) else: next_time = datetime.datetime.now() for media in medias: program = Program.add_program(channel, media, time=next_time) if program: programs.append(program) next_time = next_time + datetime.timedelta(seconds=media.duration) broadcast.broadcastNewPrograms(channel, programs, token=token)