app/browse/innertube.py

   1 # functions that deal with parsing data from youtube's internal API ("innertube")
   2
   3 from ..common.common import mkthumbs, log_unknown_card, G
   4
   5 class Select:
   6     """ |Select('foo') returns the first foo in list, |Select(all='foo') returns all foos. """
   7     def __init__(self, key=None, *, all=None):
   8         self.key = key or all
   9         self.all = all
  10     def __ror__(self, other):
  11         try: items = [ other[self.key] for other in other if self.key in other.keys() ]
  12         except: items = []
  13         return items if self.all else items|G(0)
  14 class A:
  15     """ apply """
  16     def __init__(self, f, *args):
  17         self.f = f
  18         self.args = args
  19     def __ror__(self, other):
  20         return self.f(other, *self.args)
  21     class _Int:
  22         def __ror__(self, other):
  23             try:    return int(''.join(filter(str.isdigit, other)))
  24             except: return None
  25     int = _Int()
  26
  27
  28 def prepare_searchresults(yt_results):
  29     contents = ( # from continuation token
  30         yt_results
  31         |G('onResponseReceivedCommands')
  32         |Select('appendContinuationItemsAction')
  33         |G('continuationItems')
  34     ) or ( # from page 1
  35         yt_results
  36         |G('contents')
  37         |G('twoColumnSearchResultsRenderer')
  38         |G('primaryContents')
  39         |G('sectionListRenderer')
  40         |G('contents')
  41     )
  42     items = contents|Select('itemSectionRenderer')|G('contents')
  43     items, extra = parse_result_items(items)
  44     more = contents|Select("continuationItemRenderer")|G("continuationEndpoint")|G("continuationCommand")|G("token")
  45     estimatedResults = yt_results|G("estimatedResults")
  46
  47     return items, extra, more
  48
  49 def prepare_channel(response, channel_id, channel_name):
  50     meta1 = response|G('metadata')|G('channelMetadataRenderer')
  51     meta2 = response|G('microformat')|G('microformatDataRenderer')
  52     title = meta1|G('title') or meta2|G('title') or channel_name
  53     descr = meta1|G('description') or meta2|G('description') # meta2.description is capped at 160chars
  54     thumb = mkthumbs((meta2|G('thumbnail') or meta1|G('avatar'))|G('thumbnails') or {}) # .avatar ~ 900px
  55
  56     contents = (
  57         response|G('continuationContents') or
  58         response|G('onResponseReceivedActions')
  59     )
  60     if not contents: # overran end of list
  61         return title, descr, thumb, [], False
  62
  63     unparsed = contents|G('gridContinuation')|G('items') or \
  64             contents|G('sectionListContinuation')|G('contents') or \
  65             contents|G('richGridContinuation')|G('contents') or \
  66             contents|Select('appendContinuationItemsAction')|G('continuationItems') or \
  67             contents|G(-1)|G('reloadContinuationItemsCommand')|G('continuationItems') or []
  68     items, extra = parse_channel_items(unparsed, channel_id, title)
  69
  70     more = ( # videos, livestreams
  71         unparsed
  72         |Select('continuationItemRenderer')
  73         |G('continuationEndpoint')
  74         |G('continuationCommand')
  75         |G('token')
  76     ) or ( # playlists, search
  77         contents
  78         |G('gridContinuation', 'sectionListContinuation')
  79         |G('continuations')
  80         |Select('nextContinuationData')
  81         |G('continuation')
  82     )
  83
  84     return title, descr, thumb, items, more
  85
  86 def prepare_playlist(result):
  87     contents = result['continuationContents']
  88     unparsed = contents['playlistVideoListContinuation'].get('contents',[])
  89     more = (
  90         contents
  91         |G('playlistVideoListContinuation')
  92         |G('continuations')
  93         |Select('nextContinuationData')
  94         |G('continuation')
  95     )
  96
  97     meta = result|G('sidebar')|G('playlistSidebarRenderer')|G('items')
  98     meta1 = meta|Select('playlistSidebarPrimaryInfoRenderer')
  99     meta2 = meta|Select('playlistSidebarSecondaryInfoRenderer') \
 100         |G('videoOwner')|G('videoOwnerRenderer')
 101     title = meta1|G('title')|G.text
 102     author = meta2|G('title')|G.text
 103     channel_id = meta2|G('navigationEndpoint')|G('browseEndpoint')|G('browseId')
 104
 105     return title, author, channel_id, list(filter(None, map(parse_playlist, unparsed))), more
 106
 107 def age(s):
 108     if s is None: # missing from autogen'd music, some livestreams
 109         return None
 110     # Some livestreams have "Streamed 7 hours ago"
 111     s = s.replace("Streamed ","")
 112     # Now, everything should be in the form "1 year ago"
 113     value, unit, _ = s.split(" ")
 114     suffix = dict(
 115         minute='min',
 116         minutes='min',
 117     ).get(unit, unit[0]) # first letter otherwise (e.g. year(s) => y)
 118
 119     return f"{value}{suffix}"
 120
 121 def parse_result_items(items):
 122     # TODO: use .get() for most non-essential attributes
 123     """
 124     parses youtube search response into an easier to use format.
