From 04785d70a24baff53f3fded19e8847fabf646046 Mon Sep 17 00:00:00 2001
From: Grant Slater <github@firefishy.com>
Date: Mon, 14 Sep 2026 21:19:51 +0100
Subject: [PATCH] Add some AI scraper blocks

---
 cookbooks/dev/templates/default/apache.user.erb  |  1 +
 cookbooks/mediawiki/templates/default/apache.erb | 15 +++++++++++++++
 2 files changed, 16 insertions(+)

diff --git a/cookbooks/dev/templates/default/apache.user.erb b/cookbooks/dev/templates/default/apache.user.erb
index b3e9e92a2..f7223b115 100644
--- a/cookbooks/dev/templates/default/apache.user.erb
+++ b/cookbooks/dev/templates/default/apache.user.erb
@@ -32,6 +32,7 @@ WSGIDaemonProcess <%= @user %>.dev.openstreetmap.org user=<%= @user %> processes
 
 	# Prevent abuse by an anonymous AI bot
 	RewriteCond %{REQUEST_METHOD} ^(GET|HEAD)$
+	RewriteCond %{SERVER_PROTOCOL} =HTTP/1.1
 	RewriteCond %{HTTP_REFERER} ^-?$
 	RewriteCond %{HTTP_USER_AGENT} ((CriOS|Chrome)/[1-9][0-9]?\.0\.|Chrome/1[0-4][0-9]\.0\.|(Firefox|FxiOS)/[1-6]?[0-9]\.|MSIE\ [5-9]\.0|Opera/[8-9]\.|Windows\ NT\ [3-5]\.|Version/[3-5]\.[0-1]) [NC]
 	RewriteRule ^ - [R=429,L]
diff --git a/cookbooks/mediawiki/templates/default/apache.erb b/cookbooks/mediawiki/templates/default/apache.erb
index 4c8654588..3631da8fa 100644
--- a/cookbooks/mediawiki/templates/default/apache.erb
+++ b/cookbooks/mediawiki/templates/default/apache.erb
@@ -11,6 +11,12 @@
   CustomLog /var/log/apache2/<%= @name %>-access.log combined_extended
   ErrorLog /var/log/apache2/<%= @name %>-error.log
 
+  RewriteCond %{HTTP_REFERER} ^https://<%= @name.gsub('.', '\.') %> [NC]
+  RewriteCond %{HTTPS} !=on
+  RewriteCond %{REQUEST_METHOD} =GET
+  RewriteCond %{REQUEST_URI} ^/w/
+  RewriteRule ^ - [R=429,L]
+
   RedirectPermanent /.well-known/acme-challenge/ http://acme.openstreetmap.org/.well-known/acme-challenge/
   RedirectPermanent / https://<%= @name %>/
 </VirtualHost>
@@ -49,10 +55,19 @@
   RewriteCond %{HTTP_USER_AGENT} "^Mozilla/5\.0 \(Windows NT 10\.0; Win64; x64\) AppleWebKit/537\.36 \(KHTML, like Gecko\) Chrome/105\.0\.0\.0 Safari/537\.3$" [NC]
   RewriteRule ^ - [R=429,L]
 
+  # Prevent abuse by an anonymous AI bot
   RewriteCond %{REQUEST_METHOD} ^(GET|HEAD)$
   RewriteCond %{HTTP_REFERER} =https://<%= @name %>
   RewriteRule ^ - [R=429,L]
 
+  # Temporary measure to prevent abuse by an anonymous AI bot
+  RewriteCond %{HTTP_USER_AGENT} "Chrome/1[0-4][0-9]\."
+  RewriteCond %{QUERY_STRING} "(^|&)title=Special(?::|%3A)RecentChanges(&|$)" [NC,OR]
+  RewriteCond %{QUERY_STRING} "(^|&)returnto=Special(?::|%3A|%253A)RecentChanges(&|$)" [NC,OR]
+  RewriteCond %{QUERY_STRING} "feedrecentchanges" [NC]
+  RewriteCond %{REQUEST_URI} ^/w/
+  RewriteRule ^ - [R=429,L]
+
   # Historical Compatibility Links
   RedirectMatch 301 ^/index\.php$                 /w/index.php
   RedirectMatch 301 ^/index\.php/(.*)$            /wiki/$1
-- 
2.47.3