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