 125     """
 126     results = []
 127     extras = []
 128     for item in items:
 129         key = next(iter(item.keys()), None)
 130         content = item[key]
 131         if key in ['videoRenderer', 'reelItemRenderer']:
 132             results.append({'type': 'VIDEO', 'content': {
 133                 'video_id': content['videoId'],
 134                 'title': content|G('title')|G.text or content|G('headline')|G.text,
 135                 'author': content|G('longBylineText','shortBylineText')|G.text,
 136                 'channel_id': content|G('ownerText')|G('runs')|G(0) \
 137                     |G('navigationEndpoint')|G('browseEndpoint')|G('browseId'),
 138                 'length': content|G('lengthText')|G.text, # "44:07", "1:41:50"
 139                 'views': content|G('viewCountText')|G.text|A.int or 0, # "1,234 {views|watching}", absent on 0 views
 140                 'published': content|G('publishedTimeText')|G('simpleText')|A(age),
 141                 'live': content|G('badges')|Select('metadataBadgeRenderer')|G('style')=='BADGE_STYLE_TYPE_LIVE_NOW',
 142             }})
 143         elif key in ['playlistRenderer', 'radioRenderer', 'showRenderer']: # radio == "Mix" playlist, show == normal playlist, specially displayed
 144             results.append({'type': 'PLAYLIST', 'content': {
 145                 'playlist_id': content['navigationEndpoint']|G('watchEndpoint')|G('playlistId'),
 146                 'video_id': content['navigationEndpoint']|G('watchEndpoint')|G('videoId'),
 147                 'title': content['title']|G.text,
 148                 'author': content|G('longBylineText','shortBylineText')|G.text,
 149                 'channel_id': content|G('longBylineText','shortBylineText')|G('runs')|G(0) \
 150                     |G('navigationEndpoint')|G('browseEndpoint')|G('browseId'),
 151                 'n_videos': content|G('videoCount')|A.int or \
 152                     content|G('videoCountShortText','videoCountText')|G.text, # "Mix" playlists
 153             }})
 154         elif key == 'channelRenderer':
 155             results.append({'type': 'CHANNEL', 'content': {
 156                 'channel_id': content['channelId'],
 157                 'title': content['title']|G.text,
 158                 'icons': content['thumbnail']['thumbnails']|A(mkthumbs),
 159                 'subscribers': content|G('subscriberCountText')|G('simpleText'), # "2.47K subscribers"
 160             }})
 161         elif key == 'shelfRenderer':
 162             subkey = next(iter(content['content'].keys()), None) #verticalListRenderer/horizontalMovieListRenderer
 163             r, e = parse_result_items(content['content'][subkey]['items'])
 164             results.extend(r)
 165             extras.extend(e)
 166         elif key in ["reelShelfRenderer"]:
 167             r, e = parse_result_items(content['items'])
 168             results.extend(r)
 169             extras.extend(e)
 170         elif key in ['movieRenderer', 'gridMovieRenderer']: # movies to buy/rent
 171             pass # gMR.{videoId,title.runs[].text,lengthText.simpleText}
 172         elif key in ['carouselAdRenderer','searchPyvRenderer','promotedSparklesTextSearchRenderer',
 173                 'promotedSparklesWebRenderer','compactPromotedItemRenderer', 'adSlotRenderer']: # haha, no.
