Block more contacted sites from tiles

[chef.git] / cookbooks / tilecache / templates / default / nginx_tile.conf.erb

diff --git a/cookbooks/tilecache/templates/default/nginx_tile.conf.erb b/cookbooks/tilecache/templates/default/nginx_tile.conf.erb
index 7d56776eb0ebc7387c7741d5700d0efe81520978..b8b0b5321f79b59fe99b91574dba64a72e940393 100644 (file)
--- a/cookbooks/tilecache/templates/default/nginx_tile.conf.erb
+++ b/cookbooks/tilecache/templates/default/nginx_tile.conf.erb
@@ -52,33 +52,48 @@ map $http_user_agent $approved_scraper {
 map $http_user_agent $denied_scraper {
   default                0; # Not denied
   ''                     1; # No User-Agent Set
-  '~^Python\-urllib\/'   1; # Library Default
-  '~^python\-requests\/' 1; # Library Default
-  '~^node\-fetch\/'      1; # Library Default
-  '~^R$'                 1; # Library Default
-  '~^Java\/'             1; # Library Default
-  '~^tiles$'             1; # Library Default
-  '~^okhttp\/'           1; # Library Default
-  '~^Microsoft-ATL-Native\/' 1; # Library Default
-  '/n software IPWorks HTTP/S Component - www.nsoftware.com' 1; #Library default
-  '~^Wget\/'             1; # Library Default
-  'C# TilesDownloader'   1; # Downloader
-  'MapDownloader'        1; # Downloader
-  '~^staticmaps'         1; # Downloader
-  'Android'              1; # Default or fake
-  'kc_android'           1; # Default or fake
-  'host'                 1; # Default or fake
-  '~^maptestapp'         1; # Default or fake
-  'Other'                1; # Default or fake
-  'Mozilla/4.0'          1; # Fake
-  'Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1)' 1;  # Fake
-  '~^runtastic'          1; # App
-  '~^Where\ my\ children' 1; # App
-  'nossoonibusjp.android.crosswalk' 1; # App
-  'br.com.concisoti.potybus' 1; # App
+  '-'                    1;
+
+  # Library defaults
+  '~^Python\-urllib\/'   1;
+  '~^python\-requests\/' 1;
+  '~^node\-fetch\/'      1;
+  '~^R$'                 1;
+  '~^Java\/'             1;
+  '~^tiles$'             1;
+  '~^okhttp\/'           1;
+  '~^Microsoft-ATL-Native\/' 1;
+  '/n software IPWorks HTTP/S Component - www.nsoftware.com' 1;
+  '~^Wget\/'             1;
+  'java'                 1;
+
+  # Library defaults or fakes
+  'Android'              1;
+  'kc_android'           1;
+  'host'                 1;
+  '~^maptestapp'         1;
+  'Other'                1;
+  'osmdroid'             1;
+  '~^tilelive-http'      1;
+  '~^Java-http-client'   1;
+
+  # Fakes
+  'Mozilla/4.0'          1;
+  'Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1)' 1;
+
+  # Bulk downloaders
+  'C# TilesDownloader'   1;
+  'MapDownloader'        1;
+  '~^staticmaps'         1;
+
+  # Overusage apps
+  '~^runtastic'          1;
+  '~^Where\ my\ children' 1;
+  'nossoonibusjp.android.crosswalk' 1;
+  'br.com.concisoti.potybus' 1;
   'com.soft373.taptaxi'  1;
   'com.kradac.ktxcore'   1;
-  'ru.crowdsystems.topcontrol.knd' 1;
+  '~^ru.crowdsystems.topcontrol.knd' 1;
   'pl.itaxi.driver'      1;
   'net.uztaxi.driver'    1;
   'OSMDroid/2.1 (its; rutaxi 3.28.0)' 1;
@@ -86,46 +101,57 @@ map $http_user_agent $denied_scraper {
   'ru.taximaster.www'    1;
   'com.arobs.trackgps'   1;
   'com.helleniccomms.asteras.driver' 1;
+
+  # Malware
+  'Agent Smith'          1;
   # '~[a-f0-9]{8}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{12}' 1; # Fake UA
 }
 
