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