Merge pull request #4908 from tomhughes/changeset-size-limit

Add the ability to limit changeset size

app/models/changeset.rb

@@ -130,6 +130,8 @@ def bbox_valid?
   def update_bbox!(bbox_update)
     bbox.expand!(bbox_update)
 
+    raise OSM::APISizeLimitExceeded if bbox.linear_size > size_limit
+
     # update active record. rails 2.1's dirty handling should take care of
     # whether this object needs saving or not.
     self.min_lon, self.min_lat, self.max_lon, self.max_lat = @bbox.to_a.collect(&:round) if bbox.complete?
@@ -225,4 +227,10 @@ def unsubscribe(user)
   def subscribed?(user)
     subscribers.exists?(user.id)
   end
+
+  def size_limit
+    @size_limit ||= ActiveRecord::Base.connection.select_value(
+      "SELECT api_size_limit($1)", "api_size_limit", [user_id]
+    )
+  end
 end

config/settings.yml

@@ -73,6 +73,13 @@ max_changes_per_hour: 100000
 days_to_max_changes: 7
 importer_changes_per_hour: 1000000
 moderator_changes_per_hour: 1000000
+# Size limit for changes
+min_size_limit: 10000000
+initial_size_limit: 30000000
+max_size_limit: 5400000000
+days_to_max_size_limit: 28
+importer_size_limit: 5400000000
+moderator_size_limit: 5400000000
 # Domain for handling message replies
 #messages_domain: "messages.openstreetmap.org"
 # MaxMind GeoIPv2 database

db/migrate/20240618193051_api_size_limit.rb

@@ -0,0 +1,13 @@
+class ApiSizeLimit < ActiveRecord::Migration[7.1]
+  def up
+    safety_assured do
+      execute DatabaseFunctions::API_SIZE_LIMIT
+    end
+  end
+
+  def down
+    safety_assured do
+      execute "DROP FUNCTION api_size_limit(bigint)"
+    end
+  end
+end

db/structure.sql

@@ -176,6 +176,63 @@ CREATE FUNCTION public.api_rate_limit(user_id bigint) RETURNS integer
     $$;
 
 
+--
+-- Name: api_size_limit(bigint); Type: FUNCTION; Schema: public; Owner: -
+--
+
+CREATE FUNCTION public.api_size_limit(user_id bigint) RETURNS bigint
+    LANGUAGE plpgsql STABLE
+    AS $$
+    DECLARE
+      min_size_limit int8 := 10000000;
+      initial_size_limit int8 := 30000000;
+      max_size_limit int8 := 5400000000;
+      days_to_max_size_limit int4 := 28;
+      importer_size_limit int8 := 5400000000;
+      moderator_size_limit int8 := 5400000000;
+      roles text[];
+      last_block timestamp without time zone;
+      first_change timestamp without time zone;
+      active_reports int4;
+      time_since_first_change double precision;
+      size_limit int8;
+    BEGIN
+      SELECT ARRAY_AGG(user_roles.role) INTO STRICT roles FROM user_roles WHERE user_roles.user_id = api_size_limit.user_id;
+
+      IF 'moderator' = ANY(roles) THEN
+        size_limit := moderator_size_limit;
+      ELSIF 'importer' = ANY(roles) THEN
+        size_limit := importer_size_limit;
+      ELSE
+        SELECT user_blocks.created_at INTO last_block FROM user_blocks WHERE user_blocks.user_id = api_size_limit.user_id ORDER BY user_blocks.created_at DESC LIMIT 1;
+
+        IF FOUND THEN
+          SELECT changesets.created_at INTO first_change FROM changesets WHERE changesets.user_id = api_size_limit.user_id AND changesets.created_at > last_block ORDER BY changesets.created_at LIMIT 1;
+        ELSE
+          SELECT changesets.created_at INTO first_change FROM changesets WHERE changesets.user_id = api_size_limit.user_id ORDER BY changesets.created_at LIMIT 1;
+        END IF;
+
+        IF NOT FOUND THEN
+          first_change := CURRENT_TIMESTAMP AT TIME ZONE 'UTC';
+        END IF;
+
+        SELECT COUNT(*) INTO STRICT active_reports
+        FROM issues INNER JOIN reports ON reports.issue_id = issues.id
+        WHERE issues.reported_user_id = api_size_limit.user_id AND issues.status = 'open' AND reports.updated_at >= COALESCE(issues.resolved_at, '1970-01-01');
+
+        time_since_first_change := EXTRACT(EPOCH FROM CURRENT_TIMESTAMP AT TIME ZONE 'UTC' - first_change);
+
+        size_limit := max_size_limit * POWER(time_since_first_change, 2) / POWER(days_to_max_size_limit * 24 * 60 * 60, 2);
+        size_limit := GREATEST(initial_size_limit, LEAST(max_size_limit, FLOOR(size_limit)));
+        size_limit := size_limit / POWER(2, active_reports);
+        size_limit := GREATEST(min_size_limit, LEAST(max_size_limit, size_limit));
+      END IF;
+
+      RETURN size_limit;
+    END;
+    $$;
+
+
 SET default_tablespace = '';
 
