The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.
package Email::MIME::Kit::Renderer::TestRenderer;
# ABSTRACT:  extremely simple renderer for testing purposes only
$Email::MIME::Kit::Renderer::TestRenderer::VERSION = '3.000001';
use Moose;
with 'Email::MIME::Kit::Role::Renderer';

#pod =head1 WARNING
#pod
#pod Seriously, this is horrible code.  If you want, look at it.  It's swell for
#pod testing simple things, but if you use this for real mkits, you're going to be
#pod upset by something horrible soon.
#pod
#pod =head1 DESCRIPTION
#pod
#pod The test renderer is like a version of Template Toolkit 2 that has had a crayon
#pod shoved up its nose and into its brain.  It can only do a very few things, but
#pod it does them well enough to test simple kits.
#pod
#pod Given the following template:
#pod
#pod   This will say "I love pie": [% actor %] [% m_obj.verb() %] [% z_by("me") %]
#pod
#pod ...and the following set of variables:
#pod
#pod   {
#pod     actor => 'I',
#pod     m_obj => $object_whose_verb_method_returns_love,
#pod     z_by  => sub { 'me' },
#pod   }
#pod
#pod ..then it will be a true statement.
#pod
#pod In method calls, the parens are B<not> optional.  Anything between them (or
#pod between the parens in a coderef call) is evaluated like perl code.  For
#pod example, this will actually get the OS:
#pod
#pod   [% z_by($^O) %]
#pod
#pod =cut

sub render {
  my ($self, $content_ref, $stash) = @_;

  my $output = $$content_ref;
  for my $key (sort %$stash) {
    $output =~
      s<\[%\s+\Q$key\E(?:(?:\.(\w+))?\((.*?)\))?\s+%\]>
      [ defined $2
        ? ($1 ? $stash->{$key}->$1(eval $2) : $stash->{$key}->(eval $2))
        : $stash->{$key}
      ]ge;
  }

  return \$output;
}

no Moose;
1;

__END__

=pod

=encoding UTF-8

=head1 NAME

Email::MIME::Kit::Renderer::TestRenderer - extremely simple renderer for testing purposes only

=head1 VERSION

version 3.000001

=head1 DESCRIPTION

The test renderer is like a version of Template Toolkit 2 that has had a crayon
shoved up its nose and into its brain.  It can only do a very few things, but
it does them well enough to test simple kits.

Given the following template:

  This will say "I love pie": [% actor %] [% m_obj.verb() %] [% z_by("me") %]

...and the following set of variables:

  {
    actor => 'I',
    m_obj => $object_whose_verb_method_returns_love,
    z_by  => sub { 'me' },
  }

..then it will be a true statement.

In method calls, the parens are B<not> optional.  Anything between them (or
between the parens in a coderef call) is evaluated like perl code.  For
example, this will actually get the OS:

  [% z_by($^O) %]

=head1 WARNING

Seriously, this is horrible code.  If you want, look at it.  It's swell for
testing simple things, but if you use this for real mkits, you're going to be
upset by something horrible soon.

=head1 AUTHOR

Ricardo Signes <rjbs@cpan.org>

=head1 COPYRIGHT AND LICENSE

This software is copyright (c) 2014 by Ricardo Signes.

This is free software; you can redistribute it and/or modify it under
the same terms as the Perl 5 programming language system itself.

=cut