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