Use perl::critic in test programs
Test one file:
use Test::Perl::Critic; use Test::More tests => 1; critic_ok($file);
Or test all files in one or more directories:
use Test::Perl::Critic; all_critic_ok($dir_1, $dir_2, $dir_N );
Or test all files in a distribution:
use Test::Perl::Critic; all_critic_ok();
Recommended usage for \s-1CPAN\s0 distributions:
use strict; use warnings; use File::Spec; use Test::More; use English qw(-no_match_vars);
if ( not $ENV{TEST_AUTHOR} ) { my $msg = 'Author test. Set $ENV{TEST_AUTHOR} to a true value to run.'; plan( skip_all => $msg ); }
eval { require Test::Perl::Critic; };
if ( $EVAL_ERROR ) { my $msg = 'Test::Perl::Critic required to criticise code'; plan( skip_all => $msg ); }
my $rcfile = File::Spec->catfile( 't', 'perlcriticrc' ); Test::Perl::Critic->import( -profile => $rcfile ); all_critic_ok();
Test::Perl::Critic wraps the Perl::Critic engine in a convenient subroutine suitable for test programs written using the Test::More framework. This makes it easy to integrate coding-standards enforcement into the build process. For ultimate convenience (at the expense of some flexibility), see the criticism pragma.
If you have an large existing code base, you might prefer to use Test::Perl::Critic::Progressive.
If you'd like to try Perl::Critic without installing anything, there is a web-service available at <http://perlcritic.com>. The web-service does not yet support all the configuration features that are available in the native Perl::Critic \s-1API\s0, but it should give you a good idea of what it does. You can also invoke the perlcritic web-service from the command line by doing an HTTP-post, such as one of these:
$> POST http://perlcritic.com/perl/critic.pl < MyModule.pm $> lwp-request -m POST http://perlcritic.com/perl/critic.pl < MyModule.pm $> wget -q -O - --post-file=MyModule.pm http://perlcritic.com/perl/critic.pl
Please note that the perlcritic web-service is still alpha code. The \s-1URL\s0 and interface to the service are subject to change.
Okays the test if Perl::Critic does not find any violations in $FILE. If it does, the violations will be reported in the test diagnostics. The optional second argument is the name of test, which defaults to \*(L"Perl::Critic test for $FILE\*(R". If you use this form, you should emit your own Test::More plan first. Runs \*(C`critic_ok()\*(C' for all Perl files beneath the given list of @DIRECTORIES. If @DIRECTORIES is empty or not given, this function tries to find all Perl files in the blib/ directory. If the blib/ directory does not exist, then it tries the lib/ directory. Returns true if all files are okay, or false if any file fails. This subroutine emits its own Test::More plan, so you do not need to specify an expected number of tests yourself.
\s-1DEPRECATED:\s0 Use the \*(C`all_perl_files\*(C' subroutine that is exported by Perl::Critic::Utils instead. Returns a list of all the Perl files found beneath each \s-1DIRECTORY\s0, If @DIRECTORIES is an empty list, defaults to blib/. If blib/ does not exist, it tries lib/. Skips any files in \s-1CVS\s0 or Subversion directories. A Perl file is:
Any file that ends in .PL, .pl, .pm, or .t
Any file that has a first line with a shebang containing 'perl'
Perl::Critic is highly configurable. By default, Test::Perl::Critic invokes Perl::Critic with its default configuration. But if you have developed your code against a custom Perl::Critic configuration, you will want to configure Test::Perl::Critic to do the same.
Any arguments passed through the \*(C`use\*(C' pragma (or via \*(C`Test::Perl::Critic->import()\*(C' )will be passed into the Perl::Critic constructor. So if you have developed your code using a custom ~/.perlcriticrc file, you can direct Test::Perl::Critic to use your custom file too.
use Test::Perl::Critic (-profile => 't/perlcriticrc'); all_critic_ok();
Now place a copy of your own ~/.perlcriticrc file in the distribution as t/perlcriticrc. Then, \*(C`critic_ok()\*(C' will be run on all Perl files in this distribution using this same Perl::Critic configuration. See the Perl::Critic documentation for details on the .perlcriticrc file format.
Any argument that is supported by the Perl::Critic constructor can be passed through this interface. For example, you can also set the minimum severity level, or include & exclude specific policies like this:
use Test::Perl::Critic (-severity => 2, -exclude => ['RequireRcsKeywords']); all_critic_ok();
See the Perl::Critic documentation for complete details on its options and arguments.
