active_record_data_loader 1.0.2 → 1.3.1

data/README.md

@@ -1,6 +1,6 @@
-# ActiveRecord Data Loader
+# active_record_data_loader
 
-[![Build Status](https://travis-ci.org/abeiderman/active_record_data_loader.svg?branch=master)](https://travis-ci.org/abeiderman/active_record_data_loader)
+[![Build Status](https://github.com/abeiderman/active_record_data_loader/actions/workflows/build.yml/badge.svg)](https://github.com/abeiderman/active_record_data_loader/actions/workflows/build.yml)
 [![Coverage Status](https://coveralls.io/repos/github/abeiderman/active_record_data_loader/badge.svg?branch=master&service=github)](https://coveralls.io/github/abeiderman/active_record_data_loader?branch=master)
 [![Maintainability](https://api.codeclimate.com/v1/badges/338904b3f7e8d19a3cb1/maintainability)](https://codeclimate.com/github/abeiderman/active_record_data_loader/maintainability)
 
@@ -10,6 +10,10 @@ Efficiently bulk load data for your ActiveRecord models with a simple DSL.
 
 Load, performance, and stress tests often require setting up a realistic amount of data in your database. This gem is intended to help organize that data load and make it more maintainable than having a collection of SQL scripts.
 
+#### How is this different from using _factory_bot_?
+
+This gem is not a replacement for [factory_bot](https://github.com/thoughtbot/factory_bot). It solves a different use case. While _factory_bot_ is great for organizing test data and reducing duplication in your functional tests, _active_record_data_loader_ is focused around bulk loading data for performance tests. The purpose of _active_record_data_loader_ is loading large amounts of data as efficiently as possible while providing a DSL that helps with maintainability.
+
 ## Installation
 
 Add this line to your application's Gemfile:
@@ -37,6 +41,7 @@ Polymorphic associations need to be defined explicitly as shown in [Polymorphic
 ### Basic usage
 
 Let's say you have the following models:
+
 ```ruby
 class Customer < ApplicationRecord
 end
@@ -47,6 +52,7 @@ end
 ```
 
 The following code will create 10,000 customers and 100,000 orders, and will associate the orders to those customers evenly:
+
 ```ruby
 data_loader = ActiveRecordDataLoader.define do
   model Customer do |m|
@@ -63,6 +69,7 @@ data_loader.load_data
 
 #### Overriding column values
 To provide your own values for columns your can provide a lambda or a constant value:
+
 ```ruby
 data_loader = ActiveRecordDataLoader.define do
   model Customer do |m|
@@ -87,7 +94,7 @@ In this example, we are creating 25K orders for customers in CAN with a CAD curr
 data_loader = ActiveRecordDataLoader.define do
   model Customer do |m|
     m.count 10_000
-    m.column :country, -> { %w[CAN MXN USA].sample }
+    m.column :country, -> { %w[CAN MEX USA].sample }
   end
 
   model Order do |m|
@@ -95,13 +102,13 @@ data_loader = ActiveRecordDataLoader.define do
     m.column :currency, "CAD"
     m.belongs_to :customer, eligible_set: -> { Customer.where(country: "CAN") }
   end
-  
+
   model Order do |m|
     m.count 25_000
     m.column :currency, "MXN"
     m.belongs_to :customer, eligible_set: -> { Customer.where(country: "MEX") }
   end
-  
+
    model Order do |m|
     m.count 50_000
     m.column :currency, "USD"
@@ -117,6 +124,7 @@ data_loader.load_data
 If you have a polymorphic `belongs_to` association, you will need to define that explicitly for it to be populated.
 
