]>
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 *
14 frontend
= Blueprint('youtube', __name__
,
15 template_folder
='templates',
16 static_folder
='static',
17 static_url_path
='/static/yt')
21 return redirect(url_for('.feed'), code
=302)
23 @frontend.route('/feed/subscriptions')
24 # disabled for guest user: @login_required
26 if current_user
.is_anonymous
:
28 if 'welcome_message' in cf
['frontend']:
29 flash(cf
['frontend']['welcome_message'], "welcome")
31 token
= current_user
.token
32 page
= int(request
.args
.get('page', 0))
33 with sqlite3
.connect(cf
['global']['database']) as conn
:
36 SELECT videos.id, channel_id, name, title, length, livestream, published, playlist_videos.playlist_id, display
38 JOIN channels ON videos.channel_id = channels.id
39 LEFT JOIN playlist_videos ON (videos.id = playlist_videos.video_id)
40 LEFT JOIN flags ON (videos.id = flags.video_id) AND (flags.user = ?)
41 WHERE (channel_id IN (SELECT channel_id FROM subscriptions WHERE user=? AND type = 'channel')
42 OR playlist_videos.playlist_id IN (SELECT channel_id FROM subscriptions WHERE user=? AND type = 'playlist')
43 OR flags.display = 'pinned')
44 AND flags.display IS NOT 'hidden'
45 ORDER BY (display = 'pinned') DESC, crawled DESC
47 OFFSET 36*?""", (token
, token
, token
, page
))
50 'channel_id': channel_id
,
54 'livestream': livestream
,
55 'published': published
,
57 'pinned': display
== 'pinned',
58 } for (video_id
, channel_id
, author
, title
, length
, livestream
, published
, playlist
, display
) in c
.fetchall()]
59 return render_template('index.html.j2', rows
=rows
, page
=page
)
61 @frontend.route('/watch')
63 if current_user
.is_anonymous
:
66 token
= current_user
.token
68 if not 'v' in request
.args
:
69 return "missing video id", 400
72 'content-type': 'text/plain',
73 'Link': "<data:text/css,body%7Bcolor:%23eee;background:%23333%7D>; rel=stylesheet;"
76 video_id
= request
.args
.get('v')
77 sts
, algo
= get_cipher()
78 video_url
, stream_map
, metadata
, error
, errdetails
= get_video_info(video_id
, sts
, algo
)
80 extra
= {'geolocked':'local=1', 'livestream':'raw=0'}.get(error
,'')
81 invidious_url
= f
"https://invidious.snopyta.org/watch?v={video_id}&{extra}"
83 'banned': "Instance is being rate limited.",
84 'malformed': "Video ID is invalid.",
85 'geolocked': "This video is geolocked.",
86 'livestream': "Livestreams not yet supported.",
87 'exhausted': errdetails
or "Couldn't extract video URLs.",
91 # if the video is geolocked, and the proxy is enabled, we can still play it:
93 if error
== 'geolocked':
94 videoplayback
= url_for('proxy.videoplayback')
96 query
= urlparse(video_url
).query
97 video_url
= f
"{videoplayback}?{query}"
98 for s
in stream_map
['adaptive']:
99 query
= urlparse(s
['url']).query
100 s
['url'] = f
"{videoplayback}?{query}"
101 for s
in stream_map
['muxed']:
102 query
= urlparse(s
['url']).query
103 s
['url'] = f
"{videoplayback}?{query}"
108 # if the instance is blocked, try submitting a job to the anti captcha service:
110 if error
== 'banned' and request
.args
.get('tobi'):
111 from ..anticaptcha
import lib
as anticaptcha
112 r2
= requests
.get(f
'https://www.youtube.com/watch?v={video_id}&hl=en&gl=US')
113 anticaptcha
.check_captcha_or_raise(r2
)
114 except BadGateway
as e
: raise e
115 except Exception as e
: pass # unavailable (TODO: check if blueprint is loaded)
117 show
= request
.args
.get("show")
120 msg
= errdetails
if error
=='player' else f
"{error.upper()}: {errdetails}"
121 return f
"{msg}\n\nRedirecting to Invidious.", 502, {
122 'Refresh': f
'2; URL={invidious_url}&raw=1',
124 return redirect(video_url
, code
=307)
126 if error
and not metadata
:
127 return {'error': True, error
: errdetails
}, 400 # TODO: better (test _CpR4o81XQc)
128 return jsonify(metadata
)
129 elif show
== "audio":
130 # sorting: we want to prioritize mp4a over opus, and sort by highest quality first
131 # todo: geolocking; prefer open format?
132 if error
and not metadata
:
133 msg
= errdetails
if error
=='player' else f
"{error.upper()}: {errdetails}"
134 return msg
, 400, plaintextheaders
# TODO: nicer
135 stream
= next(iter(sorted([
136 stream
for stream
in stream_map
['adaptive']
137 if stream
['mimeType'].startswith('audio/')],
138 key
=lambda stream
: ('opus' not in stream
['mimeType'],
139 stream
['bitrate']), reverse
=True)
141 return redirect(stream
)
143 if error
and not metadata
: # e.g. malformed, private/deleted video, ...
