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, current_app
   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 = request.args.get('page', 0, type=int)
  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     if len(request.args.get('v')) != 11:
  72         return "malformed video id", 400
  73
  74     plaintextheaders = {
  75         'content-type': 'text/plain',
  76         'Link': "<data:text/css,body%7Bcolor:%23eee;background:%23333%7D>; rel=stylesheet;"
  77     }
  78
  79     video_id = request.args.get('v')
  80     sts, algo = get_cipher()
  81     video_url, stream_map, metadata, error, errdetails = get_video_info(video_id, sts, algo)
  82
  83     extra = {'geolocked':'local=1', 'livestream':'raw=0'}.get(error,'')
  84     invidious_url = f"https://invidious.snopyta.org/watch?v={video_id}&{extra}"
  85     errdetails = {
  86         'banned':     "Instance is being rate limited.",
  87         'malformed':  "Video ID is invalid.",
  88         'geolocked':  "This video is geolocked.",
  89         'livestream': "Livestreams not supported on this instance.",
  90         'agegated':   "Unable to bypass age-restriction.",
  91         'exhausted':  errdetails or "Couldn't extract video URLs.",
  92         'player':     errdetails,
  93     }.get(error)
  94
  95     # if the video is geolocked, and the proxy is enabled, we can still play
  96     # it, if the video is available in the instance server's region:
  97     if error == 'geolocked' and video_url and 'proxy' in current_app.blueprints.keys():
  98         videoplayback = url_for('proxy.videoplayback')
  99         query = urlparse(video_url).query
 100         video_url = f"{videoplayback}?{query}"
 101         for s in stream_map['adaptive']:
 102             query = urlparse(s['url']).query
 103             s['url'] = f"{videoplayback}?{query}"
 104         for s in stream_map['muxed']:
 105             query = urlparse(s['url']).query
 106             s['url'] = f"{videoplayback}?{query}"
 107         error = None
 108
 109     # if the proxy is enabled, we can also play livestreams:
 110     if error == 'livestream' and 'proxy' in current_app.blueprints.keys():
 111         # Note: hlsManifestUrl's hostname will be replaced client-side
 112         video_url = stream_map['hlsManifestUrl']
 113         error = None
 114
 115     # if the instance is blocked, try submitting a job to the anti captcha service:
 116     if error == 'banned' and cf['captcha']['api_key']:
 117         r2 = requests.get(f'https://www.youtube.com/watch?v={video_id}&hl=en&gl=US')
 118         status = submit_captcha(r2)
 119         if status is False:
 120             raise Exception("we are banned, but captcha wasn't triggered!")
 121         else:
 122             message = "right now" if status is True else f"{int(status)} seconds ago"
 123             raise BadGateway(f"""
 124                 {errdetails} An attempt at getting unblocked has been made {message}.
 125                 Please try again in 30 seconds.
 126             """)
 127
 128     show = request.args.get("show")
 129     if show == "raw":
 130         if error:
 131             msg = errdetails if error=='player' else f"{error.upper()}: {errdetails}"
 132             return f"{msg}\n\nRedirecting to Invidious.", 502, {
 133                 'Refresh': f'2; URL={invidious_url}&raw=1',
 134                 **plaintextheaders}
 135         return redirect(video_url, code=307)
 136     elif show == "json":
 137         if error and not metadata:
 138             return {'error': True, error: errdetails}, 400 # TODO: better (test _CpR4o81XQc)
 139         return jsonify(metadata)
 140     elif show == "audio":
 141         # sorting: we want to prioritize mp4a over opus, and sort by highest quality first
 142         # todo: geolocking; prefer open format?
 143         if error and not stream_map:
 144             msg = errdetails if error=='player' else f"{error.upper()}: {errdetails}"
 145             return msg, 400, plaintextheaders # TODO: nicer
 146         stream = next(iter(sorted([
 147                     stream for stream in stream_map['adaptive']
 148                     if stream['mimeType'].startswith('audio/')],
 149                 key=lambda stream: ('opus' not in stream['mimeType'],
 150                     stream['bitrate']), reverse=True)
 151             ),{}).get('url')
 152         return redirect(stream)
 153     else:
 154         if error and not metadata: # e.g. malformed, private/deleted video, ...
