text processing - I have two arrays roles added and roles deleted, want to merge them and write in this order in perl -

i have scenario.
file has following fields:

username | roles | type |date | abc|admin |added | 01072015 abc|developer |deleted |01072015 abc|deploy |added |01072015 xyz |admin |deleted |01072015 xyz| deploy|deleted|01072015 cdf|deploy|added|01072015 

note here, date going same day, no change now, want printed as

username |roles_added |roles deleted |date abc |admin,deploy |developer |01072015  xyz ||admin,deploy |01072015 cdf |deploy||01072015 

i tried below approach given, didn't work out me. please guide me.

#!/usr/bin/perl  open(fil,"report.txt") or die("$!");  %k=();  while (my $line=<fil>) {     ($user,$roles,$type,$dt)=split(/\|/,$line);      $k{$user}{$roles}=1;     $k{$user}{$type}=1; }  @names=(sort keys(%k));  foreach $name (@names) {     foreach $value ( (keys(%{$k{$name}})) ){         print "$value ";     }     print "$name\n"; }  print " here \n"; while( ($k, $v) = each %$k ) {     print "key: $k, value: $v.\n"; } 

when dealing complex data structures, data::dumper friend. try adding use data::dumper code , print dumper \%k after you've finished building %k hash. you'll see this:

$var1 = {           'xyz' => {                      ' deploy' => 1,                      'deleted' => 1                    },           'username ' => {                            ' type ' => 1,                            ' roles ' => 1                          },           'xyz ' => {                       'deleted ' => 1,                       'admin ' => 1                     },           'cdf' => {                      'deploy' => 1,                      'added' => 1                    },           'abc' => {                      'deleted ' => 1,                      'deploy ' => 1,                      'developer ' => 1,                      'added ' => 1,                      'admin ' => 1                    }         }; 

see how keys in sub-hashes names after 2 types of things. half of them role names , half of them "added" or "deleted". it's going hard useful out of data structure, let's try different.

where had:

$k{$user}{$roles}=1; $k{$user}{$type}=1; 

try instead:

push @{$k{$user}{type}}, $role; 

now our data structure looks this:

$var1 = {           'xyz' => {                      'deleted' => [                                     ' deploy'                                   ]                    },           'xyz ' => {                       'deleted ' => [                                       'admin '                                     ]                     },           'cdf' => {                      'added' => [                                   'deploy'                                 ]                    },           'abc' => {                      'deleted ' => [                                      'developer '                                    ],                      'added ' => [                                    'admin ',                                    'deploy '                                  ]                    }         }; 

i think can see it's far easier information want out of data structure. it's basically:

foreach (@names) {    print join ',', @{$k{$_}{added}};    print join ',', @{$k{$_}{deleted}}; } 

you'll need change code little want.

oh, , please habit of adding use strict , use warnings of code. have shown why debugging output @ end of original code wasn't working.

update: threw complete solution.

#!/usr/bin/perl  use strict; use warnings; use 5.010;  %k;  <>; # skip header  while (<>) {   chomp;   ($user, $roles, $type, $dt) = split(/\s*\|\s*/);    push @{$k{$user}{$type}}, $roles;   $k{$user}{date} = $dt; }  'username |roles_added |roles deleted |date';  foreach $name (sort keys %k) {   "$name |",       join(',',@{$k{$name}{added} || []}), ' |',       join(',',@{$k{$name}{deleted} || []}), ' |',       $k{$name}{date}; } 

you need pass name of input file command line argument.