144 return errdetails
+ f
"<br>Try visiting <a href='{invidious_url}'>Invidious</a>.",400 # TODO: nicer
145 meta
= prepare_metadata(metadata
)
146 with sqlite3
.connect(cf
['global']['database']) as conn
:
150 SELECT 1 FROM subscriptions WHERE channel_id = ? AND user = ?
152 SELECT 1 FROM flags WHERE video_id = ? AND display = 'pinned' AND user = ?
153 ))""", (meta
['channel_id'], token
, video_id
, token
))
154 (is_subscribed
, is_pinned
) = c
.fetchone()
155 return render_template('watch.html.j2',
156 video_id
=video_id
, video_url
=video_url
, stream_map
=stream_map
,
157 video_error
=error
, errdetails
=errdetails
, invidious_url
=invidious_url
,
158 is_pinned
=is_pinned
, is_subscribed
=is_subscribed
,
161 @frontend.route('/embed/videoseries')
162 def embed_videoseries():
163 return redirect(url_for('.playlist', list=request
.args
.get('list')))
164 @frontend.route('/embed/<video_id>')
166 if video_id
== "videoseries":
167 return redirect(url_for('.playlist', list=request
.args
.get('list')))
169 return redirect(url_for('.watch', v
=video_id
, t
=request
.args
.get('start')))
171 @frontend.route('/<something>', strict_slashes
=False)
172 def plain_user_or_video(something
):
173 # yt.com interprets this as a username, but we also want to catch youtu.be
174 # short-urls. so we check if it's a channel by querying the RSS feed (this
175 # shoudn't be rate-limited); if that fails, check if it looks like a video
176 # id; or finally give up.
177 if '.' not in something
and channel_exists(something
):
178 # periods are not valid in usernames, vanity urls or ucids, but common
179 # in urls that get crawled by bots (e.g. index.php). failing early
180 # reduces the amount of invalid channel names getting looked up.
181 return redirect(url_for('.channel', channel_id
=something
))
182 elif re
.match(r
"^[-_0-9A-Za-z]{11}$", something
): # looks like a video id
183 return redirect(url_for('.watch', v
=something
, t
=request
.args
.get('t')))
185 # XXX: something == 'thethoughtemporium' -> 404s
186 raise NotFound("Note: some usernames not recognized; try searching it")
188 @frontend.route('/channel/<channel_id>/<subpage>')
189 @frontend.route('/user/<channel_id>/<subpage>')
190 @frontend.route('/c/<channel_id>/<subpage>')
191 @frontend.route('/channel/<channel_id>/')
192 @frontend.route('/user/<channel_id>/')
193 @frontend.route('/c/<channel_id>/')
194 def channel(channel_id
, _
=None):
195 token
= getattr(current_user
, 'token', 'guest')
197 if re
.match(r
"(UC[A-Za-z0-9_-]{22})", channel_id
):
198 xmlfeed
= fetch_xml("channel_id", channel_id
)
200 xmlfeed
= fetch_xml("user", channel_id
)
203 return "not found or something", 404 # XXX
204 title
, author
, videos
, channel_id
, _
= parse_xml(xmlfeed
)
206 with sqlite3
.connect(cf
['global']['database']) as conn
:
