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         '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 an <a href='{invidious_url}'>Invidious instance</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('/<video_id>')
 153 def raw_video_id_url(video_id):
 154     # a "just-a-variable-endpoint" has lowest priority, but we check if it
 155     # looks like a video id anyways, so we can have more than one such endpoint
 156     # in the future.
 157     if not re.match(r"^[-_0-9A-Za-z]{11}$", video_id): # not actually a video id
 158         return fallback_route()
 159     return redirect(url_for('youtube.watch', v=video_id))
 160
 161 @frontend.route('/channel/<channel_id>/<subpage>')
 162 @frontend.route('/user/<channel_id>/<subpage>')
 163 @frontend.route('/c/<channel_id>/<subpage>')
 164 @frontend.route('/channel/<channel_id>/')
 165 @frontend.route('/user/<channel_id>/')
 166 @frontend.route('/c/<channel_id>/')
 167 def channel(channel_id, _=None):
 168     token = getattr(current_user, 'token', 'guest')
 169
 170     if re.match(r"(UC[A-Za-z0-9_-]{22})", channel_id):
 171         xmlfeed = fetch_xml("channel_id", channel_id)
 172     else:
 173         xmlfeed = fetch_xml("user", channel_id)
 174
 175     if not xmlfeed:
 176         return "not found or something", 404 # XXX
 177     title, author, videos, channel_id, _ = parse_xml(xmlfeed)
 178
 179     with sqlite3.connect(cf['global']['database']) as conn:
 180         c = conn.cursor()
 181         c.execute("""
 182             SELECT COUNT(*)
 183               FROM subscriptions
 184              WHERE channel_id = ? AND user = ?
 185         """, (channel_id, token))
 186         (is_subscribed,) = c.fetchone()
 187
 188     return render_template('xmlfeed.html.j2', title=author, rows=videos,
 189         is_subscribed=is_subscribed, channel_id=channel_id)
 190
 191 @frontend.route('/playlist')
 192 def playlist():
 193     playlist_id = request.args.get('list')
 194     if not playlist_id:
 195         return "bad list id", 400 # todo
 196
 197     xmlfeed = fetch_xml("playlist_id", playlist_id)
 198     if not xmlfeed:
 199         return "not found or something", 404 # XXX
 200     title, author, videos, _, _ = parse_xml(xmlfeed)
 201     return render_template('xmlfeed.html.j2', title=f"{title} by {author}", rows=videos)
 202
 203 @frontend.route('/api/timedtext')
 204 def timedtext():
 205     r = requests.get("https://www.youtube.com/api/timedtext", request.args.to_dict())
 206     # Note: in srv1 format, xmlentities are double-encoded m( a smart quote is
 207     # even worse: it's '&amp;39;<smartquote>' wtf!? (at least vvt seems ok)
 208     if not r.ok:
 209         return "error: {r.text}", 400 # TODO: better
 210     retval = r.text
 211     if request.args.get('fmt') == 'vtt' and request.args.get('kind') == 'asr':
 212         # autocaptions are extremely confusing, and stuck in the lower-left corner. fix it up a bit
 213         retval = re.sub(r"<.+?>", "", retval) # remove inline html-like markup that times each word/adds styles
 214         retval = retval.replace("align:start position:0%", "") # let browser position the text itself
 215         # each subtitle-line is repeated twice (first on the lower line, then
 216         # on the next "frame" on the upper line). we want to remove the
 217         # repetition, as that's confusing without word and line animations:
 218         lines = retval.split('\n')
 219         retval = '\n'.join([line for line, prev in zip(lines, ['']+lines) if not " --> " in prev])
 220     return retval, {'Content-Type': r.headers.get("Content-Type")}
 221
 222 @frontend.route('/manage/subscriptions')
 223 # disabled for guest user: @login_required
 224 def subscription_manager():
 225     if current_user.is_anonymous:
 226         token = 'guest'
 227     else:
 228         token = current_user.token
 229     with sqlite3.connect(cf['global']['database']) as conn:
 230         #with conn.cursor() as c:
 231             c = conn.cursor()
 232             c.execute("""
 233                   SELECT subscriptions.channel_id, name, type,
 234                          (subscribed_until < datetime('now')) AS obsolete
 235                     FROM subscriptions
 236                     LEFT JOIN (SELECT name, id FROM channels
 237                        UNION
 238                        SELECT name, id FROM playlists
 239                       ) AS channels ON channels.id = subscriptions.channel_id