 155             return render_template('video-error.html.j2', video_id=video_id,
 156                 video_error=error, errdetails=errdetails, invidious_url=invidious_url)
 157         meta = prepare_metadata(metadata)
 158         with sqlite3.connect(cf['global']['database']) as conn:
 159             c = conn.cursor()
 160             c.execute("""
 161             SELECT COUNT((
 162                 SELECT 1 FROM subscriptions WHERE channel_id = ? AND user = ?
 163             )), COUNT((
 164                 SELECT 1 FROM flags WHERE video_id = ? AND display = 'pinned' AND user = ?
 165             ))""", (meta['channel_id'], token, video_id, token))
 166             (is_subscribed, is_pinned) = c.fetchone()
 167             return render_template('watch.html.j2',
 168                 video_id=video_id, video_url=video_url, stream_map=stream_map,
 169                 video_error=error, errdetails=errdetails, invidious_url=invidious_url,
 170                 is_pinned=is_pinned, is_subscribed=is_subscribed,
 171                 **meta)
 172
 173 @frontend.route('/embed/videoseries')
 174 def embed_videoseries():
 175     return redirect(url_for('.playlist', list=request.args.get('list')))
 176 @frontend.route('/embed/<video_id>', strict_slashes=False)
 177 def embed(video_id):
 178     if video_id == "videoseries":
 179         return redirect(url_for('.playlist', list=request.args.get('list')))
 180
 181     return redirect(url_for('.watch', v=video_id, t=request.args.get('start')))
 182
 183 @frontend.route('/<something>', strict_slashes=False)
 184 def plain_user_or_video(something):
 185     # yt.com interprets this as a username, but we also want to catch youtu.be
 186     # short-urls. so we check if it's a channel by querying the RSS feed (this
 187     # shoudn't be rate-limited); if that fails, check if it looks like a video
 188     # id; or finally give up.
 189     if '.' not in something and channel_exists(something):
 190         # periods are not valid in usernames, vanity urls or ucids, but common
 191         # in urls that get crawled by bots (e.g. index.php). failing early
 192         # reduces the amount of invalid channel names getting looked up.
 193         return redirect(url_for('.channel', channel_id=something))
 194     elif re.match(r"^[-_0-9A-Za-z]{11}$", something): # looks like a video id
 195         return redirect(url_for('.watch', v=something, t=request.args.get('t')))
 196     else: # ¯\_(ツ)_/¯
 197         # XXX: something == 'thethoughtemporium' -> 404s
 198         raise NotFound("Note: some usernames not recognized; try searching it")
 199
 200 @frontend.route('/channel/<channel_id>/<subpage>')
 201 @frontend.route('/user/<channel_id>/<subpage>')
 202 @frontend.route('/c/<channel_id>/<subpage>')
 203 @frontend.route('/channel/<channel_id>/')
 204 @frontend.route('/user/<channel_id>/')
 205 @frontend.route('/c/<channel_id>/')
 206 def channel(channel_id, _=None):
 207     token = getattr(current_user, 'token', 'guest')
 208
 209     if re.match(r"(UC[A-Za-z0-9_-]{22})", channel_id):
 210         xmlfeed = fetch_xml("channel_id", channel_id)
 211     else:
 212         xmlfeed = fetch_xml("user", channel_id)
 213
 214     if not xmlfeed:
 215         return "not found or something", 404 # XXX
 216     title, author, videos, channel_id, _ = parse_xml(xmlfeed)
 217
 218     with sqlite3.connect(cf['global']['database']) as conn:
 219         c = conn.cursor()
 220         c.execute("""
