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