Tilecache: block more fake google UA

[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 95d7b8b8c062f559c22d3de12534da4740e90094..4f6680233d68eb2482aa4534ed7140926cdee626 100644 (file)
--- a/cookbooks/tilecache/templates/default/nginx_tile.conf.erb
+++ b/cookbooks/tilecache/templates/default/nginx_tile.conf.erb
@@ -1,71 +1,278 @@
 # DO NOT EDIT - This file is being maintained by Chef
 
 upstream tile_cache_backend {
 # DO NOT EDIT - This file is being maintained by Chef
 
 upstream tile_cache_backend {

-    server 127.0.0.1;
+  server 127.0.0.1:8080;
+  server 127.0.0.2:8080;

 
 

-    keepalive 32;
+  # Add the other caches to relieve pressure if local squid failing
+  # Balancer: round-robin
+<% @caches.each do |cache| -%>
+<% if cache[:hostname] != node[:hostname] -%>
+<% cache.ipaddresses(:family => :inet, :role => :external).sort.each do |address| -%>
+  server <%= address %>:80 backup; # Server <%= cache[:hostname] %>
+<% end -%>
+<% end -%>
+<% end -%>
+
+  keepalive 512;
+  keepalive_requests 1024;

 }
 
 }
 

-# Rates table based on current cookie value
-map $cookie_qos_token $limit_rate_qos {
-  include /etc/nginx/conf.d/tile_qos_rates.map;
+# Geo Map of tile caches
+geo $tile_cache {
+  default 0;
+<% @caches.each do |cache| -%>
+<% cache.ipaddresses(:family => :inet, :role => :external).sort.each do |address| -%>
+  <%= address %> 1; # <%= cache[:hostname] %>
+<% end -%>
+<% end -%>

 }
 
 }
 

+# Rates table based on current cookie value
+# map $cookie__osm_totp_token $limit_rate_qos {
+#  include /etc/nginx/conf.d/tile_qos_rates.map;
+# }
+

 # Set-Cookie table based on current cookie value
 # Set-Cookie table based on current cookie value

-map $cookie_qos_token $cookie_qos_token_set {
-  include /etc/nginx/conf.d/tile_qos_cookies.map;
-}
+# map $cookie__osm_totp_token $cookie_qos_token_set {
+#  include /etc/nginx/conf.d/tile_qos_cookies.map;
+# }

 
 map $http_user_agent $approved_scraper {
 
 map $http_user_agent $approved_scraper {

-  default '';                       # Not approved
-  '~^JOSM\/' 'JOSM';
-  '~^Mozilla\/5\.0\ QGIS\/' 'QGIS';
+  default                   0; # Not approved
+  '~^JOSM\/'                1; # JOSM
+  '~^Mozilla\/5\.0\ QGIS\/' 1; # QGIS
+}
+
+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
+  'Android'              1; # Default or fake
+  'kc_android'           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
+  'com.soft373.taptaxi'  1;
+  'com.kradac.ktxcore'   1;
+  'ru.crowdsystems.topcontrol.knd' 1;
+}
+
+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
+  '~^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
+}
+
+map $http_referer $osm_referer {
+  default                                 '';    # False
+  '~^https:\/\/www\.openstreetmap\.org\/' 'osm'; # True

 }
 
 # Limit Cache-Control header to only approved User-Agents
 }
 
 # Limit Cache-Control header to only approved User-Agents

-map $http_user_agent $limit_http_cache_control {
-  default '';                              # Unset Header
-  '~^Mozilla\/5\.0\ QGIS\/' '';            # Unset Header
-  '~^Mozilla\/5\.0\ ' $http_cache_control; # Pass Header
+map $osm_referer$http_user_agent $limit_http_cache_control {
+  default '';                                # Unset Header
+  '~^osmMozilla\/5\.0\ QGIS\/' '';            # Unset Header
+  '~^osmMozilla\/5\.0\ ' $http_cache_control; # Pass Header

