001   #!/usr/bin/perl -w
002   use strict;
003   my %USERS;
004   foreach my $file (<passwd.*>) {
005       open(PASSWD,$file);
006       my $strip=$file;
007       $strip =~ s/passwd\.//;
008       while(<PASSWD>) {
009           my($login,$gcos) = (split(':',$_))[0,4];
010   
011           if(exists $USERS{$login}) {
012               push(@{$USERS{$login}},$strip);
013           } else {
014               $USERS{$login} = [$gcos,$strip];
015           }
016       }
017       close(PASSWD);
018   }
019   
020   open(NEWFILE,">endstrx") || print "Can't open servup: $!\n";
021   foreach my $login (sort keys %USERS) {
022       print NEWFILE "$login:".shift(@{$USERS{$login}}).":";
023       print NEWFILE join(':',@{$USERS{$login}})."\n";
024   }
025   close(NEWFILE);

001 - 002 = Get the script ready to run turning warnings and strict on
003       = Get the hash %USERS ready for later
004       = loop threw all files matching the glob 'passwd.*'
005       = open the file found and give it the handle PASSWD
006 - 007 = copy the file name and remove the part before the first .
008       = loop over the lines of the file
009       = split the line on :, take the 1st and 4th element and putt
+ing
            them in $login, and $gcos respectivly
011       = check if an element already exists in the hash with that u
+ser name
012       =   if so then push this file name onto the list at that key
+.
014       =   otherwise start a list at this key with the gcos and the
+ filename
017       = close the file

020       = open a new file for output with teh name endstrx and the h
+andle NEWFILE
021       = loop over the keys of the USER hash, sorting them alphabet
+icaly
022       = print the users name, followed by the first element of the
+ list at that key,
023       = then join all the file names together and print them as we
+ll
025       = close the file
[download]

_{janitored by ybiC: s/pre/code/ as per Monastery convention}

Well, first of all, your title says "hash of hashes" but what you are working with here is a hash of arrays (actually, hash of anonymous array references).

Rather than walk you through your example code line by line (I assume you must understand some of it and are only really worried about the hash of arrays).

    my($login,$gcos) = (split(':',$_))[0,4];

    if(exists $USERS{$login}) {
        push(@{$USERS{$login}},$strip);
    } else {
        $USERS{$login} = [$gcos,$strip];
    }
[download]

The above is the core code that builds your structure. You obtain a login and gcos from a line in a passwd file. Next, you test %USERS to see if that login exists as a key in that hash. If it *doesn't* we assign an anonymous array reference to that key and populate it with the gcos and the current servername stripped from the filename. If the login does exists (we found it in a previous file), then we just push the current server ($strip) onto the array reference that already exists for that key:

    push( @{  $USERS{$login}  }, $strip );
[download]

Here, $USERS{$login} resolves to an array reference and we are de-referencing it: @{ arrayref } in order to push another value onto that array.

In the final section of code, you simply loop over the sorted keys (login ids) of the hash, and print out each login followed by its gcos and the list of servers (contained in the anonymous array) with colon separators. That loop could be simplified to:

foreach my $login (sort keys %USERS) {
    print NEWFILE join(':', $login, @{$USERS{$login}}), "\n";
}
[download]

Hope it helps.