A framework to allow Ruby applications to generate file/folder stubs (like the ‘rails` command does for Ruby on Rails, and the ’script/generate’ command within a Rails application during development).
RubiGen is originally extracted from Ruby on Rails (specifically the rails_generator from its railties gem).
The rails_generator was hardcoded with Rails-specific dependencies (‘RAILS_ROOT`), Rails generators (’app’ = Rails application; ‘model’ = Rails model+tests+migration), and generally assumed it was the only generator framework within the Ruby world (it was). So, any RubyGem whose name ended with ‘_generator’ was assumed to be a generator for a Rails application.
But if you are developing an Adhearsion application, then you may want a different set of generators. If you are developing a RubyGem, then you will want a different set of generators.
RubiGen exists to give different development environments their own generator framework.
Jeremy Kemper wrote the original Rails Generator, which is 95% of the basis of RubiGen. He’s awesome.
RubiGen is only required at development time, and normally isn’t required at deployment time (unless your application uses it to generate files at runtime).
On your development machine:
sudo gem install rubigen
RubiGen comes with a stand-alone executable to allow you to invoke generators:
For example, to run the rails’ ‘model` generator:
rubigen rails model Person name:string
would replace the normal
script/generate model Person name:string
RubiGen has been traditionally integrated into another project, such as ‘rails`, `newgem` or `camping`, rather than be used on its own.
These frameworks might use RubiGen for two reasons:
-
To generate an initial stub for developers, e.g. ‘rails` generated a stub to write a Rails application. `newgem` generates a stub to write a RubyGem. BTW - RubiGen has a builtin application `ruby_app` which generates a bare-bones Ruby application stub (lib, test, and script folders, plus a Rakefile, and a script/generate script)
-
To generate components within their development areas, e.g. Rails had its ‘script/generate` script within each Rails application, which hooked back into the rails_generator to lookup and execute generators.
So, there are two steps to integrating RubiGen into your framework:
-
Use it to generate an initial stub for the developers of your framework. This would create the folders (‘lib/app`, `test`, `script`, `doc`, `log`, etc) and starting files (`Rakefile`, `README.txt`, `test/test_helper.rb` etc). Importantly, it would generate a `script/generate` file. The `script/generate` file (example below) will allow developers of your framework to generate components/extensions within the framework. RubiGen allows you to restrict which generators are available. For example, within RubyGem development environment (as generated by `newgem`), the `script/generator` only shows `rubygem`-related generators. Rails could restrict `script/generator` to only show Rails related generators
-
Your framework RubyGem (e.g. ‘newgem` or `rails`) needs to add `rubigen` as a dependency, so that users of your RubyGem can access the generator framework.
There are two types of generators:
-
Application Generators - used by developers of your framework to get started. Generally, you will create one Application Generator for your framework. It generates a base stub (such as the ‘rails` stub for new Rails applications) for your framework users.
-
Component Generators - used by developers to extend their application. You may include 1+ built-in generators with your framework. Developers can also write generators for your framework, and like Rails’ generator install them in various places and have access to their via RubiGen.
Without RubiGen, to give your users a head start and create a stub for them, you will copiously use ‘mkdir_p` and `File.open`. Your script will either be primitive (only create the bare folders and very few files) or it will be very long and unreadable (ok, perhaps I’m just talking about the ‘newgem` script, which I am dubiously responsible for… :P).
With RubiGen, you can create stubs using powerful, yet simple, syntax. Templates are extracted into a ‘templates` folder, and activating the generator from a script requires only a few lines of code.
These are the ‘newgem` files related to its Application Generator.
bin/ bin/newgem # Application Generator script; Usage: newgem gemname [options] app_generators/ app_generators/newgem/ app_generators/newgem/newgem_generator.rb app_generators/newgem/USAGE app_generators/newgem/templates/ app_generators/newgem/templates/app.rb app_generators/newgem/templates/History.txt app_generators/newgem/templates/... lots and lots of templates
The ‘bin/newgem` script is very simple, and looks like:
require 'rubygems' require 'rubigen' if %w(-v --version).include? ARGV.first require 'newgem/version' puts "#{File.basename($0)} #{Newgem::VERSION}" exit(0) end require 'rubigen/scripts/generate' RubiGen::Base.use_application_sources! RubiGen::Scripts::Generate.new.run(ARGV, :generator => 'newgem')
You can copy and paste this for your own generator script, and place it in your RubyGem’s ‘bin` folder. Change `newgem` to your RubyGem’s name in the script above (and in all the folders listed above too)
NOTE: If you leave ‘newgem` there, then it will execute the `newgem_generator.rb` generator; as the generators are loaded from all RubyGem’s having ‘/app_generators` folders.
So, for your RubyGem, you need to keep the ‘/app_generators` folder (as you are creating an Application Generator, not a Component Generator), but change `newgem` to `your gem name` in all the subfolders and files. ESPECIALLY `newgem_generator.rb` -> `yourgem_generator.rb`, as this is how the generator is discovered (via `RubiGen::Base.use_application_sources!`).
All the generator work is performed within ‘yourgem_generator.rb`. A stub for it will be:
require 'rbconfig' class YourgemGenerator < RubiGen::Base DEFAULT_SHEBANG = File.join(Config::CONFIG['bindir'], Config::CONFIG['ruby_install_name']) default_options :shebang => DEFAULT_SHEBANG, :an_option => 'some_default' attr_reader :app_name, :module_name def initialize(runtime_args, runtime_options = {}) super usage if args.empty? @destination_root = args.shift @app_name = File.basename(File.expand_path(@destination_root)) @module_name = app_name.camelize extract_options end def manifest # Use /usr/bin/env if no special shebang was specified script_options = { :chmod => 0755, :shebang => options[:shebang] == DEFAULT_SHEBANG ? nil : options[:shebang] } windows = (RUBY_PLATFORM =~ /dos|win32|cygwin/i) || (RUBY_PLATFORM =~ /(:?mswin|mingw)/) record do |m| # Root directory and all subdirectories. m.directory '' BASEDIRS.each { |path| m.directory path } # Root m.template_copy_each %w( Rakefile ) m.file_copy_each %w( README.txt ) # Test helper m.template "test_helper.rb", "test/test_helper.rb" # Scripts %w( generate ).each do |file| m.template "script/#{file}", "script/#{file}", script_options m.template "script/win_script.cmd", "script/#{file}.cmd", :assigns => { :filename => file } if windows end end end protected def banner <<-EOS Create a stub for #{File.basename $0} to get started. Usage: #{File.basename $0} /path/to/your/app [options]" EOS end def add_options!(opts) opts.separator '' opts.separator "#{File.basename $0} options:" opts.on("-v", "--version", "Show the #{File.basename($0)} version number and quit.") end # Installation skeleton. Intermediate directories are automatically # created so don't sweat their absence here. BASEDIRS = %w( doc lib log script test tmp ) end
Easy peasy.