app/youtube/__init__.py

   1 import re
   2 import time
   3 import sqlite3
   4 import requests
   5 from urllib.parse import urlparse
   6 #from flask_login import current_user, login_required
   7 from flask_login import LoginManager, UserMixin, current_user, login_user, logout_user, login_required
   8 from flask import Blueprint, render_template, request, redirect, flash, url_for, jsonify, g, current_app
   9 from werkzeug.exceptions import NotFound, BadGateway
  10
  11 from ..common.common import *
  12 from ..common.anticaptcha import submit_captcha
  13 from .lib import *
  14
  15 frontend = Blueprint('youtube', __name__,
  16     template_folder='templates',
  17     static_folder='static',
  18     static_url_path='/static/yt')
  19
  20 @frontend.route('/')
  21 def index():
  22     return redirect(url_for('.feed'), code=302)
  23
  24 @frontend.route('/feed/subscriptions')
  25 # disabled for guest user: @login_required
  26 def feed():
  27     if current_user.is_anonymous:
  28         token = 'guest'
  29         settings = {}
  30         if 'welcome_message' in cf['frontend']:
  31             flash(cf['frontend']['welcome_message'], "welcome")
  32     else:
  33         token = current_user.token
  34         settings = current_user.get_settings()
  35     page = request.args.get('page', 0, type=int)
  36     with sqlite3.connect(cf['global']['database']) as conn:
  37         c = conn.cursor()
  38
  39         c.execute("""
  40        SELECT videos.id, channel_id, name, title, length, livestream, premiere, shorts, published > datetime('now') as upcoming, published, playlist_videos.playlist_id, display
  41              FROM videos
  42              JOIN channels ON videos.channel_id = channels.id
  43     LEFT JOIN playlist_videos ON (videos.id = playlist_videos.video_id)
  44         LEFT JOIN flags ON (videos.id = flags.video_id) AND (flags.user = ?)
  45         WHERE (channel_id IN (SELECT channel_id FROM subscriptions WHERE user=? AND type = 'channel')
  46                OR playlist_videos.playlist_id IN (SELECT channel_id FROM subscriptions WHERE user=? AND type = 'playlist')
  47                    OR flags.display = 'pinned')
  48                   AND flags.display IS NOT 'hidden'
  49               AND (flags.display = 'pinned' OR not ? or shorts is null or not shorts)
  50          ORDER BY (display = 'pinned') DESC, crawled DESC
  51             LIMIT 36
  52            OFFSET 36*?""", (token, token, token, settings.get('noshorts', False), page))
  53         rows = [{
  54             'video_id': video_id,
  55             'channel_id': channel_id,
  56             'author': author,
  57             'title': title,
  58             'length': length,
  59             'livestream': livestream,
  60             'premiere': premiere,
  61             'shorts': shorts,
  62             'upcoming': upcoming,
  63             'published': published,
  64             'playlist': playlist,
  65             'pinned': display == 'pinned',
  66         } for (video_id, channel_id, author, title, length, livestream, premiere, shorts, upcoming, published, playlist, display) in c.fetchall()]
  67     return render_template('index.html.j2', rows=rows, page=page)
  68
  69 @frontend.route('/watch')
  70 def watch():
  71     if current_user.is_anonymous:
  72         token = 'guest'
  73     else:
  74         token = current_user.token
  75
  76     if not 'v' in request.args:
  77         return "missing video id", 400
  78     if len(request.args.get('v')) != 11:
  79         return "malformed video id", 400
  80
  81     plaintextheaders = {
  82         'content-type': 'text/plain',
  83         'Link': "<data:text/css,body%7Bcolor:%23eee;background:%23333%7D>; rel=stylesheet;"
  84     }
  85
  86     video_id = request.args.get('v')
  87     playlist = request.args.get('list')
  88     show = request.args.get("show")
  89     itag = request.args.get("itag")
  90     video_url, stream_map, metadata, error, errdetails = get_video_info(video_id, metaOnly=(show == 'meta'))
  91
