transform urls matched in messages

We add the transformed urls to the set to account for the way discord renders relative urls in cases like Twitter's

1 files changed, 9 insertions, 2 deletions

diff --git a/bot/exts/filters/filtering.py b/bot/exts/filters/filtering.py
index e4df0b1fd..db13df9b2 100644
--- a/bot/exts/filters/filtering.py
+++ b/bot/exts/filters/filtering.py
@@ -23,10 +23,12 @@ from bot.constants import Bot as BotConfig, Channels, Colours, Filter, Guild, Ic
 from bot.exts.events.code_jams._channels import CATEGORY_NAME as JAM_CATEGORY_NAME
 from bot.exts.moderation.modlog import ModLog
 from bot.log import get_logger
+from bot.utils.helpers import remove_subdomain_from_url
 from bot.utils.messages import format_user
 
 log = get_logger(__name__)
 
+
 # Regular expressions
 CODE_BLOCK_RE = re.compile(
     r"(?P<delim>``?)[^`]+?(?P=delim)(?!`+)"  # Inline codeblock
@@ -583,7 +585,7 @@ class Filtering(Cog):
         """
         text = self.clean_input(text)
 
-        # Remove backslashes to prevent escape character aroundfuckery like
+        # Remove backslashes to prevent escape character around fuckery like
         # discord\.gg/gdudes-pony-farm
         text = text.replace("\\", "")
 
@@ -648,7 +650,12 @@ class Filtering(Cog):
         if msg.embeds:
             for embed in msg.embeds:
                 if embed.type == "rich":
-                    urls = URL_RE.findall(msg.content)
+                    urls = set(URL_RE.findall(msg.content))
+                    # This is due to way discord renders relative urls in Embdes
+                    # if we send the following url: https://mobile.twitter.com/something
+                    # Discord renders it as https://twitter.com/something
+                    for url in urls:
+                        urls.add(remove_subdomain_from_url(url))
                     if not embed.url or embed.url not in urls:
                         # If `embed.url` does not exist or if `embed.url` is not part of the content
                         # of the message, it's unlikely to be an auto-generated embed by Discord.