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         settings = {}
  30         if 'welcome_message' in cf['frontend']:
  31             flash(cf['frontend']['welcome_message'], "welcome")
  32     else:
  33         token = current_user.token
  34         settings = current_user.get_settings()
  35     page = request.args.get('page', 0, type=int)
  36     with sqlite3.connect(cf['global']['database']) as conn:
  37         c = conn.cursor()
  38
  39         c.execute("""
  40        SELECT videos.id, channel_id, name, title, length, livestream, premiere, shorts, published > datetime('now') as upcoming, published, playlist_videos.playlist_id, display
  41              FROM videos
  42              JOIN channels ON videos.channel_id = channels.id
  43     LEFT JOIN playlist_videos ON (videos.id = playlist_videos.video_id)
  44         LEFT JOIN flags ON (videos.id = flags.video_id) AND (flags.user = ?)
  45         WHERE (channel_id IN (SELECT channel_id FROM subscriptions WHERE user=? AND type = 'channel')
  46                OR playlist_videos.playlist_id IN (SELECT channel_id FROM subscriptions WHERE user=? AND type = 'playlist')
  47                    OR flags.display = 'pinned')
  48                   AND flags.display IS NOT 'hidden'
  49               AND (flags.display = 'pinned' OR not ? or shorts is null or not shorts)
  50          ORDER BY (display = 'pinned') DESC, crawled DESC
  51             LIMIT 36
  52            OFFSET 36*?""", (token, token, token, settings.get('noshorts', False), page))
  53         rows = [{
  54             'video_id': video_id,
  55             'channel_id': channel_id,
  56             'author': author,
  57             'title': title,
  58             'length': length,
  59             'livestream': livestream,
  60             'premiere': premiere,
  61             'shorts': shorts,
  62             'upcoming': upcoming,
  63             'published': published,
  64             'playlist': playlist,
  65             'pinned': display == 'pinned',
  66         } for (video_id, channel_id, author, title, length, livestream, premiere, shorts, upcoming, published, playlist, display) in c.fetchall()]
  67     return render_template('index.html.j2', rows=rows, page=page)
  68
  69 @frontend.route('/watch')
  70 def watch():
  71     if current_user.is_anonymous:
  72         token = 'guest'
  73     else:
  74         token = current_user.token
  75
  76     if not 'v' in request.args:
  77         return "missing video id", 400
  78     if len(request.args.get('v')) != 11:
  79         return "malformed video id", 400
  80
  81     plaintextheaders = {
  82         'content-type': 'text/plain',
  83         'Link': "<data:text/css,body%7Bcolor:%23eee;background:%23333%7D>; rel=stylesheet;"
  84     }
  85
  86     video_id = request.args.get('v')
  87     show = request.args.get("show")
  88     itag = request.args.get("itag")
  89     video_url, stream_map, metadata, error, errdetails = get_video_info(video_id, metaOnly=(show == 'meta'))
  90
  91     if itag:  # if itag is specified and available, force its use even if not highest resolution
  92         video_url = next((e['url'] for e in stream_map['muxed'] if 'url' in e and e['itag'] == int(itag)), video_url)
  93
  94     proxy_on = getattr(g, 'proxy_on', None)
  95     proxy_msg = {
  96         False: "Log in to use proxy.", # proxy for logged-in users only
  97         None: "", # proxy disabled globally
  98         True: "", # proxy is available to this user
  99     }[proxy_on]
 100     errdetails = {
 101         'banned':     "Instance is being rate limited.",
 102         'malformed':  "Video ID is invalid.",
 103         'geolocked':  f"This video is geolocked. {proxy_msg}",
 104         'livestream': f"Livestreams disabled. {proxy_msg}",
 105         'throttled':  "Stream is severely throttled.",
 106         'agegated':   "Unable to bypass age-restriction.",
 107         'scrambled':  "VEVO style cyphering not supported.",
 108         'no-url':     "No muxed stream available.",
 109         'exhausted':  errdetails or "Couldn't extract video URLs.",
 110         'player':     errdetails,
 111     }.get(error, error)
 112
 113     # if the video is geolocked, and the proxy is enabled, we can still play
 114     # it, if the video is available in the instance server's region:
 115     if error == 'geolocked' and video_url and proxy_on:
 116         videoplayback = url_for('proxy.videoplayback')
 117         query = urlparse(video_url).query
 118         video_url = f"{videoplayback}?{query}"
 119         for t in ('adaptive_video', 'adaptive_audio', 'muxed'):
 120             for s in stream_map[t]:
 121                 query = urlparse(s['url']).query
 122                 s['url'] = f"{videoplayback}?{query}"
 123         error = None
 124
 125     # if the proxy is enabled, we can also play livestreams:
 126     if error == 'livestream' and proxy_on:
 127         # Note: hlsManifestUrl's hostname will be replaced client-side
 128         video_url = stream_map['hlsManifestUrl']
 129         error = None
 130
 131     # if the instance is blocked, try submitting a job to the anti captcha service:
 132     if error == 'banned' and cf['captcha']['api_key']:
 133         r2 = requests.get(f'https://www.youtube.com/watch?v={video_id}&hl=en&gl=US')
 134         status = submit_captcha(r2)
 135         if status is False:
 136             raise Exception("we are banned, but captcha wasn't triggered!")
