]> git.gir.st - subscriptionfeed.git/blob - app/youtube/__init__.py
integrate anticaptcha into common and clean it up a bit
[subscriptionfeed.git] / app / youtube / __init__.py
1 import re
2 import time
3 import sqlite3
4 import requests
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
10
11 from ..common.common import *
12 from ..common.anticaptcha import submit_captcha
13 from .lib import *
14
15 frontend = Blueprint('youtube', __name__,
16 template_folder='templates',
17 static_folder='static',
18 static_url_path='/static/yt')
19
20 @frontend.route('/')
21 def index():
22 return redirect(url_for('.feed'), code=302)
23
24 @frontend.route('/feed/subscriptions')
25 # disabled for guest user: @login_required
26 def feed():
27 if current_user.is_anonymous:
28 token = 'guest'
29 if 'welcome_message' in cf['frontend']:
30 flash(cf['frontend']['welcome_message'], "welcome")
31 else:
32 token = current_user.token
33 page = int(request.args.get('page', 0))
34 with sqlite3.connect(cf['global']['database']) as conn:
35 c = conn.cursor()
36 c.execute("""
37 SELECT videos.id, channel_id, name, title, length, livestream, published, playlist_videos.playlist_id, display
38 FROM videos
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
47 LIMIT 36
48 OFFSET 36*?""", (token, token, token, page))
49 rows = [{
50 'video_id': video_id,
51 'channel_id': channel_id,
52 'author': author,
53 'title': title,
54 'length': length,
55 'livestream': livestream,
56 'published': published,
57 'playlist': playlist,
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)
61
62 @frontend.route('/watch')
63 def watch():
64 if current_user.is_anonymous:
65 token = 'guest'
66 else:
67 token = current_user.token
68
69 if not 'v' in request.args:
70 return "missing video id", 400
71
72 plaintextheaders = {
73 'content-type': 'text/plain',
74 'Link': "<data:text/css,body%7Bcolor:%23eee;background:%23333%7D>; rel=stylesheet;"
75 }
76
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)
80
81 extra = {'geolocked':'local=1', 'livestream':'raw=0'}.get(error,'')
82 invidious_url = f"https://invidious.snopyta.org/watch?v={video_id}&{extra}"
83 errdetails = {
84 'banned': "Instance is being rate limited.",
85 'malformed': "Video ID is invalid.",
86 'geolocked': "This video is geolocked.",
87 'livestream': "Livestreams not yet supported.",
88 'exhausted': errdetails or "Couldn't extract video URLs.",
89 'player': errdetails,
90 }.get(error)
91
92 # if the video is geolocked, and the proxy is enabled, we can still play it:
93 try:
94 if error == 'geolocked':
95 videoplayback = url_for('proxy.videoplayback')
96
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}"
105
106 error = None
107 except: pass
108
109 # if the instance is blocked, try submitting a job to the anti captcha service:
110 if error == 'banned' and cf['captcha']['api_key']:
111 r2 = requests.get(f'https://www.youtube.com/watch?v={video_id}&hl=en&gl=US')
112 status = submit_captcha(r2)
113 if status is False:
114 raise Exception("we are banned, but captcha wasn't triggered!")
115 else:
116 message = "right now" if status is True else f"{int(status)} seconds ago"
117 raise BadGateway(f"""
118 {errdetails} An attempt at getting unblocked has been made {message}.
119 Please try again in 30 seconds.
120 """)
121
122 show = request.args.get("show")
123 if show == "raw":
124 if error:
125 msg = errdetails if error=='player' else f"{error.upper()}: {errdetails}"
126 return f"{msg}\n\nRedirecting to Invidious.", 502, {
127 'Refresh': f'2; URL={invidious_url}&raw=1',
128 **plaintextheaders}
129 return redirect(video_url, code=307)
130 elif show == "json":
131 if error and not metadata:
132 return {'error': True, error: errdetails}, 400 # TODO: better (test _CpR4o81XQc)
133 return jsonify(metadata)
134 elif show == "audio":
135 # sorting: we want to prioritize mp4a over opus, and sort by highest quality first
136 # todo: geolocking; prefer open format?
