I have a whole load of Test::Unit::TestCase class files. Each one contains a whole load of test_* methods. If I want to run one or two individual test_ methods from a selection of  TestCase classes, how do I do it? Simple. Test::Unit has a feature that allows me to create TestSuites from TestCases just by specifying the name of the test when I create an instance of the relevant TestCase class.

I create a file to contain my custom TestSuite (‘sanity_tests.rb‘), and add a require for each TestCase that I contains a test_ method that I want to run as part of this custom TestSuite. I then create a new TestSuite giving it a name. Then, for each test_ method I want to add to my custom TestSuite, I create an instance of its containing TestCase class, passing in a string representation of the test_ method I want to execute. The initialization returns a TestSuite object which is appended to the custom TestSuite that I just created.

Once I’ve done this for every test_ method that I’m interested in, I execute my custom TestSuite and wait for the results! Here’s a code example. The comments should make it clear…

#relevant test::unit requires
require 'test/unit'
require 'test/unit/ui/console/testrunner'

#TestCase classes that contain the methods I want to execute
require 'One/test_one'
require 'Two/test_two'

#create a new empty TestSuite, giving it a name
my_tests = Test::Unit::TestSuite.new("My Special Tests")

#add the test method called 'test_one' defined
#in the TestOne TestCase class to the my_tests
#test suite defined above
my_tests << TestOne.new('test_one')

#add the test method called 'test_me' defined
#in the TestTwo TestCase class to the my_tests
#test suite defined above
my_tests << TestTwo.new('test_me')

#run the suite
Test::Unit::UI::Console::TestRunner.run(my_tests)

And the output is as expected:

Loaded suite My Special Tests
Started
..
Finished in 1.000935 seconds.
2 tests, 2 assertions, 0 failures, 0 errors


I’m using Ruby’s Test::Unit with Watir and Firewatir for a project I’m working on. Tests are divided up into test classes, each test class being in one file. I then have files that aggregate the test case classes using the magical AutoRunner functionality (ie: ‘require’ all the test case classes, and they’ll be automatically run). These ‘test suites’ are very useful when I want to run a whole load of tests, but sometimes I just want to know how many tests I have in total – I need to provide this statistic daily and don’t want to have to do the count manually.

After browsing through the Test::Unit code (I was hoping it would send me to sleep – it was 3am), I figured out how the test case aggregation function worked. Turns out I could use the same concept to do my count, if I could stop the tests from running when they get ‘require’d!

Here’s the contents of a class you can use that will search the directory it lives in for all files whose names match ‘all_test*.rb’; requires the relevant files, performs a count on each one, and displays the results including a total:

#stop the tests from being run
require "test/unit"
Test::Unit.run=(true)

#read in only my 'all_test*.rb' files, and
Dir.glob(File.join(File.dirname(__FILE__),'all_test*.rb')) do |f|
  require f
end

#populate an array with all classes that inherit from TestCase
testcases = []
ObjectSpace.each_object(Class) do |obj|
  testcases << obj if(obj < Test::Unit::TestCase)
end

#sort by name
testcases.sort! { |a,b| a.name <=> b.name }

#print results!
total_tests = 0
testcases.each do |testcase|
  puts "#{testcase.name}: #{testcase.suite.size}"
  total_tests += testcase.suite.size
end
puts "Total: #{total_tests}"