]>
git.gir.st - subscriptionfeed.git/blob - app/youtube/__init__.py
5 from urllib
.parse
import urlparse
6 #from flask_login import current_user, login_required
7 from flask_login
import LoginManager
, UserMixin
, current_user
, login_user
, logout_user
, login_required
8 from flask
import Blueprint
, render_template
, request
, redirect
, flash
, url_for
, jsonify
, g
9 from werkzeug
.exceptions
import NotFound
, BadGateway
11 from ..common
.common
import *
12 from ..common
.anticaptcha
import submit_captcha
15 frontend
= Blueprint('youtube', __name__
,
16 template_folder
='templates',
17 static_folder
='static',
18 static_url_path
='/static/yt')
22 return redirect(url_for('.feed'), code
=302)
24 @frontend.route('/feed/subscriptions')
25 # disabled for guest user: @login_required
27 if current_user
.is_anonymous
:
29 if 'welcome_message' in cf
['frontend']:
30 flash(cf
['frontend']['welcome_message'], "welcome")
32 token
= current_user
.token
33 page
= int(request
.args
.get('page', 0))
34 with sqlite3
.connect(cf
['global']['database']) as conn
:
37 SELECT videos.id, channel_id, name, title, length, livestream, published, playlist_videos.playlist_id, display
39 JOIN channels ON videos.channel_id = channels.id
40 LEFT JOIN playlist_videos ON (videos.id = playlist_videos.video_id)
41 LEFT JOIN flags ON (videos.id = flags.video_id) AND (flags.user = ?)
42 WHERE (channel_id IN (SELECT channel_id FROM subscriptions WHERE user=? AND type = 'channel')
43 OR playlist_videos.playlist_id IN (SELECT channel_id FROM subscriptions WHERE user=? AND type = 'playlist')
44 OR flags.display = 'pinned')
45 AND flags.display IS NOT 'hidden'
46 ORDER BY (display = 'pinned') DESC, crawled DESC
48 OFFSET 36*?""", (token
, token
, token
, page
))
51 'channel_id': channel_id
,
55 'livestream': livestream
,
56 'published': published
,
58 'pinned': display
== 'pinned',
59 } for (video_id
, channel_id
, author
, title
, length
, livestream
, published
, playlist
, display
) in c
.fetchall()]
60 return render_template('index.html.j2', rows
=rows
, page
=page
)
62 @frontend.route('/watch')
64 if current_user
.is_anonymous
:
67 token
= current_user
.token
69 if not 'v' in request
.args
:
70 return "missing video id", 400
73 'content-type': 'text/plain',
74 'Link': "<data:text/css,body%7Bcolor:%23eee;background:%23333%7D>; rel=stylesheet;"
77 video_id
= request
.args
.get('v')
78 sts
, algo
= get_cipher()
79 video_url
, stream_map
, metadata
, error
, errdetails
= get_video_info(video_id
, sts
, algo
)
81 extra
= {'geolocked':'local=1', 'livestream':'raw=0'}.get(error
,'')
82 invidious_url
= f
"https://invidious.snopyta.org/watch?v={video_id}&{extra}"
84 'banned': "Instance is being rate limited.",
85 'malformed': "Video ID is invalid.",
86 'geolocked': "This video is geolocked.",
87 'livestream': "Livestreams not supported on this instance.",
88 'exhausted': errdetails
or "Couldn't extract video URLs.",
92 # if the video is geolocked, and the proxy is enabled, we can still play it:
94 if error
== 'geolocked':
95 videoplayback
= url_for('proxy.videoplayback')
97 query
= urlparse(video_url
).query
98 video_url
= f
"{videoplayback}?{query}"
99 for s
in stream_map
['adaptive']:
100 query
= urlparse(s
['url']).query
101 s
['url'] = f
"{videoplayback}?{query}"
102 for s
in stream_map
['muxed']:
103 query
= urlparse(s
['url']).query
104 s
['url'] = f
"{videoplayback}?{query}"
110 if error
== 'livestream':
111 proxy_enabled
= url_for('proxy.videoplayback') # will raise if disabled, i think
112 video_url
= urlparse(stream_map
['hlsManifestUrl']).path
117 # if the instance is blocked, try submitting a job to the anti captcha service:
118 if error
== 'banned' and cf
['captcha']['api_key']:
119 r2
= requests
.get(f
'https://www.youtube.com/watch?v={video_id}&hl=en&gl=US')
120 status
= submit_captcha(r2
)
122 raise Exception("we are banned, but captcha wasn't triggered!")
