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, premiere, 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             'premiere': premiere and (# only if it hasn't yet premiered:
  57                 datetime.strptime(published+'+0000', "%Y-%m-%d %H:%M:%S%z")>datetime.now(tz=timezone.utc)
  58             ),
  59             'published': published,
  60             'playlist': playlist,
  61             'pinned': display == 'pinned',
  62         } for (video_id, channel_id, author, title, length, livestream, premiere, published, playlist, display) in c.fetchall()]
  63     return render_template('index.html.j2', rows=rows, page=page)
  64
  65 @frontend.route('/watch')
  66 def watch():
  67     if current_user.is_anonymous:
  68         token = 'guest'
  69     else:
  70         token = current_user.token
  71
  72     if not 'v' in request.args:
  73         return "missing video id", 400
  74     if len(request.args.get('v')) != 11:
  75         return "malformed video id", 400
  76
  77     plaintextheaders = {
  78         'content-type': 'text/plain',
  79         'Link': "<data:text/css,body%7Bcolor:%23eee;background:%23333%7D>; rel=stylesheet;"
  80     }
  81
  82     video_id = request.args.get('v')
  83     show = request.args.get("show")
  84     video_url, stream_map, metadata, error, errdetails = get_video_info(video_id, metaOnly=(show == 'meta'))
  85
  86     proxy_on = getattr(g, 'proxy_on', None)
  87     proxy_msg = {
  88         False: "Log in to use proxy.", # proxy for logged-in users only
  89         None: "", # proxy disabled globally
  90         True: "", # proxy is available to this user
  91     }[proxy_on]
  92     extra = {'geolocked':'local=1', 'livestream':'raw=0'}.get(error,'')
  93     invidious_url = f"https://invidious.snopyta.org/watch?v={video_id}&{extra}"
  94     errdetails = {
  95         'banned':     "Instance is being rate limited.",
  96         'malformed':  "Video ID is invalid.",
  97         'geolocked':  f"This video is geolocked. {proxy_msg}",
  98         'livestream': f"Livestreams disabled. {proxy_msg}",
  99         'agegated':   "Unable to bypass age-restriction.",
 100         'no-url':     "No muxed stream available.",
 101         'exhausted':  errdetails or "Couldn't extract video URLs.",
 102         'player':     errdetails,
 103     }.get(error, error)
 104
 105     # if the video is geolocked, and the proxy is enabled, we can still play
 106     # it, if the video is available in the instance server's region:
 107     if error == 'geolocked' and video_url and proxy_on:
 108         videoplayback = url_for('proxy.videoplayback')
 109         query = urlparse(video_url).query
 110         video_url = f"{videoplayback}?{query}"
 111         for t in ('adaptive_video', 'adaptive_audio', 'muxed'):
 112             for s in stream_map[t]:
 113                 query = urlparse(s['url']).query
 114                 s['url'] = f"{videoplayback}?{query}"
 115         error = None
 116
 117     # if the proxy is enabled, we can also play livestreams:
 118     if error == 'livestream' and proxy_on:
 119         # Note: hlsManifestUrl's hostname will be replaced client-side
 120         video_url = stream_map['hlsManifestUrl']
 121         error = None
 122
 123     # if the instance is blocked, try submitting a job to the anti captcha service:
 124     if error == 'banned' and cf['captcha']['api_key']:
 125         r2 = requests.get(f'https://www.youtube.com/watch?v={video_id}&hl=en&gl=US')
 126         status = submit_captcha(r2)
 127         if status is False:
 128             raise Exception("we are banned, but captcha wasn't triggered!")
 129         else:
 130             message = "right now" if status is True else f"{int(status)} seconds ago"
 131             raise BadGateway(f"""
 132                 {errdetails} An attempt at getting unblocked has been made {message}.
