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