124 message
= "right now" if status
is True else f
"{int(status)} seconds ago"
125 raise BadGateway(f
"""
126 {errdetails} An attempt at getting unblocked has been made {message}.
127 Please try again in 30 seconds.
130 show
= request
.args
.get("show")
133 msg
= errdetails
if error
=='player' else f
"{error.upper()}: {errdetails}"
134 return f
"{msg}\n\nRedirecting to Invidious.", 502, {
135 'Refresh': f
'2; URL={invidious_url}&raw=1',
137 return redirect(video_url
, code
=307)
139 if error
and not metadata
:
140 return {'error': True, error
: errdetails
}, 400 # TODO: better (test _CpR4o81XQc)
141 return jsonify(metadata
)
142 elif show
== "audio":
143 # sorting: we want to prioritize mp4a over opus, and sort by highest quality first
144 # todo: geolocking; prefer open format?
145 if error
and not metadata
:
146 msg
= errdetails
if error
=='player' else f
"{error.upper()}: {errdetails}"
147 return msg
, 400, plaintextheaders
# TODO: nicer
148 stream
= next(iter(sorted([
149 stream
for stream
in stream_map
['adaptive']
150 if stream
['mimeType'].startswith('audio/')],
151 key
=lambda stream
: ('opus' not in stream
['mimeType'],
152 stream
['bitrate']), reverse
=True)
154 return redirect(stream
)
156 if error
and not metadata
: # e.g. malformed, private/deleted video, ...
157 return errdetails
+ f
"<br>Try visiting <a href='{invidious_url}'>Invidious</a>.",400 # TODO: nicer
158 meta
= prepare_metadata(metadata
)
159 with sqlite3
.connect(cf
['global']['database']) as conn
:
163 SELECT 1 FROM subscriptions WHERE channel_id = ? AND user = ?
165 SELECT 1 FROM flags WHERE video_id = ? AND display = 'pinned' AND user = ?
166 ))""", (meta
['channel_id'], token
, video_id
, token
))
167 (is_subscribed
, is_pinned
) = c
.fetchone()
168 return render_template('watch.html.j2',
169 video_id
=video_id
, video_url
=video_url
, stream_map
=stream_map
,
170 video_error
=error
, errdetails
=errdetails
, invidious_url
=invidious_url
,
171 is_pinned
=is_pinned
, is_subscribed
=is_subscribed
,
174 @frontend.route('/embed/videoseries')
175 def embed_videoseries():
176 return redirect(url_for('.playlist', list=request
.args
.get('list')))
177 @frontend.route('/embed/<video_id>')
179 if video_id
== "videoseries":
180 return redirect(url_for('.playlist', list=request
.args
.get('list')))
182 return redirect(url_for('.watch', v
=video_id
, t
=request
.args
.get('start')))
184 @frontend.route('/<something>', strict_slashes
=False)
185 def plain_user_or_video(something
):
186 # yt.com interprets this as a username, but we also want to catch youtu.be
187 # short-urls. so we check if it's a channel by querying the RSS feed (this
188 # shoudn't be rate-limited); if that fails, check if it looks like a video
189 # id; or finally give up.
190 if '.' not in something
and channel_exists(something
):
191 # periods are not valid in usernames, vanity urls or ucids, but common
192 # in urls that get crawled by bots (e.g. index.php). failing early
193 # reduces the amount of invalid channel names getting looked up.