 133                 Please try again in 30 seconds.
 134             """)
 135
 136     if show == "raw":
 137         if error:
 138             msg = errdetails if error=='player' else f"{error.upper()}: {errdetails}"
 139             return f"{msg}\n\nRedirecting to Invidious.", 502, {
 140                 'Refresh': f'2; URL={invidious_url}&raw=1',
 141                 **plaintextheaders}
 142         return redirect(video_url, code=307)
 143     elif show == "json":
 144         if error and not metadata:
 145             return {'error': True, error: errdetails}, 400 # TODO: better (test _CpR4o81XQc)
 146         return jsonify(metadata)
 147     elif show == "audio":
 148         # sorting: we want to prioritize mp4a over opus, and sort by highest quality first
 149         # todo: geolocking; prefer open format?
 150         if error and not stream_map:
 151             msg = errdetails if error=='player' else f"{error.upper()}: {errdetails}"
 152             return msg, 400, plaintextheaders # TODO: nicer
 153         stream = next(iter(sorted(
 154             stream_map['adaptive_audio'],
 155             key=lambda e: ('opus' not in e['mimeType'], e['bitrate']),
 156             reverse=True
 157         )),{}).get('url')
 158         return redirect(stream)
 159     elif show == "meta":
 160         # this is the subset of (useful) keys that are not present in the
 161         # Android API response. the special key '_' contains ready-to-use
 162         # parsed versions of that data.
 163         if error and not metadata:
 164             return {'error': True, error: errdetails}, 400 # TODO: better (test _CpR4o81XQc)
 165         parsed = microformat_parser(metadata)
 166         return {'microformat': metadata.get('microformat'),'cards':metadata.get('cards'), '_':parsed}
 167     else:
 168         if error and not metadata: # e.g. malformed, private/deleted video, ...
 169             return render_template('video-error.html.j2', video_id=video_id,
 170                 video_error=error, errdetails=errdetails, invidious_url=invidious_url)
 171         meta = prepare_metadata(metadata)
 172         with sqlite3.connect(cf['global']['database']) as conn:
 173             c = conn.cursor()
 174             c.execute("""
 175             SELECT COUNT((
 176                 SELECT 1 FROM subscriptions WHERE channel_id = ? AND user = ?
 177             )), COUNT((
 178                 SELECT 1 FROM flags WHERE video_id = ? AND display = 'pinned' AND user = ?
 179             ))""", (meta['channel_id'], token, video_id, token))
 180             (is_subscribed, is_pinned) = c.fetchone()
 181             return render_template('watch.html.j2',
 182                 video_id=video_id, video_url=video_url, stream_map=stream_map,
 183                 video_error=error, errdetails=errdetails, invidious_url=invidious_url,
 184                 is_pinned=is_pinned, is_subscribed=is_subscribed,
 185                 **meta)
 186
 187 @frontend.route('/embed/videoseries')
 188 def embed_videoseries():
 189     return redirect(url_for('.playlist', list=request.args.get('list')))
 190 @frontend.route('/embed/<video_id>', strict_slashes=False)
 191 def embed(video_id):
 192     if video_id == "videoseries":
 193         return redirect(url_for('.playlist', list=request.args.get('list')))
 194
 195     return redirect(url_for('.watch', v=video_id, t=request.args.get('start')))
 196
 197 @frontend.route('/<something>', strict_slashes=False)
 198 def plain_user_or_video(something):
 199     # yt.com interprets this as a username, but we also want to catch youtu.be
 200     # short-urls. so we check if it's a channel by querying the RSS feed (this
 201     # shoudn't be rate-limited); if that fails, check if it looks like a video
 202     # id; or finally give up.
 203     if '.' not in something and channel_exists(something):
 204         # periods are not valid in usernames, vanity urls or ucids, but common
 205         # in urls that get crawled by bots (e.g. index.php). failing early
 206         # reduces the amount of invalid channel names getting looked up.
 207         return redirect(url_for('.channel', channel_id=something))
 208     elif re.match(r"^[-_0-9A-Za-z]{11}$", something): # looks like a video id
 209         return redirect(url_for('.watch', v=something, t=request.args.get('t')))
 210     else: # ¯\_(ツ)_/¯
 211         # XXX: something == 'thethoughtemporium' -> 404s
 212         raise NotFound("Note: some usernames not recognized; try searching it")
 213
 214 @frontend.route('/c/<channel_id>/<subpage>')
 215 @frontend.route('/c/<channel_id>/')
 216 @frontend.route('/user/<channel_id>/<subpage>')
 217 @frontend.route('/user/<channel_id>/')
 218 def channel_redirect(channel_id, subpage=None):
 219     # Note: we can't check /c/, so we have to assume it is the same as /user/,
 220     # which is sometimes wrong.
