Testing Wrapped LDAP Classes

[yulivee07]

Hello fellow Perl-Monks,

I am expanding my knowledge in testing right now. I want to test some functions of mine which interact with an LDAP-Server.

This might sound like a stupid question, but what is the best way to deal with Wrapper-Classes?

Context:
I write a module, which gets some Users from LDAP.
To communicate with the LDAP, I use a provided module from my workplace which wraps some system-specific parameters in the LDAP-methods (Utils::Ldap::CompanyLdap) and provides some additional methods.
This module again is a subclass of another LDAP-Wrapper Module (Utils::Ldap::Base) which finally uses Net::LDAP.

Now, in my program I use the first Ldap-Module e.g.

package MyAwesomeProgram;
use Utils::Ldap::CompanyLdap;

sub read_users {
   # code that reads users from Ldap
   # searchGetEntries/get_value is provided by Utils::Ldap::CompanyLda
+p
   foreach my $entry ( $ldap->searchGetEntries() ) {
      my $uid = $entry->get_value( 'uid' );
      #... do something more
   }

}

1;

I want to test the read_users function of my program, which gets its users via Utils::Ldap::CompanyLdap. How do I mock this? As you see, I do not directly interact with a Net::LDAP-Object

I have already understood that there is Test::Net::LDAP::Mock/Test::Net::LDAP::Utils qw(ldap_mockify) which can mock an Ldap Connection. What I do not understand yet: How can I use this in conjunction with my Wrapper Module? My module uses the code and the functions of our Ldap-Wrappers. Would the correct way be to write mocking code for the wrappers (e.g. Utils::Ldap::CompanyLdap::Mock), or is there a way of overwriting the Net::LDAP Object which is used by the Base-Wrapper Class Utils::Ldap::Base? I am not shure how to best approach this.

Second Question:
In case I get this working, is there a way of taking a Net::LDAP object and feeding it into my Mock-Object? I'd like to copy the datastructe of our LDAP to the Mock-Object.

Kind regards,

yulivee

[stevieb]

There are a few mocking distributions on the CPAN, but I'll show an example using my Mock::Sub. You can mock out subs, then tell it to do something (side_effect()), or return something (return_value()). Instead of using a method to set them, you can also specify them in the constructor if you choose (then remove/modify them with the methods later):

use warnings;
use strict;

use lib '.';

use Mock::Sub;
use Test::More;
use Utils::Ldap::CompanyLdap;

my $m = Mock::Sub->new;
my $ldap = Utils::Ldap::CompanyLdap->new;

my $mocked_sub = $m->mock(
    'Utils::Ldap::CompanyLdap::searchGetEntries'
);
$mocked_sub->return_value(qw(steve mike dave));

read_users();

is $mocked_sub->called, 1, "searchGetEntries() called ok";

sub read_users { 
    for my $entry ($ldap->searchGetEntries()){
        print "$entry\n";
    }
}

done_testing();

Output:

steve
mike
dave
ok 1 - searchGetEntries() called ok
1..1

[yulivee07]

Update:
The Ldap Object looks like this before the searchUser (some details obfuscated):


$VAR1 = bless( {
                 'OPT' => {
                            'base' => somebase,
                            'idprefix' => 'cn',
                            'userid' => 'admin',
                            'charset' => 'ISO-8859-15',
                            'bindretry' => '3',
                            'ldaps' => 1,
                            'timeout' => '60',
                            'server' => someip,
                            'waitretry' => '5',
                           }
               }, 'Utils::Ldap::CompanyLdap' );


And it looks like that after the searchUser:


bless( {
   's_result' => bless( {
      'parent' => bless( {
         'net_ldap_version' => '3',
         'net_ldap_scheme' => 'ldaps',
         'net_ldap_debug' => 0,
         'net_ldap_onerror' => sub { "DUMMY" },
         'net_ldap_host' => someip,
         'net_ldap_uri' => someip,
         'net_ldap_resp' => {},
         'net_ldap_async' => 0,
         'net_ldap_port' => 636,
         'net_ldap_refcnt' => 0
         }, 'Net::LDAPS' ),
            'entries' => [
               bless( {                                               
+                                                                     
+                                
                  'changes' => [],
                  'changetype' => 'modify',
                  'asn' => {
                      'objectName' => 'uid=bla,ou=People,ou=foo,dc=bar
+,dc=baz,dc=com',
                      'attributes' => [
                         {'type' => 'uid',   'vals' => [ someuid ] },
                         {'type' => 'cnum',  'vals' => [ 'L07345897' ]
+ },
                         {'type' => 'freeze','vals' => [ 'false' ] },
                         {'type' => 'dn',    'vals' => ['uid=bla,c=us,
+ou=otherldap,o=baz.com' ] },
                         {'type' => 'email', 'vals' => [ [e-mail]'] },
                                       ]
                           }
                        }, 'Net::LDAP::Entry' ),
                          ],
           'errorMessage' => '',
           'ctrl_hash' => undef,
           'resultCode' => 0,
           'callback' => undef,
           'matchedDN' => '',
           'mesgid' => 2,
           'controls' => undef,
           'raw' => undef
           }, 'Net::LDAP::Search' ),
     'BIND' => 0,
     'errors' => '',
     'error' => '',
      'OPT' => {
               'base' => somebase,
               'idprefix' => 'cn',
               'userid' => 'admin',
               'charset' => 'ISO-8859-15',
               'bindretry' => 3,
               'ldaps' => 1,
               'timeout' => '60',
               'server' => someip,
               'waitretry' => '5',
               },
   }, 'Utils::Ldap::CompanyLdap' );

