Sort the column values and search the value - perl

Question

Input to my script is this file which contains data as below.
A food 75
B car 136
A car 69
A house 179
B food 75
C car 136
C food 85
For each distinct value of the second column, I want to print any line where the number in the third column is different.
Example output
C food 85
A car 69
Here is my Perl code.
#! /usr/local/bin/perl
use strict;
use warning;
my %data = ();
open FILE, '<', 'data.txt' or die $!;
while ( <FILE> ) {
chomp;
$data{$1} = $2 while /\s*(\S+),(\S+)/g;
}
close FILE;
print $_, '-', $data{$_}, $/ for keys %data;
I am able to print the hash keys and values, but not able to get the desired output.
Any pointers on how to do that using Perl?

Answer 1

As far as I can tell from your question, you want a list of all the lines where there is an "odd one out" with the same item type and a different number in the third column from all the rest
I think this is what you need
It reads all the data into hash %data, so that $data{$type}{$n} is a (reference to an) array of all the data lines that use that object type and number
Then the hash is scanned again, looking for and printing all instances that have only a single line with the given type/number and where there are other values for the same object type (otherwise it would be the only entry and not an "odd one out")
use strict;
use warnings 'all';
use autodie;
my %data;
open my $fh, '<', 'data.txt';
while ( <$fh> ) {
my ( $label, $type, $n) = split;
push #{ $data{$type}{$n} }, $_;
}
for my $type ( keys %data ) {
my $items = $data{$type};
next unless keys %$items > 1;
for my $n ( keys %$items ) {
print $items->{$n}[0] if #{ $items->{$n} } == 1;
}
}
output
C food 85
A car 69
Note that this may print multiple lines for a given object type if the input looks like, say
B car 22
A car 33
B car 136
C car 136
This has two "odd ones out" that appear only once for the given object type, so both B car 22 and A car 33 will be printed

Answer 2

Here are the pointers:
First, you need to remember lines somewhere before outputting them.
Second, you need to discard previously remembered line for the object according to the rules you set.
In your case, the rule is to discard when the number for the object differs from the previous remembered.
Both tasks can be accomplished with the hash.
For each line:
my ($letter, $object, $number)=split /\s+/, $line;
if (!defined($hash{$object}) || $hash{$object}[0]!=$number) {
$hash{$object}=[$number, $line];
}
Third, you need to output the hash:
for my $object(keys %hash) {
print $hash{$object}[1];
}
But there is the problem: a hash is an unordered structure, it won't return its keys in the order you put them into the hash.
So, the fourth: you need to add the ordering to your hash data, which can be accomplished like this:
$hash{$object}=[$number,$line,$.]; # $. is the row number over all the input files or STDIN, we use it for sorting
And in the output part you sort with the stored row number
(see sort for details about $a, $b variables):
for my $object(sort { $hash{$a}[2]<=>$hash{$b}[2] } keys %hash) {
print $hash{$object}[1];
}
Regarding the comments
I am certain that my code does not contain any errors.
If we look at the question before it was edited by some high rep users, it states:
[cite]
Now where if Numeric column(Third) has different value (Where in 2nd column matches) ...Then print only the mismatched number line. example..
A food 75
B car 136
A car 69
A house 179
B food 75
B car 136
C food 85
Example output (As number columns are not matching)
C food 85
[/cite]
I can only interpret that print only the mismatched number line as: to print the last line for the object where the number changed. That clearly matches the example the OP provided.
Even so, in my answer I addressed the possibility of misinterpretation, by stating that line omitting is done according to whatever rules the OP wants.
And below that I indicated what was the rule by that time in my opinion.
I think it well addressed the OP problem, because, after all, the OP wanted the pointers.
And now my answer is critiqued because it does not match the edited (long after and not by OP) requirements.
I disagree.
Regarding the whitespace: specifying /\s+/ for split is not an error here, despite of some comments trying to assert that.
While I agree that " " is common for split, I would disagree that there are a lot of cases where you must use " " instead of /\s+/.
/\s+/ is a regular expression which is the conventional argument for split, while " " is the shorthand, that actually masks the meaning.
With that I decided to use explicit split /\s+/, $line in my example instead of just split " ", $line or just split specifically to show the innerworkings of perl.
I think it is important to any one new to perl.
It is perfectly ok to use /\s+/, but be careful if you expect to have leading whitespace in your data, consult perldoc -f split and decide whether /\s+/ suits your needs or not.