 Let's assume the following models where an order could belong to either a person or a business:
+
 ```ruby
 class Person < ApplicationRecord
   has_many :orders
@@ -132,6 +140,7 @@ end
 ```
 
 In order to populate the `customer` association in orders, you would specify them like this:
+
 ```ruby
 data_loader = ActiveRecordDataLoader.define do
   model Person do |m|
@@ -144,7 +153,7 @@ data_loader = ActiveRecordDataLoader.define do
 
   model Order do |m|
     m.count 100_000
-    
+
     m.polymorphic :customer do |c|
       c.model Person
       c.model Business
@@ -156,6 +165,7 @@ data_loader.load_data
 ```
 
 You can also provide a `weight` to each of the target models if you want to control how they are distributed. If you wanted to have twice as many orders for `Person` than for `Business`, it would look like this:
+
 ```ruby
 data_loader = ActiveRecordDataLoader.define do
   model Person do |m|
@@ -168,7 +178,7 @@ data_loader = ActiveRecordDataLoader.define do
 
   model Order do |m|
     m.count 100_000
-    
+
     m.polymorphic :customer do |c|
       c.model Person, weight: 2
       c.model Business, weight: 1
@@ -180,6 +190,7 @@ data_loader.load_data
 ```
 
 Additionaly, you can also provide an `eligible_set` to control which records to limit the association to:
+
 ```ruby
 data_loader = ActiveRecordDataLoader.define do
   model Person do |m|
@@ -193,7 +204,7 @@ data_loader = ActiveRecordDataLoader.define do
 
   model Order do |m|
     m.count 100_000
-    
+
     m.polymorphic :customer do |c|
       c.model Person, weight: 2
       c.model Business, weight: 1, eligible_set: -> { Business.where(country: "USA") }
@@ -204,6 +215,148 @@ end
 data_loader.load_data
 ```
 
+### Unique indexes
+
+Unique indexes will be detected automatically and the data generator will attempt to generate unique values for each row. The generator keeps track of unique values previously generated and retries rows with repeating values. Because some columns could be generating random values, retrying can eventually be successful.
+
+There are a couple of behaviors you can control regarding preventing duplicates. The first is the number of times to retry a given row with duplicate values (that would fail the unique index/constraint). The second is what to do if a unique value cannot be generated after the retries are exhausted.
+
+By default, there will be 5 retries per row and the row will be skipped after all retries are unsuccessful. This means fewer rows than requested may end up being populated on that table.
+
+Alternatively, you can choose to raise an error if a unique row cannot be generated. You can also set the number of retries to 0 to not retry at all. If the table in question is a primary target for your testing and will be loaded with a lot of data, you will likely not want to have retries since it could potentially slow down data generation significantly.
+
+Here is how to adjust these settings. Here let's assyme that `daily_notes` has a unique index on both `date` and `person_id`:
+
+```ruby
+class Person < ApplicationRecord
+end
+
+class DailyNotes < ApplicationRecord
+  belongs_to :person
+end
+
+data_loader = ActiveRecordDataLoader.define do
+  model Person do |m|
+    m.count 500
+  end
+
+  model DailyNotes do |m|
+    m.count 10_000
+    m.max_duplicate_retries 10
+    m.do_not_raise_on_duplicates
+
+    m.column :date, -> { Date.today - rand(20) }
+  end
+end
+
+data_loader.load_data
+```
+
+In the case above, retrying could be a reasonable choice since the date is generated at random and it's a small number of rows being generated.
+
+If you want to disable retrying duplicates altogether and raise an error to fail fast you can specify it like this:
+
+```ruby
+class Person < ApplicationRecord
+end
+
+class Skill < ApplicationRecord
+end
+
+class SkillRating < ApplicationRecord
+  belongs_to :person
+  belongs_to :skill
+end
+
+data_loader = ActiveRecordDataLoader.define do
+  model Person do |m|
+    m.count 100_000
+  end
+
+  model Skill do |m|
+    m.count 100
+  end
+
+  model SkillRating do |m|
+    m.count 10_000_000
+    m.max_duplicate_retries 0
+    m.raise_on_duplicates
+
+    m.column :rating, -> { rand(1..10) }
+  end
+end
+
+data_loader.load_data
+```
+
+
+### Configuration options
+
+You can define global configuration options like this:
+
+```ruby
+ActiveRecordDataLoader.configure do |c|
+  c.logger = ActiveSupport::Logger.new("my_file.log", level: :debug)
+  c.statement_timeout = "5min"
+end
+```
+
+Or you can create a configuration object for the specific data loader instance rather than globally:
+
+```ruby
+config = ActiveRecordDataLoader::Configuration.new(
+  c.logger = ActiveSupport::Logger.new("my_file.log", level: :debug)
+  c.statement_timeout = "5min"
+)
+loader = ActiveRecordDataLoader.define(config) do
+  model Company do |m|
+    m.count 10
+  end
+
+  # ... more definitions
+end
+```
+
+#### statement_timeout
+
+This is currently only used for Postgres connections to adjust the `statement_timeout` value for the connection. The default is `2min`. Depending on the size of the batches you are loading and overall size of the tables you may need to increase this value:
+
+```ruby
+ActiveRecordDataLoader.configure do |c|
+  c.statement_timeout = "5min"
+end
+```
+
+#### connection_factory
+
+The `connection_factory` option accepts a lambda that should return a connection object whenever executed. If not specified, the default behavior is to retrieve a connection using `ActiveRecord::Base.connection`. You can configure it like this:
+
+```ruby
+ActiveRecordDataLoader.configure do |c|
+  c.connection_factory = -> { MyCustomConnectionHandler.open_connection }
+end
+```
+
+#### output
+
+The `output` option accepts an optional file name to write a SQL script with the data loading statements. This script file can then be executed manually to load the data. This can be helpful if you need to load the same data multiple times. For example if you are profiling different alternatives in your code and you want to see how each performs with a fully loaded database. In that case you would want to have the same data starting point for each alternative you evaluate. By generating the script file, it would be significantly faster to load that data over and over by executing the existing script.
+
+If `output` is nil or empty, no script file will be written.
+
+Example usage:
+
+```ruby
+ActiveRecordDataLoader.configure do |c|
+  c.output = "./my_script.sql"  # Outputs to the provided file
+end
+```
+
+When using an output script file with Postgres, the resulting script will have `\COPY` commands which reference CSV files that contain the data batches to be copied. The CSV files will be created along side the SQL script and will have a naming convention of using the table name and the rows range for the given batch. For example `./my_script_customers_1_to_1000.csv`. Each `\COPY` command in the SQL file will reference the corresponding CSV file so all you need to do is execute the SQL file using `psql`:
+
+```bash
+psql -h my-db-host -U my_user -f my_script.sql
+```
+
 ## Development
 