 137         else:
 138             message = "right now" if status is True else f"{int(status)} seconds ago"
 139             raise BadGateway(f"""
 140                 {errdetails} An attempt at getting unblocked has been made {message}.
 141                 Please try again in 30 seconds.
 142             """)
 143
 144     if show == "raw":
 145         if error:
 146             msg = errdetails if error=='player' else f"{error.upper()}: {errdetails}"
 147             return msg, 400, plaintextheaders # TODO: nicer
 148         return redirect(video_url, code=307)
 149     elif show == "json":
 150         if error and not metadata:
 151             return {'error': True, error: errdetails}, 400 # TODO: better (test _CpR4o81XQc)
 152         return jsonify(metadata)
 153     elif show == "audio":
 154         # sorting: we want to prioritize mp4a over opus, and sort by highest quality first
 155         # todo: geolocking; prefer open format?
 156         if error and not stream_map:
 157             msg = errdetails if error=='player' else f"{error.upper()}: {errdetails}"
 158             return msg, 400, plaintextheaders # TODO: nicer
 159         stream = next(iter(sorted(
 160             stream_map['adaptive_audio'],
 161             key=lambda e: ('opus' not in e['mimeType'], e['bitrate']),
 162             reverse=True
 163         )),{}).get('url')
 164         return redirect(stream)
 165     elif show == "meta":
 166         # this is the subset of (useful) keys that are not present in the
 167         # Android API response. the special key '_' contains ready-to-use
 168         # parsed versions of that data.
 169         if error and not metadata:
 170             return {'error': True, error: errdetails}, 400 # TODO: better (test _CpR4o81XQc)
 171         parsed = microformat_parser(metadata)
 172         return {'microformat': metadata.get('microformat'),'cards':metadata.get('cards'), '_':parsed}
 173     else:
 174         extra = {'geolocked':'local=1', 'livestream':'raw=0'}.get(error,'')
 175         invidious_url = f"https://redirect.invidious.io/watch?v={video_id}&{extra}"
 176         if error and not metadata: # e.g. malformed, private/deleted video, ...
 177             return render_template('video-error.html.j2', video_id=video_id,
 178                 video_error=error, errdetails=errdetails, invidious_url=invidious_url)
 179         meta = prepare_metadata(metadata)
 180         with sqlite3.connect(cf['global']['database']) as conn:
 181             c = conn.cursor()
 182             c.execute("""
 183             SELECT COUNT((
 184                 SELECT 1 FROM subscriptions WHERE channel_id = ? AND user = ?
 185             )), COUNT((
 186                 SELECT 1 FROM flags WHERE video_id = ? AND display = 'pinned' AND user = ?
 187             ))""", (meta['channel_id'], token, video_id, token))
 188             (is_subscribed, is_pinned) = c.fetchone()
 189             return render_template('watch.html.j2',
 190                 video_id=video_id, video_url=video_url, stream_map=stream_map,
 191                 video_error=error, errdetails=errdetails, invidious_url=invidious_url,
 192                 is_pinned=is_pinned, is_subscribed=is_subscribed,
 193                 **meta)
 194
 195 @frontend.route('/embed/videoseries')
 196 def embed_videoseries():
 197     return redirect(url_for('.playlist', list=request.args.get('list')))
 198 @frontend.route('/embed/<video_id>', strict_slashes=False)
 199 def embed(video_id):
 200     if video_id == "videoseries":
 201         return redirect(url_for('.playlist', list=request.args.get('list')))
 202
 203     return redirect(url_for('.watch', v=video_id, t=request.args.get('start')))
 204
 205 @frontend.route('/live/<video_id>', strict_slashes=False)
 206 @frontend.route('/shorts/<video_id>', strict_slashes=False)
 207 def shorts_or_live(video_id):
 208     return redirect(url_for('.watch', v=video_id, t=request.args.get('start')))
 209
 210 @frontend.route('/<something>', strict_slashes=False)
 211 def plain_user_or_video(something):
 212     # yt.com interprets this as a username, but we also want to catch youtu.be
 213     # short-urls. so we check if it's a channel by querying the RSS feed (this
 214     # shoudn't be rate-limited); if that fails, check if it looks like a video
 215     # id; or finally give up.
