Spectator

Spectator is a fully-featured spec-based test framework for Crystal. It mimics features from RSpec. Developers coming from Ruby and RSpec will feel right at home. Spectator provides additional functionality to make testing easier and more fluent.

Goal:

Spectator is designed to:

Installation

Add this to your application's shard.yml:

development_dependencies:
  spectator:
    gitlab: arctic-fox/spectator
    version: ~> 0.12.0

Usage

If it doesn't exist already, create a spec/spec_helper.cr file. In it, place the following:

require "../src/*"
require "spectator"

This will include Spectator and the source code for your shard. Now you can start writing your specs. The syntax is the same as what you would expect from modern RSpec. The "expect" syntax is recommended and the default, however the "should" syntax is also available. Your specs must be wrapped in a Spectator.describe block. All other blocks inside the top-level block may use describe and context without the Spectator. prefix.

Here's a minimal spec to demonstrate:

require "./spec_helper"

Spectator.describe String do
  subject { "foo" }

  describe "#==" do
    context "with the same value" do
      let(value) { subject.dup }

      it "is true" do
        is_expected.to eq(value)
      end
    end

    context "with a different value" do
      let(value) { "bar" }

      it "is false" do
        is_expected.to_not eq(value)
      end
    end
  end
end

If you find yourself trying to shoehorn in functionality or unsure how to write a test, please create an issue for it. The goal is to make it as easy as possible to write specs and keep your code clean. We may come up with a solution or even introduce a feature to support your needs.

NOTE: Due to the way this shard uses macros, you may find that some code you would expect to work, or works in other spec libraries, creates syntax errors. If you run into this, please create an issue so that we may try to resolve it.

Features

Spectator has all of the basic functionality for BDD. For full documentation on what it can do, please visit the wiki.

Contexts

The DSL supports arbitrarily nested contexts. Contexts can have values defined for multiple tests (let and subject). Additionally, hooks can be used to ensure any initialization or cleanup is done (before, after, and around). Pre- and post-conditions can be used to ensure code contracts are kept.

# Initialize the database before running the tests in this context.
before_all { Database.init }

# Teardown the database and cleanup after tests in the is context finish.
after_all { Database.cleanup }

# Before each test, add some rows to the database.
let(row_count) { 5 }
before_each do
  row_count.times { Database.insert_row }
end

# Remove the rows after the test to get a clean slate.
after_each { Database.clear }

describe "#row_count" do
  it "returns the number of rows" do
    expect(Database.row_count).to eq(row_count)
  end
end

Spectator has different types of contexts to reduce boilerplate. One is the sample context. This context type repeats all tests (and contexts within) for a set of values. For instance, some feature should behave the same for different input. However, some inputs might cause problems, but should behave the same. An example is various strings (empty strings, quoted strings, strings with non-ASCII, etc), and numbers (positive, negative, zero, NaN, infinity).

# List of integers to test against.
def various_integers
  [-7, -1, 0, 1, 42]
end

# Repeat nested tests for every value in `#various_integers`.
sample various_integers do |int|
  # Example that checks if a fictitious method `#format` converts to strings.
  it "formats correctly" do
    expect(format(int)).to eq(int.to_s)
  end
end

Another context type is provided. This context drastically reduces the amount of code needed in some scenarios. It can be used where one (or more inputs) changes the output of multiple methods. The provided context gives a concise syntax for this use case.

subject(user) { User.new(age) }

# Each expression in the `provided` block is its own test.
provided age = 10 do
  expect(user.can_drive?).to be_false
  expect(user.can_vote?).to be_false
end

provided age = 16 do
  expect(user.can_drive?).to be_true
  expect(user.can_vote?).to be_false
end

provided age = 18 do
  expect(user.can_drive?).to be_true
  expect(user.can_vote?).to be_true
end

Assertions

Spectator supports two formats for assertions (expectations). The preferred format is the "expect syntax". This takes the form:

expect(THIS).to eq(THAT)

The other format, "should syntax" is used by Crystal's default Spec.

THIS.should eq(THAT)

The first format doesn't monkey-patch the Object type. And as a bonus, it captures the expression or variable passed to expect(). For instance, compare these two tests:

foo = "Hello world"
foo.size.should eq(12) # Wrong on purpose!

