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
   9 from werkzeug.exceptions import NotFound
  10
  11 from ..common.common import *
  12 from .lib import *
  13
  14 frontend = Blueprint('youtube', __name__,
  15     template_folder='templates',
  16     static_folder='static',
  17     static_url_path='/static/yt')
  18
  19 @frontend.route('/')
  20 def index():
  21     return redirect(url_for('.feed'), code=302)
  22
  23 @frontend.route('/feed/subscriptions')
  24 # disabled for guest user: @login_required
  25 def feed():
  26     if current_user.is_anonymous:
  27         token = 'guest'
  28         if 'welcome_message' in cf['frontend']:
  29             flash(cf['frontend']['welcome_message'], "welcome")
  30     else:
  31         token = current_user.token
  32     page = int(request.args.get('page', 0))
  33     with sqlite3.connect(cf['global']['database']) as conn:
  34         c = conn.cursor()
  35         c.execute("""
  36            SELECT videos.id, channel_id, name, title, length, livestream, published, playlist_videos.playlist_id, display
  37              FROM videos
  38              JOIN channels ON videos.channel_id = channels.id
  39     LEFT JOIN playlist_videos ON (videos.id = playlist_videos.video_id)
  40         LEFT JOIN flags ON (videos.id = flags.video_id) AND (flags.user = ?)
  41         WHERE (channel_id IN (SELECT channel_id FROM subscriptions WHERE user=? AND type = 'channel')
  42                OR playlist_videos.playlist_id IN (SELECT channel_id FROM subscriptions WHERE user=? AND type = 'playlist')
  43                    OR flags.display = 'pinned')
  44                   AND flags.display IS NOT 'hidden'
  45          ORDER BY (display = 'pinned') DESC, crawled DESC
  46             LIMIT 36
  47            OFFSET 36*?""", (token, token, token, page))
  48         rows = [{
  49             'video_id': video_id,
  50             'channel_id': channel_id,
  51             'author': author,
  52             'title': title,
  53             'length': length,
  54             'livestream': livestream,
  55             'published': published,
  56             'playlist': playlist,
  57             'pinned': display == 'pinned',
  58         } for (video_id, channel_id, author, title, length, livestream, published, playlist, display) in c.fetchall()]
  59     return render_template('index.html.j2', rows=rows, page=page)
  60
  61 @frontend.route('/watch')
  62 def watch():
  63     if current_user.is_anonymous:
  64         token = 'guest'
  65     else:
  66         token = current_user.token
  67
  68     if not 'v' in request.args:
  69         return "missing video id", 400
  70
  71     plaintextheaders = {
  72         'content-type': 'text/plain',
  73         'Link': "<data:text/css,body%7Bcolor:%23eee;background:%23333%7D>; rel=stylesheet;"
  74     }
  75
  76     video_id = request.args.get('v')
  77     sts, algo = get_cipher()
  78     video_url, stream_map, metadata, error, errdetails = get_video_info(video_id, sts, algo)
  79
  80     extra = {'geolocked':'local=1', 'livestream':'raw=0'}.get(error,'')
  81     invidious_url = f"https://invidious.snopyta.org/watch?v={video_id}&{extra}"
  82     errdetails = {
  83         'banned':     "Instance is being rate limited.",
  84         'malformed':  "Video ID is invalid.",
  85         'geolocked':  "This video is geolocked.",
  86         'livestream': "Livestreams not yet supported.",
  87         'exhausted':  errdetails or "Couldn't extract video URLs.",
  88         'player':     errdetails,
  89     }.get(error)
  90
  91     # if the video is geolocked, and the proxy is enabled, we can still play it:
  92     try:
  93         if error == 'geolocked':
  94             videoplayback = url_for('proxy.videoplayback')
  95
  96             query = urlparse(video_url).query
  97             video_url = f"{videoplayback}?{query}"
  98             for s in stream_map['adaptive']:
  99                 query = urlparse(s['url']).query
 100                 s['url'] = f"{videoplayback}?{query}"
 101             for s in stream_map['muxed']:
 102                 query = urlparse(s['url']).query
 103                 s['url'] = f"{videoplayback}?{query}"
 104
 105             error = None
 106     except: pass
 107
 108     show = request.args.get("show")
 109     if show == "raw":
 110         if error:
 111             msg = errdetails if error=='player' else f"{error.upper()}: {errdetails}"
 112             return f"{msg}\n\nRedirecting to Invidious.", 502, {
 113                 'Refresh': f'2; URL={invidious_url}&raw=1',
 114                 **plaintextheaders}
 115         return redirect(video_url, code=307)
 116     elif show == "json":
 117         if error and not metadata:
 118             return {'error': True, error: errdetails}, 400 # TODO: better (test _CpR4o81XQc)
 119         return jsonify(metadata)
 120     elif show == "audio":
 121         # sorting: we want to prioritize mp4a over opus, and sort by highest quality first
 122         # todo: geolocking; prefer open format?
