Clean up job for search logs

app/jobs/scheduled/clean_up_search_logs.rb

@@ -0,0 +1,9 @@
+module Jobs
+  class CleanUpSearchLogs < Jobs::Scheduled
+    every 1.week
+
+    def execute(args)
+      SearchLog.clean_up
+    end
+  end
+end

app/models/search_log.rb

@@ -44,4 +44,11 @@ class SearchLog < ActiveRecord::Base
       [:updated, rows[0]['id'].to_i]
     end
   end
+
+  def self.clean_up
+    search_id = SearchLog.order(:id).offset(SiteSetting.search_query_log_max_size).limit(1).pluck(:id)
+    if search_id.present?
+      SearchLog.where('id < ?', search_id[0]).delete_all
+    end
+  end
 end

config/locales/server.en.yml

@@ -948,6 +948,7 @@ en:
     search_prefer_recent_posts: "If searching your large forum is slow, this option tries an index of more recent posts first"
     search_recent_posts_size: "How many recent posts to keep in the index"
     log_search_queries: "Log search queries performed by users"
+    search_query_log_max_size: "Maximum amount of search queries to keep"
     allow_uncategorized_topics: "Allow topics to be created without a category. WARNING: If there are any uncategorized topics, you must recategorize them before turning this off."
     allow_duplicate_topic_titles: "Allow topics with identical, duplicate titles."
     unique_posts_mins: "How many minutes before a user can make a post with the same content again"

config/site_settings.yml

@@ -1155,6 +1155,7 @@ search:
   search_prefer_recent_posts: false
   search_recent_posts_size: 100000
   log_search_queries: true
+  search_query_log_max_size: 1000000
 
 uncategorized:
   version_checks:

spec/models/search_log_spec.rb

@@ -137,6 +137,26 @@ RSpec.describe SearchLog, type: :model do
       end
 
     end
+  end
+
+  context "clean_up" do
+
+    it "will remove old logs" do
+      SearchLog.log(term: 'jawa', search_type: :header, ip_address: '127.0.0.1')
+      SearchLog.log(term: 'jedi', search_type: :header, ip_address: '127.0.0.1')
+      SearchLog.log(term: 'rey', search_type: :header, ip_address: '127.0.0.1')
+      SearchLog.log(term: 'finn', search_type: :header, ip_address: '127.0.0.1')
+
+      SiteSetting.search_query_log_max_size = 5
+      SearchLog.clean_up
+      expect(SearchLog.count).to eq(4)
+
+      SiteSetting.search_query_log_max_size = 2
+      SearchLog.clean_up
+      expect(SearchLog.count).to eq(2)
+      expect(SearchLog.where(term: 'rey').first).to be_present
+      expect(SearchLog.where(term: 'finn').first).to be_present
+    end
 
   end