Unescape HTML in URLs parsed from user input

Merge pull request #275 from mweinelt/url-unescape-html
2024-11-23 13:56:17 +00:00 · 2021-05-16 16:30:50 +02:00 · 2021-05-16 15:04:50 +02:00 · 2021-05-16 15:04:25 +02:00 · 2021-05-16 05:51:42 +02:00
2 changed files with 13 additions and 5 deletions
--- a/media/url_from_playlist.py
+++ b/media/url_from_playlist.py
@ -1,3 +1,4 @@
+import logging
 import youtube_dl
 from constants import tr_cli as tr
 import variables as var
@ -5,8 +6,10 @@ from media.item import item_builders, item_loaders, item_id_generators
 from media.url import URLItem, url_item_id_generator


+log = logging.getLogger("bot")
+
+
 def get_playlist_info(url, start_index=0, user=""):
-    items = []
    ydl_opts = {
        'extract_flat': 'in_playlist',
        'verbose': var.config.getboolean('debug', 'youtube_dl')
@ -23,6 +26,7 @@ def get_playlist_info(url, start_index=0, user=""):
    with youtube_dl.YoutubeDL(ydl_opts) as ydl:
        attempts = var.config.getint('bot', 'download_attempts', fallback=2)
        for i in range(attempts):
+            items = []
            try:
                info = ydl.extract_info(url, download=False)
                # # if url is not a playlist but a video
@ -55,8 +59,10 @@ def get_playlist_info(url, start_index=0, user=""):
                    }

                    items.append(music)
-            except:  # todo need to be specified
-                pass
+
+            except Exception as ex:
+                log.exception(ex, exc_info=True)
+                continue

            return items

--- a/util.py
+++ b/util.py
@ -2,6 +2,7 @@
 # coding=utf-8

 import hashlib
+import html
 import magic
 import os
 import io
@ -311,7 +312,8 @@ def get_url_from_input(string):
    match = re.search("(http|https)://(\S*)?/(\S*)", string, flags=re.IGNORECASE)
    if match:
        url = match[1].lower() + "://" + match[2].lower() + "/" + match[3]
-        return url
+        # https://github.com/mumble-voip/mumble/issues/4999
+        return html.unescape(url)
    else:
        return ""
Author	SHA1	Message	Date
azlux	1c283654fd	Unescape HTML in URLs parsed from user input Merge pull request #275 from mweinelt/url-unescape-html	2021-05-16 16:30:50 +02:00
Martin Weinelt	faddf4c6c2	Unescape HTML in URLs parsed from user input Escapes tokens like & to &, so GET parameters are correctly preserved. Fixes: #274	2021-05-16 15:04:50 +02:00
azlux	b0e3e65caf	Fix playlist items over multiple attemps Merge pull request #276 from mweinelt/fix-playback-attempts	2021-05-16 15:04:25 +02:00
Martin Weinelt	e6d12e36c7	Don't accumulate playlist items over multiple attemps Playlist items would previously have been added once for each attempt, due to items being scoped differently than the playlist parsing attempts. Each attempt would add to the playlist every time, since we only returned outside the attempts loop.	2021-05-16 05:51:42 +02:00