From 27498a912bf182f6840b50d5b92c0bfbf2cfe287 Mon Sep 17 00:00:00 2001
From: Neil Lalonde <neillalonde@gmail.com>
Date: Wed, 31 Jul 2013 14:57:31 -0400
Subject: [PATCH] Add indexes to help IncomingLinksReport, but not much.
 Refactor to limit result set size, but it doesn't help perf much either.
 SELECT COUNT(DISTINCT ...) is still slow.

---
 app/models/incoming_links_report.rb           | 22 ++++++++++---------
 ...3035_add_report_index_to_incoming_links.rb |  6 +++++
 2 files changed, 18 insertions(+), 10 deletions(-)
 create mode 100644 db/migrate/20130731163035_add_report_index_to_incoming_links.rb

diff --git a/app/models/incoming_links_report.rb b/app/models/incoming_links_report.rb
index 34a984af2..77129ec89 100644
--- a/app/models/incoming_links_report.rb
+++ b/app/models/incoming_links_report.rb
@@ -62,8 +62,8 @@ class IncomingLinksReport
     report.y_titles[:num_users] = I18n.t("reports.#{report.type}.num_users")
 
     num_clicks  = link_count_per_domain
-    num_topics = topic_count_per_domain
-    num_users  = user_count_per_domain
+    num_topics = topic_count_per_domain(num_clicks.keys)
+    num_users  = user_count_per_domain(num_clicks.keys)
     report.data = []
     num_clicks.keys.each do |domain|
       report.data << {domain: domain, num_clicks: num_clicks[domain], num_topics: num_topics[domain], num_users: num_users[domain]}
@@ -71,20 +71,22 @@ class IncomingLinksReport
     report.data = report.data.sort_by {|x| x[:num_clicks]}.reverse[0,10]
   end
 
-  def self.per_domain
-    @per_domain_query ||= IncomingLink.where('created_at > ? AND domain IS NOT NULL', 30.days.ago).group('domain')
+  def self.link_count_per_domain(limit=10)
+    IncomingLink.where('created_at > ? AND domain IS NOT NULL', 30.days.ago).group('domain').order('count_all DESC').limit(limit).count
   end
 
-  def self.link_count_per_domain
-    per_domain.count
+  def self.per_domain(domains)
+    IncomingLink.where('created_at > ? AND domain IN (?)', 30.days.ago, domains).group('domain')
   end
 
-  def self.topic_count_per_domain
-    per_domain.count('topic_id', distinct: true)
+  def self.topic_count_per_domain(domains)
+    # COUNT(DISTINCT) is slow
+    per_domain(domains).count('topic_id', distinct: true)
   end
 
-  def self.user_count_per_domain
-    per_domain.count('user_id', distinct: true)
+  def self.user_count_per_domain(domains)
+    # COUNT(DISTINCT) is slow
+    per_domain(domains).count('user_id', distinct: true)
   end
 
 
diff --git a/db/migrate/20130731163035_add_report_index_to_incoming_links.rb b/db/migrate/20130731163035_add_report_index_to_incoming_links.rb
new file mode 100644
index 000000000..7d0254169
--- /dev/null
+++ b/db/migrate/20130731163035_add_report_index_to_incoming_links.rb
@@ -0,0 +1,6 @@
+class AddReportIndexToIncomingLinks < ActiveRecord::Migration
+  def change
+    add_index :incoming_links, [:created_at, :user_id]
+    add_index :incoming_links, [:created_at, :domain]
+  end
+end