 After checking out the repo, run `bin/setup` to install dependencies. Then, run `rake spec` to run the tests. You can also run `bin/console` for an interactive prompt that will allow you to experiment.
@@ -220,4 +373,4 @@ The gem is available as open source under the terms of the [MIT License](https:/
 
 ## Code of Conduct
 
-Everyone interacting in the ActiveRecord Data Loader project’s codebases, issue trackers, chat rooms and mailing lists is expected to follow the [code of conduct](https://github.com/abeiderman/active_record_data_loader/blob/master/CODE_OF_CONDUCT.md).
+Everyone interacting in the _active_record_data_loader_ project’s codebases, issue trackers, chat rooms and mailing lists is expected to follow the [code of conduct](https://github.com/abeiderman/active_record_data_loader/blob/master/CODE_OF_CONDUCT.md).

data/Rakefile

@@ -3,10 +3,16 @@
 require "bundler/gem_tasks"
 require "rspec/core/rake_task"
 require "rubocop/rake_task"
-require "coveralls/rake/task"
 
 RSpec::Core::RakeTask.new(:spec)
 RuboCop::RakeTask.new(:rubocop)
-Coveralls::RakeTask.new
 
 task default: [:spec, :rubocop]
+
+task :wait_for_test_db do
+  require "active_record_data_loader"
+  require "./spec/active_record_helper"
+
+  ActiveRecordHelper.wait_for_mysql
+  ActiveRecordHelper.wait_for_postgres
+end

data/active_record_data_loader.gemspec

@@ -8,7 +8,7 @@ Gem::Specification.new do |spec|
   spec.name          = "active_record_data_loader"
   spec.version       = ActiveRecordDataLoader::VERSION
   spec.authors       = ["Alejandro Beiderman"]
-  spec.email         = ["abeiderman@gmail.com"]
+  spec.email         = ["active_record_data_loader@ossprojects.dev"]
 