 216     if '.' not in something and channel_exists(something):
 217         # periods are not valid in usernames, vanity urls or ucids, but common
 218         # in urls that get crawled by bots (e.g. index.php). failing early
 219         # reduces the amount of invalid channel names getting looked up.
 220         return redirect(url_for('.channel', channel_id=something))
 221     elif re.match(r"^[-_0-9A-Za-z]{11}$", something): # looks like a video id
 222         return redirect(url_for('.watch', v=something, t=request.args.get('t')))
 223     else: # ¯\_(ツ)_/¯
 224         # XXX: something == 'thethoughtemporium' -> 404s
 225         raise NotFound("Note: some usernames not recognized; try searching it")
 226
 227 @frontend.route('/attribution_link', strict_slashes=False)
 228 def attribution_link():
 229     # /attribution_link?a=anything&u=/channel/UCZYTClx2T1of7BRZ86-8fow
 230     # /attribution_link?a=JdfC0C9V6ZI&u=%2Fwatch%3Fv%3DEhxJLojIE_o%26feature%3Dshare
 231     return redirect(request.args.get('u') or '/')
 232
 233 @frontend.route('/c/<channel_id>/<subpage>')
 234 @frontend.route('/c/<channel_id>/')
 235 @frontend.route('/user/<channel_id>/<subpage>')
 236 @frontend.route('/user/<channel_id>/')
 237 def channel_redirect(channel_id, subpage=None):
 238     # Note: we can't check /c/, so we have to assume it is the same as /user/,
 239     # which is sometimes wrong.
 240     xmlfeed = fetch_xml("user", channel_id)
 241
 242     if not xmlfeed:
 243         raise NotFound("unknown channel name")
 244
 245     _, _, _, channel_id, _ = parse_xml(xmlfeed)
 246
 247     return redirect(url_for('.channel', channel_id=channel_id))
 248
 249 @frontend.route('/channel/<channel_id>/<subpage>')
 250 @frontend.route('/channel/<channel_id>/')
 251 def channel(channel_id, _=None):
 252     token = getattr(current_user, 'token', 'guest')
 253     sort = request.args.get("sort", "newest")
 254
 255     if not re.match(r"(UC[A-Za-z0-9_-]{22})", channel_id):
 256         # canonicalize channel id, otherwise popular won't work
 257         return redirect(url_for('.channel_redirect', channel_id=channel_id))
 258
 259     if sort == "popular":
 260         xmlfeed = fetch_xml("playlist_id", f"PU{channel_id[2:]}")
 261     else:
 262         xmlfeed = fetch_xml("channel_id", channel_id)
 263         #^note: could also use playlist_id=UU...
 264
 265     if not xmlfeed:
 266         raise NotFound("unknown channel id")
 267
 268     title, author, videos, _, _ = parse_xml(xmlfeed)
 269
 270     with sqlite3.connect(cf['global']['database']) as conn:
 271         c = conn.cursor()
 272         c.execute("""