137 if error and not metadata:
138 msg = errdetails if error=='player' else f"{error.upper()}: {errdetails}"
139 return msg, 400, plaintextheaders # TODO: nicer
140 stream = next(iter(sorted([
141 stream for stream in stream_map['adaptive']
142 if stream['mimeType'].startswith('audio/')],
143 key=lambda stream: ('opus' not in stream['mimeType'],
144 stream['bitrate']), reverse=True)
145 ),{}).get('url')
146 return redirect(stream)
147 else:
148 if error and not metadata: # e.g. malformed, private/deleted video, ...
149 return errdetails + f"<br>Try visiting <a href='{invidious_url}'>Invidious</a>.",400 # TODO: nicer
150 meta = prepare_metadata(metadata)
151 with sqlite3.connect(cf['global']['database']) as conn:
152 c = conn.cursor()
153 c.execute("""
154 SELECT COUNT((
155 SELECT 1 FROM subscriptions WHERE channel_id = ? AND user = ?
156 )), COUNT((
157 SELECT 1 FROM flags WHERE video_id = ? AND display = 'pinned' AND user = ?
158 ))""", (meta['channel_id'], token, video_id, token))
159 (is_subscribed, is_pinned) = c.fetchone()
160 return render_template('watch.html.j2',
161 video_id=video_id, video_url=video_url, stream_map=stream_map,
162 video_error=error, errdetails=errdetails, invidious_url=invidious_url,
163 is_pinned=is_pinned, is_subscribed=is_subscribed,
164 **meta)
165
166 @frontend.route('/embed/videoseries')
167 def embed_videoseries():
168 return redirect(url_for('.playlist', list=request.args.get('list')))
169 @frontend.route('/embed/<video_id>')
170 def embed(video_id):
171 if video_id == "videoseries":
172 return redirect(url_for('.playlist', list=request.args.get('list')))
173
174 return redirect(url_for('.watch', v=video_id, t=request.args.get('start')))
175
176 @frontend.route('/<something>', strict_slashes=False)
177 def plain_user_or_video(something):
178 # yt.com interprets this as a username, but we also want to catch youtu.be
179 # short-urls. so we check if it's a channel by querying the RSS feed (this
180 # shoudn't be rate-limited); if that fails, check if it looks like a video
181 # id; or finally give up.
182 if '.' not in something and channel_exists(something):
183 # periods are not valid in usernames, vanity urls or ucids, but common
184 # in urls that get crawled by bots (e.g. index.php). failing early
185 # reduces the amount of invalid channel names getting looked up.
186 return redirect(url_for('.channel', channel_id=something))
187 elif re.match(r"^[-_0-9A-Za-z]{11}$", something): # looks like a video id
188 return redirect(url_for('.watch', v=something, t=request.args.get('t')))
189 else: # ¯\_(ツ)_/¯
190 # XXX: something == 'thethoughtemporium' -> 404s
191 raise NotFound("Note: some usernames not recognized; try searching it")
192
193 @frontend.route('/channel/<channel_id>/<subpage>')
194 @frontend.route('/user/<channel_id>/<subpage>')
195 @frontend.route('/c/<channel_id>/<subpage>')
196 @frontend.route('/channel/<channel_id>/')
197 @frontend.route('/user/<channel_id>/')
198 @frontend.route('/c/<channel_id>/')
199 def channel(channel_id, _=None):
200 token = getattr(current_user, 'token', 'guest')
201
202 if re.match(r"(UC[A-Za-z0-9_-]{22})", channel_id):
203 xmlfeed = fetch_xml("channel_id", channel_id)
204 else:
205 xmlfeed = fetch_xml("user", channel_id)
206
207 if not xmlfeed:
208 return "not found or something", 404 # XXX
209 title, author, videos, channel_id, _ = parse_xml(xmlfeed)
210
211 with sqlite3.connect(cf['global']['database']) as conn:
212 c = conn.cursor()
213 c.execute("""