194 return redirect(url_for('.channel', channel_id
=something
))
195 elif re
.match(r
"^[-_0-9A-Za-z]{11}$", something
): # looks like a video id
196 return redirect(url_for('.watch', v
=something
, t
=request
.args
.get('t')))
197 else: # ¯\_(ツ)_/¯
198 # XXX: something == 'thethoughtemporium' -> 404s
199 raise NotFound("Note: some usernames not recognized; try searching it")
201 @frontend.route('/channel/<channel_id>/<subpage>')
202 @frontend.route('/user/<channel_id>/<subpage>')
203 @frontend.route('/c/<channel_id>/<subpage>')
204 @frontend.route('/channel/<channel_id>/')
205 @frontend.route('/user/<channel_id>/')
206 @frontend.route('/c/<channel_id>/')
207 def channel(channel_id
, _
=None):
208 token
= getattr(current_user
, 'token', 'guest')
210 if re
.match(r
"(UC[A-Za-z0-9_-]{22})", channel_id
):
211 xmlfeed
= fetch_xml("channel_id", channel_id
)
213 xmlfeed
= fetch_xml("user", channel_id
)
216 return "not found or something", 404 # XXX
217 title
, author
, videos
, channel_id
, _
= parse_xml(xmlfeed
)
219 with sqlite3
.connect(cf
['global']['database']) as conn
:
224 WHERE channel_id = ? AND user = ?
225 """, (channel_id
, token
))
226 (is_subscribed
,) = c
.fetchone()
228 return render_template('xmlfeed.html.j2', title
=author
, rows
=videos
,
229 is_subscribed
=is_subscribed
, channel_id
=channel_id
)
231 @frontend.route('/playlist')
233 playlist_id
= request
.args
.get('list')
235 return "bad list id", 400 # todo
237 xmlfeed
= fetch_xml("playlist_id", playlist_id
)
239 return "not found or something", 404 # XXX
240 title
, author
, videos
, _
, _
= parse_xml(xmlfeed
)
241 return render_template('xmlfeed.html.j2', title
=f
"{title} by {author}", rows
=videos
)
243 @frontend.route('/api/timedtext')
245 r
= requests
.get("https://www.youtube.com/api/timedtext", request
.args
.to_dict())
246 # Note: in srv1 format, xmlentities are double-encoded m( a smart quote is
247 # even worse: it's '&39;<smartquote>' wtf!? (at least vvt seems ok)
249 return "error: {r.text}", 400 # TODO: better
251 if request
.args
.get('fmt') == 'vtt' and request
.args
.get('kind') == 'asr':
252 # autocaptions are extremely confusing, and stuck in the lower-left corner. fix it up a bit
253 retval
= re
.sub(r
"<.+?>", "", retval
) # remove inline html-like markup that times each word/adds styles
254 retval
= retval
.replace("align:start position:0%", "") # let browser position the text itself
255 # each subtitle-line is repeated twice (first on the lower line, then
256 # on the next "frame" on the upper line). we want to remove the
257 # repetition, as that's confusing without word and line animations:
258 lines
= retval
.split('\n')
259 retval
= '\n'.join([line
for line
, prev
in zip(lines
, ['']+lines
) if not " --> " in prev
])
260 return retval
, {'Content-Type': r
.headers
.get("Content-Type")}
262 @frontend.route('/manage/subscriptions')
263 # disabled for guest user: @login_required
264 def subscription_manager():
265 if current_user
.is_anonymous
:
268 token
= current_user
.token
269 with sqlite3
.connect(cf
['global']['database']) as conn
:
270 #with conn.cursor() as c:
273 SELECT subscriptions.channel_id, name, type,
274 (subscribed_until < datetime('now')) AS obsolete
276 LEFT JOIN (SELECT name, id FROM channels
278 SELECT name, id FROM playlists
279 ) AS channels ON channels.id = subscriptions.channel_id
280 left JOIN websub ON channels.id = websub.channel_id
282 AND subscriptions.type IN ('channel', 'playlist')
283 ORDER BY obsolete=0, name COLLATE NOCASE ASC""", (token
,))
285 'channel_id': channel_id
,
286 'author': author
or channel_id
,
288 'subscribed_until': subscribed_until
289 } for (channel_id
, author
, type, subscribed_until
) in c
.fetchall()]
290 return render_template('subscription_manager.html.j2', rows
=rows
)
292 @frontend.route('/feed/subscriptions', methods
=['POST'])
295 token
= current_user
.token
296 action
= next(request
.form
.keys(), None)
297 if action
in ['pin', 'unpin', 'hide', 'unhide']:
298 video_id
= request
.form
.get(action
)
305 with sqlite3
.connect(cf
['global']['database']) as conn
:
307 store_video_metadata(video_id
) # only needed for pinning
309 INSERT OR REPLACE INTO flags (user, video_id, display)
311 """, (token
, video_id
, display
))
312 undo_flash(video_id
, action
)
314 flash("unsupported action", "error")
315 return redirect(request
.url
, code
=303)
317 @frontend.route('/manage/subscriptions', methods
=['POST'])
319 def manage_subscriptions():
320 token
= current_user
.token
321 if 'subscribe' in request
.form
:
322 some_id
= request
.form
.get("subscribe")
323 match
= re
.search(r
"(UC[A-Za-z0-9_-]{22})", some_id
)
325 some_id
= match
.group(1)
328 match
= re
.search(r
"((?:PL|LL|EC|UU|FL|UL|OL)[A-Za-z0-9_-]{10,})", some_id
)
329 if match
: # NOTE: PL-playlists are 32chars, others differ in length.