 174             pass
 175         elif key == 'horizontalCardListRenderer':
 176             # suggested searches: .cards[].searchRefinementCardRenderer.query.runs[].text
 177             pass
 178         elif key == 'emergencyOneboxRenderer': # suicide prevention hotline
 179             pass
 180         elif key in ['clarificationRenderer', 'infoPanelContainerRenderer']: # COVID-19/conspiracy theory infos
 181             pass
 182         elif key == 'webAnswerRenderer': # "Result from the web"
 183             pass
 184         elif key == 'infoPanelContentRenderer': # "These results may be new or changing quickly"
 185             pass
 186         elif key == 'hashtagTileRenderer': # link to '/hashtag/<search_query>'
 187             pass
 188         elif key in ['didYouMeanRenderer', 'showingResultsForRenderer', 'includingResultsForRenderer']:
 189             extras.append({
 190                 'type': 'spelling',
 191                 'query': content['correctedQueryEndpoint']['searchEndpoint']['query'], # non-misspelled query
 192                 'autocorrected': key in ['showingResultsForRenderer', 'includingResultsForRenderer'],
 193             })
 194         elif key == 'messageRenderer': # "No more results"
 195             extras.append({
 196                 'type': 'message',
 197                 'message': content|G('title','text')|G.text,
 198             })
 199         elif key == 'backgroundPromoRenderer': # e.g. "no results"
 200             extras.append({
 201                 'type': content['icon']['iconType'],
 202                 'message': content['title']|G.text,
 203             })
 204         else:
 205             log_unknown_card(item)
 206     return results, extras
 207
 208 def parse_channel_items(items, channel_id, author):
 209     result = []
 210     extra = []
 211     for item in items:
 212         key = next(iter(item.keys()), None)
 213         content = item[key]
 214         if key in ["gridVideoRenderer", "videoRenderer", "videoCardRenderer", 'reelItemRenderer']: # reel==youtube-shorts
 215             # only videoCardRenderer (topic channels) has author and channel, others fall back to supplied ones.
 216             result.append({'type': 'VIDEO', 'content': {
 217                 'video_id': content['videoId'],
 218                 'title': content|G('title')|G.text or content|G('headline')|G.text,
 219                 'author': content|G('bylineText')|G.text or author,
 220                 'channel_id': (content|G('bylineText')|G('runs')
 221                     |Select('navigationEndpoint')
 222                     |G('browseEndpoint')|G('browseId') or channel_id),
 223                 'length': (content|G('lengthText')|G.text or # topic channel
 224                     content|G('thumbnailOverlays')
 225                     |Select('thumbnailOverlayTimeStatusRenderer')
 226                     |G('text')|G.text),
 227                 # topic channel: .metadataText.simpleText = "22M views \u00b7 2 months ago"
 228                 'views': content|G('viewCountText')|G.text|A.int,
 229                 'published': content|G('publishedTimeText')|G.text|A(age),
 230             }})
 231         elif key in ["gridPlaylistRenderer", "playlistRenderer", "gridRadioRenderer"]:
 232             result.append({'type': 'PLAYLIST', 'content': {
 233                 'playlist_id': content|G('navigationEndpoint')|G('watchEndpoint')|G('playlistId'),
 234                 'video_id': content|G('navigationEndpoint')|G('watchEndpoint')|G('videoId'),
 235                 'title': content|G('title')|G.text,
 236                 'author': author, # Note: gridRadioRenderer is by 'Youtube' without channel_id, ignoring that.