 123         if error and not metadata:
 124             msg = errdetails if error=='player' else f"{error.upper()}: {errdetails}"
 125             return msg, 400, plaintextheaders # TODO: nicer
 126         stream = next(iter(sorted([
 127                     stream for stream in stream_map['adaptive']
 128                     if stream['mimeType'].startswith('audio/')],
 129                 key=lambda stream: ('opus' not in stream['mimeType'],
 130                     stream['bitrate']), reverse=True)
 131             ),{}).get('url')
 132         return redirect(stream)
 133     else:
 134         if error and not metadata: # e.g. malformed, private/deleted video, ...
 135             return errdetails + f"<br>Try visiting <a href='{invidious_url}'>Invidious</a>.",400 # TODO: nicer
 136         meta = prepare_metadata(metadata)
 137         with sqlite3.connect(cf['global']['database']) as conn:
 138             c = conn.cursor()
 139             c.execute("""
 140             SELECT COUNT((
 141                 SELECT 1 FROM subscriptions WHERE channel_id = ? AND user = ?
 142             )), COUNT((
 143                 SELECT 1 FROM flags WHERE video_id = ? AND display = 'pinned' AND user = ?
 144             ))""", (meta['channel_id'], token, video_id, token))
 145             (is_subscribed, is_pinned) = c.fetchone()
 146             return render_template('watch.html.j2',
 147                 video_id=video_id, video_url=video_url, stream_map=stream_map,
 148                 video_error=error, errdetails=errdetails, invidious_url=invidious_url,
 149                 is_pinned=is_pinned, is_subscribed=is_subscribed,
 150                 **meta)
 151
 152 @frontend.route('/embed/videoseries')
 153 def embed_videoseries():
 154     return redirect(url_for('.playlist', list=request.args.get('list')))
 155 @frontend.route('/embed/<video_id>')
 156 def embed(video_id):
 157     if video_id == "videoseries":
 158         return redirect(url_for('.playlist', list=request.args.get('list')))
 159
 160     return redirect(url_for('.watch', v=video_id, t=request.args.get('start')))
 161
 162 @frontend.route('/<something>', strict_slashes=False)
 163 def plain_user_or_video(something):
 164     # yt.com interprets this as a username, but we also want to catch youtu.be
 165     # short-urls. so we check if it's a channel by querying the RSS feed (this
 166     # shoudn't be rate-limited); if that fails, check if it looks like a video
 167     # id; or finally give up.
 168     if '.' not in something and channel_exists(something):
 169         # periods are not valid in usernames, vanity urls or ucids, but common
 170         # in urls that get crawled by bots (e.g. index.php). failing early
 171         # reduces the amount of invalid channel names getting looked up.
 172         return redirect(url_for('.channel', channel_id=something))
 173     elif re.match(r"^[-_0-9A-Za-z]{11}$", something): # looks like a video id
 174         return redirect(url_for('.watch', v=something, t=request.args.get('t')))
 175     else: # ¯\_(ツ)_/¯
 176         # XXX: something == 'thethoughtemporium' -> 404s
 177         raise NotFound("Note: some usernames not recognized; try searching it")
 178
 179 @frontend.route('/channel/<channel_id>/<subpage>')
 180 @frontend.route('/user/<channel_id>/<subpage>')
 181 @frontend.route('/c/<channel_id>/<subpage>')
 182 @frontend.route('/channel/<channel_id>/')
 183 @frontend.route('/user/<channel_id>/')
 184 @frontend.route('/c/<channel_id>/')
 185 def channel(channel_id, _=None):
 186     token = getattr(current_user, 'token', 'guest')
 187
 188     if re.match(r"(UC[A-Za-z0-9_-]{22})", channel_id):
 189         xmlfeed = fetch_xml("channel_id", channel_id)
 190     else:
 191         xmlfeed = fetch_xml("user", channel_id)
 192
 193     if not xmlfeed:
 194         return "not found or something", 404 # XXX
 195     title, author, videos, channel_id, _ = parse_xml(xmlfeed)
 196
 197     with sqlite3.connect(cf['global']['database']) as conn:
 198         c = conn.cursor()
 199         c.execute("""
