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('/channel/<channel_id>/<subpage>')
 215 @frontend.route('/user/<channel_id>/<subpage>')
 216 @frontend.route('/c/<channel_id>/<subpage>')
 217 @frontend.route('/channel/<channel_id>/')
 218 @frontend.route('/user/<channel_id>/')
 219 @frontend.route('/c/<channel_id>/')
 220 def channel(channel_id, _=None):
 221     token = getattr(current_user, 'token', 'guest')
 222
 223     if re.match(r"(UC[A-Za-z0-9_-]{22})", channel_id):
 224         xmlfeed = fetch_xml("channel_id", channel_id)
 225     else:
 226         xmlfeed = fetch_xml("user", channel_id)
 227
 228     if not xmlfeed:
 229         return "not found or something", 404 # XXX
 230     title, author, videos, channel_id, _ = parse_xml(xmlfeed)
 231
 232     with sqlite3.connect(cf['global']['database']) as conn:
 233         c = conn.cursor()
 234         c.execute("""
 235             SELECT COUNT(*)
 236               FROM subscriptions
 237              WHERE channel_id = ? AND user = ?
 238         """, (channel_id, token))
 239         (is_subscribed,) = c.fetchone()
 240
 241     return render_template('xmlfeed.html.j2', title=author, rows=videos,
 242         is_subscribed=is_subscribed, channel_id=channel_id)
 243
 244 @frontend.route('/playlist')
 245 def playlist():
 246     playlist_id = request.args.get('list')
 247     if not playlist_id:
 248         return "bad list id", 400 # todo
 249
 250     xmlfeed = fetch_xml("playlist_id", playlist_id)
 251     if not xmlfeed:
 252         return "not found or something", 404 # XXX
 253     title, author, videos, _, _ = parse_xml(xmlfeed)
 254     return render_template('xmlfeed.html.j2', title=f"{title} by {author}", rows=videos)
 255
 256 @frontend.route('/api/timedtext')
 257 def timedtext():
 258     r = requests.get("https://www.youtube.com/api/timedtext", request.args.to_dict())
 259     # Note: in srv1 format, xmlentities are double-encoded m( a smart quote is
 260     # even worse: it's '&amp;39;<smartquote>' wtf!? (at least vvt seems ok)
 261     if not r.ok:
 262         return "error: {r.text}", 400 # TODO: better
 263     retval = r.text
 264     if request.args.get('fmt') == 'vtt' and request.args.get('kind') == 'asr':
 265         # autocaptions are extremely confusing, and stuck in the lower-left corner. fix it up a bit
 266         retval = re.sub(r"<.+?>", "", retval) # remove inline html-like markup that times each word/adds styles
 267         retval = retval.replace("align:start position:0%", "") # let browser position the text itself
 268         # each subtitle-line is repeated twice (first on the lower line, then
 269         # on the next "frame" on the upper line). we want to remove the
 270         # repetition, as that's confusing without word and line animations:
 271         lines = retval.split('\n')
 272         retval = '\n'.join([line for line, prev in zip(lines, ['']+lines) if not " --> " in prev])
 273     return retval, {'Content-Type': r.headers.get("Content-Type")}
 274
 275 @frontend.route('/manage/subscriptions')
 276 # disabled for guest user: @login_required
 277 def subscription_manager():
 278     if current_user.is_anonymous:
 279         token = 'guest'
 280     else:
 281         token = current_user.token
 282     with sqlite3.connect(cf['global']['database']) as conn:
 283         #with conn.cursor() as c:
 284             c = conn.cursor()
 285             c.execute("""
 286                   SELECT subscriptions.channel_id, name, type,
 287                          (subscribed_until < datetime('now')) AS obsolete
 288                     FROM subscriptions
 289                     LEFT JOIN (SELECT name, id FROM channels
 290                        UNION
 291                        SELECT name, id FROM playlists
 292                       ) AS channels ON channels.id = subscriptions.channel_id
 293                     left JOIN websub ON channels.id = websub.channel_id
 294                    WHERE user = ?
 295                      AND subscriptions.type IN ('channel', 'playlist')
 296                 ORDER BY obsolete=0, name COLLATE NOCASE ASC""", (token,))
 297             rows = [{
 298                 'channel_id': channel_id,
 299                 'author': author or channel_id,
 300                 'type': type,
 301                 'subscribed_until': subscribed_until
 302             } for (channel_id, author, type, subscribed_until) in c.fetchall()]
 303     return render_template('subscription_manager.html.j2', rows=rows)
 304
 305 @frontend.route('/feed/subscriptions', methods=['POST'])
 306 @login_required
 307 def feed_post():
 308     token = current_user.token
 309     action = next(request.form.keys(), None)
 310     if action in ['pin', 'unpin', 'hide', 'unhide']:
 311         video_id = request.form.get(action)
 312         display = {
 313             'pin': 'pinned',
 314             'unpin': None,
 315             'hide': 'hidden',
 316             'unhide': None,
 317         }[action]
 318         with sqlite3.connect(cf['global']['database']) as conn:
 319             c = conn.cursor()
 320             store_video_metadata(video_id) # only needed for pinning
 321             c.execute("""
 322                 INSERT OR REPLACE INTO flags (user, video_id, display)
 323                 VALUES (?, ?, ?)
 324             """, (token, video_id, display))
 325         undo_flash(video_id, action)
 326     else:
 327         flash("unsupported action", "error")
 328     return redirect(request.url, code=303)
 329
 330 @frontend.route('/manage/subscriptions', methods=['POST'])
 331 @login_required
 332 def manage_subscriptions():
 333     token = current_user.token
 334     if 'subscribe' in request.form:
 335         some_id = request.form.get("subscribe")
 336         match = re.search(r"(UC[A-Za-z0-9_-]{22})", some_id)
 337         if match:
 338             some_id = match.group(1)
 339             id_type = "channel"
 340         else:
 341             match = re.search(r"((?:PL|LL|EC|UU|FL|UL|OL)[A-Za-z0-9_-]{10,})", some_id)
 342             if match:  # NOTE: PL-playlists are 32chars, others differ in length.
