parsing metadata

[Anonymous Monk]

Hello monks,

Could some one help me with parsing this metadata. I have the metadata slurped in to an array .

my @results = qx(/software/bin/imeta ls -d $path);
my @created_date = grep ( /^(value: )(\d{4}-\d{2}-\d{2}T\w+)/,@results
+);

But I want to get more data from the @results. The metadata actually looks like this.

AVUs defined for dataObj 3a/73/2c/metadata.csv:
attribute: dcterms:created
value: 2014-08-13T00:00:10
units:
----
attribute: control
value: 0
units:
----
attribute: md5
value: 3a732cd0fddaa80fa65fcf28664eaf6d
units:
----

I want to get both the dcterms:created as well as md5. I guess having them in a hash with their attributes would help but I don't know how to do. Any help or suggestions please?

[toolic]

This works for the small snippet of your input:

use warnings;
use strict;

my %atts;
my $att;
while (<DATA>) {
    chomp;
    $att = $1 if /^attribute:\s+(.+)/;
    $atts{$att} = $1 if /^value:\s+(.+)/;
}

use Data::Dumper;
$Data::Dumper::Sortkeys=1;
print Dumper(\%atts);

__DATA__
AVUs defined for dataObj 3a/73/2c/metadata.csv:
attribute: dcterms:created
value: 2014-08-13T00:00:10
units:
----
attribute: control
value: 0
units:
----
attribute: md5
value: 3a732cd0fddaa80fa65fcf28664eaf6d
units:
----

Outputs:

$VAR1 = {
          'control' => '0',
          'dcterms:created' => '2014-08-13T00:00:10',
          'md5' => '3a732cd0fddaa80fa65fcf28664eaf6d'
        };

Your input probably has several records, in which case you can look at perldsc.

Of course, if this is some standard format, there's probably a better solution on CPAN.

[AppleFritter]

Alternatively, you could write the loop like this:

my @interesting = ();
while(<DATA>) {
    chomp;
    push @interesting, $1 if m/^(?:attribute|value): (.*)$/;
}
my %attributes = @interesting; # magic

It's a kind of magic. :) You could even turn this into a oneliner, using e.g. grep and map, and employing some more dirty tricks along the way:

my %attributes = grep { length } map { m/^(?:attribute|value): (.*)$/ 
+and $1 } <DATA>;

Anyhow, going back to the while loop version, adding support for parsing the attributes of several objects at the same time is also fairly straightforward. E.g.:

my %dataObjs = ();

my @interesting = ();
my $dataObj;
while(<DATA>) {
    chomp;

    if(m/^AVUs defined for dataObj (.*):$/) {
        defined $dataObj and $dataObjs{$dataObj} = { @interesting };
        $dataObj = $1;
    }

    push @interesting, $1 if m/^(?:attribute|value): (.*)$/;
}
$dataObjs{$dataObj} = { @interesting };

Though in this case I'd use your solution instead, since it leads to much simpler code:

my %dataObjs = ();

my @interesting = ();
my $dataObj;
my $att;
while(<DATA>) {
    chomp;
    $dataObj                    = $1 if m/^AVUs defined for dataObj (.
+*):$/;
    $att                        = $1 if m/^attribute:\s+(.+)/;
    $dataObjs{$dataObj}->{$att} = $1 if /^value:\s+(.+)/;
}

No magic, alas. :)

Side note - according to my favorite WWW search engine, this sort of data is generated by iRODS. CPAN doesn't have any related modules, so here's a good opportunity to contribute to the Perl ecosystem for anyone who works with that system.

[Anonymous Monk]

Hi Both , Thanks !

Yes they are iRods meta data. But they are printed on screen and I don't want to push them to a file and create more overhead. I was trying something along like this:

foreach my $re(@results){                            
next if $re =~/^AVUs defined/; 
warn $re; 
my $attribute= $1 if($re =~ /^attribute:\s+(.*)/);                    
+                         
my $value = $1 if($re =~ /^value: (.*)/);                             
+                          |  warn $attribute,$value;

And I get

attribute: md5

EVAL_ERROR: Use of uninitialized value $value in warn at Access.pl lin
+e 125.

A problem occurred at /nfs/users/nfs_a/aj6/CGP/Fluidigm/perl/scripts/L
+oadGenotypingResults.pl line 61.