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