 240                     left JOIN websub ON channels.id = websub.channel_id
 241                    WHERE user = ?
 242                      AND subscriptions.type IN ('channel', 'playlist')
 243                 ORDER BY obsolete=0, name COLLATE NOCASE ASC""", (token,))
 244             rows = [{
 245                 'channel_id': channel_id,
 246                 'author': author or channel_id,
 247                 'type': type,
 248                 'subscribed_until': subscribed_until
 249             } for (channel_id, author, type, subscribed_until) in c.fetchall()]
 250     return render_template('subscription_manager.html.j2', rows=rows)
 251
 252 @frontend.route('/feed/subscriptions', methods=['POST'])
 253 @login_required
 254 def feed_post():
 255     token = current_user.token
 256     action = next(request.form.keys(), None)
 257     if action in ['pin', 'unpin', 'hide']:
 258         video_id = request.form.get(action)
 259         display = {
 260             'pin': 'pinned',
 261             'unpin': None,
 262             'hide': 'hidden',
 263         }[action]
 264         with sqlite3.connect(cf['global']['database']) as conn:
 265             c = conn.cursor()
 266             store_video_metadata(video_id) # only needed for pinning
 267             c.execute("""
 268                 INSERT OR REPLACE INTO flags (user, video_id, display)
 269                 VALUES (?, ?, ?)
 270             """, (token, video_id, display))
 271     else:
 272         flash("unsupported action", "error")
 273     return redirect(request.url, code=303)
 274
 275 @frontend.route('/manage/subscriptions', methods=['POST'])
 276 @login_required
 277 def manage_subscriptions():
 278     token = current_user.token
 279     if 'subscribe' in request.form:
 280         some_id = request.form.get("subscribe")
 281         match = re.search(r"(UC[A-Za-z0-9_-]{22})", some_id)
 282         if match:
 283             some_id = match.group(1)
 284             id_type = "channel"
 285         else:
 286             match = re.search(r"((?:PL|LL|EC|UU|FL|UL|OL)[A-Za-z0-9_-]{10,})", some_id)
 287             if match:  # NOTE: PL-playlists are 32chars, others differ in length.
 288                 some_id = match.group(1)
 289                 id_type = "playlist"
 290             else:
 291                 flash("not a valid/subscribable URI", "error")
 292                 return redirect(request.url, code=303)
 293         with sqlite3.connect(cf['global']['database']) as conn:
 294             #with conn.cursor() as c:
 295                 c = conn.cursor()
 296                 c.execute("""
 297                         INSERT OR IGNORE INTO subscriptions (user, channel_id, type)
 298                         VALUES (?, ?, ?)
 299                 """, (token, some_id, id_type))
 300                 # TODO: sql-error-handling, asynchronically calling update-subs.pl
 301
 302     elif 'unsubscribe' in request.form:
 303         some_id = request.form.get("unsubscribe")
 304         with sqlite3.connect(cf['global']['database']) as conn:
 305             #with conn.cursor() as c:
 306                 c = conn.cursor()
 307                 c.execute("""
 308                         DELETE FROM subscriptions
 309                         WHERE user = ? AND channel_id = ?
 310                 """, (token, some_id))
 311                 # TODO: sql-error-handling, report success
 312
 313     else:
 314         flash("unsupported action", "error")
 315
 316     return redirect(request.url, code=303)
 317
 318 def get_cipher():
 319     # reload cipher from database every 1 hour
 320     if 'cipher' not in g or time.time() - g.get('cipher_updated', 0) > 1 * 60 * 60:
 321         with sqlite3.connect(cf['global']['database']) as conn:
 322             c = conn.cursor()
 323             c.execute("SELECT sts, algorithm FROM cipher")
 324             g.cipher = c.fetchone()
 325             g.cipher_updated = time.time()
 326
 327     return g.cipher
 328
 329 #@frontend.teardown_appcontext
 330 #def teardown_db():
 331 #    db = g.pop('db', None)
 332 #
 333 #    if db is not None:
 334 #        db.close()
 335
 336
 337 @frontend.app_template_filter('format_date')
 338 def format_date(s):
 339     import datetime # can't import at top level, because it is inherited from common
 340     (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'
 341     M = '_ Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec'.split()
 342     if y == datetime.datetime.now().year:
 343         return f"{d} {M[m]}"
 344     else:
 345         return f"{M[m]} '{y%100}"
 346
 347 @frontend.app_template_filter('format_time')
 348 def format_time(i):
 349     if i is None:
 350         return None
 351     h = i // (60*60)
 352     m = i // 60 % 60
 353     s = i % 60
 354     return '%d:%02d:%02d' % (h,m,s) if h else '%02d:%02d' % (m,s)