]>
git.gir.st - subscriptionfeed.git/blob - app/common/innertube.py
1 # functions that deal with parsing data from youtube's internal API ("innertube")
3 from urllib
. parse
import parse_qs
, urlparse
8 null-coalescing version of dict.get() that also works on lists.
10 the | operator is overloaded to achieve similar looking code to jq(1) filters.
11 the first found key is used: dict(foo=1)|G('bar','foo') returns 1.
13 def __init__ ( self
, * keys
):
15 def __ror__ ( self
, other
):
17 try : return other
[ key
]
21 """ parses youtube's .runs[].text and .simpleText variants """
22 def __ror__ ( self
, other
): # Note: only returning runs[0], not concat'ing all!
23 return other|
G ( 'simpleText' ) or other|
G ( 'runs' ) |
G ( 0 ) |
G ( 'text' )
26 """ |Select('foo') returns the first foo in list, |Select(all='foo') returns all foos. """
27 def __init__ ( self
, key
= None , *, all
= None ):
30 def __ror__ ( self
, other
):
31 try : items
= [ other
[ self
. key
] for other
in other
if self
. key
in other
. keys () ]
33 return items
if self
. all
else items|
G ( 0 )
36 def __init__ ( self
, f
, * args
):
39 def __ror__ ( self
, other
):
40 return self
. f ( other
, * self
. args
)
42 def __ror__ ( self
, other
):
43 try : return int ( '' . join ( filter ( str . isdigit
, other
)))
48 def prepare_searchresults ( yt_results
):
49 contents
= ( # from continuation token
51 |
G ( 'onResponseReceivedCommands' )
52 |
Select ( 'appendContinuationItemsAction' )
53 |
G ( 'continuationItems' )
57 |
G ( 'twoColumnSearchResultsRenderer' )
59 |
G ( 'sectionListRenderer' )
62 items
= contents|
Select ( 'itemSectionRenderer' ) |
G ( 'contents' )
63 items
, extra
= parse_result_items ( items
)
64 more
= contents|
Select ( "continuationItemRenderer" ) |
G ( "continuationEndpoint" ) |
G ( "continuationCommand" ) |
G ( "token" )
65 estimatedResults
= yt_results|
G ( "estimatedResults" )
67 return items
, extra
, more
69 def prepare_infocards ( metadata
):
70 cards
= metadata
. get ( 'cards' ,{}). get ( 'cardCollectionRenderer' ,{}). get ( 'cards' ,[])
71 return list ( filter ( None , map ( parse_infocard
, cards
)))
73 def prepare_endcards ( metadata
):
74 endsc
= metadata
. get ( 'endscreen' ,{}). get ( 'endscreenRenderer' ,{}). get ( 'elements' ,[])
75 return list ( filter ( None , map ( parse_endcard
, endsc
)))
77 def prepare_channel ( response
, channel_id
):
78 meta1
= response|
G ( 'metadata' ) |
G ( 'channelMetadataRenderer' )
79 meta2
= response|
G ( 'microformat' ) |
G ( 'microformatDataRenderer' )
80 title
= meta1|
G ( 'title' ) or meta2|
G ( 'title' )
81 descr
= meta1|
G ( 'description' ) or meta2|
G ( 'description' ) # meta2.description is capped at 160chars
82 thumb
= mkthumbs (( meta2|
G ( 'thumbnail' ) or meta1|
G ( 'avatar' )) |