  92     if itag:  # if itag is specified and available, force its use even if not highest resolution
  93         video_url = next((e['url'] for e in stream_map['muxed'] if 'url' in e and e['itag'] == int(itag)), video_url)
  94
  95     proxy_on = getattr(g, 'proxy_on', None)
  96     proxy_msg = {
  97         False: "Log in to use proxy.", # proxy for logged-in users only
  98         None: "", # proxy disabled globally
  99         True: "", # proxy is available to this user
 100     }[proxy_on]
 101     errdetails = {
 102         'banned':     "Instance is being rate limited.",
 103         'malformed':  "Video ID is invalid.",
 104         'geolocked':  f"This video is geolocked. {proxy_msg}",
 105         'livestream': f"Livestreams disabled. {proxy_msg}",
 106         'throttled':  "Stream is severely throttled.",
 107         'agegated':   "Unable to bypass age-restriction.",
 108         'scrambled':  "VEVO style cyphering not supported.",
 109         'no-url':     "No muxed stream available.",
 110         'exhausted':  errdetails or "Couldn't extract video URLs.",
 111         'player':     errdetails,
 112     }.get(error, error)
 113
 114     # if the video is geolocked, and the proxy is enabled, we can still play
 115     # it, if the video is available in the instance server's region:
 116     if error == 'geolocked' and video_url and proxy_on:
 117         videoplayback = url_for('proxy.videoplayback')
 118         query = urlparse(video_url).query
 119         video_url = f"{videoplayback}?{query}"
 120         for t in ('adaptive_video', 'adaptive_audio', 'muxed'):
 121             for s in stream_map[t]:
 122                 query = urlparse(s['url']).query
 123                 s['url'] = f"{videoplayback}?{query}"
 124         error = None
 125
 126     # if the proxy is enabled, we can also play livestreams:
 127     if error == 'livestream' and proxy_on:
 128         # Note: hlsManifestUrl's hostname will be replaced client-side
 129         video_url = stream_map['hlsManifestUrl']
 130         error = None
 131
 132     # if the instance is blocked, try submitting a job to the anti captcha service:
 133     if error == 'banned' and cf['captcha']['api_key']:
 134         r2 = requests.get(f'https://www.youtube.com/watch?v={video_id}&hl=en&gl=US')
 135         status = submit_captcha(r2)
 136         if status is False:
 137             raise Exception("we are banned, but captcha wasn't triggered!")
 138         else:
 139             message = "right now" if status is True else f"{int(status)} seconds ago"
 140             raise BadGateway(f"""
 141                 {errdetails} An attempt at getting unblocked has been made {message}.
 142                 Please try again in 30 seconds.
 143             """)
 144
 145     if show == "raw":
 146         if error:
 147             msg = errdetails if error=='player' else f"{error.upper()}: {errdetails}"
 148             return msg, 400, plaintextheaders # TODO: nicer
 149         return redirect(video_url, code=307)
 150     elif show == "json":
 151         if error and not metadata:
 152             return {'error': True, error: errdetails}, 400 # TODO: better (test _CpR4o81XQc)
 153         return jsonify(metadata)
 154     elif show == "audio":
 155         # sorting: we want to prioritize mp4a over opus, and sort by highest quality first
 156         # todo: geolocking; prefer open format?
 157         if error and not stream_map:
 158             msg = errdetails if error=='player' else f"{error.upper()}: {errdetails}"
 159             return msg, 400, plaintextheaders # TODO: nicer
 160         stream = next(iter(sorted(
 161             stream_map['adaptive_audio'],
 162             key=lambda e: ('opus' not in e['mimeType'], e['bitrate']),
 163             reverse=True
 164         )),{}).get('url')
 165         return redirect(stream)
 166     elif show == "meta":
 167         # this is the subset of (useful) keys that are not present in the
 168         # Android API response. the special key '_' contains ready-to-use
 169         # parsed versions of that data.
