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