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