 }
 
 # Limit Pragma header to only approved User-Agents
 }
 
 # Limit Pragma header to only approved User-Agents

-map $http_user_agent $limit_http_pragma {
-  default '';                       # Unset Header
-  '~^Mozilla\/5\.0\ QGIS\/' '';     # Unset Header
-  '~^Mozilla\/5\.0\ ' $http_pragma; # Pass Header
+map $osm_referer$http_user_agent $limit_http_pragma {
+  default '';                          # Unset Header
+  '~^osmMozilla\/5\.0\ QGIS\/' '';     # Unset Header
+  '~^osmMozilla\/5\.0\ ' $http_pragma; # Pass Header

 }
 
 server {
 }
 
 server {

-    listen       443 ssl fastopen=2048 http2 default_server;
+    # IPv4
+    listen       80 deferred backlog=16384 reuseport fastopen=2048 default_server;
+    listen       443 ssl deferred backlog=16384 reuseport fastopen=2048 http2 default_server;
+    # IPv6
+    listen       [::]:80 deferred backlog=16384 reuseport fastopen=2048 default_server;
+    listen       [::]:443 ssl deferred backlog=16384 reuseport fastopen=2048 http2 default_server;

     server_name  localhost;
 
     proxy_buffers 8 64k;
     server_name  localhost;
 
     proxy_buffers 8 64k;

+    proxy_busy_buffers_size 64k;

 
     ssl_certificate      /etc/ssl/certs/tile.openstreetmap.org.pem;
     ssl_certificate_key  /etc/ssl/private/tile.openstreetmap.org.key;
 
 
     ssl_certificate      /etc/ssl/certs/tile.openstreetmap.org.pem;
     ssl_certificate_key  /etc/ssl/private/tile.openstreetmap.org.key;
 

+    # Requests sent within early data are subject to replay attacks.
+    # See: http://nginx.org/en/docs/http/ngx_http_ssl_module.html#ssl_early_data
+    ssl_early_data on;
+
+    # Immediately 404 layers we do not support
+<% for i in 20..99 do %>
+    location /<%= i %>/ {
+      set $limit_rate 512;
+      return 404;
+    }
+<% end %>
+
+    # Immediately 404 silly tile requests
+    location = /0/0/-1.png {
+      set $limit_rate 512;
+      return 404;
+    }
+    location = /1/0/-1.png {
+      set $limit_rate 512;
+      return 404;
+    }
+    location = /1/-1/0.png {
+      set $limit_rate 512;
+      return 404;
+    }
+    location = /1/-1/1.png {
+      set $limit_rate 512;
+      return 404;
+    }
+    location = /1/-1/-1.png {
+      set $limit_rate 512;
+      return 404;
+    }
+    location = /1/-1/2.png {
+      set $limit_rate 512;
+      return 404;
+    }
+    location = /1/1/-1.png {
+      set $limit_rate 512;
+      return 404;
+    }
+    location = /1/2/-1.png {
+      set $limit_rate 512;
+      return 404;
+    }
+    location = /2/0/-1.png {
+      set $limit_rate 512;
+      return 404;
+    }
+    location = /2/-1/0.png {
+      set $limit_rate 512;
+      return 404;
+    }
+    location = /2/-1/1.png {
+      set $limit_rate 512;
+      return 404;
+    }
+    location = /2/1/-1.png {
+      set $limit_rate 512;
+      return 404;
+    }
+    location = /2/-1/2.png {
+      set $limit_rate 512;
+      return 404;
+    }
+    location = /2/-1/3.png {
+      set $limit_rate 512;
+      return 404;
+    }
+
+<% for i in 0..16 do %>
+<% if i == 0 -%>
+    # Default Fallback Location Handler (lowest)

     location / {
     location / {

+<% elsif -%>
+    # Dedicated zoom handler for caching
+    location /<%= i %>/ {
+<% end %>
+      # Only allow GET / HEAD / OPTIONS (CORS) requests
+      limit_except GET HEAD OPTIONS {
+        deny all;
+      }
+

       proxy_pass http://tile_cache_backend;
       proxy_set_header X-Forwarded-For $remote_addr;
       proxy_http_version 1.1;
       proxy_set_header Connection '';
 
       proxy_pass http://tile_cache_backend;
       proxy_set_header X-Forwarded-For $remote_addr;
       proxy_http_version 1.1;
       proxy_set_header Connection '';
 

-      proxy_connect_timeout 5s;
+      proxy_connect_timeout 20s;

 
 

-      # Preserve host header.
-      proxy_set_header Host $host;
+      # Replace host header.
+      proxy_set_header Host 'tile.openstreetmap.org';