 170         if error and not metadata:
 171             return {'error': True, error: errdetails}, 400 # TODO: better (test _CpR4o81XQc)
 172         parsed = microformat_parser(metadata)
 173         return {'microformat': metadata.get('microformat'),'cards':metadata.get('cards'), '_':parsed}
 174     else:
 175         extra = {'geolocked':'local=1', 'livestream':'raw=0'}.get(error,'')
 176         invidious_url = f"https://redirect.invidious.io/watch?v={video_id}&{extra}"
 177         if error and not metadata: # e.g. malformed, private/deleted video, ...
 178             return render_template('video-error.html.j2', video_id=video_id,
 179                 video_error=error, errdetails=errdetails, invidious_url=invidious_url)
 180         meta = prepare_metadata(metadata)
 181         with sqlite3.connect(cf['global']['database']) as conn:
 182             c = conn.cursor()
 183             c.execute("""
 184             SELECT COUNT((
 185                 SELECT 1 FROM subscriptions WHERE channel_id = ? AND user = ?
 186             )), COUNT((
 187                 SELECT 1 FROM flags WHERE video_id = ? AND display = 'pinned' AND user = ?
 188             ))""", (meta['channel_id'], token, video_id, token))
 189             (is_subscribed, is_pinned) = c.fetchone()
 190             return render_template('watch.html.j2',
 191                 video_id=video_id, video_url=video_url, stream_map=stream_map,
 192                 video_error=error, errdetails=errdetails, invidious_url=invidious_url,
 193                 playlist=playlist, is_pinned=is_pinned, is_subscribed=is_subscribed,
 194                 **meta)
 195
 196 @frontend.route('/embed/videoseries')
 197 def embed_videoseries():
 198     return redirect(url_for('.playlist', list=request.args.get('list')))
 199 @frontend.route('/embed/<video_id>', strict_slashes=False)
 200 def embed(video_id):
 201     if video_id == "videoseries":
 202         return redirect(url_for('.playlist', list=request.args.get('list')))
 203
 204     return redirect(url_for('.watch', v=video_id, t=request.args.get('start')))
 205
 206 @frontend.route('/live/<video_id>', strict_slashes=False)
 207 @frontend.route('/shorts/<video_id>', strict_slashes=False)
 208 def shorts_or_live(video_id):
 209     return redirect(url_for('.watch', v=video_id, t=request.args.get('start')))
 210
 211 @frontend.route('/<something>', strict_slashes=False)
 212 def plain_user_or_video(something):
 213     # yt.com interprets this as a username, but we also want to catch youtu.be
 214     # short-urls. so we check if it's a channel by querying the RSS feed (this
 215     # shoudn't be rate-limited); if that fails, check if it looks like a video
 216     # id; or finally give up.
 217     if '.' not in something and channel_exists(something):
 218         # periods are not valid in usernames, vanity urls or ucids, but common
 219         # in urls that get crawled by bots (e.g. index.php). failing early
 220         # reduces the amount of invalid channel names getting looked up.
 221         return redirect(url_for('.channel', channel_id=something))
 222     elif re.match(r"^[-_0-9A-Za-z]{11}$", something): # looks like a video id
 223         return redirect(url_for('.watch', v=something, t=request.args.get('t')))
 224     else: # ¯\_(ツ)_/¯
 225         # XXX: something == 'thethoughtemporium' -> 404s
 226         raise NotFound("Note: some usernames not recognized; try searching it")
 227
 228 @frontend.route('/attribution_link', strict_slashes=False)
 229 def attribution_link():
 230     # /attribution_link?a=anything&u=/channel/UCZYTClx2T1of7BRZ86-8fow
 231     # /attribution_link?a=JdfC0C9V6ZI&u=%2Fwatch%3Fv%3DEhxJLojIE_o%26feature%3Dshare
 232     return redirect(request.args.get('u') or '/')
 233
 234 @frontend.route('/c/<channel_id>/<subpage>')
 235 @frontend.route('/c/<channel_id>/')
 236 @frontend.route('/user/<channel_id>/<subpage>')
 237 @frontend.route('/user/<channel_id>/')
 238 def channel_redirect(channel_id, subpage=None):
 239     # Note: we can't check /c/, so we have to assume it is the same as /user/,
 240     # which is sometimes wrong.