 200             SELECT COUNT(*)
 201               FROM subscriptions
 202              WHERE channel_id = ? AND user = ?
 203         """, (channel_id, token))
 204         (is_subscribed,) = c.fetchone()
 205
 206     return render_template('xmlfeed.html.j2', title=author, rows=videos,
 207         is_subscribed=is_subscribed, channel_id=channel_id)
 208
 209 @frontend.route('/playlist')
 210 def playlist():
 211     playlist_id = request.args.get('list')
 212     if not playlist_id:
 213         return "bad list id", 400 # todo
 214
 215     xmlfeed = fetch_xml("playlist_id", playlist_id)
 216     if not xmlfeed:
 217         return "not found or something", 404 # XXX
 218     title, author, videos, _, _ = parse_xml(xmlfeed)
 219     return render_template('xmlfeed.html.j2', title=f"{title} by {author}", rows=videos)
 220
 221 @frontend.route('/api/timedtext')
 222 def timedtext():
 223     r = requests.get("https://www.youtube.com/api/timedtext", request.args.to_dict())
 224     # Note: in srv1 format, xmlentities are double-encoded m( a smart quote is
 225     # even worse: it's '&amp;39;<smartquote>' wtf!? (at least vvt seems ok)
 226     if not r.ok:
 227         return "error: {r.text}", 400 # TODO: better
 228     retval = r.text
 229     if request.args.get('fmt') == 'vtt' and request.args.get('kind') == 'asr':
 230         # autocaptions are extremely confusing, and stuck in the lower-left corner. fix it up a bit
 231         retval = re.sub(r"<.+?>", "", retval) # remove inline html-like markup that times each word/adds styles
 232         retval = retval.replace("align:start position:0%", "") # let browser position the text itself
 233         # each subtitle-line is repeated twice (first on the lower line, then
 234         # on the next "frame" on the upper line). we want to remove the
 235         # repetition, as that's confusing without word and line animations:
 236         lines = retval.split('\n')
 237         retval = '\n'.join([line for line, prev in zip(lines, ['']+lines) if not " --> " in prev])
 238     return retval, {'Content-Type': r.headers.get("Content-Type")}
 239
 240 @frontend.route('/manage/subscriptions')
 241 # disabled for guest user: @login_required
 242 def subscription_manager():
 243     if current_user.is_anonymous:
 244         token = 'guest'
 245     else:
 246         token = current_user.token
 247     with sqlite3.connect(cf['global']['database']) as conn:
 248         #with conn.cursor() as c:
 249             c = conn.cursor()
 250             c.execute("""
 251                   SELECT subscriptions.channel_id, name, type,
 252                          (subscribed_until < datetime('now')) AS obsolete
 253                     FROM subscriptions
 254                     LEFT JOIN (SELECT name, id FROM channels
 255                        UNION
 256                        SELECT name, id FROM playlists
 257                       ) AS channels ON channels.id = subscriptions.channel_id
 258                     left JOIN websub ON channels.id = websub.channel_id
 259                    WHERE user = ?
 260                      AND subscriptions.type IN ('channel', 'playlist')
 261                 ORDER BY obsolete=0, name COLLATE NOCASE ASC""", (token,))
 262             rows = [{
 263                 'channel_id': channel_id,
 264                 'author': author or channel_id,
 265                 'type': type,
 266                 'subscribed_until': subscribed_until
 267             } for (channel_id, author, type, subscribed_until) in c.fetchall()]
 268     return render_template('subscription_manager.html.j2', rows=rows)
 269
 270 @frontend.route('/feed/subscriptions', methods=['POST'])
 271 @login_required
 272 def feed_post():
 273     token = current_user.token
 274     action = next(request.form.keys(), None)
 275     if action in ['pin', 'unpin', 'hide']:
 276         video_id = request.form.get(action)
 277         display = {
 278             'pin': 'pinned',
 279             'unpin': None,
 280             'hide': 'hidden',
 281         }[action]
 282         with sqlite3.connect(cf['global']['database']) as conn:
 283             c = conn.cursor()
 284             store_video_metadata(video_id) # only needed for pinning
 285             c.execute("""