330 some_id
= match
.group(1)
333 flash("not a valid/subscribable URI", "error")
334 return redirect(request
.url
, code
=303)
335 with sqlite3
.connect(cf
['global']['database']) as conn
:
336 #with conn.cursor() as c:
339 INSERT OR IGNORE INTO subscriptions (user, channel_id, type)
341 """, (token
, some_id
, id_type
))
342 # TODO: sql-error-handling, asynchronically calling update-subs.pl
343 undo_flash(some_id
, 'subscribe')
345 elif 'unsubscribe' in request
.form
:
346 some_id
= request
.form
.get("unsubscribe")
347 with sqlite3
.connect(cf
['global']['database']) as conn
:
348 #with conn.cursor() as c:
351 DELETE FROM subscriptions
352 WHERE user = ? AND channel_id = ?
353 """, (token
, some_id
))
354 # TODO: sql-error-handling, report success
355 undo_flash(some_id
, 'unsubscribe')
358 flash("unsupported action", "error")
360 return redirect(request
.url
, code
=303)
363 # reload cipher from database every 1 hour
364 if 'cipher' not in g
or time
.time() - g
.get('cipher_updated', 0) > 1 * 60 * 60:
365 with sqlite3
.connect(cf
['global']['database']) as conn
:
367 c
.execute("SELECT sts, algorithm FROM cipher")
368 g
.cipher
= c
.fetchone()
369 g
.cipher_updated
= time
.time()
373 #@frontend.teardown_appcontext
375 # db = g.pop('db', None)
380 def undo_flash(thing_id
, action
):
381 undo_action
, past_action
= {
382 'pin': ('unpin', 'pinned'),
383 'unpin': ('pin', 'unpinned'),
384 'hide': ('unhide', 'hidden'),
385 'unhide': ('hide', 'unhidden'),
386 'subscribe': ('unsubscribe', 'subscribed'),
387 'unsubscribe': ('subscribe', 'unsubscribed'),
389 if 'subscribe' in action
and thing_id
.startswith('UC'):
391 thing_url
= url_for('.channel', channel_id
=thing_id
)
392 elif 'subscribe' in action
:
394 thing_url
= url_for('.playlist', playlist_id
=thing_id
)
397 thing_url
= url_for('.watch', v
=thing_id
)
399 <form method=post><input type=hidden name="{undo_action}" value="{thing_id}">
400 <a href="{thing_url}">{thing}</a> {past_action}.
401 <label><input type="submit" hidden>
402 <span style="text-decoration:underline;cursor:pointer">undo</span>.
403 </label></form>''', "info")
405 @frontend.app_template_filter('format_date')
407 import datetime
# can't import at top level, because it is inherited from common
408 (y
,m
,d
) = (int(n
) for n
in s
.split('T')[0].split(' ')[0].split('-')) # iso-dates can seperate date from time with space or 'T'
409 M
= '_ Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec'.split()
410 if y
== datetime
.datetime
.now().year
:
413 return f
"{M[m]} '{y%100}"
415 @frontend.app_template_filter('format_time')
422 return '%d:%02d:%02d' % (h
,m
,s
) if h
else '%02d:%02d' % (m
,s
)
424 @frontend.app_template_filter('timeoffset')
428 match
= re
.match(r
"^(\d+)s?$", s
) # e.g. 2040s
430 return match
.group(1)
431 match
= re
.match(r
"^(?:(\d+)h)?(?:(\d+)m)?(?:(\d+)s)?$", s
) # e.g. 34m, 1h23s
433 return ":".join([n
.zfill(2) for n
in match
.groups('0')])