 241     xmlfeed = fetch_xml("user", channel_id)
 242
 243     if not xmlfeed:
 244         raise NotFound("unknown channel name")
 245
 246     _, _, _, channel_id, _ = parse_xml(xmlfeed)
 247
 248     return redirect(url_for('.channel', channel_id=channel_id))
 249
 250 @frontend.route('/channel/<channel_id>/<subpage>')
 251 @frontend.route('/channel/<channel_id>/')
 252 def channel(channel_id, _=None):
 253     token = getattr(current_user, 'token', 'guest')
 254     sort = request.args.get("sort", "newest")
 255
 256     if not re.match(r"(UC[A-Za-z0-9_-]{22})", channel_id):
 257         # canonicalize channel id, otherwise popular won't work
 258         return redirect(url_for('.channel_redirect', channel_id=channel_id))
 259
 260     if sort == "popular":
 261         xmlfeed = fetch_xml("playlist_id", f"PU{channel_id[2:]}")
 262     else:
 263         xmlfeed = fetch_xml("channel_id", channel_id)
 264         #^note: could also use playlist_id=UU...
 265
 266     if not xmlfeed:
 267         raise NotFound("unknown channel id")
 268
 269     title, author, videos, _, _ = parse_xml(xmlfeed)
 270
 271     with sqlite3.connect(cf['global']['database']) as conn:
 272         c = conn.cursor()
 273         c.execute("""
 274             SELECT COUNT(*)
 275               FROM subscriptions
 276              WHERE channel_id = ? AND user = ?
 277         """, (channel_id, token))
 278         (is_subscribed,) = c.fetchone()
 279
 280     return render_template('xmlfeed.html.j2', title=author, rows=videos,
 281         is_subscribed=is_subscribed, channel_id=channel_id)
 282
 283 @frontend.route('/playlist')
 284 def playlist():
 285     playlist_id = request.args.get('list')
 286     if not playlist_id:
 287         return "bad list id", 400 # todo
 288
 289     xmlfeed = fetch_xml("playlist_id", playlist_id)
 290     if not xmlfeed:
 291         return "not found or something", 404 # XXX
 292     title, author, videos, _, _ = parse_xml(xmlfeed)
 293     return render_template('xmlfeed.html.j2', title=f"{title} by {author}", rows=videos)
 294
 295 @frontend.route('/api/timedtext')
 296 def timedtext():
 297     r = requests.get("https://www.youtube.com/api/timedtext", request.args.to_dict())
 298     # Note: in srv1 format, xmlentities are double-encoded m( a smart quote is
 299     # even worse: it's '&amp;39;<smartquote>' wtf!? (at least vvt seems ok)
 300     if not r.ok:
 301         return "error: {r.text}", 400 # TODO: better
 302     retval = r.text
 303     if request.args.get('fmt') == 'vtt' and request.args.get('kind') == 'asr':
 304         # autocaptions are extremely confusing, and stuck in the lower-left corner. fix it up a bit
 305         retval = re.sub(r"<.+?>", "", retval) # remove inline html-like markup that times each word/adds styles
 306         retval = retval.replace("align:start position:0%", "") # let browser position the text itself
 307         # each subtitle-line is repeated twice (first on the lower line, then
 308         # on the next "frame" on the upper line). we want to remove the
 309         # repetition, as that's confusing without word and line animations:
 310         lines = retval.split('\n')
 311         retval = '\n'.join([line for line, prev in zip(lines, ['']+lines) if not " --> " in prev])
 312     return retval, {'Content-Type': r.headers.get("Content-Type")}
 313
 314 @frontend.route('/manage/subscriptions')
 315 # disabled for guest user: @login_required
 316 def subscription_manager():
 317     if current_user.is_anonymous:
 318         token = 'guest'
 319     else:
 320         token = current_user.token
 321     with sqlite3.connect(cf['global']['database']) as conn:
 322         #with conn.cursor() as c:
 323             c = conn.cursor()
 324             c.execute("""
 325                   SELECT subscriptions.channel_id, name, type,
 326                          (subscribed_until < datetime('now')) AS obsolete
 327                     FROM subscriptions
 328                     LEFT JOIN (SELECT name, id FROM channels
 329                        UNION
 330                        SELECT name, id FROM playlists
 331                       ) AS channels ON channels.id = subscriptions.channel_id