   spec.summary       = "A utility to bulk load test data for performance testing."
   spec.description   = "A utility to bulk load test data for performance testing."
@@ -20,7 +20,7 @@ Gem::Specification.new do |spec|
     spec.metadata["source_code_uri"] = "https://github.com/abeiderman/active_record_data_loader"
   else
     raise "RubyGems 2.0 or newer is required to protect against " \
-      "public gem pushes."
+          "public gem pushes."
   end
 
   spec.files = `git ls-files -z`.split("\x0").reject do |f|
@@ -30,20 +30,21 @@ Gem::Specification.new do |spec|
   spec.executables   = spec.files.grep(%r{^bin/}) { |f| File.basename(f) }
   spec.require_paths = ["lib"]
 
-  spec.required_ruby_version = ">= 2.3.0"
+  spec.required_ruby_version = ">= 2.5.0"
 
-  spec.add_dependency "activerecord", ">= 4.0"
+  spec.add_dependency "activerecord", ">= 5.0"
 
   spec.add_development_dependency "appraisal"
   spec.add_development_dependency "bundler", ">= 1.16"
-  spec.add_development_dependency "coveralls"
   spec.add_development_dependency "mysql2"
   spec.add_development_dependency "pg"
   spec.add_development_dependency "pry"
-  spec.add_development_dependency "rake", "~> 12.0"
+  spec.add_development_dependency "rake", "~> 13.0"
   spec.add_development_dependency "rspec", "~> 3.0"
   spec.add_development_dependency "rspec-collection_matchers"
   spec.add_development_dependency "rubocop"
+  spec.add_development_dependency "simplecov"
+  spec.add_development_dependency "simplecov-lcov"
   spec.add_development_dependency "sqlite3"
   spec.add_development_dependency "timecop"
 end

data/config/database.yml

@@ -1,6 +1,7 @@
 postgres:
   adapter: "postgresql"
   host: "127.0.0.1"
+  port: "2345"
   database: "test"
   username: "test"
   password: "test"
@@ -12,6 +13,7 @@ sqlite3:
 mysql:
   adapter: "mysql2"
   host: "127.0.0.1"
+  port: "3306"
   database: "test"
   username: "test"
   password: "test"

data/docker-compose.yml

@@ -0,0 +1,18 @@
+version: "3.9"
+services:
+  postgres:
+    image: postgres:11
+    ports:
+      - "2345:5432"
+    environment:
+      - POSTGRES_USER=test
+      - POSTGRES_PASSWORD=test
+  mysql:
+    image: mysql:5
+    ports:
+      - "3306:3306"
+    environment:
+      - MYSQL_ROOT_PASSWORD=test
+      - MYSQL_USER=test
+      - MYSQL_PASSWORD=test
+      - MYSQL_DATABASE=test

data/gemfiles/activerecord_6.gemfile

@@ -2,6 +2,6 @@
 
 source "https://rubygems.org"
 
-gem "activerecord", "6.0.0.rc1"
+gem "activerecord", "~>6.1"
 
 gemspec path: "../"

data/lib/active_record_data_loader/active_record/{belongs_to_configuration.rb → belongs_to_data_provider.rb}

@@ -2,30 +2,31 @@
 
 module ActiveRecordDataLoader
   module ActiveRecord
-    class BelongsToConfiguration
-      def self.config_for(ar_association:, query: nil)
+    class BelongsToDataProvider
+      def self.provider_for(ar_association:, query: nil, strategy: :random)
         raise "#{name} does not support polymorphic associations" if ar_association.polymorphic?
 
-        { ar_association.join_foreign_key.to_sym => new(ar_association, query).foreign_key_func }
+        { ar_association.join_foreign_key.to_sym => new(ar_association, query, strategy).foreign_key_func }
       end
 
-      def initialize(ar_association, query)
+      def initialize(ar_association, query, strategy)
         @ar_association = ar_association
         @query = query
+        @strategy = strategy
       end
 
       def foreign_key_func
-        -> { possible_values.sample }
+        -> { possible_values.next }
       end
 
       private
 
       def possible_values
-        @possible_values ||= base_query.pluck(@ar_association.join_primary_key).to_a
+        @possible_values ||= List.for(base_query.pluck(@ar_association.join_primary_key), strategy: @strategy)
       end
 
       def base_query
-        if @query&.respond_to?(:call)
+        if @query.respond_to?(:call)
           @query.call.all
         else
           @ar_association.klass.all

data/lib/active_record_data_loader/active_record/{column_configuration.rb → column_data_provider.rb}

