Stefan Hornburg (Racke)

It is a best practice to use namespace::clean in your class or your role after use ... statements. Otherwise imported functions double as methods for your class, which can have unexpected side effects which are really hard to track down.

Using Moo

Types for Moo

We can recommend to use Type::Tiny instead of MooX::Types::Mooselike. The latter is not Moose-friendly, while Type::Tiny types get inflated to full Moose types if you happen to play with them in Moose. Also they are faster and we get better coercion in Moo for free.

One thing you may not have spotted for Type::Tiny which is useful... it overloads | and & for use in isa - very tidy.

use Moo;
use Types::Standard qw/ArrayRef HashRef InstanceOf/;

has fields => (
    is => 'ro',
    isa => ArrayRef [ InstanceOf ['Template::Flute::Form::Field'] ],
);

ArrayRef

has config_files => (
    is      => 'ro',
    isa     => ArrayRef,
    default => sub { [] },
);

Maybe

Sometimes it makes sense to allow undef as a value, e.g. if the builder isn't able to produce the promised type:

has config => (
    is      => 'ro',
    isa     => Maybe[HashRef],
    lazy    => 1,
    builder => '_build_config',
);

Speed ups

UTF-8

Ensure that me use UTF-8 encoding for the standard input/output streams:

use open ':std', ':encoding(utf-8)';

Using DateTime

Converting JSON dates

APIs often using JSON as output. A JSON date looks like that:

Formatting

DateTime->now->strftime('%Y-%m-%d');

Week

This function determines the first day of the week for a given year:

sub first_day_of_week
{
  my ($year, $week) = @_;

  # Week 1 is defined as the one containing January 4:
  DateTime
    ->new( year => $year, month => 1, day => 4 )
    ->add( weeks => ($week - 1) )
    ->truncate( to => 'week' );
}

Parsing

Text

Split text into an array of lines:

my @lines = split /\r?\n/, $content;
my %email_headers;

for my $line (@lines) {
  if ($line =~ /^(From|To):\s(.*)/) {
    $email_headers{lc($1)} = $2;
  }
}

List of arguments

The simplest way to parse a list of arguments separated by whitespace is the split function:

my $arglist = 'foo bar baz';

my @args = split (/\s+/, $arglist));

This doesn't work with a list of files when it includes filenames with whitespaces:

'foo bar'
'baz'

You can use the shellwords function from the Text::ParseWords module in this case:

use Text::ParseWords;
my $arglist = '"foo bar" baz';

my @args = shellwords($arglist);

Regular expresssions

Turn blank seperated text into a CSV

perl -pe 's/\p{Blank}+/;/g' dns2.txt > dns.csv

Wrap all urls in a text into a HTML link

$text =~s#(https?://\S+?)(\.*[\s\<,])#<a href="$1">$1</a>$2#igs;

Recipes

Print version of installed Perl modules

There is a very useful module called V for printing out version information about installed Perl modules:

perl -MV=Dancer2

Dancer2
    /home/racke/perl5/perlbrew/perls/perl-5.24.0/lib/site_perl/5.24.1/Dancer2.pm: 0.205000
    /home/racke/perl5/perlbrew/perls/perl-5.24.0/lib/site_perl/5.24.0/Dancer2.pm: 0.204001

Convert images with Image::Magick

Resize by percents

$ convert mypic.jpg -resize 25% mypicsmall.jpg

Create transparent picture

$ convert -size 1000x1000 xc:transparent any-square.png

CPAN modules - the good and the bad

File::Slurp is deprecated, use for example Path::Tiny instead.

Charts

Chart::Clicker

Writing tests

Test for warnings

Always use Test::Warnings to capture unexpected warnings in your tests:

use Test::Warnings;

Test whether result is one of multiple values

use Test::Deep;

my $status = $mws->GetServiceStatus;

cmp_deeply $status, any(qw/GREEN GREEN_I YELLOW RED/),
   "Test response of GetServiceStatus API method";

Skipping tests

Often tests are using an optional module or external resources. In that case we just skip these tests if we haven't all pieces in the right place.

For example, if an environment variable isn't set properly:

if ($ENV{SOLR_URL}) {
    $solr_url = $ENV{SOLR_URL};
}
else {
    plan skip_all => "Please set environment variable SOLR_URL.";
}

Provide useful information with `diag`

Even if you think your tests won't fail, they might. In this case it is very helpful to provide additional information to trace down the source of the problem:

# check output

my @matches = $out =~ /Blue ball/g;

ok (@matches == 2, 'Test replacement in both lists')
    || diag "Matches: ", scalar(@matches), "Output: $out";

A good alternative is to use is instead of ok, which automatically provides diagnostics:

# check output

my @matches = $out =~ /Blue ball/g;

is (scalar(@matches), 2, 'Test replacement in both lists');

Also other test functions like isa_ok print further information.

