Adding ignored domains list to urlspamcog

README.md

@@ -4,7 +4,7 @@ Discord bot for detecting and dealing with jerks. This bot is Rocketsoup's fault
 
 ## Requirements
 
-* Python 3.9
+* Python 3.10
 
 ## Setup
 

rocketbot/cogs/urlspamcog.py

@@ -5,7 +5,7 @@ import re
 from datetime import timedelta
 from typing import Literal
 
-from discord import Member, Message
+from discord import Guild, Member, Message
 from discord import utils as discordutils
 from discord.ext.commands import Cog
 from discord.utils import escape_markdown
@@ -69,6 +69,14 @@ class URLSpamCog(BaseCog, name='URL Spam'):
 		enum_values={'nothing', 'modwarn', 'modwarndelete',
 					 'chatwarn', 'chatwarndelete', 'delete', 'kick', 'ban'},
 	)
+	SETTING_IGNORED_DOMAINS = CogSetting(
+		'ignoreddomains',
+		str,
+		default_value='tenor.com;giphy.com;imgur.com',
+		brief='List of allowed domains to exclude from detecting',
+		description='A semicolon-delimited list of domains that can be posted '
+					'without triggering a warning. E.g. GIF hosting sites.'
+	)
 
 	def __init__(self, bot):
 		super().__init__(
@@ -80,79 +88,115 @@ class URLSpamCog(BaseCog, name='URL Spam'):
 		self.add_setting(URLSpamCog.SETTING_ACTION)
 		self.add_setting(URLSpamCog.SETTING_JOIN_AGE)
 		self.add_setting(URLSpamCog.SETTING_DECEPTIVE_ACTION)
+		self.add_setting(URLSpamCog.SETTING_IGNORED_DOMAINS)
 
 	@Cog.listener()
 	async def on_message(self, message: Message):
 		"""Event listener"""
 		if message.author is None or \
-				message.author.bot or \
 				message.guild is None or \
 				message.channel is None or \
 				message.content is None:
+			self.__trace("Missing message data")
+			return
+		if message.author.bot:
+			self.__trace("Message from bot")
+			return
+		if message.channel.permissions_for(message.author).ban_members:
+			self.__trace("User exempt")
 			return
 		if not self.get_guild_setting(message.guild, self.SETTING_ENABLED):
+			self.__trace("Cog disabled")
+			return
+		urls = self.__find_urls(message.content)
+		if len(urls) == 0:
+			self.__trace("Contains no URLs")
+			return
+		if self.__all_ignored_domains(message.guild, urls):
+			self.__trace("All URLs are ignored domains")
 			return
 		await self.check_message_recency(message)
 		await self.check_deceptive_links(message)
 