G ( 'thumbnails' ) or {}) # .avatar ~ 900px
84 contents
= response|
G ( 'continuationContents' )
85 if not contents
: # overran end of list
86 return title
, descr
, thumb
, [], False
88 unparsed
= contents|
G ( 'gridContinuation' ) |
G ( 'items' ) or \
89 contents|
G ( 'sectionListContinuation' ) |
G ( 'contents' ) or []
90 items
, extra
= parse_channel_items ( unparsed
, channel_id
, title
)
93 |
G ( 'gridContinuation' , 'sectionListContinuation' )
95 |
Select ( 'nextContinuationData' )
99 return title
, descr
, thumb
, items
, more
101 def prepare_playlist ( result
):
102 contents
= result
[ 'continuationContents' ]
103 unparsed
= contents
[ 'playlistVideoListContinuation' ]. get ( 'contents' ,[])
106 |
G ( 'playlistVideoListContinuation' )
108 |
Select ( 'nextContinuationData' )
112 meta
= result|
G ( 'sidebar' ) |
G ( 'playlistSidebarRenderer' ) |
G ( 'items' )
113 meta1
= meta|
Select ( 'playlistSidebarPrimaryInfoRenderer' )
114 meta2
= meta|
Select ( 'playlistSidebarSecondaryInfoRenderer' ) \
115 |
G ( 'videoOwner' ) |
G ( 'videoOwnerRenderer' )
116 title
= meta1|
G ( 'title' ) |G
. text
117 author
= meta2|
G ( 'title' ) |G
. text
118 channel_id
= meta2|
G ( 'navigationEndpoint' ) |
G ( 'browseEndpoint' ) |
G ( 'browseId' )
120 return title
, author
, channel_id
, list ( filter ( None , map ( parse_playlist
, unparsed
))), more
122 def mkthumbs ( thumbs
):
123 output
= { str ( e
[ 'height' ]): e
[ 'url' ] for e
in thumbs
}
124 largest
= next ( iter ( sorted ( output
. keys (), reverse
= True , key
= int )), None )
125 return {** output
, 'largest' : largest
}
128 # externals URLs are redirected through youtube.com/redirect, but we
129 # may encounter internal URLs, too
130 return parse_qs ( urlparse ( url
). query
). get ( 'q' ,[ url
])[ 0 ]
132 def toInt ( s
, fallback
= 0 ):
136 return int ( '' . join ( filter ( str . isdigit
, s
)))
140 # Remove left-/rightmost word from string:
141 delL
= lambda s
: s
. partition ( ' ' )[ 2 ]
144 if s
is None : # missing from autogen'd music, some livestreams
146 # Some livestreams have "Streamed 7 hours ago"
147 s
= s
. replace ( "Streamed " , "" )
148 # Now, everything should be in the form "1 year ago"
149 value
, unit
, _
= s
. split ( " " )
153 ). get ( unit
, unit
[ 0 ]) # first letter otherwise (e.g. year(s) => y)
155 return f
" {value}{suffix} "
157 def log_unknown_card ( data
):
160 from flask
import request
162 except : source
= "unknown"
163 with
open ( "/tmp/innertube.err" , "a" , encoding
= "utf-8" , errors
= "backslashreplace" ) as f
:
164 f
. write ( f
" \n /***** {source} *****/ \n " )
165 json
. dump ( data
, f
, indent
= 2 )
167 def parse_result_items ( items
):
168 # TODO: use .get() for most non-essential attributes
170 parses youtube search response into an easier to use format.