211 WHERE channel_id = ? AND user = ?
212 """, (channel_id
, token
))
213 (is_subscribed
,) = c
.fetchone()
215 return render_template('xmlfeed.html.j2', title
=author
, rows
=videos
,
216 is_subscribed
=is_subscribed
, channel_id
=channel_id
)
218 @frontend.route('/playlist')
220 playlist_id
= request
.args
.get('list')
222 return "bad list id", 400 # todo
224 xmlfeed
= fetch_xml("playlist_id", playlist_id
)
226 return "not found or something", 404 # XXX
227 title
, author
, videos
, _
, _
= parse_xml(xmlfeed
)
228 return render_template('xmlfeed.html.j2', title
=f
"{title} by {author}", rows
=videos
)
230 @frontend.route('/api/timedtext')
232 r
= requests
.get("https://www.youtube.com/api/timedtext", request
.args
.to_dict())
233 # Note: in srv1 format, xmlentities are double-encoded m( a smart quote is
234 # even worse: it's '&39;<smartquote>' wtf!? (at least vvt seems ok)
236 return "error: {r.text}", 400 # TODO: better
238 if request
.args
.get('fmt') == 'vtt' and request
.args
.get('kind') == 'asr':
239 # autocaptions are extremely confusing, and stuck in the lower-left corner. fix it up a bit
240 retval
= re
.sub(r
"<.+?>", "", retval
) # remove inline html-like markup that times each word/adds styles
241 retval
= retval
.replace("align:start position:0%", "") # let browser position the text itself
242 # each subtitle-line is repeated twice (first on the lower line, then
243 # on the next "frame" on the upper line). we want to remove the
244 # repetition, as that's confusing without word and line animations:
245 lines
= retval
.split('\n')
246 retval
= '\n'.join([line
for line
, prev
in zip(lines
, ['']+lines
) if not " --> " in prev
])
247 return retval
, {'Content-Type': r
.headers
.get("Content-Type")}
249 @frontend.route('/manage/subscriptions')
250 # disabled for guest user: @login_required
251 def subscription_manager():
252 if current_user
.is_anonymous
:
255 token
= current_user
.token
256 with sqlite3
.connect(cf
['global']['database']) as conn
:
257 #with conn.cursor() as c:
260 SELECT subscriptions.channel_id, name, type,
261 (subscribed_until < datetime('now')) AS obsolete
263 LEFT JOIN (SELECT name, id FROM channels
265 SELECT name, id FROM playlists
266 ) AS channels ON channels.id = subscriptions.channel_id
267 left JOIN websub ON channels.id = websub.channel_id
269 AND subscriptions.type IN ('channel', 'playlist')
270 ORDER BY obsolete=0, name COLLATE NOCASE ASC""", (token
,))
272 'channel_id': channel_id
,
273 'author': author
or channel_id
,
275 'subscribed_until': subscribed_until
276 } for (channel_id
, author
, type, subscribed_until
) in c
.fetchall()]
277 return render_template('subscription_manager.html.j2', rows
=rows
)
279 @frontend.route('/feed/subscriptions', methods
=['POST'])
282 token
= current_user
.token
283 action
= next(request
.form
.keys(), None)
284 if action
in ['pin', 'unpin', 'hide']:
285 video_id
= request
.form
.get(action
)
291 with sqlite3
.connect(cf
['global']['database']) as conn
:
293 store_video_metadata(video_id
) # only needed for pinning
295 INSERT OR REPLACE INTO flags (user, video_id, display)
297 """, (token
, video_id
, display
))
299 flash("unsupported action", "error")
300 return redirect(request
.url
, code
=303)
302 @frontend.route('/manage/subscriptions', methods
=['POST'])
304 def manage_subscriptions():
305 token
= current_user
.token
306 if 'subscribe' in request
.form
:
307 some_id
= request
.form
.get("subscribe")
308 match
= re
.search(r
"(UC[A-Za-z0-9_-]{22})", some_id
)
310 some_id
= match
.group(1)
313 match
= re
.search(r
"((?:PL|LL|EC|UU|FL|UL|OL)[A-Za-z0-9_-]{10,})", some_id
)
314 if match
: # NOTE: PL-playlists are 32chars, others differ in length.
315 some_id
= match
.group(1)
318 flash("not a valid/subscribable URI", "error")
319 return redirect(request
.url
, code
=303)
320 with sqlite3
.connect(cf
['global']['database']) as conn
:
321 #with conn.cursor() as c:
324 INSERT OR IGNORE INTO subscriptions (user, channel_id, type)
326 """, (token
, some_id
, id_type
))
327 # TODO: sql-error-handling, asynchronically calling update-subs.pl
329 elif 'unsubscribe' in request
.form
:
330 some_id
= request
.form
.get("unsubscribe")
331 with sqlite3
.connect(cf
['global']['database']) as conn
:
332 #with conn.cursor() as c:
335 DELETE FROM subscriptions
336 WHERE user = ? AND channel_id = ?
337 """, (token
, some_id
))
338 # TODO: sql-error-handling, report success
341 flash("unsupported action", "error")
343 return redirect(request
.url
, code
=303)
346 # reload cipher from database every 1 hour
347 if 'cipher' not in g
or time
.time() - g
.get('cipher_updated', 0) > 1 * 60 * 60:
348 with sqlite3
.connect(cf
['global']['database']) as conn
:
350 c
.execute("SELECT sts, algorithm FROM cipher")
351 g
.cipher
= c
.fetchone()
352 g
.cipher_updated
= time
.time()
356 #@frontend.teardown_appcontext
358 # db = g.pop('db', None)
364 @frontend.app_template_filter('format_date')
366 import datetime
# can't import at top level, because it is inherited from common
367 (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'
368 M
= '_ Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec'.split()
369 if y
== datetime
.datetime
.now().year
:
372 return f
"{M[m]} '{y%100}"
374 @frontend.app_template_filter('format_time')
381 return '%d:%02d:%02d' % (h
,m
,s
) if h
else '%02d:%02d' % (m
,s
)
383 @frontend.app_template_filter('timeoffset')
387 match
= re
.match(r
"^(\d+)s?$", s
) # e.g. 2040s
389 return match
.group(1)
390 match
= re
.match(r
"^(?:(\d+)h)?(?:(\d+)m)?(?:(\d+)s)?$", s
) # e.g. 34m, 1h23s
392 return ":".join([n
.zfill(2) for n
in match
.groups('0')])