@@ -2,7 +2,7 @@
 
 module ActiveRecordDataLoader
   module ActiveRecord
-    class ColumnConfiguration
+    class ColumnDataProvider
       class << self
         VALUE_GENERATORS = {
           enum: EnumValueGenerator,
@@ -12,7 +12,7 @@ module ActiveRecordDataLoader
           datetime: DatetimeValueGenerator,
         }.freeze
 
-        def config_for(model_class:, ar_column:, connection_factory:)
+        def provider_for(model_class:, ar_column:, connection_factory:)
           raise_error_if_not_supported(model_class, ar_column)
 
           {

data/lib/active_record_data_loader/active_record/enum_value_generator.rb

@@ -5,34 +5,35 @@ module ActiveRecordDataLoader
     class EnumValueGenerator
       class << self
         def generator_for(model_class:, ar_column:, connection_factory:)
-          values = enum_values_for(model_class, ar_column.sql_type, connection_factory)
+          values = enum_values_for(ar_column.sql_type, connection_factory)
           -> { values.sample }
         end
 
         private
 
-        def enum_values_for(model_class, enum_type, connection_factory)
+        def enum_values_for(enum_type, connection_factory)
           connection = connection_factory.call
 
           if connection.adapter_name.downcase.to_sym == :postgresql
-            postgres_enum_values_for(model_class, enum_type)
+            postgres_enum_values_for(connection, enum_type)
           elsif connection.adapter_name.downcase.to_s.start_with?("mysql")
-            mysql_enum_values_for(model_class, enum_type)
+            mysql_enum_values_for(enum_type)
           else
             []
           end
+        ensure
+          connection&.close
         end
 
-        def postgres_enum_values_for(model_class, enum_type)
-          model_class
-            .connection
+        def postgres_enum_values_for(connection, enum_type)
+          connection
             .execute("SELECT unnest(enum_range(NULL::#{enum_type}))::text")
             .map(&:values)
             .flatten
             .compact
         end
 
-        def mysql_enum_values_for(_model_class, enum_type)
+        def mysql_enum_values_for(enum_type)
           enum_type
             .to_s
             .downcase

data/lib/active_record_data_loader/active_record/integer_value_generator.rb

@@ -5,7 +5,7 @@ module ActiveRecordDataLoader
     class IntegerValueGenerator
       class << self
         def generator_for(model_class:, ar_column:, connection_factory: nil)
-          range_limit = [(256**number_of_bytes(ar_column)) / 2 - 1, 1_000_000_000].min
+          range_limit = [((256**number_of_bytes(ar_column)) / 2) - 1, 1_000_000_000].min
 
           -> { rand(0..range_limit) }
         end

data/lib/active_record_data_loader/active_record/list.rb

@@ -0,0 +1,47 @@
+# frozen_string_literal: true
+
+module ActiveRecordDataLoader
+  module ActiveRecord
+    class List
+      def self.for(enumerable, strategy: :random)
+        if strategy == :random_cycle
+          RandomCycle.new(enumerable)
+        else
+          Random.new(enumerable)
+        end
+      end
+
+      class Random
+        def initialize(enumerable)
+          @list = enumerable
+        end
+
+        def next
+          @list.sample
+        end
+      end
+
+      class RandomCycle
+        def initialize(enumerable)
+          @enumerable = enumerable
+          @count = enumerable.count
+          reset_list
+        end
+
+        def next
+          value = @list.next
+          reset_list if (@index += 1) >= @count
+          value
+        end
+
+        private
+
+        def reset_list
+          @index = 0
+          @enumerable = @enumerable.shuffle
+          @list = @enumerable.cycle
+        end
+      end
+    end
+  end
+end