175 key
= next ( iter ( item
. keys ()), None )
177 if key
== 'videoRenderer' :
178 results
. append ({ 'type' : 'VIDEO' , 'content' : {
179 'video_id' : content
[ 'videoId' ],
180 'title' : content
[ 'title' ] |G
. text
,
181 'author' : content|
G ( 'longBylineText' , 'shortBylineText' ) |G
. text
,
182 'channel_id' : content|
G ( 'ownerText' ) |
G ( 'runs' ) |
G ( 0 ) \
183 |
G ( 'navigationEndpoint' ) |
G ( 'browseEndpoint' ) |
G ( 'browseId' ),
184 'length' : content|
G ( 'lengthText' ) |G
. text
, # "44:07", "1:41:50"
185 'views' : content|
G ( 'viewCountText' ) |G
. text|A
. int or 0 , # "1,234 {views|watching}", absent on 0 views
186 'published' : content|
G ( 'publishedTimeText' ) |
G ( 'simpleText' ) |
A ( age
),
187 'live' : content|
G ( 'badges' ) |
Select ( 'metadataBadgeRenderer' ) |
G ( 'style' )== 'BADGE_STYLE_TYPE_LIVE_NOW' ,
189 elif key
in [ 'playlistRenderer' , 'radioRenderer' , 'showRenderer' ]: # radio == "Mix" playlist, show == normal playlist, specially displayed
190 results
. append ({ 'type' : 'PLAYLIST' , 'content' : {
191 'playlist_id' : content
[ 'navigationEndpoint' ] |
G ( 'watchEndpoint' ) |
G ( 'playlistId' ),
192 'video_id' : content
[ 'navigationEndpoint' ] |
G ( 'watchEndpoint' ) |
G ( 'videoId' ),
193 'title' : content
[ 'title' ] |G
. text
,
194 'author' : content|
G ( 'longBylineText' , 'shortBylineText' ) |G
. text
,
195 'channel_id' : content|
G ( 'longBylineText' , 'shortBylineText' ) |
G ( 'runs' ) |
G ( 0 ) \
196 |
G ( 'navigationEndpoint' ) |
G ( 'browseEndpoint' ) |
G ( 'browseId' ),
197 'n_videos' : content|
G ( 'videoCount' ) |A
. int or \
198 content|
G ( 'videoCountShortText' , 'videoCountText' ) |G
. text
, # "Mix" playlists
200 elif key
== 'channelRenderer' :
201 results
. append ({ 'type' : 'CHANNEL' , 'content' : {
202 'channel_id' : content
[ 'channelId' ],
203 'title' : content
[ 'title' ] |G
. text
,
204 'icons' : content
[ 'thumbnail' ][ 'thumbnails' ] |
A ( mkthumbs
),
205 'subscribers' : content|
G ( 'subscriberCountText' ) |
G ( 'simpleText' ), # "2.47K subscribers"
207 elif key
== 'shelfRenderer' :
208 subkey
= next ( iter ( content
[ 'content' ]. keys ()), None ) #verticalListRenderer/horizontalMovieListRenderer
209 r
, e
= parse_result_items ( content
[ 'content' ][ subkey
][ 'items' ])
212 elif key
in [ 'movieRenderer' , 'gridMovieRenderer' ]: # movies to buy/rent
213 pass # gMR.{videoId,title.runs[].text,lengthText.simpleText}
214 elif key
in [ 'carouselAdRenderer' , 'searchPyvRenderer' , 'promotedSparklesTextSearchRenderer' ,
215 'promotedSparklesWebRenderer' , 'compactPromotedItemRenderer' ]: # haha, no.
217 elif key
== 'horizontalCardListRenderer' :
218 # suggested searches: .cards[].searchRefinementCardRenderer.query.runs[].text
220 elif key
== 'emergencyOneboxRenderer' : # suicide prevention hotline
222 elif key
in [ 'clarificationRenderer' , 'infoPanelContainerRenderer' ]: # COVID-19/conspiracy theory infos
224 elif key
== 'webAnswerRenderer' : # "Result from the web"
226 elif key
== 'infoPanelContentRenderer' : # "These results may be new or changing quickly"
228 elif key
== 'hashtagTileRenderer' : # link to '/hashtag/<search_query>'
230 elif key
in [ 'didYouMeanRenderer' , 'showingResultsForRenderer' , 'includingResultsForRenderer' ]:
233 'query' : content
[ 'correctedQueryEndpoint' ][ 'searchEndpoint' ][ 'query' ], # non-misspelled query
234 'autocorrected' : key
in [ 'showingResultsForRenderer' , 'includingResultsForRenderer' ],
236 elif key
== 'messageRenderer' : # "No more results"
239 'message' : content|
G ( 'title' , 'text' ) |G
. text
,
241 elif key
== 'backgroundPromoRenderer' : # e.g. "no results"
243 'type' : content
[ 'icon' ][ 'iconType' ],
244 'message' : content
[ 'title' ] |G
. text
,
247 log_unknown_card ( item
)
248 return results
, extras
250 def parse_infocard ( card
):
252 parses a single infocard into a format that's easier to handle.
