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