 332                     left JOIN websub ON channels.id = websub.channel_id
 333                    WHERE user = ?
 334                      AND subscriptions.type IN ('channel', 'playlist')
 335                 ORDER BY obsolete=0, name COLLATE NOCASE ASC""", (token,))
 336             rows = [{
 337                 'channel_id': channel_id,
 338                 'author': author or channel_id,
 339                 'type': type,
 340                 'subscribed_until': subscribed_until
 341             } for (channel_id, author, type, subscribed_until) in c.fetchall()]
 342     return render_template('subscription_manager.html.j2', rows=rows)
 343
 344 @frontend.route('/feed/subscriptions', methods=['POST'])
 345 @login_required
 346 def feed_post():
 347     token = current_user.token
 348     action = next(iter(list(request.form.keys())), None)
 349     if action in ['pin', 'unpin', 'hide', 'unhide']:
 350         video_id = request.form.get(action)
 351         display = {
 352             'pin': 'pinned',
 353             'unpin': None,
 354             'hide': 'hidden',
 355             'unhide': None,
 356         }[action]
 357         with sqlite3.connect(cf['global']['database']) as conn:
 358             c = conn.cursor()
 359             store_video_metadata(video_id) # only needed for pinning
 360             c.execute("""
 361                 INSERT OR REPLACE INTO flags (user, video_id, display)
 362                 VALUES (?, ?, ?)
 363             """, (token, video_id, display))
 364         undo_flash(video_id, action)
 365     else:
 366         flash("unsupported action", "error")
 367     return redirect(request.url, code=303)
 368
 369 @frontend.route('/manage/subscriptions', methods=['POST'])
 370 @login_required
 371 def manage_subscriptions():
 372     token = current_user.token
 373     if 'subscribe' in request.form:
 374         some_id = request.form.get("subscribe")
 375         match = re.search(r"(UC[A-Za-z0-9_-]{22})", some_id)
 376         if match:
 377             some_id = match.group(1)
 378             id_type = "channel"
 379         else:
 380             match = re.search(r"((?:PL|LL|EC|UU|FL|UL|OL)[A-Za-z0-9_-]{10,})", some_id)
 381             if match:  # NOTE: PL-playlists are 32chars, others differ in length.
 382                 some_id = match.group(1)
 383                 id_type = "playlist"
 384             else:
 385                 flash("not a valid/subscribable URI", "error")
 386                 return redirect(request.url, code=303)
 387         with sqlite3.connect(cf['global']['database']) as conn:
 388             #with conn.cursor() as c:
 389                 c = conn.cursor()
 390                 c.execute("""
 391                         INSERT OR IGNORE INTO subscriptions (user, channel_id, type)
 392                         VALUES (?, ?, ?)
 393                 """, (token, some_id, id_type))
 394                 # TODO: sql-error-handling, asynchronically calling update-subs.pl
 395         undo_flash(some_id, 'subscribe')
 396
 397     elif 'unsubscribe' in request.form:
 398         some_id = request.form.get("unsubscribe")
 399         with sqlite3.connect(cf['global']['database']) as conn:
 400             #with conn.cursor() as c:
 401                 c = conn.cursor()
 402                 c.execute("""
 403                         DELETE FROM subscriptions
 404                         WHERE user = ? AND channel_id = ?
 405                 """, (token, some_id))
 406                 # TODO: sql-error-handling, report success
 407         undo_flash(some_id, 'unsubscribe')
 408
 409     else:
 410         flash("unsupported action", "error")
 411
 412     return redirect(request.url, code=303)
 413
 414 @frontend.route('/vi/<vid>/<res>.jpg')
 415 def redirect_thumbnails(vid, res):
 416     return redirect(f"https://i.ytimg.com{request.path}", code=301)
 417
 418 @frontend.record
 419 def redirect_youtube_dot_com(state):
 420     """
 421     This is executed when the blueprint is loaded dynamically builds a number
 422     of routes so that URLs like
 423     https://subscriptions.gir.st/https://www.youtube.com/watch?v=dQw4w9WgXcQ
 424     redirect to the /watch page. Works with /watch, /embed/ and youtu.be short
 425     links, with or without protocl and/or 'www'.