 221     xmlfeed = fetch_xml("user", channel_id)
 222
 223     if not xmlfeed:
 224         raise NotFound("unknown channel name")
 225
 226     _, _, _, channel_id, _ = parse_xml(xmlfeed)
 227
 228     return redirect(url_for('.channel', channel_id=channel_id))
 229
 230 @frontend.route('/channel/<channel_id>/<subpage>')
 231 @frontend.route('/channel/<channel_id>/')
 232 def channel(channel_id, _=None):
 233     token = getattr(current_user, 'token', 'guest')
 234     sort = request.args.get("sort", "newest")
 235
 236     if not re.match(r"(UC[A-Za-z0-9_-]{22})", channel_id):
 237         # canonicalize channel id, otherwise popular won't work
 238         return redirect(url_for('.channel_redirect', channel_id=channel_id))
 239
 240     if sort == "popular":
 241         xmlfeed = fetch_xml("playlist_id", f"PU{channel_id[2:]}")
 242     else:
 243         xmlfeed = fetch_xml("channel_id", channel_id)
 244         #^note: could also use playlist_id=UU...
 245
 246     if not xmlfeed:
 247         raise NotFound("unknown channel id")
 248
 249     title, author, videos, channel_id, _ = parse_xml(xmlfeed)
 250
 251     with sqlite3.connect(cf['global']['database']) as conn:
 252         c = conn.cursor()
 253         c.execute("""
 254             SELECT COUNT(*)
 255               FROM subscriptions
 256              WHERE channel_id = ? AND user = ?
 257         """, (channel_id, token))
 258         (is_subscribed,) = c.fetchone()
 259
 260     return render_template('xmlfeed.html.j2', title=author, rows=videos,
 261         is_subscribed=is_subscribed, channel_id=channel_id)
 262
 263 @frontend.route('/playlist')
 264 def playlist():
 265     playlist_id = request.args.get('list')
 266     if not playlist_id:
 267         return "bad list id", 400 # todo
 268
 269     xmlfeed = fetch_xml("playlist_id", playlist_id)
 270     if not xmlfeed:
 271         return "not found or something", 404 # XXX
 272     title, author, videos, _, _ = parse_xml(xmlfeed)
 273     return render_template('xmlfeed.html.j2', title=f"{title} by {author}", rows=videos)
 274
 275 @frontend.route('/api/timedtext')
 276 def timedtext():
 277     r = requests.get("https://www.youtube.com/api/timedtext", request.args.to_dict())
 278     # Note: in srv1 format, xmlentities are double-encoded m( a smart quote is
 279     # even worse: it's '&amp;39;<smartquote>' wtf!? (at least vvt seems ok)
 280     if not r.ok:
 281         return "error: {r.text}", 400 # TODO: better
 282     retval = r.text
 283     if request.args.get('fmt') == 'vtt' and request.args.get('kind') == 'asr':
 284         # autocaptions are extremely confusing, and stuck in the lower-left corner. fix it up a bit
 285         retval = re.sub(r"<.+?>", "", retval) # remove inline html-like markup that times each word/adds styles
 286         retval = retval.replace("align:start position:0%", "") # let browser position the text itself
 287         # each subtitle-line is repeated twice (first on the lower line, then
 288         # on the next "frame" on the upper line). we want to remove the
 289         # repetition, as that's confusing without word and line animations:
 290         lines = retval.split('\n')
 291         retval = '\n'.join([line for line, prev in zip(lines, ['']+lines) if not " --> " in prev])
 292     return retval, {'Content-Type': r.headers.get("Content-Type")}
 293
 294 @frontend.route('/manage/subscriptions')
 295 # disabled for guest user: @login_required
 296 def subscription_manager():
 297     if current_user.is_anonymous:
 298         token = 'guest'
 299     else:
 300         token = current_user.token
 301     with sqlite3.connect(cf['global']['database']) as conn:
 302         #with conn.cursor() as c:
 303             c = conn.cursor()
 304             c.execute("""
 305                   SELECT subscriptions.channel_id, name, type,
 306                          (subscribed_until < datetime('now')) AS obsolete
 307                     FROM subscriptions
 308                     LEFT JOIN (SELECT name, id FROM channels
 309                        UNION
 310                        SELECT name, id FROM playlists
 311                       ) AS channels ON channels.id = subscriptions.channel_id
