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