app/common/innertube.py

   1 # functions that deal with parsing data from youtube's internal API ("innertube")
   2
   3 from urllib.parse import parse_qs, urlparse
   4
   5 def findall(obj, key):
   6     """
   7     given a list of dicts, where one dict contains a given key, return said key.
   8     """
   9     return [ obj[key] for obj in obj if key in obj.keys() ]
  10 def listget(obj, index, fallback=None):
  11     return next(iter(obj[index:]), fallback)
  12 flatten = lambda l: [item for sublist in l for item in sublist] # https://stackoverflow.com/a/952952
  13 first = lambda l: next(iter(l),{})
  14 listfind = lambda obj,key: first(findall(obj,key))
  15
  16 def prepare_searchresults(yt_results):
  17     contents = listfind(yt_results, 'response') \
  18         .get('contents',{})\
  19         .get('twoColumnSearchResultsRenderer',{})\
  20         .get('primaryContents',{})\
  21         .get('sectionListRenderer',{})\
  22         .get('contents',[])
  23     contents = flatten([c.get('contents',[]) for c in findall(contents, 'itemSectionRenderer')])
  24
  25     return parse_result_items(contents)
  26
  27 def prepare_infocards(metadata):
  28     cards = metadata.get('cards',{}).get('cardCollectionRenderer',{}).get('cards',[])
  29     return [parse_infocard(card) for card in cards if card is not None]
  30
  31 def prepare_endcards(metadata):
  32     endsc = metadata.get('endscreen',{}).get('endscreenRenderer',{}).get('elements',[])
  33     return [parse_endcard(card) for card in endsc if card is not None]
  34
  35 def mkthumbs(thumbs):
  36     output = {str(e['height']): e['url'] for e in thumbs}
  37     largest=next(iter(sorted(output.keys(),reverse=True,key=int)),None)
  38     return {**output, 'largest': largest}
  39
  40 def clean_url(url):
  41     # externals URLs are redirected through youtube.com/redirect, but we
  42     # may encounter internal URLs, too
  43     return parse_qs(urlparse(url).query).get('q',[url])[0]
  44
  45 def toInt(s, fallback=0):
  46     if s is None:
  47         return fallback
  48     try:
  49         return int(''.join(filter(str.isdigit, s)))
  50     except ValueError:
  51         return fallback
  52
  53 # Remove left-/rightmost word from string:
  54 delL = lambda s: s.partition(' ')[2]
  55
  56 def age(s):
  57     if s is None: # missing from autogen'd music, some livestreams
  58         return None
  59     # Some livestreams have "Streamed 7 hours ago"
  60     s = s.replace("Streamed ","")
  61     # Now, everything should be in the form "1 year ago"
  62     value, unit, _ = s.split(" ")
  63     suffix = dict(
  64         month='mn',
  65         months='mn',
  66     ).get(unit, unit[0]) # first letter otherwise (e.g. year(s) => y)
  67
  68     return f"{value}{suffix}"
  69
  70 def log_unknown_card(data):
  71     import json
  72     try:
  73         from flask import request
  74         source = request.url
  75     except: source = "unknown"
  76     with open("/tmp/innertube.err", "a") as f:
  77         f.write(f"\n/***** {source} *****/\n")
  78         json.dump(data, f, indent=2)
  79
  80 def parse_result_items(items):
  81     # TODO: use .get() for most non-essential attributes
  82     """
  83     parses youtube search response into an easier to use format.
  84     """
  85     results = []
  86     for item in items:
  87         key = next(iter(item.keys()), None)
  88         content = item[key]
  89         if key == 'videoRenderer':
  90             is_live = listfind(content.get('badges',[]), 'metadataBadgeRenderer').get('style') == 'BADGE_STYLE_TYPE_LIVE_NOW'
  91             results.append({'type': 'VIDEO', 'content': {
  92                 'video_id': content['videoId'],
  93                 'title': content['title']['runs'][0]['text'],
  94                 'author': content['longBylineText']['runs'][0]['text'] or \
  95                           content['shortBylineText']['runs'][0]['text'],
  96                 'channel_id': content['ownerText']['runs'][0] \
  97                     ['navigationEndpoint']['browseEndpoint']['browseId'],
  98                 'length': content.get('lengthText',{}).get('simpleText') \
  99                     if not is_live else 'LIVE', # "44:07", "1:41:50"
 100                 'views': toInt(content.get('viewCountText',{}).get('simpleText') or # "123,456 views"
 101                     listget(content.get('viewCountText',{}).get('runs'),0,{}).get('text')), # "1,234 watching"
 102                 'published': age(content.get('publishedTimeText',{}).get('simpleText')),
 103             }})
 104         elif key == 'playlistRenderer':
 105             results.append({'type': 'PLAYLIST', 'content': {
 106                 'playlist_id': content['navigationEndpoint']['watchEndpoint']['playlistId'],
 107                 'video_id': content['navigationEndpoint']['watchEndpoint']['videoId'],
 108                 'title': content['title']['simpleText'],
 109                 'author': content['longBylineText']['runs'][0]['text'] or
 110                      content['shortBylineText']['runs'][0]['text'],
 111                 'channel_id': content['longBylineText']['runs'][0]['navigationEndpoint']['browseEndpoint']['browseId'], # OR .shortBylineText
 112                 'n_videos': toInt(content['videoCount']),
 113             }})
 114         elif key == 'radioRenderer': # "Mix" playlists
 115             results.append({'type': 'PLAYLIST', 'content': {
 116                 'playlist_id': content['playlistId'],
 117                 'video_id': content['navigationEndpoint']['watchEndpoint']['videoId'],
 118                 'title': content['title']['simpleText'],
 119                 'author': content['longBylineText']['simpleText'] or \
 120                     content['shortBylineText']['simpleText'] , # always "YouTube"
 121                 'channel_id': None,
 122                 'n_videos': content['videoCountShortText']['runs'][0]['text'] or \
 123                     content['videoCountText']['runs'][0]['text'],
 124                     # videoCountShortText: "50+"; videoCountText: "50+ videos"
 125             }})
 126         elif key == 'channelRenderer':
 127             results.append({'type': 'CHANNEL', 'content': {
 128                 'channel_id': content['channelId'],
 129                 'title': content['title']['simpleText'],
 130                 'icons': mkthumbs(content['thumbnail']['thumbnails']),
 131                 'subscribers': content['subscriberCountText']['simpleText'], # "2.47K subscribers"
 132             }})
 133         elif key == 'shelfRenderer':
 134             results.extend([item for item in
 135                 parse_result_items(content['content']['verticalListRenderer']['items'])
 136             ])
 137         elif key == 'movieRenderer': # movies to buy/rent
 138             pass
 139         elif key == 'carouselAdRenderer': # haha, no.
