1 files changed, 31 insertions, 44 deletions
diff --git a/lib/tasks/temp.rake b/lib/tasks/temp.rake
index e49a84ecb..67fa10174 100644
--- a/lib/tasks/temp.rake
+++ b/lib/tasks/temp.rake
@@ -1,53 +1,40 @@
 namespace :temp do
 
-    desc 'Populate the request_classifications table from info_request_events'
-    task :populate_request_classifications => :environment do
-        InfoRequestEvent.find_each(:conditions => ["event_type = 'status_update'"]) do |classification|
-            RequestClassification.create!(:created_at => classification.created_at,
-                                          :user_id => classification.params[:user_id],
-                                          :info_request_event_id => classification.id)
-        end
-    end
-
-    desc "Remove plaintext passwords from post_redirect params"
-    task :remove_post_redirect_passwords => :environment do
-        PostRedirect.find_each(:conditions => ['post_params_yaml is not null']) do |post_redirect|
-              if post_redirect.post_params && post_redirect.post_params[:signchangeemail] && post_redirect.post_params[:signchangeemail][:password]
-                params = post_redirect.post_params
-                params[:signchangeemail].delete(:password)
-                post_redirect.post_params = params
-                post_redirect.save!
-              end
-        end
-    end
 
-    desc 'Remove file caches for requests that are not publicly visible or have been destroyed'
-    task :remove_obsolete_info_request_caches => :environment do
-        dryrun = ENV['DRYRUN'] == '0' ? false : true
-        verbose = ENV['VERBOSE'] == '0' ? false : true
-        if dryrun
-            puts "Running in dryrun mode"
+    desc 'Analyse rails log specified by LOG_FILE to produce a list of request volume'
+    task :request_volume => :environment do
+        example = 'rake log_analysis:request_volume LOG_FILE=log/access_log OUTPUT_FILE=/tmp/log_analysis.csv'
+        check_for_env_vars(['LOG_FILE', 'OUTPUT_FILE'],example)
+        log_file_path = ENV['LOG_FILE']
+        output_file_path = ENV['OUTPUT_FILE']
+        is_gz = log_file_path.include?(".gz")
+        urls = Hash.new(0)
+        f = is_gz ? Zlib::GzipReader.open(log_file_path) : File.open(log_file_path, 'r')
+        processed = 0
+        f.each_line do |line|
+            line.force_encoding('ASCII-8BIT') if RUBY_VERSION.to_f >= 1.9
+            if request_match = line.match(/^Started (GET|OPTIONS|POST) "(\/request\/.*?)"/)
+                next if line.match(/request\/\d+\/response/)
+                urls[request_match[2]] += 1
+                processed += 1
+            end
         end
-        request_cache_path = File.join(Rails.root, 'cache', 'views', 'request', '*', '*')
-        Dir.glob(request_cache_path) do |request_subdir|
-            info_request_id = File.basename(request_subdir)
-            puts "Looking for InfoRequest with id #{info_request_id}" if verbose
-            begin
-                info_request = InfoRequest.find(info_request_id)
-                puts "Got InfoRequest #{info_request_id}" if verbose
-                if ! info_request.all_can_view?
-                    puts "Deleting cache at #{request_subdir} for hidden/requester_only InfoRequest #{info_request_id}"
-                    if ! dryrun
-                        FileUtils.rm_rf(request_subdir)
-                    end
-                end
-            rescue ActiveRecord::RecordNotFound
-                puts "Deleting cache at #{request_subdir} for deleted InfoRequest #{info_request_id}"
-                if ! dryrun
-                    FileUtils.rm_rf(request_subdir)
-                end
+        url_counts = urls.to_a
+        num_requests_visited_n_times = Hash.new(0)
+        CSV.open(output_file_path, "wb") do |csv|
+            csv << ['URL', 'Number of visits']
+            url_counts.sort_by(&:last).each do |url, count|
+                num_requests_visited_n_times[count] +=1
+                csv << [url,"#{count}"]
             end
+            csv << ['Number of visits', 'Number of URLs']
+            num_requests_visited_n_times.to_a.sort.each do |number_of_times, number_of_requests|
+                csv << [number_of_times, number_of_requests]
+            end
+            csv << ['Total number of visits']
+            csv << [processed]
         end
+
     end
 
 end