254 card
= card
[ 'cardRenderer' ]
255 ctype
= list ( card
[ 'content' ]. keys ())[ 0 ]
256 content
= card
[ 'content' ][ ctype
]
257 if ctype
== "pollRenderer" :
258 return { 'type' : "POLL" , 'content' : {
259 'question' : content
[ 'question' ][ 'simpleText' ],
260 'answers' : [( a
[ 'text' ][ 'simpleText' ], a
[ 'numVotes' ]) \
261 for a
in content
[ 'choices' ]],
263 elif ctype
== "videoInfoCardContentRenderer" :
264 is_live
= content
. get ( 'badge' ,{}). get ( 'liveBadgeRenderer' ) is not None
265 return { 'type' : "VIDEO" , 'content' : {
266 'video_id' : content
[ 'action' ][ 'watchEndpoint' ][ 'videoId' ],
267 'title' : content
[ 'videoTitle' ][ 'simpleText' ],
268 'author' : delL ( content
[ 'channelName' ][ 'simpleText' ]),
269 'length' : content
. get ( 'lengthString' ,{}). get ( 'simpleText' ) \
270 if not is_live
else "LIVE" , # "23:03"
271 'views' : toInt ( content
. get ( 'viewCountText' ,{}). get ( 'simpleText' )),
272 # XXX: views sometimes "Starts: July 31, 2020 at 1:30 PM"
274 elif ctype
== "playlistInfoCardContentRenderer" :
275 return { 'type' : "PLAYLIST" , 'content' : {
276 'playlist_id' : content
[ 'action' ][ 'watchEndpoint' ][ 'playlistId' ],
277 'video_id' : content
[ 'action' ][ 'watchEndpoint' ][ 'videoId' ],
278 'title' : content
[ 'playlistTitle' ][ 'simpleText' ],
279 'author' : delL ( content
[ 'channelName' ][ 'simpleText' ]),
280 'n_videos' : toInt ( content
[ 'playlistVideoCount' ][ 'simpleText' ]),
282 elif ctype
== "simpleCardContentRenderer" and \
283 'urlEndpoint' in content
[ 'command' ]:
284 return { 'type' : "WEBSITE" , 'content' : {
285 'url' : clean_url ( content
[ 'command' ][ 'urlEndpoint' ][ 'url' ]),
286 'domain' : content
[ 'displayDomain' ][ 'simpleText' ],
287 'title' : content
[ 'title' ][ 'simpleText' ],
288 # XXX: no thumbnails for infocards
290 elif ctype
== "collaboratorInfoCardContentRenderer" :
291 return { 'type' : "CHANNEL" , 'content' : {
292 'channel_id' : content
[ 'endpoint' ][ 'browseEndpoint' ][ 'browseId' ],
293 'title' : content
[ 'channelName' ][ 'simpleText' ],
294 'icons' : mkthumbs ( content
[ 'channelAvatar' ][ 'thumbnails' ]),
295 'subscribers' : content
. get ( 'subscriberCountText' ,{}). get ( 'simpleText' , '' ), # "545K subscribers"
298 log_unknown_card ( card
)
301 def parse_endcard ( card
):
303 parses a single endcard into a format that's easier to handle.
305 card
= card
. get ( 'endscreenElementRenderer' , card
) #only sometimes nested
306 ctype
= card
[ 'style' ]
307 if ctype
== "CHANNEL" :
308 return { 'type' : ctype
, 'content' : {
309 'channel_id' : card
[ 'endpoint' ][ 'browseEndpoint' ][ 'browseId' ],
310 'title' : card
[ 'title' ] |G
. text
,
311 'icons' : mkthumbs ( card
[ 'image' ][ 'thumbnails' ]),
313 elif ctype
== "VIDEO" :
314 if not 'endpoint' in card
: return None # title == "This video is unavailable."