       # Do not pass cookies to backends.
       proxy_set_header Cookie '';
       # Do not pass Accept-Encoding to backends.
       proxy_set_header Accept-Encoding '';
       # Do not pass cookies to backends.
       proxy_set_header Cookie '';
       # Do not pass Accept-Encoding to backends.
       proxy_set_header Accept-Encoding '';

+      # Do not pass Accept to backends.
+      proxy_set_header Accept '';
+      # Do not pass Accept-Language to backends as unused.
+      proxy_set_header Accept-Language '';
+      proxy_set_header Accept-Charset '';
+      # Do not send origin, we allow all.
+      proxy_set_header origin '';
+      # Do not pass invalid headers to backend.
+      proxy_set_header X-Forwarded-Host '';
+      proxy_set_header X-Host '';
+      proxy_set_header Authorization '';
+      proxy_set_header Proxy-Authorization '';
+
+      # Drop partial requests
+      proxy_set_header range '';

 
       # Do not allow setting cookies from backends due to caching.
       proxy_ignore_headers Set-Cookie;
       proxy_hide_header Set-Cookie;
 
 
       # Do not allow setting cookies from backends due to caching.
       proxy_ignore_headers Set-Cookie;
       proxy_hide_header Set-Cookie;
 

+<% if i != 0 -%>
+      # Caching
+      proxy_cache "proxy_cache_zone";
+      proxy_cache_lock on;
+      proxy_cache_valid 200 2d;
+      proxy_cache_valid 404 15m;
+      # Serve stale cache on errors or if updating
+      proxy_cache_use_stale error timeout updating http_404 http_500 http_503 http_504;
+      # If in cache as stale, serve stale and update in background
+      proxy_cache_background_update on;
+      # Workaround nginx async bug which causes stale cache replies to wait for the async backend cache update reply (seen in v1.16.0)
+      keepalive_requests 0;
+      # Enable revalidation using If-Modified-Since and If-None-Match for stale items
+      proxy_cache_revalidate on;
+      proxy_cache_min_uses 4;
+
+      add_header x-cache-status $upstream_cache_status;
+<% end -%>
+

       # Set a QoS cookie if none presented (uses nginx Map)
       # Set a QoS cookie if none presented (uses nginx Map)

-      add_header Set-Cookie $cookie_qos_token_set;
+      # add_header Set-Cookie $cookie_qos_token_set;

 <% if node[:ssl][:strict_transport_security] -%>
       # Ensure Strict-Transport-Security header is removed from proxied server responses
       proxy_hide_header Strict-Transport-Security;
 <% if node[:ssl][:strict_transport_security] -%>
       # Ensure Strict-Transport-Security header is removed from proxied server responses
       proxy_hide_header Strict-Transport-Security;


@@ -75,11 +282,20 @@ server {
 <% end -%>
 
       # QoS Traffic Rate see $limit_rate on http://nginx.org/en/docs/http/ngx_http_core_module.html
 <% end -%>
 
       # QoS Traffic Rate see $limit_rate on http://nginx.org/en/docs/http/ngx_http_core_module.html

-      set $limit_rate $limit_rate_qos;
+      # set $limit_rate $limit_rate_qos;

 
       # Allow Higher Traffic Rate from Approved User-Agents which do not support cookies (uses nginx Map)
 
       # Allow Higher Traffic Rate from Approved User-Agents which do not support cookies (uses nginx Map)

-      if ($approved_scraper) {
-        set $limit_rate 32768;
+      # if ($approved_scraper) {
+      #   set $limit_rate 65536;
+      # }
+
+      if ($denied_scraper) {
+        set $limit_rate 512;
+        return 429;
+      }
+      if ($denied_referer) {
+        set $limit_rate 512;
+        return 418;

       }
 
       # Strip any ?query parameters from urls
       }
 
       # Strip any ?query parameters from urls


@@ -89,4 +305,5 @@ server {
       proxy_set_header Cache-Control $limit_http_cache_control;
       proxy_set_header Pragma $limit_http_pragma;
     }
       proxy_set_header Cache-Control $limit_http_cache_control;
       proxy_set_header Pragma $limit_http_pragma;
     }

+<% end %>

 }
 }