[brand] darmarit.org/searx - BLOCK known bots

Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>

Author: return42

searx/plugins/limiter.py

@@ -40,6 +40,13 @@
     + r')'
 )
 
+ip_block_list = ['35.84.41.175', '183.93.70.31', '209.198.8.138', '58.97.160.13']
+
+query_block_list = re.compile(
+    r'('
+    + r"site:zhidao\.baidu\.com .*"
+    + r')'
+)
 
 def is_accepted_request() -> bool:
     # pylint: disable=too-many-return-statements
@@ -50,6 +57,15 @@ def is_accepted_request() -> bool:
     if request.path == '/healthz':
         return True
 
+    if x_forwarded_for in ip_block_list:
+        logger.debug("BLOCK %s: %s --> IP is on block list: %s" % (x_forwarded_for, request.path, user_agent))
+        return False
+
+    q = request.form.get('q')
+    if q and query_block_list.match(q):
+        logger.debug("BLOCK %s: %s --> query is on block list: %s" % (x_forwarded_for, request.path, q))
+        return False
+
     if block_user_agent.match(user_agent):
         logger.debug("BLOCK %s: %s --> detected User-Agent: %s" % (x_forwarded_for, request.path, user_agent))
         return False