 273             SELECT COUNT(*)
 274               FROM subscriptions
 275              WHERE channel_id = ? AND user = ?
 276         """, (channel_id, token))
 277         (is_subscribed,) = c.fetchone()
 278
 279     return render_template('xmlfeed.html.j2', title=author, rows=videos,
 280         is_subscribed=is_subscribed, channel_id=channel_id)
 281
 282 @frontend.route('/playlist')
 283 def playlist():
 284     playlist_id = request.args.get('list')
 285     if not playlist_id:
 286         return "bad list id", 400 # todo
 287
 288     xmlfeed = fetch_xml("playlist_id", playlist_id)
 289     if not xmlfeed:
 290         return "not found or something", 404 # XXX
 291     title, author, videos, _, _ = parse_xml(xmlfeed)
 292     return render_template('xmlfeed.html.j2', title=f"{title} by {author}", rows=videos)
 293
 294 @frontend.route('/api/timedtext')
 295 def timedtext():
 296     r = requests.get("https://www.youtube.com/api/timedtext", request.args.to_dict())
 297     # Note: in srv1 format, xmlentities are double-encoded m( a smart quote is
 298     # even worse: it's '&amp;39;<smartquote>' wtf!? (at least vvt seems ok)
 299     if not r.ok:
 300         return "error: {r.text}", 400 # TODO: better
 301     retval = r.text
 302     if request.args.get('fmt') == 'vtt' and request.args.get('kind') == 'asr':
 303         # autocaptions are extremely confusing, and stuck in the lower-left corner. fix it up a bit
 304         retval = re.sub(r"<.+?>", "", retval) # remove inline html-like markup that times each word/adds styles
 305         retval = retval.replace("align:start position:0%", "") # let browser position the text itself
 306         # each subtitle-line is repeated twice (first on the lower line, then
 307         # on the next "frame" on the upper line). we want to remove the
 308         # repetition, as that's confusing without word and line animations:
 309         lines = retval.split('\n')
 310         retval = '\n'.join([line for line, prev in zip(lines, ['']+lines) if not " --> " in prev])
 311     return retval, {'Content-Type': r.headers.get("Content-Type")}
 312
 313 @frontend.route('/manage/subscriptions')
 314 # disabled for guest user: @login_required
 315 def subscription_manager():
 316     if current_user.is_anonymous:
 317         token = 'guest'
 318     else:
 319         token = current_user.token
 320     with sqlite3.connect(cf['global']['database']) as conn:
 321         #with conn.cursor() as c:
 322             c = conn.cursor()
 323             c.execute("""
 324                   SELECT subscriptions.channel_id, name, type,
 325                          (subscribed_until < datetime('now')) AS obsolete
 326                     FROM subscriptions
 327                     LEFT JOIN (SELECT name, id FROM channels
 328                        UNION
 329                        SELECT name, id FROM playlists
 330                       ) AS channels ON channels.id = subscriptions.channel_id
 331                     left JOIN websub ON channels.id = websub.channel_id
 332                    WHERE user = ?
 333                      AND subscriptions.type IN ('channel', 'playlist')
 334                 ORDER BY obsolete=0, name COLLATE NOCASE ASC""", (token,))
 335             rows = [{
 336                 'channel_id': channel_id,
 337                 'author': author or channel_id,
 338                 'type': type,
 339                 'subscribed_until': subscribed_until
 340             } for (channel_id, author, type, subscribed_until) in c.fetchall()]
 341     return render_template('subscription_manager.html.j2', rows=rows)
 342
 343 @frontend.route('/feed/subscriptions', methods=['POST'])
 344 @login_required
 345 def feed_post():
 346     token = current_user.token
 347     action = next(iter(list(request.form.keys())), None)
 348     if action in ['pin', 'unpin', 'hide', 'unhide']:
 349         video_id = request.form.get(action)
 350         display = {
 351             'pin': 'pinned',
 352             'unpin': None,
 353             'hide': 'hidden',
 354             'unhide': None,
 355         }[action]
 356         with sqlite3.connect(cf['global']['database']) as conn:
 357             c = conn.cursor()
 358             store_video_metadata(video_id) # only needed for pinning
 359             c.execute("""
 360                 INSERT OR REPLACE INTO flags (user, video_id, display)
 361                 VALUES (?, ?, ?)
 362             """, (token, video_id, display))
 363         undo_flash(video_id, action)
 364     else:
 365         flash("unsupported action", "error")
 366     return redirect(request.url, code=303)
 367
 368 @frontend.route('/manage/subscriptions', methods=['POST'])
 369 @login_required
 370 def manage_subscriptions():
 371     token = current_user.token
 372     if 'subscribe' in request.form:
 373         some_id = request.form.get("subscribe")
 374         match = re.search(r"(UC[A-Za-z0-9_-]{22})", some_id)
 375         if match:
 376             some_id = match.group(1)
 377             id_type = "channel"
 378         else:
 379             match = re.search(r"((?:PL|LL|EC|UU|FL|UL|OL)[A-Za-z0-9_-]{10,})", some_id)
 380             if match:  # NOTE: PL-playlists are 32chars, others differ in length.
 381                 some_id = match.group(1)
 382                 id_type = "playlist"
 383             else:
 384                 flash("not a valid/subscribable URI", "error")
 385                 return redirect(request.url, code=303)
 386         with sqlite3.connect(cf['global']['database']) as conn:
 387             #with conn.cursor() as c:
 388                 c = conn.cursor()
 389                 c.execute("""
 390                         INSERT OR IGNORE INTO subscriptions (user, channel_id, type)
 391                         VALUES (?, ?, ?)
 392                 """, (token, some_id, id_type))
 393                 # TODO: sql-error-handling, asynchronically calling update-subs.pl
 394         undo_flash(some_id, 'subscribe')
 395
 396     elif 'unsubscribe' in request.form:
 397         some_id = request.form.get("unsubscribe")
 398         with sqlite3.connect(cf['global']['database']) as conn:
 399             #with conn.cursor() as c:
 400                 c = conn.cursor()
 401                 c.execute("""