315 return { 'type' : ctype
, 'content' : {
316 'video_id' : card
[ 'endpoint' ][ 'watchEndpoint' ][ 'videoId' ],
317 'title' : card
[ 'title' ] |G
. text
,
318 'length' : card|
G ( 'videoDuration' ) |G
. text
, # '12:21'
319 'views' : toInt ( card
[ 'metadata' ] |G
. text
),
320 # XXX: no channel name
322 elif ctype
== "PLAYLIST" :
323 return { 'type' : ctype
, 'content' : {
324 'playlist_id' : card
[ 'endpoint' ][ 'watchEndpoint' ][ 'playlistId' ],
325 'video_id' : card
[ 'endpoint' ][ 'watchEndpoint' ][ 'videoId' ],
326 'title' : card
[ 'title' ] |G
. text
,
327 'author' : delL ( card
[ 'metadata' ] |G
. text
),
328 'n_videos' : toInt ( card
[ 'playlistLength' ] |G
. text
),
330 elif ctype
== "WEBSITE" or ctype
== "CREATOR_MERCHANDISE" :
331 url
= clean_url ( card
[ 'endpoint' ][ 'urlEndpoint' ][ 'url' ])
332 return { 'type' : "WEBSITE" , 'content' : {
334 'domain' : urlparse ( url
). netloc
,
335 'title' : card
[ 'title' ] |G
. text
,
336 'icons' : mkthumbs ( card
[ 'image' ][ 'thumbnails' ]),
339 log_unknown_card ( card
)
342 def parse_channel_items ( items
, channel_id
, author
):
346 key
= next ( iter ( item
. keys ()), None )
348 if key
in [ "gridVideoRenderer" , "videoRenderer" , "videoCardRenderer" ]:
349 # only videoCardRenderer (topic channels) has author and channel, others fall back to supplied ones.
350 result
. append ({ 'type' : 'VIDEO' , 'content' : {
351 'video_id' : content
[ 'videoId' ],
352 'title' : content|
G ( 'title' ) |G
. text
,
353 'author' : content|
G ( 'bylineText' ) |G
. text
or author
,
354 'channel_id' : ( content|
G ( 'bylineText' ) |
G ( 'runs' )
355 |
Select ( 'navigationEndpoint' )
356 |
G ( 'browseEndpoint' ) |
G ( 'browseId' ) or channel_id
),
357 'length' : ( content|
G ( 'lengthText' ) |G
. text
or # topic channel
358 content|
G ( 'thumbnailOverlays' )
359 |
Select ( 'thumbnailOverlayTimeStatusRenderer' )
361 # topic channel: .metadataText.simpleText = "22M views \u00b7 2 months ago"
362 'views' : content|
G ( 'viewCountText' ) |G
. text|A
. int ,
363 'published' : content|
G ( 'publishedTimeText' ) |G
. text|
A ( age
),
365 elif key
in [ "gridPlaylistRenderer" , "playlistRenderer" , "gridRadioRenderer" ]:
366 result
. append ({ 'type' : 'PLAYLIST' , 'content' : {
367 'playlist_id' : content|
G ( 'navigationEndpoint' ) |
G ( 'watchEndpoint' ) |
G ( 'playlistId' ),
368 'video_id' : content|
G ( 'navigationEndpoint' ) |
G ( 'watchEndpoint' ) |
G ( 'videoId' ),
369 'title' : content|
G ( 'title' ) |G
. text
,
370 'author' : author
, # Note: gridRadioRenderer is by 'Youtube' without channel_id, ignoring that.
