Ruby + Golang Testing = Rgot

Rgot is a testing package convert from golang testing.

test/sample.rb

class Sample
  def sum(i, j)
    i + j
  end
end

test/pass_test.rb

module SampleTest
  class TypeSum < Struct.new(:left, :right, :expect)
  end

  DATA = [
    TypeSum.new(2, 3, 5),
    TypeSum.new(12, 9, 21),
    TypeSum.new(85, 42, 127),
  ]

  def test_pass(t)
    s = Sample.new
    DATA.each do |ts|
      sum = s.sum(ts.left, ts.right)
      unless sum.kind_of?(Integer)
        t.error("expect Integer got #{sum.class}")
      end
      unless sum == ts.expect
        t.error("expect #{ts.expect} got #{sum}")
      end
    end
  end
end

$ rgot -v --require ./test/sample test/pass_test.rb
=== RUN test_pass
--- PASS: test_pass (0.00003s)
PASS
ok	0.001s

I provide a very simple testing feature to you.

Rgot testing is quite different from RSpec and MiniTest etc.

Rgot carve out a new world of testing.

So, You check only bad case in testing.

You can write simple benchmark script with testing.

This benchmark to adjust the time automatically.

module FooTest
  def benchmark_something(b)
    i = 0
    while i < b.n
      something(1)
      i += 1
    end
  end
end

$ rgot foo_test.rb --bench .
benchmark_something	14400000	81 ns/op
ok	FooTest	2.782s

b.n is automatically adjusted.

$ rgot target_file_test.rb --fuzz . --fuzztime 1

Fuzzing tests are also supported. Please refer to the gloang documentation for details.

https://go.dev/security/fuzz/

module FooTest
  # To enable fuzzing, the method name
  # should be prefixed with `fuzz`.
  def fuzz_any_func(f)
    f.add(5, "hello")
    f.fuzz do |t, i, s|
      out, err = foo(i, s)
      if err != nil && out != ""
        t.errorf("%s, %s", out, err)
      end
    end
  end
end

Rgot's example feature is the best and if you want to write the sample code of your library.

While presenting the sample code, it will be able to test whether the display results match at the same time.

module FooTest
  class User
    def initialize(name)
      @name = name
    end

    def hello
      "Hello #{@name}"
    end
  end

  def example_something
    user = User.new('ksss')
    puts user.hello
    # Output:
    # Hello ksss
  end

  def example_fail
    user = User.new('ksss')
    puts user.hello
    # Output:
    # Hi ksss
  end
end

example_fail fail since output is different.

So, you can notice that the sample code is wrong.

FLAGTESTS = [
  ["%a", "[%a]"],
  ["%-a", "[%-a]"],
  ["%+a", "[%+a]"],
  ["%#a", "[%#a]"],
  ["% a", "[% a]"],
  ["%0a", "[%0a]"],
  ["%1.2a", "[%1.2a]"],
  ["%-1.2a", "[%-1.2a]"],
  ["%+1.2a", "[%+1.2a]"],
  ["%-+1.2a", "[%+-1.2a]"],
  ["%-+1.2abc", "[%+-1.2a]bc"],
  ["%-1.2abc", "[%-1.2a]bc"],
]

def test_flag_parser(t)
  FLAGTESTS.each do |input, output|
    s = Flag.print(input)
    unless s == output
      t.errorf("Flag.print(%p) => %p, want %p", input, s, output)
    end
  end
end

see https://github.com/golang/go/wiki/TableDrivenTests

Filename should be set '*_test.rb'

Module name should be set 'XxxTest'

'Xxx' can replace any string (in range of ruby module)

Testing code file can split any number.

But all file should be have one module (like golang package name).

module XxxTest
  # ...
end

Method name should be set test_* for testing.

And benchmark method should be set benchmark_*.

And fuzz method should be set fuzz_*.

And example method should be set example_*.

module XxxTest
  def test_any_name(t)
  end

  def benchmark_any_name(b)
  end

  def fuzz_any_name(f)
  end

  def example_any_name
  end
end

$ rgot -h
Usage: rgot [options]
    -v, --verbose                    log all tests
        --version                    show Rgot version
        --bench [regexp]             benchmark
        --benchtime [sec]            benchmark running time
        --timeout [sec]              set timeout sec to testing
        --cpu [count,...]            set cpu counts of comma split
        --thread [count,...]         set thread counts of comma split
        --require [path]             load some code before running
        --load-path [path]           Specify $LOAD_PATH directory
        --fuzz [regexp]              run the fuzz test matching `regexp`
        --fuzztime [sec]             time to spend fuzzing; default is to run indefinitely

$ rgot file_of_test.rb
PASS
ok	0.001s

Set filename to argument.