Produces this error output:

Failure: 11 does not equal 12

  expected: 11
    actual: 12

Which is reasonable, but where did 11 come from? Alternatively, with the "expect syntax":

foo = "Hello world"
expect(foo.size).to eq(12) # Wrong on purpose!

Produces this error output:

Failure: foo.size does not equal 12

  expected: 12
    actual: 11

This makes it clearer what was being tested and failed.

Matchers

Spectator has a variety of matchers for assertions. These are named in such a way to help tests read as plain English. Matchers can be used on any value or block.

There are typical matchers for testing equality: eq and ne. And matchers for comparison: <, <=, >, >=, be_within. There are matchers for checking contents of collections: contain, have, start_with, end_with, be_empty, have_key, and more. See the wiki for a full list of matchers.

Running

Spectator supports multiple options for running tests. "Fail fast" aborts on the first test failure. "Fail blank" fails if there are no tests. Tests can be filtered by their location and name. Additionally, tests can be randomized. Spectator can be configured with command-line arguments, a configure block in a spec_helper.cr file, and .spectator configuration file.

Spectator.configure do |config|
  config.fail_blank # Fail on no tests.
  config.randomize  # Randomize test order.
  config.profile    # Display slowest tests.
end

Mocks and Doubles

Spectator supports an extensive mocking feature set via two types - mocks and doubles. Mocks are used to override behavior in existing types. Doubles are objects that stand-in when there are no type restrictions. Stubs can be defined on both which control how methods behave.

abstract class Interface
  abstract def invoke(thing) : String
end

# Type being tested.
class Driver
  def do_something(interface : Interface, thing)
    interface.invoke(thing)
  end
end

Spectator.describe Driver do
  # Define a mock for Interface.
  mock Interface

  # Define a double that the interface will use.
  double(:my_double, foo: 42)

  it "does a thing" do
    # Create an instance of the mock interface.
    interface = mock(Interface)
    # Indicate that `#invoke` should return "test" when called.
    allow(interface).to receive(:invoke).and_return("test")

    # Create an instance of the double.
    dbl = double(:my_double)
    # Call the mock method.
    subject.do_something(interface, dbl)
    # Verify everything went okay.
    expect(interface).to have_received(:invoke).with(dbl)
  end
end

For details on mocks and doubles, see the wiki.

Output

Spectator matches Crystal's default Spec output with some minor changes. JUnit and TAP are also supported output formats. There are also highly detailed JSON and HTML outputs.

Development

This shard is still in active development. New features are being added and existing functionality improved. Spectator is well-tested, but may have some yet-to-be-found bugs.

Feature Progress

In no particular order, features that have been implemented and are planned. Items not marked as completed may have partial implementations.

How it Works (in a nutshell)

This shard makes extensive use of the Crystal macro system to build classes and modules. Each describe and context block creates a new class that inherits its parent. The it block creates an method. An instance of the group class is created to run the test. Each group class includes all test values and hooks.

Contributing

  1. Fork it (GitHub https://github.com/icy-arctic-fox/spectator/fork or GitLab https://gitlab.com/arctic-fox/spectator/fork/new)
  2. Create your feature branch (git checkout -b my-new-feature)
  3. Commit your changes (git commit -am 'Add some feature')
  4. Push to the branch (git push origin my-new-feature)
  5. Create a new Pull/Merge Request

Please make sure to run crystal tool format before submitting. The CI build checks for properly formatted code. Ameba is run to check for code style.

Documentation is automatically generated and published to GitLab pages. It can be found here: https://arctic-fox.gitlab.io/spectator

This project's home is (and primarily developed) on GitLab. A mirror is maintained to GitHub. Issues, pull requests (merge requests), and discussion are welcome on both. Maintainers will ensure your contributions make it in.

For more information, see: CONTRIBUTING.md

Testing

Tests must be written for any new functionality.

The spec/ directory contains feature tests as well as unit tests. These demonstrate small bits of functionality. The feature tests are grouped into sub directories based on their type, they are:

The other directories are for unit testing various parts of Spectator.