Can a hash key have a scalar value and also contain a hash?

[bcarroll]

Not sure how to properly word this question, or if it has already been asked using better wording...

I am parsing a file that is similar to a Windows registry file and building a hash from the key structure in the file.
The issue I am running into is the way the file is structured (pretty much identical to the Windows registry). I would like the resulting hash to mimic the file content structure, but I don't think it is possible to assign a value to a key that has subkeys under it (hash of hashes), in this case isn't the hash of the subkeys the value?

The only way I could think of that works is to create an array under the key (if it has a scalar value and subkeys) and make the "default value" the first element then any subkeys would become hashes in the other elements.

Not sure if their is a CPAN module that can tackle this task or if it is just not possible.

Here is a sample of the contents of the registry-ish file I am parsing (file.reg):

HKEY_LOCAL_MACHINE\SOFTWARE\VENDOR\PRODUCT\CurrentVersion\Reports\Name
+spaceProviders=25137
ODBC:=                  reportsodbclog;       REG_SZ
SQLBulkInsertFlushInterval=     0x3c;                   REG_DWORD
SQLBulkInsertFlushRowCount=     0x3e8;                  REG_DWORD
SYSLOG:=                reportssyslog;        REG_SZ
TEXT:=                  reportstextlog;       REG_SZ

HKEY_LOCAL_MACHINE\SOFTWARE\VENDOR\PRODUCT\CurrentVersion\SessionServe
+r=12060
MaintenancePeriod=      0x3c;                   REG_DWORD
MaintenanceQueryTimeout=        0x3c;                   REG_DWORD

HKEY_LOCAL_MACHINE\SOFTWARE\VENDOR\PRODUCT\CurrentVersion\SessionServe
+r\NamespaceProviders=32225
LDAP:=                  ssprovider_ldap;      REG_SZ
ODBC:=                  ssprovider_db;        REG_SZ

#Currently I am creating a key named 'DEF_VAL' and assigning the paren
+t key's value to that.  Any suggestions?

use warnings;
use strict;
use Data::Dumper;
print Dumper loadreg('file.reg');

sub loadreg{
  my $registry;
  open(REGISTRY,'<',shift) or die("Error loading registry: $!\n");
  my @reg = <REGISTRY>;
  close REGISTRY;
  my $head = 0;
  my $section = "";
  for my $line ( @reg ) {
    #if ( $line =~ /^\s?(HKEY.*)\n/ ) {
    if ( $line =~ /^\s?HKEY.*?CurrentVersion\\(.*)\n/ ) {
      # found start of section
      $head = 1;
      my ( $key , $def_val ) = split /=/ , $1;
      $key =~ s/\\/\//g;
      $registry->{$key}{"DEF_VAL"} = $def_val;
      $section = $key;
    } elsif ( $head == 1 and $line =~ /^\n$/ ) {
      $head = 0;
      $section = "";
    } elsif ( $head == 1 and $line =~ /^[a-z]/i ) {
      # this is a key/value line for the current section
      # MaxReferralHops=        0xa;                    REG_DWORD
      # LDAP:=                  dsldap;               REG_SZ
      $line =~ /^([^=:]+)[:=]+\s+([^;]+);\s+([^ ]+)\n$/;
      my $value = $2;
      $value = hex($value) if $value =~ /^0x/;
      $registry->{$section}{$1}{"VALUE"} = $value if $1;
      $registry->{$section}{$1}{"TYPE"} = $3 if $1;
    }
  }
  return($registry);
}

Here is the output from the above example:

$VAR1 = {
          'SessionServer/NamespaceProviders' => {
                                                  'ODBC' => {
                                                              'VALUE' 
+=> 'ssprovider_db',
                                                              'TYPE' =
+> 'REG_SZ'
                                                            },
                                                  'LDAP' => {
                                                              'VALUE' 
+=> 'ssprovider_ldap',
                                                              'TYPE' =
+> 'REG_SZ'
                                                            },
                                                  'DEF_VAL' => '32225'
                                                },
          'SessionServer' => {
                               'DEF_VAL' => '12060'
                             },
          'Reports/NamespaceProviders' => {
                                            'TEXT' => {
                                                        'VALUE' => 're
+portstextlog',
                                                        'TYPE' => 'REG
+_SZ'
                                                      },
                                            'ODBC' => {
                                                        'VALUE' => 're
+portsodbclog',
                                                        'TYPE' => 'REG
+_SZ'
                                                      },
                                            'SYSLOG' => {
                                                          'VALUE' => '
+reportssyslog',
                                                          'TYPE' => 'R
+EG_SZ'
                                                        },
                                            'DEF_VAL' => '25137'
                                          }
        };

[SuicideJunkie]

Note: Regkey and Regvalue used to maintain a distinction from the hash definition of the terms key and value.

Perhaps you want an extra step, and to have a hash of regkeys=>{...} and regvalues=>{...}. You can't have two regkeys with the same name or two regvalues with the same name, but you can have a regkey and a regvalue with the same name.

{
  'Regkeys' => {
            HKEY_CURRENT_USER => {
                   'Regkeys'=>{...},
                   'Regvalues'=>{'(default)'=>{type=>REG_SZ, value=>un
+def},
                                ...},
                   },
            HKEY_LOCAL_MACHINE => {...},
            
          },
  'Regvalues' => {'(default)'=>{type=>REG_SZ,value=>undef},
                  ...},
}

[bcarroll]

Thanks SuicideJunkie!
That is one approach I hadn't thought of.

[AppleFritter]

"but I don't think it is possible to assign a value to a key that has subkeys under it (hash of hashes), in this case isn't the hash of the subkeys the value?"

Indeed. Each key/value pair in a hash maps precisely one key to precisely one value; so it can't be both a scalar and a hash reference at the same time, though you're of course right that you could use an array there as an extra layer of indirection.

Searching CPAN immediately reveals Win32::Registry::File, and the much newer Parse::Win32Registry. Have you looked at those?

[bcarroll]

I did look at some of the Win32 registry modules, but I can't recall why I ended up not using them.

I am thinking it was because, although the file structure is similar to a Windows registry file, it is different enough to make it incompatible with the available modules, and more complicated than just writing my own parsing subroutine.

Or I may have just been overwhelmed by the module's source code and got too intimidated to try... I honestly can't remember.

[2teez]

Hi bcarroll,

I must confess I didn't get what you wanted done properly, but am sure what you are trying to do is not Impossible if properly expressed with a likely show of how the expected output is suppose to look like.
That been said, while you want to show that, you can also look at this documentation perldsc.

[bcarroll]

Here is an example of the result I was dreaming of (given the example data):

my $registry = loadreg('file.reg');
print $registry->{'SessionServer/NamespaceProviders'}; # would print 3
+2225
print $registry->{'SessionServer/NamespaceProviders'}{'ODBC'}; # would
+ print ssprovider_db
print $registry->{'SessionServer/NamespaceProviders'}{'ODBC'}{'TYPE'};
+ # would print REG_SZ