for (keys %csv) {
    $sql{"EMP_$_"} eq $cvs{$_} and next;
    print "$_ values are different:\n",
          "  SQL: ".$sql{"EMP_$_"}."\n",
          "  CSV: $sql{$_}\n\n";
}
[download]

You might find also find this node useful: CSV table diff utility

foreach my $sqlkey (keys %sql){
    foreach my $csvkey (keys %csv){
        next unless $csvkey =~ /$sqlkey\z/;  # match end of sqlkey
        # code to compare $csv{$csvkey} & $sql{$sqlkey} and do other t
+hings
    }
}
[download]

   my %sqltocsv = qw(EMP_NAME NAME EMP_JOB JOB ...);

   foreach $sqlkey (keys %sql){
       if($sql{$sqlkey} eq $csv{$sqltocsv{$sqlkey}}){
           # do things
       }
    }
[download]

Here's another method using map & grep functions. I believe this should capture all the data you need.

It only performs a single pass through the hash keys. There are no nested loops.

# hash_key_cmp

use strict;
use warnings;

# Test data

our %CVS = (
    b   => 1,
    c   => 0,
    d   => 1,
    e   => 1,
    f   => 1,
    g   => 0,
    h   => 1,
    i   => 1,
    j   => 1,
    k   => 1,
    l   => 1,
);

our %SQL = (
    x_a   => 1,
    x_b   => 1,
    x_d   => 1,
    x_e   => 1,
    x_f   => 0,
    x_g   => 1,
    x_h   => 0,
    x_i   => 1,
    x_j   => 1,
    x_l   => 1,
    x__k  => 1,
    x_m   => 0,
);


# Compare hashes

(my $prefix = (each %SQL)[0]) =~ s/^(.+_)?.*/$1/;
my $prefix_len = length $prefix;

my (@valid_in_both, @bad_in_both, @not_in_sql, @not_in_cvs);

map { exists $SQL{join '', $prefix, $_}
        ? $CVS{$_} eq $SQL{join '', $prefix, $_}
                ? push(@valid_in_both, $_)
                : push(@bad_in_both, $_)
        : push(@not_in_sql, $_)
    } keys %CVS;

@not_in_cvs = grep { ! exists($CVS{substr $_, $prefix_len}) } keys %SQ
+L;


# Output results

print 'Not in SQL:    ', "@not_in_sql", "\n";
print 'Not in CVS:    ', "@not_in_cvs", "\n";
print 'Valid in both: ', "@valid_in_both", "\n";
print 'Bad in both:   ', "@bad_in_both", "\n";
[download]

Here's the output:

[ ~/tmp ] $ perl hash_key_cmp
Not in SQL:    c k
Not in CVS:    x_a x__k x_m
Valid in both: b d e i j l
Bad in both:   f g h
[ ~/tmp ] $
[download]

Update: Changed second map to a grep. Line used to read:

map { exists($CVS{substr $_, $prefix_len}) || push(@not_in_cvs, $_) } 
+keys %SQL;
[download]

Brain stuck in push() mode, I think :-)

Why can't they have the same keys? What is preventing that? Since we don't know what constitutes the "beginning" of the field name, we can't really help you with a solution. post your field names and then maybe we could help.. however I would suspect there is a way to make them have the same keys.. which would solve the problem for you.

 
foreach my $v1 (values %sql) {
    foreach my $v2 (values %csv) {
        if ($v1 ne $v2) {
            # do something
        }
    }
}
[download]

EDIT: Thanks to cLive for pointing out that I'm totally on crack. Look at his post below for the good way to do this.

This is wrong. Very wrong. Testing each value against every other value in the other hash? So, for hashes with n keys, you're going to have at least (n-1)^2 cases where $v1 ne $v2 - and that's if they match. NAME may not match EMP_NAME, but it definitely won't match EMP_PHONE, EMP_AGE etc, so why test it?

Only n matches need to be made - assuming you can map the keys across the hashes correctly (my guess is below on this one).

I think you misunderstood the question.

cLive ;-)

You are totally right I am not sure what I was thinking when I wrote that. When I saw your post I smashed my head against the keyboard. I've had a long day...

correct.. but without making assumptions about the "beginning" of a key we can't truly know how to properly map key to key. I agree it's inefficient.. however it was the only way I saw (well agreed with actually) to do it without making assumptions.

---

Grygonos

That's true you could do values and sort on values and report any mismatches., and their field names. of course you may run into a cascading problem.. if field 2 doesn't match..then depending on what's causing the problem.. the rest of them may throw errors due to improper ordering.. or they may not. This is a decent way without fixing your data struct.. however that's still my reccomendation

---

Grygonos

my %scsv;
foreach (keys %$csv) {
    $scsv{substr($_, 4)} = %$csv->{$_};
}
&hash_match($sql, \%scsv, 'sql', 'csv');

sub hash_match {
    my ($p1, $p2, $n1, $n2) = @_;
    my ($a, $b);
    my @k1 = sort keys %$p1;
    my @k2 = sort keys %$p2;
    while ($a <= $#k1 && $b <= $#k2) {
        if (@k1[$a] eq @k2[$b]) {
            print "\$$n1\{'" . @k1[$a] . "'\} = " . $p1->{@k1[$a]} . "
+ ne " . 
                  "\$$n2\{'" . @k2[$b] . "'\} = " . $p2->{@k2[$b]} . "
+\n"
                  if $p1->{@k1[$a]} ne $p2->{@k2[$b]};
            $a++; $b++;
        }
        elsif (@k1[$a] lt @k2[$b]) {
            print "Unique \$$n1\{'" . @k1[$a] . "'\} = " . $p1->{@k1[$
+a++]} . "\n";
        }
        else {
            print "Unique \$$n2\{'" . @k2[$b] . "'\} = " . $p2->{@k2[$
+b++]} . "\n";
        }
    }
    print "Unique \$$n1\{'" . @k1[$a] . "'\} = " . $p1->{@k1[$a++]} . 
+"\n"
        while ($a <= $#k1);
    print "Unique \$$n2\{'" . @k2[$b] . "'\} = " . $p2->{@k2[$b++]} . 
+"\n"
        while ($b <= $#k2);
}
[download]