From 9d3ff8b3b742123ee03b62a40481e67a55c2f87c Mon Sep 17 00:00:00 2001
From: Gregory Soutade <gregory@soutade.fr>
Date: Wed, 14 Jun 2023 09:21:11 +0200
Subject: [PATCH] Add excluded domain option

---
 iwla.py | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/iwla.py b/iwla.py
index af3e290..2cd0edf 100755
--- a/iwla.py
+++ b/iwla.py
@@ -52,6 +52,7 @@ Conf values needed :
     locales_path
     compress_output_files
     excluded_ip
+    excluded_domain_name
 
 Output files :
     DB_ROOT/meta.db
@@ -155,6 +156,9 @@ class IWLA(object):
         self.excluded_ip = []
         for ip in conf.excluded_ip:
             self.excluded_ip += [re.compile(ip)]
+        self.excluded_domain_name = []
+        for domain_name in conf.excluded_domain_name:
+            self.excluded_domain_name += [re.compile(domain_name)]
         self.plugins = [(conf.PRE_HOOK_DIRECTORY     , conf.pre_analysis_hooks),
                         (conf.POST_HOOK_DIRECTORY    , conf.post_analysis_hooks),
                         (conf.DISPLAY_HOOK_DIRECTORY , conf.display_hooks)]
@@ -350,7 +354,6 @@ class IWLA(object):
             super_hit['last_access'] = self.meta_infos['last_time']
 
         request = hit['extract_request']
-
         uri = request.get('extract_uri', request['http_uri'])
 
         hit['is_page'] = self.isPage(uri)
@@ -707,6 +710,11 @@ class IWLA(object):
             self.logger.debug("Not in domain %s" % (hit))
             return False
 
+        for domain_name in self.excluded_domain_name:
+            if domain_name.match(hit['server_name']):
+                self.logger.debug("Domain name %s excluded" % (hit['server_name']))
+                return False
+
         t = self._decodeTime(hit)
 
         cur_time = self.meta_infos['last_time']