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