 221             SELECT COUNT(*)
 222               FROM subscriptions
 223              WHERE channel_id = ? AND user = ?
 224         """, (channel_id, token))
 225         (is_subscribed,) = c.fetchone()
 226
 227     return render_template('xmlfeed.html.j2', title=author, rows=videos,
 228         is_subscribed=is_subscribed, channel_id=channel_id)
 229
 230 @frontend.route('/playlist')
 231 def playlist():
 232     playlist_id = request.args.get('list')
 233     if not playlist_id:
 234         return "bad list id", 400 # todo
 235
 236     xmlfeed = fetch_xml("playlist_id", playlist_id)
 237     if not xmlfeed:
 238         return "not found or something", 404 # XXX
 239     title, author, videos, _, _ = parse_xml(xmlfeed)
 240     return render_template('xmlfeed.html.j2', title=f"{title} by {author}", rows=videos)
 241
 242 @frontend.route('/api/timedtext')
 243 def timedtext():
 244     r = requests.get("https://www.youtube.com/api/timedtext", request.args.to_dict())
 245     # Note: in srv1 format, xmlentities are double-encoded m( a smart quote is
 246     # even worse: it's '&amp;39;<smartquote>' wtf!? (at least vvt seems ok)
 247     if not r.ok:
 248         return "error: {r.text}", 400 # TODO: better
 249     retval = r.text
 250     if request.args.get('fmt') == 'vtt' and request.args.get('kind') == 'asr':
 251         # autocaptions are extremely confusing, and stuck in the lower-left corner. fix it up a bit
 252         retval = re.sub(r"<.+?>", "", retval) # remove inline html-like markup that times each word/adds styles
 253         retval = retval.replace("align:start position:0%", "") # let browser position the text itself
 254         # each subtitle-line is repeated twice (first on the lower line, then
 255         # on the next "frame" on the upper line). we want to remove the
 256         # repetition, as that's confusing without word and line animations:
 257         lines = retval.split('\n')
 258         retval = '\n'.join([line for line, prev in zip(lines, ['']+lines) if not " --> " in prev])
 259     return retval, {'Content-Type': r.headers.get("Content-Type")}
 260
 261 @frontend.route('/manage/subscriptions')
 262 # disabled for guest user: @login_required
 263 def subscription_manager():
 264     if current_user.is_anonymous:
 265         token = 'guest'
 266     else:
 267         token = current_user.token
 268     with sqlite3.connect(cf['global']['database']) as conn:
 269         #with conn.cursor() as c:
 270             c = conn.cursor()
 271             c.execute("""
 272                   SELECT subscriptions.channel_id, name, type,
 273                          (subscribed_until < datetime('now')) AS obsolete
 274                     FROM subscriptions
 275                     LEFT JOIN (SELECT name, id FROM channels
 276                        UNION
 277                        SELECT name, id FROM playlists
 278                       ) AS channels ON channels.id = subscriptions.channel_id
 279                     left JOIN websub ON channels.id = websub.channel_id
 280                    WHERE user = ?
 281                      AND subscriptions.type IN ('channel', 'playlist')
 282                 ORDER BY obsolete=0, name COLLATE NOCASE ASC""", (token,))
 283             rows = [{
 284                 'channel_id': channel_id,
 285                 'author': author or channel_id,
 286                 'type': type,
 287                 'subscribed_until': subscribed_until
 288             } for (channel_id, author, type, subscribed_until) in c.fetchall()]
 289     return render_template('subscription_manager.html.j2', rows=rows)
 290
 291 @frontend.route('/feed/subscriptions', methods=['POST'])
 292 @login_required
 293 def feed_post():
 294     token = current_user.token
 295     action = next(request.form.keys(), None)
 296     if action in ['pin', 'unpin', 'hide', 'unhide']:
 297         video_id = request.form.get(action)
 298         display = {
 299             'pin': 'pinned',
 300             'unpin': None,
 301             'hide': 'hidden',
 302             'unhide': None,
 303         }[action]
 304         with sqlite3.connect(cf['global']['database']) as conn:
 305             c = conn.cursor()
 306             store_video_metadata(video_id) # only needed for pinning
 307             c.execute("""
 308                 INSERT OR REPLACE INTO flags (user, video_id, display)
 309                 VALUES (?, ?, ?)
 310             """, (token, video_id, display))
 311         undo_flash(video_id, action)
 312     else:
 313         flash("unsupported action", "error")
 314     return redirect(request.url, code=303)
 315
 316 @frontend.route('/manage/subscriptions', methods=['POST'])
 317 @login_required
 318 def manage_subscriptions():
 319     token = current_user.token
 320     if 'subscribe' in request.form:
 321         some_id = request.form.get("subscribe")
 322         match = re.search(r"(UC[A-Za-z0-9_-]{22})", some_id)
 323         if match:
 324             some_id = match.group(1)
 325             id_type = "channel"
 326         else:
 327             match = re.search(r"((?:PL|LL|EC|UU|FL|UL|OL)[A-Za-z0-9_-]{10,})", some_id)
 328             if match:  # NOTE: PL-playlists are 32chars, others differ in length.