214 SELECT COUNT(*)
215 FROM subscriptions
216 WHERE channel_id = ? AND user = ?
217 """, (channel_id, token))
218 (is_subscribed,) = c.fetchone()
219
220 return render_template('xmlfeed.html.j2', title=author, rows=videos,
221 is_subscribed=is_subscribed, channel_id=channel_id)
222
223 @frontend.route('/playlist')
224 def playlist():
225 playlist_id = request.args.get('list')
226 if not playlist_id:
227 return "bad list id", 400 # todo
228
229 xmlfeed = fetch_xml("playlist_id", playlist_id)
230 if not xmlfeed:
231 return "not found or something", 404 # XXX
232 title, author, videos, _, _ = parse_xml(xmlfeed)
233 return render_template('xmlfeed.html.j2', title=f"{title} by {author}", rows=videos)
234
235 @frontend.route('/api/timedtext')
236 def timedtext():
237 r = requests.get("https://www.youtube.com/api/timedtext", request.args.to_dict())
238 # Note: in srv1 format, xmlentities are double-encoded m( a smart quote is
239 # even worse: it's '&amp;39;<smartquote>' wtf!? (at least vvt seems ok)
240 if not r.ok:
241 return "error: {r.text}", 400 # TODO: better
242 retval = r.text
243 if request.args.get('fmt') == 'vtt' and request.args.get('kind') == 'asr':
244 # autocaptions are extremely confusing, and stuck in the lower-left corner. fix it up a bit
245 retval = re.sub(r"<.+?>", "", retval) # remove inline html-like markup that times each word/adds styles
246 retval = retval.replace("align:start position:0%", "") # let browser position the text itself
247 # each subtitle-line is repeated twice (first on the lower line, then
248 # on the next "frame" on the upper line). we want to remove the
249 # repetition, as that's confusing without word and line animations:
250 lines = retval.split('\n')
251 retval = '\n'.join([line for line, prev in zip(lines, ['']+lines) if not " --> " in prev])
252 return retval, {'Content-Type': r.headers.get("Content-Type")}
253
254 @frontend.route('/manage/subscriptions')
255 # disabled for guest user: @login_required
256 def subscription_manager():
257 if current_user.is_anonymous:
258 token = 'guest'
259 else:
260 token = current_user.token
261 with sqlite3.connect(cf['global']['database']) as conn:
262 #with conn.cursor() as c:
263 c = conn.cursor()
264 c.execute("""
265 SELECT subscriptions.channel_id, name, type,
266 (subscribed_until < datetime('now')) AS obsolete
267 FROM subscriptions
268 LEFT JOIN (SELECT name, id FROM channels
269 UNION
270 SELECT name, id FROM playlists
271 ) AS channels ON channels.id = subscriptions.channel_id
272 left JOIN websub ON channels.id = websub.channel_id
273 WHERE user = ?
274 AND subscriptions.type IN ('channel', 'playlist')
275 ORDER BY obsolete=0, name COLLATE NOCASE ASC""", (token,))
276 rows = [{
277 'channel_id': channel_id,
278 'author': author or channel_id,
279 'type': type,
280 'subscribed_until': subscribed_until
281 } for (channel_id, author, type, subscribed_until) in c.fetchall()]
282 return render_template('subscription_manager.html.j2', rows=rows)
283
284 @frontend.route('/feed/subscriptions', methods=['POST'])
285 @login_required
286 def feed_post():
287 token = current_user.token
288 action = next(request.form.keys(), None)
289 if action in ['pin', 'unpin', 'hide']:
290 video_id = request.form.get(action)
291 display = {
292 'pin': 'pinned',
293 'unpin': None,
294 'hide': 'hidden',
295 }[action]
296 with sqlite3.connect(cf['global']['database']) as conn:
297 c = conn.cursor()
298 store_video_metadata(video_id) # only needed for pinning
299 c.execute("""