By default, Test::Perl::Critic displays basic information about each Policy violation in the diagnostic output of the test. You can customize the format and content of this information by using the \*(C`-verbose\*(C' option. This behaves exactly like the \*(C`-verbose\*(C' switch on the perlcritic program. For example:
use Test::Perl::Critic (-verbose => 6);
#or...
use Test::Perl::Critic (-verbose => '%f: %m at %l');
If given a number, Test::Perl::Critic reports violations using one of the predefined formats described below. If given a string, it is interpreted to be an actual format specification. If the \*(C`-verbose\*(C' option is not specified, it defaults to 3.
Verbosity Format Specification ----------- ------------------------------------------------------- 1 "%f:%l:%c:%m\n", 2 "%f: (%l:%c) %m\n", 3 "%m at %f line %l\n", 4 "%m at line %l, column %c. %e. (Severity: %s)\n", 5 "%f: %m at line %l, column %c. %e. (Severity: %s)\n", 6 "%m at line %l, near '%r'. (Severity: %s)\n", 7 "%f: %m at line %l near '%r'. (Severity: %s)\n", 8 "[%p] %m at line %l, column %c. (Severity: %s)\n", 9 "[%p] %m at line %l, near '%r'. (Severity: %s)\n", 10 "%m at line %l, column %c.\n %p (Severity: %s)\n%d\n", 11 "%m at line %l, near '%r'.\n %p (Severity: %s)\n%d\n"
Formats are a combination of literal and escape characters similar to the way \*(C`sprintf\*(C' works. See String::Format for a full explanation of the formatting capabilities. Valid escape characters are:
Escape Meaning ------- ---------------------------------------------------------------- %c Column number where the violation occurred %d Full diagnostic discussion of the violation (DESCRIPTION in POD) %e Explanation of violation or page numbers in PBP %F Just the name of the logical file where the violation occurred. %f Path to the logical file where the violation occurred. %G Just the name of the physical file where the violation occurred. %g Path to the physical file where the violation occurred. %l Logical line number where the violation occurred %L Physical line number where the violation occurred %m Brief description of the violation %P Full name of the Policy module that created the violation %p Name of the Policy without the Perl::Critic::Policy:: prefix %r The string of source code that caused the violation %C The class of the PPI::Element that caused the violation %s The severity level of the violation
Despite the convenience of using a test script to enforce your coding standards, there are some inherent risks when distributing those tests to others. Since you don't know which version of Perl::Critic the end-user has and whether they have installed any additional Policy modules, you can't really be sure that your code will pass the Test::Perl::Critic tests on another machine.
For these reasons, we strongly advise you to make your perlcritic tests optional, or exclude them from the distribution entirely.
The recommended usage in the \*(L"\s-1SYNOPSIS\s0\*(R" section illustrates one way to make your perlcritic.t test optional. Another option is to put perlcritic.t and other author-only tests in a separate directory (xt/ seems to be common), and then use a custom build action when you want to run them. Also, you should not list Test::Perl::Critic as a requirement in your build script. These tests are only relevant to the author and should not be a prerequisite for end-use.
See <http://www.chrisdolan.net/talk/index.php/2005/11/14/private-regression-tests/> for an interesting discussion about Test::Perl::Critic and other types of author-only regression tests.
critic_ok() all_critic_ok()
If you want a small performance boost, you can tell \s-1PPI\s0 to cache results from previous parsing runs. Most of the processing time is in Perl::Critic, not \s-1PPI\s0, so the speedup is not huge (only about 20%). Nonetheless, if your distribution is large, it's worth the effort.
Add a block of code like the following to your test program, probably just before the call to \*(C`all_critic_ok()\*(C'. Be sure to adjust the path to the temp directory appropriately for your system.
use File::Spec; my $cache_path = File::Spec->catdir(File::Spec->tmpdir, "test-perl-critic-cache-$ENV{USER}"); if (!-d $cache_path) { mkdir $cache_path, oct 700; } require PPI::Cache; PPI::Cache->import(path => $cache_path);
We recommend that you do \s-1NOT\s0 use this technique for tests that will go out to end-users. They're probably going to only run the tests once, so they will not see the benefit of the caching but will still have files stored in their temp directory.
If you find any bugs, please submit them to <http://rt.cpan.org/NoAuth/Bugs.html?Dist=Perl-Critic>. Thanks.
Module::Starter::PBP
Perl::Critic
Test::More
Andy Lester, whose Test::Pod module provided most of the code and documentation for Test::Perl::Critic. Thanks, Andy.
Jeffrey Ryan Thalhammer <[email protected]>
Copyright (c) 2005-2009 Imaginative Software Systems. All rights reserved.
This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself. The full text of this license can be found in the \s-1LICENSE\s0 file included with this module.