 402                         DELETE FROM subscriptions
 403                         WHERE user = ? AND channel_id = ?
 404                 """, (token, some_id))
 405                 # TODO: sql-error-handling, report success
 406         undo_flash(some_id, 'unsubscribe')
 407
 408     else:
 409         flash("unsupported action", "error")
 410
 411     return redirect(request.url, code=303)
 412
 413 @frontend.route('/vi/<vid>/<res>.jpg')
 414 def redirect_thumbnails(vid, res):
 415     return redirect(f"https://i.ytimg.com{request.path}", code=301)
 416
 417 @frontend.record
 418 def redirect_youtube_dot_com(state):
 419     """
 420     This is executed when the blueprint is loaded dynamically builds a number
 421     of routes so that URLs like
 422     https://subscriptions.gir.st/https://www.youtube.com/watch?v=dQw4w9WgXcQ
 423     redirect to the /watch page. Works with /watch, /embed/ and youtu.be short
 424     links, with or without protocl and/or 'www'.
 425     """
 426     def real_redirect_youtube_dot_com(video_id=None):
 427         if not re.match(r"^[-_0-9A-Za-z]{11}$", video_id or ''): video_id = None
 428         if not video_id: video_id = request.args.get('v')
 429         return redirect(url_for('.watch', v=video_id))
 430
 431     for protocol in ("", "http://", "https://"):
 432         for prefix in ("", "www.", "m."):
 433             for domain in ("youtube.com", "youtu.be", "youtube-nocookie.com"):
 434                 for urlpath in ("/watch", "/embed/<video_id>", "/live/<video_id>", "/shorts/<video_id>", "/<video_id>"):
 435                     if domain != "youtu.be" and urlpath == "/<video_id>":
 436                         continue # that's a channel, not a video
 437                     frontend.add_url_rule(
 438                         f"/{protocol}{prefix}{domain}{urlpath}",
 439                         view_func=real_redirect_youtube_dot_com,
 440                         strict_slashes=False
 441                     )
 442
 443 def undo_flash(thing_id, action):
 444     undo_action, past_action = {
 445         'pin': ('unpin', 'pinned'),
 446         'unpin': ('pin', 'unpinned'),
 447         'hide': ('unhide', 'hidden'),
 448         'unhide': ('hide', 'unhidden'),
 449         'subscribe': ('unsubscribe', 'subscribed'),
 450         'unsubscribe': ('subscribe', 'unsubscribed'),
 451     }.get(action)
 452     if 'subscribe' in action and thing_id.startswith('UC'):
 453         thing = "channel"
 454         thing_url = url_for('.channel', channel_id=thing_id)
 455     elif 'subscribe' in action:
 456         thing = "playlist"
 457         thing_url = url_for('.playlist', playlist_id=thing_id)
 458     else:
 459         thing = "video"
 460         thing_url = url_for('.watch', v=thing_id)
 461     flash(f'''
 462         <form method=post><input type=hidden name="{undo_action}" value="{thing_id}">
 463         <a href="{thing_url}">{thing}</a> {past_action}.
 464         <label><input type="submit" hidden>
 465         <span style="text-decoration:underline;cursor:pointer">undo</span>.
 466         </label></form>''', "info")
 467
 468 @frontend.app_template_filter('format_date')
 469 def format_date(s):
 470     import datetime # can't import at top level, because it is inherited from common
 471     (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'
 472     M = '_ Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec'.split()
 473     if y == datetime.datetime.now().year:
 474         return f"{d} {M[m]}"
 475     else:
 476         return f"{M[m]} '{y%100}"
 477
 478 @frontend.app_template_filter('format_time')
 479 def format_time(i):
 480     if i is None:
 481         return None
 482     h = i // (60*60)
 483     m = i // 60 % 60
 484     s = i % 60
 485     return '%d:%02d:%02d' % (h,m,s) if h else '%02d:%02d' % (m,s)
 486
 487 @frontend.app_template_filter('timeoffset')
 488 def timeoffset(s):
 489     if s is None:
 490         return None
 491     match = re.match(r"^(\d+)s?$", s)  # e.g. 2040s
 492     if match:
 493         return match.group(1)
 494     match = re.match(r"^(?:(\d+)h)?(?:(\d+)m)?(?:(\d+)s)?$", s)  # e.g. 34m, 1h23s
 495     if match:
 496         return ":".join([n.zfill(2) for n in match.groups('0')])
 497     return None