The 'entries' contain all the userids, I have just added one to demonstrate. Normally there would be more.

[yulivee07]

Hi stevieb, thanks for your answer! This looks really good and I tried to integrate this into my code. Yet I still have problems...

I have a subroutine in my program called like this:


sub search_ldap {                                                     
+                                                                     
+                                                                 
    my ( $self, $ldap ) = @_; 
 
    # get uid, email, ecufreeze, cnum, itimaccess from LDAP
    report LOG_INFO, "Reading from Ldap";
    unless (
             $ldap->searchUser( filter     => '(uid=*)',
                                attributes => "freeze uid dn cnum emai
+l" ) ) 
    {   
        report LOG_ERROR, "Can not list user from LDAP: " . $ldap->get
+Error();
        exit 0;
    }   
 
    return $ldap;
 
}

In this case $ldap is a Utils::Ldap::CompanyLdap object. when calling $ldap->searchUser this object writes the users into itself. So with the searchUser call, the object itself is altered.

What I tried:

my $m = Mock::Sub->new;                 
my $ldap = Utils::Ldap::CompanyLdap->new;                             
+                                                                     
+                                                                     
+       
                                     
my $mocked_sub = $m->mock(           
    'Utils::Ldap::CompanyLdap::searchUser'
);     

# This is actually a Utils::Ldap::CompanyLdap-Object I copied via Data
+::Dumper
my $return_value = bless ( ... ); 
                                     
$mocked_sub->return_value($return_value);
    
    # cache is an instance of my own object                           
+  
    ok( $cache->search_ldap($my_ldap) );
                                     
    ok( $cache->read_all_userids($my_ldap) );
                                     
is $mocked_sub->called, 1, "searchUser() called ok";

I was hoping that be putting the object the way I want it to into my return value, the solution would work. The thing is, I noticed searchUser doesn't really return anything, it only appends to the object-instance. So Mock::Sub returns the right thing, but the code continues to work with the old ldap-object. Any Ideas how to solve this?

Greetings and thanks for your insights so far,
yulivee

[stevieb]

This is what the side_effect() functionality does... allows you to do stuff (eg: modify an object) when there's no need for a return. Here's an example:

use warnings;
use strict;

package Thing; {
    sub new {
        return bless {}, shift;
    }
    sub modify {
        my ($self) = @_;
        $self->{modified} = 'modified by original sub';
    }
}

package main;

use Data::Dumper;
use Mock::Sub;
use Test::More;

my $m = Mock::Sub->new;
my $thing = Thing->new;

my $modify_sub = $m->mock('Thing::modify');

$modify_sub->side_effect(
    sub {
        my $obj = shift;
        $obj->{modified} = 'modified by mocked sub';
    }
);

print "before mocked sub called...\n\n";

print Dumper $thing;

$thing->modify;

print "\n\nafter  mocked sub called...\n\n";
print Dumper $thing;

print "\n\n";

is defined $thing->{modified}, 1, "obj was modified ok";
like $thing->{modified}, qr/mocked sub/, "obj was changed by mock";
is $modify_sub->called, 1, "mocked sub called ok";

done_testing();

Output:

before mocked sub called...

$VAR1 = bless( {}, 'Thing' );


after  mocked sub called...

$VAR1 = bless( {
                 'modified' => 'modified by mocked sub'
               }, 'Thing' );


ok 1 - obj was modified ok
ok 2 - obj was changed by mock
ok 3 - mocked sub called ok
1..3

So, there's no return anymore. The side_effect() code reference will get all parameters passed in as they were sent in to the real sub call (in this case, $self, as it's the only param on the method call. We then have the side effect add a new hash key to itself and assign it a value. After side effect is complete, the main object is updated just like the original function would have done, without having to call the real function.