From 31f1e285c667c2204194ad6db20d343dc622affd Mon Sep 17 00:00:00 2001
From: urlbot <urlbot@eagle.local.yeeer.net>
Date: Fri, 28 Nov 2014 19:13:45 +0100
Subject: [PATCH] add URL blacklist; add command_show_blacklist()

---
 local_config.py.skel |  5 +++++
 plugins.py           | 21 ++++++++++++++++++++-
 urlbot.py            | 12 ++++++++++++
 3 files changed, 37 insertions(+), 1 deletion(-)

diff --git a/local_config.py.skel b/local_config.py.skel
index 77f57f1..ef74c76 100644
--- a/local_config.py.skel
+++ b/local_config.py.skel
@@ -26,6 +26,11 @@ config['path_event_files'] = 'event_files'
 config['path_cmdfifo'] = 'cmdfifo'
 config['persistent_storage'] = 'urlbot.persistent'
 
+config['url_blacklist'] = [
+	r'^.*heise\.de/newsticker/.*$',
+	r'^.*wikipedia\.org/wiki/.*$'
+]
+
 # the "dice" feature will use more efficient random data (0) for given users
 config['enhanced-random-user'] = ( 'FIXME', 'FIXME' )
 
diff --git a/plugins.py b/plugins.py
index 9b4efaf..76515f4 100644
--- a/plugins.py
+++ b/plugins.py
@@ -419,6 +419,25 @@ def command_decode(args):
 			'msg': args['reply_user'] + ': usage: decode {single character}'
 		}
 
+def command_show_blacklist(args):
+	if 'register' == args:
+		return {
+			'name': 'show-blacklist',
+			'desc': 'show the current URL blacklist',
+			'args': ('data', 'reply_user'),
+			'ratelimit_class': RATE_GLOBAL
+		}
+
+	if 'show-blacklist' in args['data']:
+		logger('plugin', 'sent URL blacklist')
+		
+		return {
+			'msg': '\n'.join([
+				args['reply_user'] + ': URL blacklist: ' + b
+					for b in conf('url_blacklist')
+			])
+		}
+
 #def command_dummy(args):
 #	if 'register' == args:
 #		return {
@@ -518,7 +537,7 @@ funcs['parse'] = (parse_mental_ill, parse_skynet, parse_debbug)
 funcs['command'] = (
 	command_command, command_help, command_version, command_unicode,
 	command_klammer, command_source, command_dice, command_uptime, command_ping,
-	command_info, command_teatimer, command_decode
+	command_info, command_teatimer, command_decode, command_show_blacklist
 )
 
 _dir = dir()
diff --git a/urlbot.py b/urlbot.py
index 37b077d..1b3d59c 100755
--- a/urlbot.py
+++ b/urlbot.py
@@ -140,6 +140,18 @@ def extract_url(data):
 			if ratelimit_exceeded():
 				return False
 
+			flag = False
+			for b in conf('url_blacklist'):
+				if not None is re.match(b, url):
+					flag = True
+					message = 'url blacklist match for ' + url
+					logger('info', message)
+					chat_write(message)
+
+			if flag:
+				# an URL has matched the blacklist, continue to the next URL
+				continue
+
 # urllib.request is broken:
 # >>> '.'.encode('idna')
 # ....