Project

rndb

0.0
No release in over 3 years
Low commit activity in last 3 years
RnDB is an procedurally-generated mock database.
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
 Dependencies

Development

~> 1.17
~> 11.1
~> 2.16
~> 2.4

Runtime

~> 6.3
~> 0.9
 Project Readme

Ruby Style Guide test Coverage Status Gem Version Western Australia

RnDB

RnDB is a procedurally-generated fake database. Read the blog post for details.

Video Overview

IMAGE ALT TEXT HERE

Usage

First, create tables with columns that may have a pre-determined distribution of values (which can be queried on), or which may have a lambda for generating a random value, or both (such as for the weight column below).

class Widget < RnDB::Table
  column :colour, { red: 0.3, green: 0.12, brown: 0.01, blue: 0.5, orange: 0.07 }
  column :weight, { light: 0.3, medium: 0.64, heavy: 0.06 }, -> value do
    range =
      case value
      when :light
        (0.0..5.0)
      when :medium
        (6.0..9.0)
      when :heavy
        (10.0..20.0)
      end
    self.rand(range)
  end
  column :name, -> { Faker::Games::Pokemon.name }
end

Next, create a database with an optional random seed (137 in the example below), and add the table to the database, specifying the number of records to simulate (in this case, one trillion).

DB = RnDB::Database.new(137)
DB.add_table(Widget, 1e12)

Finally, fetch some records!

puts Widget.count
puts Widget[1234567890].name
puts Widget.find { |widget| (3.1415..3.1416).include?(widget.weight) }.attributes

Which will display the following:

1000000000000
Charmander
{:id=>61520, :weight=>3.1415121332762386, :colour=>:red, :name=>"Exeggcute"}

Note that the find command tested over sixty thousand records in just a second or two without needing to generate all attributes of each record first. But an even faster way of honing in on a particular record is to run a query, such as:

query = Widget.where(colour: [:brown, :orange], :weight => :heavy)

You can then retrieve random records that match the query with sample, use pluck to retrieve specific attributes without generating all of them, and use find or filter to further refine your search, like this:

puts query.count
puts query.sample.pluck(:colour, :weight)
puts query.lazy.filter { |ball| ball.name == 'Pikachu' }.map(&:id).take(10).to_a

Which will display the following:

4800000000
{:colour=>:orange, :weight=>16.096085279047017}
[429400000068, 429400000087, 429400000875, 429400000885, 429400000914, 429400001036, 429400001062, 429400001330, 429400001341, 429400001438]

Note that we used the lazy enumerator when filtering records to prevent running the block on all records before performing the map and taking the first ten results.

Release Process

  1. rake standard:fix
  2. rake version:bump:whatever
  3. rake gemspec:release BRANCH=main
  4. rake git:release BRANCH=main
  5. Create new release on GitHub to trigger ship workflow

Copyright

Copyright (c) 2021 Jason Hutchens. See LICENSE for further details.