Ordering return from fetchrow_hashref

[slayer]

Oh great monks,

I'm going round in circles on this one. Essentially I want to extract a row from a mysql database, using fetchrow_hashref, but keep the columns in the SAME ORDER as the sql statement. The returned hash seems to be in a completely random order.

I've tried:

1. Tie a hash, create a reference, then extract the data into that reference:

   my (%row_hash,$row_hash);
   tie %row_hash, "Tie::IxHash";
   my $row_hash = \%row_hash;
   while ($row_hash = $sth->fetchrow_hashref) {...
   

   ->This still jumbles the order of the returned hash.

2. Tie a hash and extract straight in to that hash: while (%row_hash = %{$sth->fetchrow_hashref}) {

   ->This returns a 'Can't use an undefined value as a HASH reference' error

3. Extract the data into a reference, then tie the hash.

   ->The data is already jumbled and it is too late.

4. Alter CPANs DBI sub fetchrow_hashref to tie the hash before returning it.

   ->Am incapable. DBI.c seems to order the rows fine...

Anymore suggestions would be greatfully received,

Cheers!

[gmax]

tye has shown you one way. However, let me question the reasons for wanting what you are asking

Hashes are data structures where you can access elements by name rather than by position, as you do with arrays. Therefore a sorted hash is a contradiction. (Although Perl Tie mechanism can change something)

Now, I see two reasons for wanting the result from a query in a hash:

• When you want to access your columns by name
• When you want to process (e.g. display) your column names with their values.

In the first case, you don't really need a sorted hash, since you are using the hash names hardcoded in your application, so that the application itself will set the order. For example

while (my $rec = $sth->fetchrow_hashref()) {
    print $rec->{xy}, $rec->{xz}, $rec->{yz};
}

If you need to process your columns according to a list of column names, here is a slight variation on tye's answer:

while (my $row = $sth->fetchrow_hashref()) {
    print map( { "$_ => $row->{$_}\t"} @{$sth->{NAME}}),  "\n";
}

#or 

while (my $row = $sth->fetchrow_hashref()) {
    for (@{$sth->{NAME}}) {
        print "$_ => $row->{$_}\t";
    }
    print "\n";
}

If you just want to use the column names together with their values, then you can use two different methods, according to your needs:

while (my $row = $sth->fetchrow_arrayref()) { # arrayref, not hash
    my $index = 0;
    print "$_ => ", $row->[$index++], "\t"  for (@{$sth->{NAME}});
    print "\n";
}

# or the "C like" way

while (my $row = $sth->fetchrow_arrayref()) { # arrayref, not hash
    for (my $index = 0; $index < $sth->{NUM_OF_FIELDS}; $index++) {
        print "$sth->{NAME}->[$index] => ", $row->[$index], "\t"
    }
    print "\n";
}

For more ideas, tips, and caveats on the same subject, see DBI Recipes.

Update
If, as you say, you need an array of hash references, then it would be much easier this idiom:

my $aref = $sth->fetchall_arrayref({});

Notice the empty hashref passed as argument.

[tye]

@row_hash{ @{sth->{NAME}} }

[eric256]

If order is what you want then you prob should do fetchrow_arrayref instead of hashref. Just my two cents, I would assume the only reason you care about order is because you want to display it out agian looking like the same table. The alternative would be to just have an array of the columns in the order you want to display them.

[rdfield]

Hashes are random. Use fetchrow_array to preserve order.

[slayer]

Yes, you were quite right, the reason I needed this order is that I will be displaying the data directly from the database to autogenerate a status report, with little user intervention and with the column order derived from the sql query.

Also, other people are ultimately relying on me to return an array of hash references, so using array ref and converting would be messier.

The solution you suggested worked perfectly, I've implemented the following routine and it works exactly as it should:

sub read_from_db {
  my %params = @_;  
  my $dbh = $params{-dbh} || "";
  my $query = $params{-query} || "";
  my @results_array;  

  if ($dbh =~ /DBI/) {
    my $sth = $dbh->prepare($query);
    my $result = $sth->execute();    
    while (my $row = $sth->fetchrow_hashref()) {
      tie my(%hash_to_return), "Tie::IxHash";
      map({$hash_to_return{$_}=$row->{$_}} @{$sth->{NAME}});
      my $ref = \%hash_to_return;   
      push(@results_array, $ref);     
    }
    $sth->finish;
  }
  return @results_array;
}

So thanks...!

[fiasudeen]

The other way to do is ... we can get all the column headers using $sth->{NAME} array. get the values using $sth->fetchrow_array() - This would give column values in an order. @{$sth->{NAME}} - This would give column names in an order.