 312                     left JOIN websub ON channels.id = websub.channel_id
 313                    WHERE user = ?
 314                      AND subscriptions.type IN ('channel', 'playlist')
 315                 ORDER BY obsolete=0, name COLLATE NOCASE ASC""", (token,))
 316             rows = [{
 317                 'channel_id': channel_id,
 318                 'author': author or channel_id,
 319                 'type': type,
 320                 'subscribed_until': subscribed_until
 321             } for (channel_id, author, type, subscribed_until) in c.fetchall()]
 322     return render_template('subscription_manager.html.j2', rows=rows)
 323
 324 @frontend.route('/feed/subscriptions', methods=['POST'])
 325 @login_required
 326 def feed_post():
 327     token = current_user.token
 328     action = next(request.form.keys(), None)
 329     if action in ['pin', 'unpin', 'hide', 'unhide']:
 330         video_id = request.form.get(action)
 331         display = {
 332             'pin': 'pinned',
 333             'unpin': None,
 334             'hide': 'hidden',
 335             'unhide': None,
 336         }[action]
 337         with sqlite3.connect(cf['global']['database']) as conn:
 338             c = conn.cursor()
 339             store_video_metadata(video_id) # only needed for pinning
 340             c.execute("""
 341                 INSERT OR REPLACE INTO flags (user, video_id, display)
 342                 VALUES (?, ?, ?)
 343             """, (token, video_id, display))
 344         undo_flash(video_id, action)
 345     else:
 346         flash("unsupported action", "error")
 347     return redirect(request.url, code=303)
 348
 349 @frontend.route('/manage/subscriptions', methods=['POST'])
 350 @login_required
 351 def manage_subscriptions():
 352     token = current_user.token
 353     if 'subscribe' in request.form:
 354         some_id = request.form.get("subscribe")
 355         match = re.search(r"(UC[A-Za-z0-9_-]{22})", some_id)
 356         if match:
 357             some_id = match.group(1)
 358             id_type = "channel"
 359         else:
 360             match = re.search(r"((?:PL|LL|EC|UU|FL|UL|OL)[A-Za-z0-9_-]{10,})", some_id)
 361             if match:  # NOTE: PL-playlists are 32chars, others differ in length.
 362                 some_id = match.group(1)
 363                 id_type = "playlist"
 364             else:
 365                 flash("not a valid/subscribable URI", "error")
 366                 return redirect(request.url, code=303)
 367         with sqlite3.connect(cf['global']['database']) as conn:
 368             #with conn.cursor() as c:
 369                 c = conn.cursor()
 370                 c.execute("""
 371                         INSERT OR IGNORE INTO subscriptions (user, channel_id, type)
 372                         VALUES (?, ?, ?)
 373                 """, (token, some_id, id_type))
 374                 # TODO: sql-error-handling, asynchronically calling update-subs.pl
 375         undo_flash(some_id, 'subscribe')
 376
 377     elif 'unsubscribe' in request.form:
 378         some_id = request.form.get("unsubscribe")
 379         with sqlite3.connect(cf['global']['database']) as conn:
 380             #with conn.cursor() as c:
 381                 c = conn.cursor()
 382                 c.execute("""
