How do I preserve the order of a hash in Perl? - perl

Question

I have a .sql file from which I am reading my input. Suppose the file contains the following input....
Message Fruits Fruit="Apple",Color="Red",Taste="Sweet";
Message Flowers Flower="Rose",Color="Red";
Now I have written a perl script to generate hash from this file..
use strict;
use Data::Dumper;
if(open(MYFILE,"file.sql")){
my #stack;
my %hash;
push #stack,\%hash;
my #file = <MYFILE>;
foreach my $row(#file){
if($row =~ /Message /){
my %my_hash;
my #words = split(" ",$row);
my #sep_words = split(",",$words[2]);
foreach my $x(#sep_words){
my($key,$value) = split("=",$x);
$my_hash{$key} = $value;
}
push #stack,$stack[$#stack]->{$words[1]} = {%my_hash};
pop #stack;
}
}
print Dumper(\%hash);
}
I am getting the following output..
$VAR1 = {
'Flowers' => {
'Flower' => '"Rose"',
'Color' => '"Red";'
},
'Fruits' => {
'Taste' => '"Sweet";',
'Fruit' => '"Apple"',
'Color' => '"Red"'
}
};
Now here the hash is not preserving the order in which the input is read.I want my hash to be in the same order as in input file.
I have found some libraries like Tie::IxHash but I want to avoid the use of any libraries.Can anybody help me out???

Answer 1

For a low key approach, you could always maintain the keys in an array, which does have an order.
foreach my $x(#sep_words){
my($key,$value) = split("=",$x);
$my_hash{$key} = $value;
push(#list_keys,$key);
}
And then to extract, iterate over the keys
foreach my $this_key (#list_keys) {
# do something with $my_hash{$this_key}
}
But that does have the issue of, you're relying on the array of keys and the hash staying in sync. You could also accidentally add the same key multiple times, if you're not careful.

Answer 2

Joel has it correct - you cannot reliably trust the order of a hash in Perl. If you need a certain order, you'll have to store your information in an array.

Answer 3

A hash is a set of key-value pairs with unique keys. A set is never ordered per se.
An array is a sequence of any number of scalars. An array is ordered per se, but uniqueness would have to be enforced externally.
Here is my take on your problem:
#!/usr/bin/perl
use strict; use warnings;
use Data::Dumper;
local $/ = ";\n";
my #messages;
while (<DATA>) {
chomp;
my ($msg, $to, $what) = split ' ', $_, 3; # limit number of fragments.
my %options;
while($what =~ /(\w+) = "((?:[^"]++|\\.)*)" (?:,|$)/xg) {
$options{$1} = $2;
}
push #messages, [$to => \%options];
}
print Dumper \#messages;
__DATA__
Message Fruits Fruit="Apple",Color="Red",Taste="Sweet";
Message Flowers Flower="Rose",Color="Red";
I put the messages into an array, because it has to be sorted. Also, I dont do weird gymnastics with a stack I don't need.
I don't split on all newlines, because you could have quoted value that contain newlines. For the same reason, I don't blindly split on , or = and use a sensible regex. It may be worth adding error detection, like die if not defined pos $what or pos($what) != length($what); at the end (requires /c flag on regex), to see if we actually processed everything or were thrown out of the loop prematurely.
This produces:
$VAR1 = [
[ 'Fruits',
{
'Taste' => 'Sweet',
'Fruit' => 'Apple',
'Color' => 'Red'
}
],
[ 'Flowers',
{
'Flower' => 'Rose',
'Color' => 'Red'
}
]
];
(with other indenting, but that's irrelevant).
One gotcha exists: The file has to be terminated by a newline, or the last semicolon isn't caught.