300 INSERT OR REPLACE INTO flags (user, video_id, display)
301 VALUES (?, ?, ?)
302 """, (token, video_id, display))
303 else:
304 flash("unsupported action", "error")
305 return redirect(request.url, code=303)
306
307 @frontend.route('/manage/subscriptions', methods=['POST'])
308 @login_required
309 def manage_subscriptions():
310 token = current_user.token
311 if 'subscribe' in request.form:
312 some_id = request.form.get("subscribe")
313 match = re.search(r"(UC[A-Za-z0-9_-]{22})", some_id)
314 if match:
315 some_id = match.group(1)
316 id_type = "channel"
317 else:
318 match = re.search(r"((?:PL|LL|EC|UU|FL|UL|OL)[A-Za-z0-9_-]{10,})", some_id)
319 if match: # NOTE: PL-playlists are 32chars, others differ in length.
320 some_id = match.group(1)
321 id_type = "playlist"
322 else:
323 flash("not a valid/subscribable URI", "error")
324 return redirect(request.url, code=303)
325 with sqlite3.connect(cf['global']['database']) as conn:
326 #with conn.cursor() as c:
327 c = conn.cursor()
328 c.execute("""
329 INSERT OR IGNORE INTO subscriptions (user, channel_id, type)
330 VALUES (?, ?, ?)
331 """, (token, some_id, id_type))
332 # TODO: sql-error-handling, asynchronically calling update-subs.pl
333
334 elif 'unsubscribe' in request.form:
335 some_id = request.form.get("unsubscribe")
336 with sqlite3.connect(cf['global']['database']) as conn:
337 #with conn.cursor() as c:
338 c = conn.cursor()
339 c.execute("""
340 DELETE FROM subscriptions
341 WHERE user = ? AND channel_id = ?
342 """, (token, some_id))
343 # TODO: sql-error-handling, report success
344
345 else:
346 flash("unsupported action", "error")
347
348 return redirect(request.url, code=303)
349
350 def get_cipher():
351 # reload cipher from database every 1 hour
352 if 'cipher' not in g or time.time() - g.get('cipher_updated', 0) > 1 * 60 * 60:
353 with sqlite3.connect(cf['global']['database']) as conn:
354 c = conn.cursor()
355 c.execute("SELECT sts, algorithm FROM cipher")
356 g.cipher = c.fetchone()
357 g.cipher_updated = time.time()
358
359 return g.cipher
360
361 #@frontend.teardown_appcontext
362 #def teardown_db():
363 # db = g.pop('db', None)
364 #
365 # if db is not None:
366 # db.close()
367
368
369 @frontend.app_template_filter('format_date')
370 def format_date(s):
371 import datetime # can't import at top level, because it is inherited from common
372 (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'
373 M = '_ Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec'.split()
374 if y == datetime.datetime.now().year:
375 return f"{d} {M[m]}"
376 else:
377 return f"{M[m]} '{y%100}"
378
379 @frontend.app_template_filter('format_time')
380 def format_time(i):
381 if i is None:
382 return None
383 h = i // (60*60)
384 m = i // 60 % 60
385 s = i % 60
386 return '%d:%02d:%02d' % (h,m,s) if h else '%02d:%02d' % (m,s)
387
388 @frontend.app_template_filter('timeoffset')
389 def timeoffset(s):
390 if s is None:
391 return None
392 match = re.match(r"^(\d+)s?$", s) # e.g. 2040s
393 if match:
394 return match.group(1)
395 match = re.match(r"^(?:(\d+)h)?(?:(\d+)m)?(?:(\d+)s)?$", s) # e.g. 34m, 1h23s
396 if match:
397 return ":".join([n.zfill(2) for n in match.groups('0')])
398 return None
Imprint / Impressum