371 'channel_id' : channel_id
,
372 'n_videos' : ( content|
G ( 'videoCount' ) |A
. int or # playlistRenderer
373 content|
G ( 'videoCountShortText' , 'videoCountText' ) |G
. text|A
. int ) # grid
375 elif key
== "showRenderer" :
376 result
. append ({ 'type' : 'PLAYLIST' , 'content' : {
377 'playlist_id' : content
[ 'navigationEndpoint' ][ 'watchEndpoint' ][ 'playlistId' ],
378 'video_id' : content
[ 'navigationEndpoint' ][ 'watchEndpoint' ][ 'videoId' ],
379 'title' : content
[ 'title' ][ 'simpleText' ],
381 'channel_id' : channel_id
,
384 elif key
in [ "gridShowRenderer" ]:
385 result
. append ({ 'type' : 'PLAYLIST' , 'content' : {
386 'playlist_id' : ( content|
G ( 'navigationEndpoint' )
387 |
G ( 'browseEndpoint' ) |
G ( 'browseId' ))[ 2 :],
388 #^: playlistId prefixed with 'VL', which must be removed
390 'title' : content|
G ( 'title' ) |G
. text
,
392 'channel_id' : channel_id
,
393 'n_videos' : content|
G ( 'thumbnailOverlays' ) |
G ( 0 )
394 |
G ( 'thumbnailOverlayBottomPanelRenderer' ) |
G ( 'text' ) |G
. text
,
396 elif key
in [ "itemSectionRenderer" , "gridRenderer" , "horizontalCardListRenderer" , "horizontalListRenderer" ]:
398 "itemSectionRenderer" : 'contents' ,
399 "gridRenderer" : 'items' ,
400 "horizontalCardListRenderer" : 'cards' ,
401 "horizontalListRenderer" : 'items' ,
403 r
, e
= parse_channel_items ( content
[ newkey
], channel_id
, author
)
406 elif key
== "shelfRenderer" :
407 r
, e
= parse_channel_items ([ content
[ 'content' ]], channel_id
, author
)
410 elif key
== "messageRenderer" :
411 # e.g. {'messageRenderer': {'text': {'runs': [{'text': 'This channel has no playlists.'}]}}}
413 elif key
== "gameCardRenderer" :
415 elif key
== "gridChannelRenderer" :
416 pass # don't care; related channels, e.g. on UCMsgXPD3wzzt8RxHJmXH7hQ
418 log_unknown_card ( item
)
422 def parse_playlist ( item
):
423 key
= next ( iter ( item
. keys ()), None )
425 if key
== "playlistVideoRenderer" :
426 if not content
. get ( 'isPlayable' , False ):
427 return None # private or deleted video
429 return { 'type' : 'VIDEO' , 'content' : {
430 'video_id' : content
[ 'videoId' ],
431 'title' : ( content
[ 'title' ]. get ( 'simpleText' ) or # playable videos
432 content
[ 'title' ]. get ( 'runs' ,[{}])[ 0 ]. get ( 'text' )), # "[Private video]"
433 'playlist_id' : content
[ 'navigationEndpoint' ][ 'watchEndpoint' ][ 'playlistId' ],
434 'index' : content
[ 'navigationEndpoint' ][ 'watchEndpoint' ]. get ( 'index' , 0 ), #or int(content['index']['simpleText']) (absent on course intros; e.g. PL96C35uN7xGJu6skU4TBYrIWxggkZBrF5)
435 # rest is missing from unplayable videos:
436 'author' : content
. get ( 'shortBylineText' ,{}). get ( 'runs' ,[{}])[ 0 ]. get ( 'text' ),
437 'channel_id' : content
. get ( 'shortBylineText' ,{}). get ( 'runs' ,[{}])[ 0 ]. get ( 'navigationEndpoint' ,{}). get ( 'browseEndpoint' ,{}). get ( 'browseId' ),
438 'length' : ( content
. get ( "lengthText" ,{}). get ( "simpleText" ) or # "8:51"
439 int ( content
. get ( "lengthSeconds" , 0 ))), # "531"
440 'starttime' : content
[ 'navigationEndpoint' ][ 'watchEndpoint' ]. get ( 'startTimeSeconds' ),
443 raise Exception ( item
) # XXX TODO