From 2d417fb3bbd155583179d2ad172e1b3c1f5e9eaf Mon Sep 17 00:00:00 2001
From: girst <tobi@isticktoit.net>
Date: Sun, 18 Oct 2020 01:13:41 +0200
Subject: [PATCH] better /<something> algorithm

note: using strict_slashes=False instread of two route decorators to
avoid useless redirect.
---
 app/youtube/__init__.py | 19 +++++++++++--------
 app/youtube/lib.py      |  8 ++++++++
 2 files changed, 19 insertions(+), 8 deletions(-)

diff --git a/app/youtube/__init__.py b/app/youtube/__init__.py
index 4421cff..899c35c 100644
--- a/app/youtube/__init__.py
+++ b/app/youtube/__init__.py
@@ -6,6 +6,7 @@ from urllib.parse import urlparse
 #from flask_login import current_user, login_required
 from flask_login import LoginManager, UserMixin, current_user, login_user, logout_user, login_required
 from flask import Blueprint, render_template, request, redirect, flash, url_for, jsonify, g
+from werkzeug.exceptions import NotFound
 
 from ..common.common import *
 from .lib import *
@@ -149,16 +150,18 @@ def watch():
 def embed(video_id):
     return redirect(url_for('.watch', v=video_id, t=request.args.get('start')))
 
-@frontend.route('/<something>')
-@frontend.route('/<something>/')
+@frontend.route('/<something>', strict_slashes=False)
 def plain_user_or_video(something):
-    # yt.com redirects those to channels, but we also want to catch invidious
-    # short-urls. a "just-a-variable-endpoint" has lowest priority, so this
-    # shouldn't interfere with anything.
-    if re.match(r"^[-_0-9A-Za-z]{11}$", something): # looks like a video id
-        return redirect(url_for('.watch', v=something, t=request.args.get('t')))
-    else:  # maybe a channel/user name?
+    # yt.com interprets this as a username, but we also want to catch youtu.be
+    # short-urls. so we check if it's a channel by querying the RSS feed (this
+    # shoudn't be rate-limited); if that fails, check if it looks like a video
+    # id; or finally give up.
+    if channel_exists(something):
         return redirect(url_for('.channel', channel_id=something))
+    elif re.match(r"^[-_0-9A-Za-z]{11}$", something): # looks like a video id
+        return redirect(url_for('.watch', v=something, t=request.args.get('t')))
+    else: # ¯\_(ツ)_/¯
+        raise NotFound
 
 @frontend.route('/channel/<channel_id>/<subpage>')
 @frontend.route('/user/<channel_id>/<subpage>')
diff --git a/app/youtube/lib.py b/app/youtube/lib.py
index d49c38b..9bc23ca 100644
--- a/app/youtube/lib.py
+++ b/app/youtube/lib.py
@@ -1,3 +1,5 @@
+import re
+import requests
 from urllib.parse import urlparse
 
 from ..common.common import video_metadata
@@ -92,3 +94,9 @@ def prepare_metadata(metadata):
         'subtitles': subtitles,
     }
 
+def channel_exists(feed_id):
+    feed_type = "channel_id" if re.match(r"^UC[A-Za-z0-9_-]{22}$", feed_id) else "user"
+    r = requests.head("https://www.youtube.com/feeds/videos.xml", params={
+        feed_type: feed_id,
+    })
+    return r.ok
-- 
2.39.3