 343                 some_id = match.group(1)
 344                 id_type = "playlist"
 345             else:
 346                 flash("not a valid/subscribable URI", "error")
 347                 return redirect(request.url, code=303)
 348         with sqlite3.connect(cf['global']['database']) as conn:
 349             #with conn.cursor() as c:
 350                 c = conn.cursor()
 351                 c.execute("""
 352                         INSERT OR IGNORE INTO subscriptions (user, channel_id, type)
 353                         VALUES (?, ?, ?)
 354                 """, (token, some_id, id_type))
 355                 # TODO: sql-error-handling, asynchronically calling update-subs.pl
 356         undo_flash(some_id, 'subscribe')
 357
 358     elif 'unsubscribe' in request.form:
 359         some_id = request.form.get("unsubscribe")
 360         with sqlite3.connect(cf['global']['database']) as conn:
 361             #with conn.cursor() as c:
 362                 c = conn.cursor()
 363                 c.execute("""
 364                         DELETE FROM subscriptions
 365                         WHERE user = ? AND channel_id = ?
 366                 """, (token, some_id))
 367                 # TODO: sql-error-handling, report success
 368         undo_flash(some_id, 'unsubscribe')
 369
 370     else:
 371         flash("unsupported action", "error")
 372
 373     return redirect(request.url, code=303)
 374
 375 @frontend.route('/vi/<vid>/<res>.jpg')
 376 def redirect_thumbnails(vid, res):
 377     return redirect(f"https://i.ytimg.com{request.path}", code=301)
 378
 379 @frontend.record
 380 def redirect_youtube_dot_com(state):
 381     """
 382     This is executed when the blueprint is loaded dynamically builds a number
 383     of routes so that URLs like
 384     https://subscriptions.gir.st/https://www.youtube.com/watch?v=dQw4w9WgXcQ
 385     redirect to the /watch page. Works with /watch, /embed/ and youtu.be short
 386     links, with or without protocl and/or 'www'.
 387     """
 388     def real_redirect_youtube_dot_com(video_id=None):
 389         if not re.match(r"^[-_0-9A-Za-z]{11}$", video_id or ''): video_id = None
 390         if not video_id: video_id = request.args.get('v')
 391         return redirect(url_for('.watch', v=video_id))
 392
 393     for protocol in ("", "http://", "https://"):
 394         for prefix in ("", "www.", "m."):
 395             for domain in ("youtube.com", "youtu.be", "youtube-nocookie.com"):
 396                 for urlpath in ("/watch", "/embed/<video_id>", "/<video_id>"):
 397                     if domain != "youtu.be" and urlpath == "/<video_id>":
 398                         continue # that's a channel, not a video
 399                     frontend.add_url_rule(
 400                         f"/{protocol}{prefix}{domain}{urlpath}",
 401                         view_func=real_redirect_youtube_dot_com,
 402                         strict_slashes=False
 403                     )
 404
 405 def undo_flash(thing_id, action):
 406     undo_action, past_action = {
 407         'pin': ('unpin', 'pinned'),
 408         'unpin': ('pin', 'unpinned'),
 409         'hide': ('unhide', 'hidden'),
 410         'unhide': ('hide', 'unhidden'),
 411         'subscribe': ('unsubscribe', 'subscribed'),
 412         'unsubscribe': ('subscribe', 'unsubscribed'),
 413     }.get(action)
 414     if 'subscribe' in action and thing_id.startswith('UC'):
 415         thing = "channel"
 416         thing_url = url_for('.channel', channel_id=thing_id)
 417     elif 'subscribe' in action:
 418         thing = "playlist"
 419         thing_url = url_for('.playlist', playlist_id=thing_id)
 420     else:
 421         thing = "video"
 422         thing_url = url_for('.watch', v=thing_id)
 423     flash(f'''
 424         <form method=post><input type=hidden name="{undo_action}" value="{thing_id}">
 425         <a href="{thing_url}">{thing}</a> {past_action}.
 426         <label><input type="submit" hidden>
 427         <span style="text-decoration:underline;cursor:pointer">undo</span>.
 428         </label></form>''', "info")
 429
 430 @frontend.app_template_filter('format_date')
 431 def format_date(s):
 432     import datetime # can't import at top level, because it is inherited from common
 433     (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'
 434     M = '_ Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec'.split()
 435     if y == datetime.datetime.now().year:
 436         return f"{d} {M[m]}"
 437     else:
 438         return f"{M[m]} '{y%100}"
 439
 440 @frontend.app_template_filter('format_time')
 441 def format_time(i):
 442     if i is None:
 443         return None
 444     h = i // (60*60)
 445     m = i // 60 % 60
 446     s = i % 60
 447     return '%d:%02d:%02d' % (h,m,s) if h else '%02d:%02d' % (m,s)
 448
 449 @frontend.app_template_filter('timeoffset')
 450 def timeoffset(s):
 451     if s is None:
 452         return None
 453     match = re.match(r"^(\d+)s?$", s)  # e.g. 2040s
 454     if match:
 455         return match.group(1)
 456     match = re.match(r"^(?:(\d+)h)?(?:(\d+)m)?(?:(\d+)s)?$", s)  # e.g. 34m, 1h23s
 457     if match:
 458         return ":".join([n.zfill(2) for n in match.groups('0')])
 459     return None