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