PERF: refactor incoming links table

app/models/incoming_domain.rb

@@ -0,0 +1,16 @@
+class IncomingDomain < ActiveRecord::Base
+end
+
+# == Schema Information
+#
+# Table name: incoming_domains
+#
+#  id    :integer          not null, primary key
+#  name  :string(100)      not null
+#  https :boolean          default(FALSE), not null
+#  port  :integer          not null
+#
+# Indexes
+#
+#  index_incoming_domains_on_name_and_https_and_port  (name,https,port) UNIQUE
+#

app/models/incoming_link.rb

@@ -89,20 +89,17 @@ end
 #
 # Table name: incoming_links
 #
-#  id              :integer          not null, primary key
-#  referer         :string(1000)
-#  domain          :string(100)
-#  topic_id        :integer
-#  post_number     :integer
-#  created_at      :datetime
-#  user_id         :integer
-#  ip_address      :inet
-#  current_user_id :integer
-#  post_id         :integer          not null
+#  id                  :integer          not null, primary key
+#  topic_id            :integer
+#  created_at          :datetime
+#  user_id             :integer
+#  ip_address          :inet
+#  current_user_id     :integer
+#  post_id             :integer          not null
+#  incoming_referer_id :integer
 #
 # Indexes
 #
-#  index_incoming_links_on_created_at_and_domain   (created_at,domain)
 #  index_incoming_links_on_created_at_and_user_id  (created_at,user_id)
 #  index_incoming_links_on_post_id                 (post_id)
 #

app/models/incoming_referer.rb

@@ -0,0 +1,16 @@
+class IncomingReferer < ActiveRecord::Base
+end
+
+# == Schema Information
+#
+# Table name: incoming_referers
+#
+#  id                 :integer          not null, primary key
+#  url                :string(1000)     not null
+#  path               :string(1000)     not null
+#  incoming_domain_id :integer          not null
+#
+# Indexes
+#
+#  index_incoming_referers_on_path_and_incoming_domain_id  (path,incoming_domain_id) UNIQUE
+#

app/models/post_timing.rb

@@ -101,6 +101,7 @@ end
 #
 # Indexes
 #
-#  post_timings_summary  (topic_id,post_number)
-#  post_timings_unique   (topic_id,post_number,user_id) UNIQUE
+#  index_post_timings_on_user_id  (user_id)
+#  post_timings_summary           (topic_id,post_number)
+#  post_timings_unique            (topic_id,post_number,user_id) UNIQUE
 #

db/migrate/20140804010803_incoming_link_normalization.rb

@@ -0,0 +1,78 @@
+class IncomingLinkNormalization < ActiveRecord::Migration
+  def up
+    remove_column :incoming_links, :post_number
+    remove_column :incoming_links, :domain
+    add_column :incoming_links, :incoming_referer_id, :integer
+
+    create_table :incoming_referers do |t|
+      t.string :url, limit: 1000, null: false
+      t.string :domain, limit: 100, null: false
+      t.string :path, limit: 1000, null: false
+      t.integer :port, null: false
+      t.boolean :https, null: false
+      t.integer :incoming_domain_id
+    end
+
+    # start the shuffle
+    #
+    execute "INSERT INTO incoming_referers(url, https, domain, port, path)
+             SELECT referer,
+                    CASE WHEN a[1] = 's' THEN true ELSE false END,
+                    a[2] as domain,
+                    CASE WHEN a[1] = 's' THEN
+                      COALESCE(a[4]::integer, 443)::integer
+                    ELSE
+                      COALESCE(a[4]::integer, 80)::integer
+                    END,
+                    COALESCE(a[5], '') path
+             FROM
+            (
+              SELECT referer, regexp_matches(referer, 'http(s)?://([^/:]+)(:(\d+))?(.*)') a
+              FROM
+              (
+               SELECT DISTINCT referer
+               FROM incoming_links WHERE referer ~ '^https?://.+'
+              ) Z
+            ) X
+          WHERE a[2] IS NOT NULL"
+
+
+    execute "UPDATE incoming_links l
+    SET incoming_referer_id = r.id
+    FROM incoming_referers r
+    WHERE r.url = l.referer"
+
+    create_table :incoming_domains do |t|
+      t.string :name, limit: 100, null: false
+      t.boolean :https, null: false, default: false
+      t.integer :port, null: false
+    end
+
+    # shuffle part 2
+    #
+    execute "INSERT INTO incoming_domains(name, port, https)
+    SELECT DISTINCT domain, port, https
+    FROM incoming_referers"
+
+    execute "UPDATE incoming_referers l
+    SET incoming_domain_id = d.id
+    FROM incoming_domains d
+    WHERE d.name = l.domain AND d.https = l.https AND d.port = l.port"
+
+
+    remove_column :incoming_referers, :domain
+    remove_column :incoming_referers, :port
+    remove_column :incoming_referers, :https
+
+    change_column :incoming_referers, :incoming_domain_id, :integer, null: false
+
+    add_index :incoming_referers, [:path, :incoming_domain_id], unique: true
+    add_index :incoming_domains, [:name, :https, :port], unique: true
+
+    remove_column :incoming_links, :referer
+  end
+
+  def down
+    raise ActiveRecord::IrreversibleMigration
+  end
+end