data/lib/active_record_data_loader/active_record/model_data_generator.rb

@@ -8,9 +8,12 @@ module ActiveRecordDataLoader
       def initialize(
         model:,
         column_settings:,
+        connection_factory:,
+        logger:,
+        raise_on_duplicates:,
+        max_duplicate_retries:,
         polymorphic_settings: [],
-        belongs_to_settings: [],
-        connection_factory:
+        belongs_to_settings: []
       )
         @model_class = model
         @table = model.table_name
@@ -18,6 +21,11 @@ module ActiveRecordDataLoader
         @polymorphic_settings = polymorphic_settings
         @belongs_to_settings = belongs_to_settings.map { |s| [s.name, s.query] }.to_h
         @connection_factory = connection_factory
+        @raise_on_duplicates = raise_on_duplicates
+        @max_duplicate_retries = max_duplicate_retries
+        @logger = logger
+        @index_tracker = UniqueIndexTracker.new(model: model, connection_factory: connection_factory)
+        @index_tracker.map_indexed_columns(column_list)
       end
 
       def column_list
@@ -25,11 +33,41 @@ module ActiveRecordDataLoader
       end
 
       def generate_row(row_number)
-        column_list.map { |c| column_data(row_number, c) }
+        @index_tracker.capture_unique_values(generate_row_with_retries(row_number))
       end
 
       private
 
+      def generate_row_with_retries(row_number)
+        retries = 0
+        while @index_tracker.repeating_unique_values?(row = generate_candidate_row(row_number))
+          if (retries += 1) > @max_duplicate_retries
+            raise DuplicateKeyError, <<~MSG if @raise_on_duplicates
+              Exhausted retries looking for unique values for row #{row_number} for '#{table}'.
+              Table '#{table}' has unique indexes that would have prevented inserting this row. If you would
+              like to skip non-unique rows instead of raising, configure `raise_on_duplicates` to be `false`.
+            MSG
+
+            @logger.warn(
+              "[ActiveRecordDataLoader] "\
+              "Exhausted retries looking for unique values. Skipping row #{row_number} for '#{table}'."
+            )
+            return nil
+          else
+            @logger.info(
+              "[ActiveRecordDataLoader] "\
+              "Retrying row #{row_number} for '#{table}' looking for unique values compliant with indexes. "\
+              "Retry number #{retries}."
+            )
+          end
+        end
+        row
+      end
+
+      def generate_candidate_row(row_number)
+        column_list.map { |c| column_data(row_number, c) }
+      end
+
       def column_data(row_number, column)
         column_value = columns[column]
         return column_value unless column_value.respond_to?(:call)
@@ -56,9 +94,9 @@ module ActiveRecordDataLoader
         @model_class
           .columns_hash
           .reject { |name| name == @model_class.primary_key }
-          .select { |_, c| ColumnConfiguration.supported?(model_class: @model_class, ar_column: c) }
+          .select { |_, c| ColumnDataProvider.supported?(model_class: @model_class, ar_column: c) }
           .map do |_, c|
-            ColumnConfiguration.config_for(
+            ColumnDataProvider.provider_for(
               model_class: @model_class,
               ar_column: c,
               connection_factory: @connection_factory
@@ -73,16 +111,33 @@ module ActiveRecordDataLoader
           .select(&:belongs_to?)
           .reject(&:polymorphic?)
           .map do |assoc|
-            BelongsToConfiguration.config_for(ar_association: assoc, query: @belongs_to_settings[assoc.name])
+            BelongsToDataProvider.provider_for(
+              ar_association: assoc,
+              query: @belongs_to_settings[assoc.name],
+              strategy: column_config_strategy(assoc)
+            )
           end
           .reduce({}, :merge)
       end
 
       def polymorphic_config
         @polymorphic_settings
-          .map { |s| PolymorphicBelongsToConfiguration.config_for(polymorphic_settings: s) }
+          .map do |s|
+            PolymorphicBelongsToDataProvider.provider_for(
+              polymorphic_settings: s,
+              strategy: column_config_strategy(s.model_class.reflect_on_association(s.name))
+            )
+          end
           .reduce({}, :merge)
       end
+
+      def column_config_strategy(column)
+        if @index_tracker.contained_in_index?(column)
+          :random_cycle
+        else
+          :random
+        end
+      end
     end
   end
 end