 map $http_referer $denied_referer {
   default                          0; # Not denied
-  'http://www.openstreetmap.org/'  1; # Faked
-  'http://www.openstreetmap.org'   1; # Faked
-  'https://www.openstreetmap.org'  1; # Faked
-  'http://openstreetmap.org/'      1; # Faked
-  'http://openstreetmap.org'       1; # Faked
-  'https://openstreetmap.org'      1; # Faked
-  'http://www.osm.org/'            1; # Faked
-  'http://www.osm.org'             1; # Faked
-  'http://osm.org/'                1; # Faked
-  'http://osm.org'                 1; # Faked
-  'http://google.com'              1; # Faked
-  'http://www.google.com'          1; # Faked
-  'http://google.com/'             1; # Faked
-  'http://www.google.com/'         1; # Faked
-  'https://google.com'             1; # Faked
-  'https://www.google.com'         1; # Faked
-  'https://google.com/'            1; # Faked
-  'https://www.google.com/'        1; # Faked
-  'http://www.microsoft.com/'      1; # Faked
-  '~^https?://pmap\.kuku\.lu/'           1; # Too much traffic
-  '~^https?://[^.]*\.pmap\.kuku\.lu/'    1; # Too much traffic
-  '~^https?://fastpokemap\.com/'         1; # Too much traffic
-  '~^https?://[^.]*\.fastpokemap\.com/'  1; # Too much traffic
-  '~^https?://pkget\.com/'               1; # Too much traffic
-  '~^https?://[^.]*\.pkget\.com/'        1; # Too much traffic
-  '~^https?://twpkinfo\.com/'            1; # Too much traffic
-  '~^https?://[^.]*\.twpkinfo\.com/'     1; # Too much traffic
-  '~^https?://9db\.jp/'                  1; # Too much traffic
-  '~^https?://[^.]*\.9db\.jp/'           1; # Too much traffic
-  '~^https?://clustrmaps\.com/'          1; # Too much traffic
-  '~^https?://[^.]*\.clustrmaps\.com/'   1; # Too much traffic
-  '~^https?://geoportal360\.pl/'         1; # Too much traffic
-  '~^https?://skelbiu\.lt/'              1; # Too much traffic
-  '~^https?://[^.]*\.skelbiu\.lt/'       1; # Too much traffic
-  # '~^https?://[^.]*\.wialon.com/'        1; # Too much traffic (hold per 2020-04-10 email)
+  # Faked sites
+  'http://www.openstreetmap.org/'  1;
+  'http://www.openstreetmap.org'   1;
+  'https://www.openstreetmap.org'  1;
+  'http://openstreetmap.org/'      1;
+  'http://openstreetmap.org'       1;
+  'https://openstreetmap.org'      1;
+  'http://www.osm.org/'            1;
+  'http://www.osm.org'             1;
+  'http://osm.org/'                1;
+  'http://osm.org'                 1;
+  'http://google.com'              1;
+  'http://www.google.com'          1;
+  'http://google.com/'             1;
+  'http://www.google.com/'         1;
+  'https://google.com'             1;
+  'https://www.google.com'         1;
+  'https://google.com/'            1;
+  'https://www.google.com/'        1;
+  'http://www.microsoft.com/'      1;
+
+  # Overusing websites
+  '~^https?://pmap\.kuku\.lu/'           1;
+  '~^https?://[^.]*\.pmap\.kuku\.lu/'    1;
+  '~^https?://fastpokemap\.com/'         1;
+  '~^https?://[^.]*\.fastpokemap\.com/'  1;
+  '~^https?://pkget\.com/'               1;
+  '~^https?://[^.]*\.pkget\.com/'        1;
+  '~^https?://twpkinfo\.com/'            1;
+  '~^https?://[^.]*\.twpkinfo\.com/'     1;
+  '~^https?://9db\.jp/'                  1;
+  '~^https?://[^.]*\.9db\.jp/'           1;
+  '~^https?://clustrmaps\.com/'          1;
+  '~^https?://[^.]*\.clustrmaps\.com/'   1;
+  '~^https?://geoportal360\.pl/'         1;
+  '~^https?://skelbiu\.lt/'              1;
+  '~^https?://[^.]*\.skelbiu\.lt/'       1;
+  '~^https?://wialon\.[^.]*/'            1; # wialon has many domains, so block them all
+  '~^https?://[^.]*\.wialon\.[^.]*/'     1;
+  '~^https?://gps-trace\.com/'           1;
+  '~^https?://[^.]*\.gps-trace\.com/'    1;
+  '~^https?://cellmapper\.net/'          1;
+  '~^https?://[^.]*\.cellmapper\.net/'   1;
 }
 
 map $http_referer $osm_referer {