 329                 some_id = match.group(1)
 330                 id_type = "playlist"
 331             else:
 332                 flash("not a valid/subscribable URI", "error")
 333                 return redirect(request.url, code=303)
 334         with sqlite3.connect(cf['global']['database']) as conn:
 335             #with conn.cursor() as c:
 336                 c = conn.cursor()
 337                 c.execute("""
 338                         INSERT OR IGNORE INTO subscriptions (user, channel_id, type)
 339                         VALUES (?, ?, ?)
 340                 """, (token, some_id, id_type))
 341                 # TODO: sql-error-handling, asynchronically calling update-subs.pl
 342         undo_flash(some_id, 'subscribe')
 343
 344     elif 'unsubscribe' in request.form:
 345         some_id = request.form.get("unsubscribe")
 346         with sqlite3.connect(cf['global']['database']) as conn:
 347             #with conn.cursor() as c:
 348                 c = conn.cursor()
 349                 c.execute("""
 350                         DELETE FROM subscriptions
 351                         WHERE user = ? AND channel_id = ?
 352                 """, (token, some_id))
 353                 # TODO: sql-error-handling, report success
 354         undo_flash(some_id, 'unsubscribe')
 355
 356     else:
 357         flash("unsupported action", "error")
 358
 359     return redirect(request.url, code=303)
 360
 361 def get_cipher():
 362     # reload cipher from database every 1 hour
 363     if 'cipher' not in g or time.time() - g.get('cipher_updated', 0) > 1 * 60 * 60:
 364         with sqlite3.connect(cf['global']['database']) as conn:
 365             c = conn.cursor()
 366             c.execute("SELECT sts, algorithm FROM cipher")
 367             g.cipher = c.fetchone()
 368             g.cipher_updated = time.time()
 369
 370     return g.cipher
 371
 372 #@frontend.teardown_appcontext
 373 #def teardown_db():
 374 #    db = g.pop('db', None)
 375 #
 376 #    if db is not None:
 377 #        db.close()
 378
 379 def undo_flash(thing_id, action):
 380     undo_action, past_action = {
 381         'pin': ('unpin', 'pinned'),
 382         'unpin': ('pin', 'unpinned'),
 383         'hide': ('unhide', 'hidden'),
 384         'unhide': ('hide', 'unhidden'),
 385         'subscribe': ('unsubscribe', 'subscribed'),
 386         'unsubscribe': ('subscribe', 'unsubscribed'),
 387     }.get(action)
 388     if 'subscribe' in action and thing_id.startswith('UC'):
 389         thing = "channel"
 390         thing_url = url_for('.channel', channel_id=thing_id)
 391     elif 'subscribe' in action:
 392         thing = "playlist"
 393         thing_url = url_for('.playlist', playlist_id=thing_id)
 394     else:
 395         thing = "video"
 396         thing_url = url_for('.watch', v=thing_id)
 397     flash(f'''
 398         <form method=post><input type=hidden name="{undo_action}" value="{thing_id}">
 399         <a href="{thing_url}">{thing}</a> {past_action}.
 400         <label><input type="submit" hidden>
 401         <span style="text-decoration:underline;cursor:pointer">undo</span>.
 402         </label></form>''', "info")
 403
 404 @frontend.app_template_filter('format_date')
 405 def format_date(s):
 406     import datetime # can't import at top level, because it is inherited from common
 407     (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'
 408     M = '_ Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec'.split()
 409     if y == datetime.datetime.now().year:
 410         return f"{d} {M[m]}"
 411     else:
 412         return f"{M[m]} '{y%100}"
 413
 414 @frontend.app_template_filter('format_time')
 415 def format_time(i):
 416     if i is None:
 417         return None
 418     h = i // (60*60)
 419     m = i // 60 % 60
 420     s = i % 60
 421     return '%d:%02d:%02d' % (h,m,s) if h else '%02d:%02d' % (m,s)
 422
 423 @frontend.app_template_filter('timeoffset')
 424 def timeoffset(s):
 425     if s is None:
 426         return None
 427     match = re.match(r"^(\d+)s?$", s)  # e.g. 2040s
 428     if match:
 429         return match.group(1)
 430     match = re.match(r"^(?:(\d+)h)?(?:(\d+)m)?(?:(\d+)s)?$", s)  # e.g. 34m, 1h23s
 431     if match:
 432         return ":".join([n.zfill(2) for n in match.groups('0')])
 433     return None