Chapter 3: Implement SQL query normalizer

lib/active_monitoring/engine.rb

@@ -30,12 +30,13 @@ class Engine < ::Rails::Engine
       end
 
       ActiveMonitoring::Notifications.subscribe("sql.active_record") do |name, start, finish, _id, payload|
-        if ActiveMonitoring::SqlQuery.new(name: payload[:name], query: payload[:sql]).track?
+        query = ActiveMonitoring::SqlQuery.new(name: payload[:name].dup, query: payload[:sql].dup)
+        if query.track?
           Metric.create(
             name: name,
             value: finish - start,
             request_id: Current.request_id,
-            sql_query: payload[:sql],
+            sql_query: query.normalized_query,
             location: Current.location,
             created_at: finish
           )

lib/active_monitoring/sql_normalizer.rb

@@ -0,0 +1,23 @@
+module ActiveMonitoring
+  class SqlNormalizer
+    def initialize(query:)
+      @query = query
+    end
+
+    def to_s
+      query.squish!
+      query.gsub!(/(\s(=|>|<|>=|<=|<>|!=)\s)('[^']+'|[\$\+\-\w\.]+)/, '\1xxx')
+      query.gsub!(/(\sIN\s)\([^\(\)]+\)/i, '\1(xxx)')
+      regex = /(\sBETWEEN\s)('[^']+'|[\+\-\w\.]+)(\sAND\s)('[^']+'|[\+\-\w\.]+)/i
+      query.gsub!(regex, '\1xxx\3xxx')
+      query.gsub!(/(\sVALUES\s)\(.+\)/i, '\1(xxx)')
+      query.gsub!(/(\s(LIKE|ILIKE|SIMILAR TO|NOT SIMILAR TO)\s)('[^']+')/i, '\1xxx')
+      query.gsub!(/(\s(LIMIT|OFFSET)\s)(\d+)/i, '\1xxx')
+      query
+    end
+
+    private
+
+      attr_reader :query
+  end
+end

lib/active_monitoring/sql_query.rb

@@ -1,4 +1,6 @@
 require_relative "sql_tracker"
+require_relative "sql_normalizer"
+
 module ActiveMonitoring
   class SqlQuery
     def initialize(query:, name:)
@@ -10,6 +12,10 @@ def track?
       SqlTracker.new(query: query, name: name).track?
     end
 
+    def normalized_query
+      SqlNormalizer.new(query: query).to_s
+    end
+
     private
 
       attr_reader :query, :name

spec/requests/metrics_spec.rb

@@ -34,7 +34,7 @@
         name: "sql.active_record",
         value: 0,
         created_at: Time.zone.local(2020, 1, 1),
-        sql_query: %(INSERT INTO "books" ("name", "created_at", "updated_at") VALUES (?, ?, ?)),
+        sql_query: %(INSERT INTO "books" ("name", "created_at", "updated_at") VALUES (xxx)),
         request_id: response.headers["X-Request-Id"],
         location: "BooksController#create"
       )