 426     """
 427     def real_redirect_youtube_dot_com(video_id=None):
 428         if not re.match(r"^[-_0-9A-Za-z]{11}$", video_id or ''): video_id = None
 429         if not video_id: video_id = request.args.get('v')
 430         return redirect(url_for('.watch', v=video_id))
 431
 432     for protocol in ("", "http://", "https://"):
 433         for prefix in ("", "www.", "m."):
 434             for domain in ("youtube.com", "youtu.be", "youtube-nocookie.com"):
 435                 for urlpath in ("/watch", "/embed/<video_id>", "/live/<video_id>", "/shorts/<video_id>", "/<video_id>"):
 436                     if domain != "youtu.be" and urlpath == "/<video_id>":
 437                         continue # that's a channel, not a video
 438                     frontend.add_url_rule(
 439                         f"/{protocol}{prefix}{domain}{urlpath}",
 440                         view_func=real_redirect_youtube_dot_com,
 441                         strict_slashes=False
 442                     )
 443
 444 def undo_flash(thing_id, action):
 445     undo_action, past_action = {
 446         'pin': ('unpin', 'pinned'),
 447         'unpin': ('pin', 'unpinned'),
 448         'hide': ('unhide', 'hidden'),
 449         'unhide': ('hide', 'unhidden'),
 450         'subscribe': ('unsubscribe', 'subscribed'),
 451         'unsubscribe': ('subscribe', 'unsubscribed'),
 452     }.get(action)
 453     if 'subscribe' in action and thing_id.startswith('UC'):
 454         thing = "channel"
 455         thing_url = url_for('.channel', channel_id=thing_id)
 456     elif 'subscribe' in action:
 457         thing = "playlist"
 458         thing_url = url_for('.playlist', playlist_id=thing_id)
 459     else:
 460         thing = "video"
 461         thing_url = url_for('.watch', v=thing_id)
 462     flash(f'''
 463         <form method=post><input type=hidden name="{undo_action}" value="{thing_id}">
 464         <a href="{thing_url}">{thing}</a> {past_action}.
 465         <label><input type="submit" hidden>
 466         <span style="text-decoration:underline;cursor:pointer">undo</span>.
 467         </label></form>''', "info")
 468
 469 @frontend.app_template_filter('format_date')
 470 def format_date(s):
 471     import datetime # can't import at top level, because it is inherited from common
 472     (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'
 473     M = '_ Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec'.split()
 474     if y == datetime.datetime.now().year:
 475         return f"{d} {M[m]}"
 476     else:
 477         return f"{M[m]} '{y%100}"
 478
 479 @frontend.app_template_filter('format_time')
 480 def format_time(i):
 481     if i is None:
 482         return None
 483     h = i // (60*60)
 484     m = i // 60 % 60
 485     s = i % 60
 486     return '%d:%02d:%02d' % (h,m,s) if h else '%02d:%02d' % (m,s)
 487
 488 @frontend.app_template_filter('timeoffset')
 489 def timeoffset(s):
 490     if s is None:
 491         return None
 492     match = re.match(r"^(\d+)s?$", s)  # e.g. 2040s
 493     if match:
 494         return match.group(1)
 495     match = re.match(r"^(?:(\d+)h)?(?:(\d+)m)?(?:(\d+)s)?$", s)  # e.g. 34m, 1h23s
 496     if match:
 497         return ":".join([n.zfill(2) for n in match.groups('0')])
 498     return None