 SET default_table_access_method = heap;
@@ -3521,6 +3578,7 @@ INSERT INTO "schema_migrations" (version) VALUES
 ('23'),
 ('22'),
 ('21'),
+('20240618193051'),
 ('20240605134916'),
 ('20240405083825'),
 ('20240307181018'),

lib/bounding_box.rb

@@ -88,6 +88,14 @@ def area
     end
   end
 
+  def linear_size
+    if complete?
+      (max_lon - min_lon) + (max_lat - min_lat)
+    else
+      0
+    end
+  end
+
   def complete?
     to_a.exclude?(nil)
   end

lib/database_functions.rb

@@ -55,4 +55,58 @@ module DatabaseFunctions
     END;
     $$ LANGUAGE plpgsql STABLE;
   ).freeze
+
+  API_SIZE_LIMIT = %(
+    CREATE OR REPLACE FUNCTION api_size_limit(user_id int8)
+      RETURNS int8
+      AS $$
+    DECLARE
+      min_size_limit int8 := #{Settings.min_size_limit};
+      initial_size_limit int8 := #{Settings.initial_size_limit};
+      max_size_limit int8 := #{Settings.max_size_limit};
+      days_to_max_size_limit int4 := #{Settings.days_to_max_size_limit};
+      importer_size_limit int8 := #{Settings.importer_size_limit};
+      moderator_size_limit int8 := #{Settings.moderator_size_limit};
+      roles text[];
+      last_block timestamp without time zone;
+      first_change timestamp without time zone;
+      active_reports int4;
+      time_since_first_change double precision;
+      size_limit int8;
+    BEGIN
+      SELECT ARRAY_AGG(user_roles.role) INTO STRICT roles FROM user_roles WHERE user_roles.user_id = api_size_limit.user_id;
+
+      IF 'moderator' = ANY(roles) THEN
+        size_limit := moderator_size_limit;
+      ELSIF 'importer' = ANY(roles) THEN
+        size_limit := importer_size_limit;
+      ELSE
+        SELECT user_blocks.created_at INTO last_block FROM user_blocks WHERE user_blocks.user_id = api_size_limit.user_id ORDER BY user_blocks.created_at DESC LIMIT 1;
+
+        IF FOUND THEN
+          SELECT changesets.created_at INTO first_change FROM changesets WHERE changesets.user_id = api_size_limit.user_id AND changesets.created_at > last_block ORDER BY changesets.created_at LIMIT 1;
+        ELSE
+          SELECT changesets.created_at INTO first_change FROM changesets WHERE changesets.user_id = api_size_limit.user_id ORDER BY changesets.created_at LIMIT 1;
+        END IF;
+
+        IF NOT FOUND THEN
+          first_change := CURRENT_TIMESTAMP AT TIME ZONE 'UTC';
+        END IF;
+
+        SELECT COUNT(*) INTO STRICT active_reports
+        FROM issues INNER JOIN reports ON reports.issue_id = issues.id
+        WHERE issues.reported_user_id = api_size_limit.user_id AND issues.status = 'open' AND reports.updated_at >= COALESCE(issues.resolved_at, '1970-01-01');
+
+        time_since_first_change := EXTRACT(EPOCH FROM CURRENT_TIMESTAMP AT TIME ZONE 'UTC' - first_change);
+
+        size_limit := max_size_limit * POWER(time_since_first_change, 2) / POWER(days_to_max_size_limit * 24 * 60 * 60, 2);
+        size_limit := GREATEST(initial_size_limit, LEAST(max_size_limit, FLOOR(size_limit)));
+        size_limit := size_limit / POWER(2, active_reports);
+        size_limit := GREATEST(min_size_limit, LEAST(max_size_limit, size_limit));
+      END IF;
+
+      RETURN size_limit;
+    END;
+    $$ LANGUAGE plpgsql STABLE;
+  ).freeze
 end

lib/osm.rb

@@ -364,6 +364,17 @@ def status
     end
   end
 
+  # Raised when a size limit is exceeded
+  class APISizeLimitExceeded < APIError
+    def initialize
+      super("Size limit exceeded")
+    end
+
+    def status
+      :payload_too_large
+    end
+  end
+
   # Helper methods for going to/from mercator and lat/lng.
   class Mercator
     include Math

lib/tasks/update_functions.rake

@@ -2,5 +2,6 @@ namespace :db do
   desc "Update database function definitions"
   task :update_functions => :environment do
     ActiveRecord::Base.connection.execute DatabaseFunctions::API_RATE_LIMIT
+    ActiveRecord::Base.connection.execute DatabaseFunctions::API_SIZE_LIMIT
   end
 end