 237                 'channel_id': channel_id,
 238                 'n_videos': (content|G('videoCount')|A.int or # playlistRenderer
 239                     content|G('videoCountShortText','videoCountText')|G.text|A.int) # grid
 240             }})
 241         elif key == "showRenderer":
 242             result.append({'type': 'PLAYLIST', 'content': {
 243                 'playlist_id': content['navigationEndpoint']['watchEndpoint']['playlistId'],
 244                 'video_id': content['navigationEndpoint']['watchEndpoint']['videoId'],
 245                 'title': content['title']['simpleText'],
 246                 'author': author,
 247                 'channel_id': channel_id,
 248                 'n_videos': None,
 249             }})
 250         elif key in ["gridShowRenderer"]:
 251             result.append({'type': 'PLAYLIST', 'content': {
 252                 'playlist_id': (content|G('navigationEndpoint')
 253                     |G('browseEndpoint')|G('browseId'))[2:],
 254                 #^: playlistId prefixed with 'VL', which must be removed
 255                 'video_id': None,
 256                 'title': content|G('title')|G.text,
 257                 'author': author,
 258                 'channel_id': channel_id,
 259                 'n_videos': content|G('thumbnailOverlays')|G(0)
 260                     |G('thumbnailOverlayBottomPanelRenderer')|G('text')|G.text,
 261             }})
 262         elif key in ["itemSectionRenderer", "gridRenderer", "horizontalCardListRenderer", "horizontalListRenderer"]:
 263                 newkey = {
 264                     "itemSectionRenderer": 'contents',
 265                     "gridRenderer": 'items',
 266                     "horizontalCardListRenderer": 'cards',
 267                     "horizontalListRenderer": 'items',
 268                 }.get(key)
 269                 r, e = parse_channel_items(content[newkey], channel_id, author)
 270                 result.extend(r)
 271                 extra.extend(e)
 272         elif key in ["shelfRenderer", "richItemRenderer"]:
 273                 r, e = parse_channel_items([content['content']], channel_id, author)
 274                 result.extend(r)
 275                 extra.extend(e)
 276         elif key in ["reelShelfRenderer"]:
 277             r, e = parse_channel_items(content['items'], channel_id, author)
 278             result.extend(r)
 279             extra.extend(e)
 280         elif key == "messageRenderer":
 281             # e.g. {'messageRenderer': {'text': {'runs': [{'text': 'This channel has no playlists.'}]}}}
 282             pass
 283         elif key == "gameCardRenderer":
 284             pass
 285         elif key == "gridChannelRenderer":
 286             pass # don't care; related channels, e.g. on UCMsgXPD3wzzt8RxHJmXH7hQ
 287         elif key == 'continuationItemRenderer': # handled in parent function
 288             pass
 289         else:
 290             log_unknown_card(item)
 291
 292     return result, extra
 293
 294 def parse_playlist(item):
 295     key = next(iter(item.keys()), None)
 296     content = item[key]
 297     if key == "playlistVideoRenderer":
 298         if not content.get('isPlayable', False):
 299             return None # private or deleted video
 300
 301         return {'type': 'VIDEO', 'content': {
 302             'video_id': content['videoId'],
 303             'title': (content['title'].get('simpleText') or # playable videos
 304                 content['title'].get('runs',[{}])[0].get('text')), # "[Private video]"
 305             'playlist_id': content['navigationEndpoint']['watchEndpoint']['playlistId'],
 306             'index': content['navigationEndpoint']['watchEndpoint'].get('index',0), #or int(content['index']['simpleText']) (absent on course intros; e.g. PL96C35uN7xGJu6skU4TBYrIWxggkZBrF5)
 307             # rest is missing from unplayable videos:
 308             'author': content.get('shortBylineText',{}).get('runs',[{}])[0].get('text'),
 309             'channel_id':content.get('shortBylineText',{}).get('runs',[{}])[0].get('navigationEndpoint',{}).get('browseEndpoint',{}).get('browseId'),
 310             'length': (content.get("lengthText",{}).get("simpleText") or # "8:51"
 311                 int(content.get("lengthSeconds", 0))), # "531"
 312             'starttime': content['navigationEndpoint']['watchEndpoint'].get('startTimeSeconds'),
 313         }}
 314     else:
 315         raise Exception(item) # XXX TODO