 286                 INSERT OR REPLACE INTO flags (user, video_id, display)
 287                 VALUES (?, ?, ?)
 288             """, (token, video_id, display))
 289     else:
 290         flash("unsupported action", "error")
 291     return redirect(request.url, code=303)
 292
 293 @frontend.route('/manage/subscriptions', methods=['POST'])
 294 @login_required
 295 def manage_subscriptions():
 296     token = current_user.token
 297     if 'subscribe' in request.form:
 298         some_id = request.form.get("subscribe")
 299         match = re.search(r"(UC[A-Za-z0-9_-]{22})", some_id)
 300         if match:
 301             some_id = match.group(1)
 302             id_type = "channel"
 303         else:
 304             match = re.search(r"((?:PL|LL|EC|UU|FL|UL|OL)[A-Za-z0-9_-]{10,})", some_id)
 305             if match:  # NOTE: PL-playlists are 32chars, others differ in length.
 306                 some_id = match.group(1)
 307                 id_type = "playlist"
 308             else:
 309                 flash("not a valid/subscribable URI", "error")
 310                 return redirect(request.url, code=303)
 311         with sqlite3.connect(cf['global']['database']) as conn:
 312             #with conn.cursor() as c:
 313                 c = conn.cursor()
 314                 c.execute("""
 315                         INSERT OR IGNORE INTO subscriptions (user, channel_id, type)
 316                         VALUES (?, ?, ?)
 317                 """, (token, some_id, id_type))
 318                 # TODO: sql-error-handling, asynchronically calling update-subs.pl
 319
 320     elif 'unsubscribe' in request.form:
 321         some_id = request.form.get("unsubscribe")
 322         with sqlite3.connect(cf['global']['database']) as conn:
 323             #with conn.cursor() as c:
 324                 c = conn.cursor()
 325                 c.execute("""
 326                         DELETE FROM subscriptions
 327                         WHERE user = ? AND channel_id = ?
 328                 """, (token, some_id))
 329                 # TODO: sql-error-handling, report success
 330
 331     else:
 332         flash("unsupported action", "error")
 333
 334     return redirect(request.url, code=303)
 335
 336 def get_cipher():
 337     # reload cipher from database every 1 hour
 338     if 'cipher' not in g or time.time() - g.get('cipher_updated', 0) > 1 * 60 * 60:
 339         with sqlite3.connect(cf['global']['database']) as conn:
 340             c = conn.cursor()
 341             c.execute("SELECT sts, algorithm FROM cipher")
 342             g.cipher = c.fetchone()
 343             g.cipher_updated = time.time()
 344
 345     return g.cipher
 346
 347 #@frontend.teardown_appcontext
 348 #def teardown_db():
 349 #    db = g.pop('db', None)
 350 #
 351 #    if db is not None:
 352 #        db.close()
 353
 354
 355 @frontend.app_template_filter('format_date')
 356 def format_date(s):
 357     import datetime # can't import at top level, because it is inherited from common
 358     (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'
 359     M = '_ Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec'.split()
 360     if y == datetime.datetime.now().year:
 361         return f"{d} {M[m]}"
 362     else:
 363         return f"{M[m]} '{y%100}"
 364
 365 @frontend.app_template_filter('format_time')
 366 def format_time(i):
 367     if i is None:
 368         return None
 369     h = i // (60*60)
 370     m = i // 60 % 60
 371     s = i % 60
 372     return '%d:%02d:%02d' % (h,m,s) if h else '%02d:%02d' % (m,s)
 373
 374 @frontend.app_template_filter('timeoffset')
 375 def timeoffset(s):
 376     if s is None:
 377         return None
 378     match = re.match(r"^(\d+)s?$", s)  # e.g. 2040s
 379     if match:
 380         return match.group(1)
 381     match = re.match(r"^(?:(\d+)h)?(?:(\d+)m)?(?:(\d+)s)?$", s)  # e.g. 34m, 1h23s
 382     if match:
 383         return ":".join([n.zfill(2) for n in match.groups('0')])
 384     return None