class Puppet::Test::TestHelper

This class is intended to provide an API to be used by external projects

when they are running tests that depend on puppet core.  This should
allow us to vary the implementation details of managing puppet's state
for testing, from one version of puppet to the next--without forcing
the external projects to do any of that state management or be aware of
the implementation details.

For now, this consists of a few very simple signatures. The plan is

that it should be the responsibility of the puppetlabs_spec_helper
to broker between external projects and this API; thus, if any
hacks are required (e.g. to determine whether or not a particular)
version of puppet supports this API, those hacks will be consolidated in
one place and won't need to be duplicated in every external project.

This should also alleviate the anti-pattern that we’ve been following,

wherein each external project starts off with a copy of puppet core's
test_helper.rb and is exposed to risk of that code getting out of
sync with core.

Since this class will be “library code” that ships with puppet, it does

not use API from any existing test framework such as rspec.  This should
theoretically allow it to be used with other unit test frameworks in the
future, if desired.

Note that in the future this API could potentially be expanded to handle

other features such as "around_test", but we didn't see a compelling
reason to deal with that right now.

Public Class Methods

after_all_tests() click to toggle source

Call this method once, at the end of a test run, when no more tests

will be run.

@return nil

# File lib/puppet/test/test_helper.rb, line 47
def self.after_all_tests()

end

after_each_test() click to toggle source

Call this method once per test, after execution of each individual test. @return nil

# File lib/puppet/test/test_helper.rb, line 92
def self.after_each_test()
  Puppet.settings.send(:clear_everything_for_tests)

  Puppet::Util::Storage.clear
  Puppet::Util::ExecutionStub.reset

  Puppet.clear_deprecation_warnings

  # uncommenting and manipulating this can be useful when tracking down calls to deprecated code
  #Puppet.log_deprecations_to_file("deprecations.txt", /^Puppet::Util.exec/)

  # Restore the indirector configuration.  See before hook.
  indirections = Puppet::Indirector::Indirection.send(:class_variable_get, :@@indirections)
  indirections.each do |indirector|
    $saved_indirection_state.fetch(indirector.name, {}).each do |variable, value|
      indirector.instance_variable_set(variable, value)
    end
  end
  $saved_indirection_state = nil

  # Restore the global process environment.  Can't just assign because this
  # is a magic variable, sadly, and doesn't do that™.  It is sufficiently
  # faster to use the compare-then-set model to avoid excessive work that it
  # justifies the complexity.  --daniel 2012-03-15
  unless ENV.to_hash == $old_env
    ENV.clear
    $old_env.each {|k, v| ENV[k] = v }
  end


  # Some tests can cause us to connect, in which case the lingering
  # connection is a resource that can cause unexpected failure in later
  # tests, as well as sharing state accidentally.
  # We're testing if ActiveRecord::Base is defined because some test cases
  # may stub Puppet.features.rails? which is how we should normally
  # introspect for this functionality.
  ActiveRecord::Base.remove_connection if defined?(ActiveRecord::Base)

  # Restore the load_path late, to avoid messing with stubs from the test.
  $LOAD_PATH.clear
  $old_load_path.each {|x| $LOAD_PATH << x }

end

before_all_tests() click to toggle source

Call this method once, when beginning a test run–prior to running

any individual tests.

@return nil

# File lib/puppet/test/test_helper.rb, line 40
def self.before_all_tests()

end

before_each_test() click to toggle source

Call this method once per test, prior to execution of each invididual test. @return nil

# File lib/puppet/test/test_helper.rb, line 53
def self.before_each_test()
  # We need to preserve the current state of all our indirection cache and
  # terminus classes.  This is pretty important, because changes to these
  # are global and lead to order dependencies in our testing.
  #
  # We go direct to the implementation because there is no safe, sane public
  # API to manage restoration of these to their default values.  This
  # should, once the value is proved, be moved to a standard API on the
  # indirector.
  #
  # To make things worse, a number of the tests stub parts of the
  # indirector.  These stubs have very specific expectations that what
  # little of the public API we could use is, well, likely to explode
  # randomly in some tests.  So, direct access.  --daniel 2011-08-30
  $saved_indirection_state = {}
  indirections = Puppet::Indirector::Indirection.send(:class_variable_get, :@@indirections)
  indirections.each do |indirector|
    $saved_indirection_state[indirector.name] = {
        :@terminus_class => indirector.instance_variable_get(:@terminus_class),
        :@cache_class    => indirector.instance_variable_get(:@cache_class)
    }
  end

  # The process environment is a shared, persistent resource.
  $old_env = ENV.to_hash

  # So is the load_path
  $old_load_path = $LOAD_PATH.dup

  initialize_settings_before_each()

  Puppet::Node::Environment.clear
  Puppet::Parser::Functions.reset

  Puppet.clear_deprecation_warnings
end

initialize() click to toggle source

Call this method once, as early as possible, such as before loading tests that call Puppet. @return nil

# File lib/puppet/test/test_helper.rb, line 33
def self.initialize()
  initialize_settings_before_each
end