Test coverage

Devel::Cover makes test coverage easy in your distribution:

cpanm Devel::Cover
perl Makefile.PL
make
cover -test

You can find more information in the blog post from Neil Bower.

Installing CPAN modules

cpanm

You can install Perl modules from Git repositories, e.g.

cpanm git@github.com:interchange/Amazon-MWS.git@topic/amazon-pay-2020-01-28

The part after the @ is a branch, tag or hashsum.

Debian prerequisites

Here we show you which libraries you need to install for binary Perl modules (Debian and Ubuntu).

DBD::MySQL

This works for MySQL and MariaDB:

$ apt install libmysqlclient-dev

DBD::ODBC

$ apt install unixodbc-dev

DBD::Pg

$ apt install libpq-dev

File::LibMagic

$ apt install libmagic-dev

Imager

You need to install a number of development packages for the different image types before installing Imager, e.g. on Debian:

$ apt install libjpeg-dev libpng-dev libgif-dev libtiff-dev libfreetype6-dev

Net::LibIDN2

$ apt install libidn2-dev

Net::SSLeay, Crypt::OpenSSL::Random, Crypt::OpenSSL::X509, ...

$ apt install libssl-dev

IO::Socket::SSL

$ apt install zlib1g-dev

XML::Parser, XML::Twig, ...

$ apt install libexpat1-dev

XML::LibXML

$ apt install libxml2-dev

Tips for other modules

Dancer2

For more speed, install recommended modules as well:

cpanm --installdeps --with-recommends Dancer2

These are XS modules replacing pure Perl modules, e.g. HTTP::XSHeaders or Type::Tiny::XS.

Patching CPAN modules

If you want to make a patch for a CPAN module which doesn't provide a Git repository, Git::CPAN::Patch comes in very handy:

% git-cpan clone WebService::Xero
creating WebService-Xero
created tag 'v0.10' (7dbe0fdd5c54307ce6ea6d6943ddf529e1a7ab8c)
created tag 'v0.11' (082ce463c0dec679710d0eecfbbaf47262526bff)

Create a branch for your patch:

% git checkout -b topic/enable-put-method

After done with patching, you can submit the changes as follows:

add your changes and commit
push patch to CPAN

For example:

% git add -u
% git commit -m "Enable PUT method in Xero agent."
[topic/enable-put-method 76c60c1] Enable PUT method in Xero agent.
%

Writing CPAN modules

Resources

Please add resources for meta::cpan to your Makefile.PL, e.g.

META_MERGE => {
    resources => {
        repository  => 'https://github.com/interchange/interchange6-schema.git',
        bugtracker => 'https://github.com/interchange/interchange6-schema/issues',
        IRC => 'irc://irc.freenode.net/#interchange',
    },
},

Prerequisites and minimum versions

Test::Deep

0.114: if you use noneof

Type::Tiny

0.008: if you use InstanceOf

Travis

Before you commit and push your .travis.yml file, make sure that file is valid:

travis lint .travis.yml

Perl versions

The lastest Perl version supported by Travis is 5.30 as of September 2020.

Coverage reports

language: perl
perl:
  - "5.10"
  - "5.12"
  - "5.14"
  - "5.16"
  - "5.18"
  - "5.20"
  - "5.22"
  - "5.24"
  - "5.28"
  - "dev"     # installs latest developer release of perl
  - "blead"   # builds perl from git
matrix:
  include:
    - perl: 5.28
      env: COVERAGE=1   # enables coverage+coveralls reporting
  allow_failures:
    - perl: blead       # ignore failures for blead perl
sudo: false             # faster builds as long as you don't need sudo access
before_install:
  - eval $(curl https://travis-perl.github.io/init) --auto

Please see Travis helpers for complete reference.

Scripts

Add missing columns with default values to CSV file:

$ perl -i -pe 's/$/;0;/' data.csv

This one-liner removes carriage returns in a file:

$ perl -i -pe 's/\r//g' data.csv

Turn file with multiples lines into a comma separated list:

$ cat myfile.txt
FOO
BAR
BAZ

$ perl -pe 's/\r?\n/,/g' myfile.txt
FOO,BAR,BAZ