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