Database Dependency Graph (DDG)

ddg is a tool for building and manipulating database dependency graphs. Hence, the origin of the abbreviation.

The database dependency graph is defined as follows: a directed, acyclic graph (DAG) where nodes represent tables and edges represent dependencies based on foreign key constraint(s). It currently supports two behaviors:

1. Determining the evaluation order and optionally performing a task on each node in the evaluation order.
2. Generating an image file containing a graph diagram of the database dependency graph.

The data stores that ddg supports are the following:

• MySQL
• PostgreSQL
• Redshift

Add this line to your application's Gemfile:

gem 'ddg'

And then execute:

$ bundle install

Or install it yourself as:

$ gem install ddg

To initialize a dependency graph:

require 'dependency_graph'

graph = DependencyGraph.new(
  adapter: :postgresql,
  database: 'database',
  user: 'user',
  password: 'password',
  port: 12345,
  host: 'www.example.com'
)

To print the evaluation order:

puts(graph.evaluation_order)

# >> [users, reports, user_reports]

To perform an action on each node, in the evaluation order:

graph.evaluation_order do |node|
  ETL.incremental_load(node)
end

To generate a graph.png image of the dependency graph:

graph.visualize

To use the dependency graph with a CLI:

$ ddg -a postgresql -d dev -u dev_ro -x password123 -p 16379 -h dev.com --evaluation-order
# >> [users, reports, user_reports]

To use the dependency graph with a Rake command:

$ bundle exec rake ddg:evaluation_order
# >> [users, reports, user_reports]

• Lazy initialization, with an option to override. Depending on the number of tables and foreign keys in the data store, the operation of building the graph is expensive. Delay until absolutely necessary.
• Duck-typing over an abstract class hierarchy or interface.
• Extensible data store adapter interface.
• Factory pattern for instantiating adapters.
• Measured performance and benchmark testing.
• CI with Travis-CI.
• SemVer and CHANGELOG compliant.
• Example application that uses this gem and it's functionality.

In theory, any data store that supports the information schema (ANSI-standard) is supported by the existing SQL used to extract table to foreign key mappings. However, if a data store is not currently supported, an adapter can be written rather easily by creating a new class inheriting from DDG::Adapter::Base.

To add an adapter that supports the information schema:

# frozen_string_literal: true

require 'ddg/adapter/base'

module DDG
  module Adapter
    class DataStore < Base
      # TODO: Implement the constructor by assigning a client to @db.
      def initialize(config)
        @db = DataStoreClient.new(
          host: config[:host],
          port: config[:port],
          username: config[:user],
          password: config[:password],
          database: config[:database]
        )
      end

      # TODO: Implement the select function, which executes
      # the SQL contained in the `sql` parameter and
      # returns an Array of Hashes, representing named rows.
      def select(sql)
        [{}]
      end
    end
  end
end

To add an adapter that does NOT support the information schema:

# frozen_string_literal: true

require 'ddg/adapter/base'

module DDG
  module Adapter
    class DataStore < Base
      # TODO: Implement the constructor by assigning a client to @db.
      def initialize(config)
        @db = DataStoreClient.new
      end

      # TODO: Implement the following instance method.
      #
      # Retrieves a mapping from table to a
      # set of its foreign key constraints.
      #
      # Example:
      #   {
      #     user_reports: [
      #       :users,
      #       :reports,
      #       ...
      #     ],
      #     ...
      #   }
      #
      # @return [Hash]
      def tables_with_foreign_keys; end
    end
  end
end

From the PostgreSQL documentation on information schema querying: "Only those constraints are shown for which the current user has write access to the referencing table (by way of being the owner or having some privilege other than SELECT)."

The SQL to extract the foreign keys of each table was heavily influenced, barring minor changes, from this article: https://msdn.microsoft.com/en-us/library/aa175805(SQL.80).aspx.

Make sure to copy the hooks/ directory into your local .git/hooks/ directory to ensure the git hooks for this project run. Most notably, there exists a pre-commit hook that runs tests and a linter over the code, failing the commit if the tests or linter fails.

Bug reports and pull requests are welcome on GitHub at https://github.com/brandon-powers/ddg.

The gem is available as open source under the terms of the MIT License.

• Fix integration with Travis-CI.
• Write/Share a blog post on DDG for Medium.
• Test additional cases (cycles, larger data models, etc.).
• Consider adding wrapper functions supported by RGL, which is an easy, low-hanging fruit improvement to make this tool more useful. The goal is to connect a database dependency graph to the hierarchical structure and operations of the graph data structure.