 383                         DELETE FROM subscriptions
 384                         WHERE user = ? AND channel_id = ?
 385                 """, (token, some_id))
 386                 # TODO: sql-error-handling, report success
 387         undo_flash(some_id, 'unsubscribe')
 388
 389     else:
 390         flash("unsupported action", "error")
 391
 392     return redirect(request.url, code=303)
 393
 394 @frontend.route('/vi/<vid>/<res>.jpg')
 395 def redirect_thumbnails(vid, res):
 396     return redirect(f"https://i.ytimg.com{request.path}", code=301)
 397
 398 @frontend.record
 399 def redirect_youtube_dot_com(state):
 400     """
 401     This is executed when the blueprint is loaded dynamically builds a number
 402     of routes so that URLs like
 403     https://subscriptions.gir.st/https://www.youtube.com/watch?v=dQw4w9WgXcQ
 404     redirect to the /watch page. Works with /watch, /embed/ and youtu.be short
 405     links, with or without protocl and/or 'www'.
 406     """
 407     def real_redirect_youtube_dot_com(video_id=None):
 408         if not re.match(r"^[-_0-9A-Za-z]{11}$", video_id or ''): video_id = None
 409         if not video_id: video_id = request.args.get('v')
 410         return redirect(url_for('.watch', v=video_id))
 411
 412     for protocol in ("", "http://", "https://"):
 413         for prefix in ("", "www.", "m."):
 414             for domain in ("youtube.com", "youtu.be", "youtube-nocookie.com"):
 415                 for urlpath in ("/watch", "/embed/<video_id>", "/<video_id>"):
 416                     if domain != "youtu.be" and urlpath == "/<video_id>":
 417                         continue # that's a channel, not a video
 418                     frontend.add_url_rule(
 419                         f"/{protocol}{prefix}{domain}{urlpath}",
 420                         view_func=real_redirect_youtube_dot_com,
 421                         strict_slashes=False
 422                     )
 423
 424 def undo_flash(thing_id, action):
 425     undo_action, past_action = {
 426         'pin': ('unpin', 'pinned'),
 427         'unpin': ('pin', 'unpinned'),
 428         'hide': ('unhide', 'hidden'),
 429         'unhide': ('hide', 'unhidden'),
 430         'subscribe': ('unsubscribe', 'subscribed'),
 431         'unsubscribe': ('subscribe', 'unsubscribed'),
 432     }.get(action)
 433     if 'subscribe' in action and thing_id.startswith('UC'):
 434         thing = "channel"
 435         thing_url = url_for('.channel', channel_id=thing_id)
 436     elif 'subscribe' in action:
 437         thing = "playlist"
 438         thing_url = url_for('.playlist', playlist_id=thing_id)
 439     else:
 440         thing = "video"
 441         thing_url = url_for('.watch', v=thing_id)
 442     flash(f'''
 443         <form method=post><input type=hidden name="{undo_action}" value="{thing_id}">
 444         <a href="{thing_url}">{thing}</a> {past_action}.
 445         <label><input type="submit" hidden>
 446         <span style="text-decoration:underline;cursor:pointer">undo</span>.
 447         </label></form>''', "info")
 448
 449 @frontend.app_template_filter('format_date')
 450 def format_date(s):
 451     import datetime # can't import at top level, because it is inherited from common
 452     (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'
 453     M = '_ Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec'.split()
 454     if y == datetime.datetime.now().year:
 455         return f"{d} {M[m]}"
 456     else:
 457         return f"{M[m]} '{y%100}"
 458
 459 @frontend.app_template_filter('format_time')
 460 def format_time(i):
 461     if i is None:
 462         return None
 463     h = i // (60*60)
 464     m = i // 60 % 60
 465     s = i % 60
 466     return '%d:%02d:%02d' % (h,m,s) if h else '%02d:%02d' % (m,s)
 467
 468 @frontend.app_template_filter('timeoffset')
 469 def timeoffset(s):
 470     if s is None:
 471         return None
 472     match = re.match(r"^(\d+)s?$", s)  # e.g. 2040s
 473     if match:
 474         return match.group(1)
 475     match = re.match(r"^(?:(\d+)h)?(?:(\d+)m)?(?:(\d+)s)?$", s)  # e.g. 34m, 1h23s
 476     if match:
 477         return ":".join([n.zfill(2) for n in match.groups('0')])
 478     return None