 140             pass
 141         elif key == 'horizontalCardListRenderer':
 142             # suggested searches: .cards[].searchRefinementCardRenderer.query.runs[].text
 143             pass
 144         else:
 145             log_unknown_card(item)
 146     return results
 147
 148 def parse_infocard(card):
 149     """
 150     parses a single infocard into a format that's easier to handle.
 151     """
 152     card = card['cardRenderer']
 153     ctype = list(card['content'].keys())[0]
 154     content = card['content'][ctype]
 155     if ctype == "pollRenderer":
 156         return {'type': "POLL", 'content': {
 157             'question': content['question']['simpleText'],
 158             'answers': [(a['text']['simpleText'],a['numVotes']) \
 159                 for a in content['choices']],
 160         }}
 161     elif ctype == "videoInfoCardContentRenderer":
 162         is_live = content.get('badge',{}).get('liveBadgeRenderer') is not None
 163         return {'type': "VIDEO", 'content': {
 164             'video_id': content['action']['watchEndpoint']['videoId'],
 165             'title': content['videoTitle']['simpleText'],
 166             'author': delL(content['channelName']['simpleText']),
 167             'length': content.get('lengthString',{}).get('simpleText') \
 168                 if not is_live else "LIVE", # "23:03"
 169             'views': toInt(content.get('viewCountText',{}).get('simpleText')),
 170                 # XXX: views sometimes "Starts: July 31, 2020 at 1:30 PM"
 171         }}
 172     elif ctype == "playlistInfoCardContentRenderer":
 173         return {'type': "PLAYLIST", 'content': {
 174             'playlist_id': content['action']['watchEndpoint']['playlistId'],
 175             'video_id': content['action']['watchEndpoint']['videoId'],
 176             'title': content['playlistTitle']['simpleText'],
 177             'author': delL(content['channelName']['simpleText']),
 178             'n_videos': toInt(content['playlistVideoCount']['simpleText']),
 179         }}
 180     elif ctype == "simpleCardContentRenderer" and \
 181             'urlEndpoint' in content['command']:
 182         return {'type': "WEBSITE", 'content': {
 183             'url': clean_url(content['command']['urlEndpoint']['url']),
 184             'domain': content['displayDomain']['simpleText'],
 185             'title': content['title']['simpleText'],
 186             # XXX: no thumbnails for infocards
 187         }}
 188     elif ctype == "collaboratorInfoCardContentRenderer":
 189         return {'type': "CHANNEL", 'content': {
 190             'channel_id': content['endpoint']['browseEndpoint']['browseId'],
 191             'title': content['channelName']['simpleText'],
 192             'icons': mkthumbs(content['channelAvatar']['thumbnails']),
 193             'subscribers': content.get('subscriberCountText',{}).get('simpleText',''), # "545K subscribers"
 194         }}
 195     else:
 196         log_unknown_card(card)
 197         return None
 198
 199 def parse_endcard(card):
 200     """
 201     parses a single endcard into a format that's easier to handle.
 202     """
 203     card = card.get('endscreenElementRenderer', card) #only sometimes nested
 204     ctype = card['style']
 205     if ctype == "CHANNEL":
 206         return {'type': ctype, 'content': {
 207             'channel_id': card['endpoint']['browseEndpoint']['browseId'],
 208             'title': card['title']['simpleText'],
 209             'icons': mkthumbs(card['image']['thumbnails']),
 210         }}
 211     elif ctype == "VIDEO":
 212         return {'type': ctype, 'content': {
 213             'video_id': card['endpoint']['watchEndpoint']['videoId'],
 214             'title': card['title']['simpleText'],
 215             'length': card['videoDuration']['simpleText'],  # '12:21'
 216             'views': toInt(card['metadata']['simpleText']),
 217             # XXX: no channel name
 218         }}
 219     elif ctype == "PLAYLIST":
 220         return {'type': ctype, 'content': {
 221             'playlist_id': card['endpoint']['watchEndpoint']['playlistId'],
 222             'video_id': card['endpoint']['watchEndpoint']['videoId'],
 223             'title': card['title']['simpleText'],
 224             'author': delL(card['metadata']['simpleText']),
 225             'n_videos': toInt(card['playlistLength']['simpleText']),
 226         }}
 227     elif ctype == "WEBSITE" or ctype == "CREATOR_MERCHANDISE":
 228         url = clean_url(card['endpoint']['urlEndpoint']['url'])
 229         return {'type': "WEBSITE", 'content': {
 230             'url': url,
 231             'domain': urlparse(url).netloc,
 232             'title': card['title']['simpleText'],
 233             'icons': mkthumbs(card['image']['thumbnails']),
 234         }}
 235     else:
 236         log_unknown_card(card)
 237         return None