Just only start testing file_of_test.rb.

$ rgot sample
PASS
ok	0.002s

And set dirname to argument, run all case of testing under this dir.

$ rgot -v target_file_test.rb
=== RUN test_pass
--- PASS: test_pass (0.00005s)
PASS
ok	0.001s

Show all log and more detail information of testing.

$ rgot target_file_test.rb --bench .

Run testing with benchmark.

. means match all string for regexp.

Set someone if you only run benchmark to match someone method.(e.g. benchmark_someone_1)

Benchmark for parallel performance.

--cpu option set process counts (default Etc.nprocessors).

And --thread option set thread counts (default 1).

Benchmark fork, run and report each by process counts.

(process and thread means ruby/linux native process and thread)

module FooTest
  def benchmark_any_func(b)
    b.run_parallel do |pb|
      # pb is instance of Rgot::PB
      # call some time by b.n
      while pb.next
        some_func()
      end
    end
  end
end

$ rgot foo_test.rb --bench . --cpu=2,4 --thread=2,4
benchmark_any_func-2(2)	40	13363604 ns/op
benchmark_any_func-2(4)	160	7125845 ns/op
benchmark_any_func-4(2)	160	7224815 ns/op
benchmark_any_func-4(4)	320	3652431 ns/op
ok	FooTest	3.061s

$ rgot target_file_test.rb --timeout 3

You can set timeout sec for testing (default 0).

Fail testing and print raised exception message to STDERR if timeout.

Run benchmark function without framework.

result = Rgot.benchmark do |b|
  i = 0
  while i < b.n
    some_func()
    i += 1
  end
end
puts result #=> 100000	100 ns/op

Check running with option verbose true/false.

Main method run first on testing.

And this is default virtual main code.

module TestSomeCode
  def test_main(m)
    m.run
  end
end

Main method should be set 'test_main' only.

Variable m is a instance of Rgot::M class means Main.

Rgot::M#run start all testing methods.

And return code of process end status.

If you want to run before/after all testing method, You can write like this.

module TestSomeCode
  def test_main(m)
    the_before_running_some_code
    code = m.run
    the_after_running_some_code
    code
  end
end

Rgot::Common is inherited to Rgot::T and Rgot::B

Rgot::Common have some logging method.

t.log("wooooo", 1, 2, 3)

Write any log message.

But this message to show need -v option.

Write any log message like sprintf.

t.logf("%d-%s", 10, "foo")

t.error("expect #{a} got #{b}")

Test fail and show some error message.

Fail loggin same as logf

Testing stop and fail with log.

t.fatal("fatal error!")

Fatal logging same as logf

t.skip("this method was skipped")

Skip current testing method.

And run to next testing method.

Skip logging same as logf

Testing is a main usage of this package.

module TestSomeCode
  def test_some_1(t)
  end
end

The t variable is instance of Rgot::T class means Testing.

For Benchmark class.

Can use log methods same as Rgot::T class

Automatic number calculated by running time.

Recommend to this idiom.

def benchmark_something(b)
  i = 0
  while i < b.n
    something()
    i += 1
  end
end

Reset benchmark timer

def benchmark_something(b)
  obj = heavy_prepare_method()
  b.reset_timer # you can ignore time of havy_prepare_method()
  i = 0
  while i < b.n
    obj.something()
    i += 1
  end
end

Start benchmark timer

Stop benchmark timer

Start parallel benchmark using fork and Thread.new.

This method should be call with block.

The block argument is instance of Rgot::PB.

Should be call this when parallel benchmark.

Repeat while return false.

Recommend this idiom.

def benchmark_foo(b)
  b.run_parallel do |pb|
    while pb.next
      some_func()
    end
  end
end

Set the sample value with #add. This value is also used as a test. It guesses the type from the value and generates a random value.

Generate the random value generated by #fuzz and execute the code. The t becomes an instance of Rgot::T and the test can be run as usual.

def fuzz_foo(f)
  f.add(100, "hello")
  f.fuzz do |t, i, s|
    i #=> 100, 84, 17, 9, 66, ...
    s #=> "hello", "Y\xD5\xAB\xBA\x8E", "r\x95D\xA5\xF7", "\xCEj=\x9C\xBD", ...
    if !foo(i, s)
      t.error("fail with i=#{i}, s=#{s}")
    end
  end
end

• Support to save and load fuzzing data

• Support sub testing
• Fix duration argument unit
• Refactoring
  • Fix M#initialize argument
  • Fix internal class API