+	def __all_ignored_domains(self, guild: Guild, urls: list[str]) -> bool:
+		domain_list: str = self.get_guild_setting(guild, self.SETTING_IGNORED_DOMAINS)
+		self.__trace(f"Domain list is {domain_list}")
+		domains = domain_list.split(';')
+		self.__trace(f"Split into {len(domains)} domains")
+		for url in urls:
+			if not self.__is_ignored_url(url, domains):
+				return False
+		return True
+
+	def __is_ignored_url(self, url: str, ignored_domains: list[str]) -> bool:
+		p = re.compile(r'https?://([\w\.]+)')
+		domain = re.search(p, url).group(1)
+		for ignored_domain in ignored_domains:
+			if domain.endswith(ignored_domain.strip()):
+				self.__trace(f"URL {url} matches ignored domain {ignored_domain}")
+				return True
+		self.__trace("URL {url} not ignored")
+		return False
+
 	async def check_message_recency(self, message: Message):
 		"""Checks if the message was sent too recently by a new user"""
 		action = self.get_guild_setting(message.guild, self.SETTING_ACTION)
 		join_seconds = self.get_guild_setting(message.guild, self.SETTING_JOIN_AGE)
 		min_join_age = timedelta(seconds=join_seconds)
 		if action == 'nothing':
-			return
-		if not self.__contains_url(message.content):
+			self.__trace("Configured action is nothing")
 			return
 		join_age = message.created_at - message.author.joined_at
 		join_age_str = describe_timedelta(join_age)
-		if join_age < min_join_age:
-			context = URLSpamContext(message)
-			needs_attention = False
-			if action == 'modwarn':
-				needs_attention = not self.was_warned_recently(message.author)
-				self.log(message.guild, f'New user {message.author.name} ' + \
-					f'({message.author.id}) posted URL {join_age_str} after ' + \
-					'joining.' + (' Mods alerted.' if needs_attention else ''))
-			elif action == 'delete':
-				await message.delete()
-				context.is_deleted = True
-				self.log(message.guild, f'New user {message.author.name} ' + \
-					f'({message.author.id}) posted URL {join_age_str} after ' + \
-					'joining. Message deleted.')
-			elif action == 'kick':
-				await message.delete()
-				context.is_deleted = True
-				await message.author.kick(
-					reason=f'Rocketbot: Posted a link {join_age_str} after joining')
-				context.is_kicked = True
-				self.log(message.guild, f'New user {message.author.name} ' + \
-					f'({message.author.id}) posted URL {join_age_str} after ' + \
-					'joining. User kicked.')
-			elif action == 'ban':
-				await message.author.ban(
-					reason=f'Rocketbot: User posted a link {join_age_str} after joining',
-					delete_message_days=1)
-				context.is_deleted = True
-				context.is_kicked = True
-				context.is_banned = True
-				self.log(message.guild, f'New user {message.author.name} ' + \
-					f'({message.author.id}) posted URL {join_age_str} after ' + \
-					'joining. User banned.')
-			bm = BotMessage(
-					message.guild,
-					f'User {message.author.mention} posted a URL ' + \
-					f'{join_age_str} after joining: {message.jump_url}',
-					type = BotMessage.TYPE_MOD_WARNING if needs_attention else BotMessage.TYPE_INFO,
-					context = context)
-			bm.quote = discordutils.remove_markdown(message.clean_content)
-			await bm.set_reactions(BotMessageReaction.standard_set(
-				did_delete=context.is_deleted,
-				did_kick=context.is_kicked,
-				did_ban=context.is_banned))
-			await self.post_message(bm)
-			if needs_attention:
-				self.record_warning(message.author)
+		if join_age > min_join_age:
+			self.__trace("User has been member long enough")
+			return
+		context = URLSpamContext(message)
+		needs_attention = False
+		if action == 'modwarn':
+			needs_attention = not self.was_warned_recently(message.author)
+			self.log(message.guild, f'New user {message.author.name} ' + \
+				f'({message.author.id}) posted URL {join_age_str} after ' + \
+				'joining.' + (' Mods alerted.' if needs_attention else ''))
+		elif action == 'delete':
+			await message.delete()
+			context.is_deleted = True
+			self.log(message.guild, f'New user {message.author.name} ' + \
+				f'({message.author.id}) posted URL {join_age_str} after ' + \
+				'joining. Message deleted.')
+		elif action == 'kick':
+			await message.delete()
+			context.is_deleted = True
+			await message.author.kick(
+				reason=f'Rocketbot: Posted a link {join_age_str} after joining')
+			context.is_kicked = True
+			self.log(message.guild, f'New user {message.author.name} ' + \
+				f'({message.author.id}) posted URL {join_age_str} after ' + \
+				'joining. User kicked.')
+		elif action == 'ban':
+			await message.author.ban(
+				reason=f'Rocketbot: User posted a link {join_age_str} after joining',
+				delete_message_days=1)
+			context.is_deleted = True
+			context.is_kicked = True
+			context.is_banned = True
+			self.log(message.guild, f'New user {message.author.name} ' + \
+				f'({message.author.id}) posted URL {join_age_str} after ' + \
+				'joining. User banned.')
+		bm = BotMessage(
+				message.guild,
+				f'User {message.author.mention} posted a URL ' + \
+				f'{join_age_str} after joining: {message.jump_url}',
+				type = BotMessage.TYPE_MOD_WARNING if needs_attention else BotMessage.TYPE_INFO,
+				context = context)
+		bm.quote = discordutils.remove_markdown(message.clean_content)
+		await bm.set_reactions(BotMessageReaction.standard_set(
+			did_delete=context.is_deleted,
+			did_kick=context.is_kicked,
+			did_ban=context.is_banned))
+		await self.post_message(bm)
+		if needs_attention:
+			self.record_warning(message.author)
 
 	async def check_deceptive_links(self, message: Message):
 		"""
@@ -302,7 +346,16 @@ class URLSpamCog(BaseCog, name='URL Spam'):
 			did_kick=context.is_kicked,
 			did_ban=context.is_banned))
 
+	def __trace(self, message: str):
+		# print(f'URLSpamCog: {message}')
+		pass
+
 	@classmethod
 	def __contains_url(cls, text: str) -> bool:
 		p = re.compile(r'https?://\S+')
 		return p.search(text) is not None
+
+	@classmethod
+	def __find_urls(cls, text: str) -> list[str]:
+		p = re.compile(r'https?://\S+')
+		return re.findall(p, text)