File Coverage

blib/lib/NCBIx/Geo/Base.pm

Criterion	Covered	Total	%
statement	30	105	28.5
branch	0	28	0.0
condition			n/a
subroutine	10	19	52.6
pod	0	4	0.0
total	40	156	25.6

package NCBIx::Geo::Base;
use Class::Std;
use Class::Std::Utils;
use LWP::Simple;
use Data::Dumper;

use warnings;
use strict;
use Carp;

use version; our $VERSION = qv('1.0.0');

use constant GEO_FTP_URL  => 'ftp://ftp.ncbi.nih.gov/pub/geo/DATA/supplementary/samples/';
use constant GEO_ACC_URL  => 'http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?view=data&db=GeoDb_blob01&acc=';

{
        my %debug_of  :ATTR( :get   :set   :default<'0'>    :init_arg );

        sub START {
                my ($self, $ident, $arg_ref) = @_;

		if ( defined $arg_ref->{debug} ) { $self->set_debug( $arg_ref->{debug} ); }

                return;
        }

	sub debug { my ( $self ) = @_; return $self->get_debug(); }

	sub get_accn_type {
		my ( $self, $arg_ref ) = @_;

		my $accn     = defined $arg_ref->{accn} ? $arg_ref->{accn} : '';
		   $accn     =~ m/^(\w{3})/;
		return $1;
	}

	sub get_sample_data {
		my ( $self, $arg_ref ) = @_;
		my $sample_accn        = defined $arg_ref->{accn} ? $arg_ref->{accn} : '';
		my @file_exts          = defined $arg_ref->{exts} ? @{ $arg_ref->{exts} } : ();

		# Calculate the NCBI GSM directory
		my $gsm_dir  = $sample_accn;
		   $gsm_dir  =~ s/.{3}$/nnn/;
		my $gsm_url  = GEO_FTP_URL . "$gsm_dir/$sample_accn/";

		# Get the raw data
		foreach my $ext ( @file_exts ) {
			if ( $ext eq 'TXT' ) { $ext = 'txt'; }

			# Calculate file_name and file_path_name
			my $file      = $sample_accn . '.' . $ext . '.gz';
			my $path_name = $self->get_data_dir() . $file;

			# Download if it doesn't exist
			if (! -s $path_name ) {
				my $url    = $gsm_url . $file;
				$self->_debug( "DOWNLOAD: $url" );
				my $result = get( $url );
				$self->_debug( "SAVE FILE: $path_name" );
				$self->_set_file_text({ file => $path_name, text => $result });
			} else {
				$self->_debug( "FILE EXISTS: $path_name" );
			}
		}

		# Calculate call data file_name and file_path_name
		my $file      = $sample_accn . '.html';
		my $path_name = $self->get_data_dir() . $file;

		# Download call data if it doesn't exist
		if (! -s $path_name ) {
			my $acc_url = GEO_ACC_URL . $sample_accn;
			$self->_debug( "DOWNLOAD: $acc_url" );
			my $result  = get( $acc_url );
			$self->_debug( "SAVE FILE: $path_name" );
			$self->_set_file_text({ file => $path_name, text => $result });
		} else {
			$self->_debug( "FILE EXISTS: $path_name" );
		}
	}

	sub _set_file_text {
		my ( $self, $arg_ref ) = @_;
		my $file = defined $arg_ref->{file} ? $arg_ref->{file} : '';
		my $text = defined $arg_ref->{text} ? $arg_ref->{text} : '';

		if ( $file ) {
			open( OUTFILE, '>', $file );
			print OUTFILE $text;
			close( OUTFILE );
		}

                return;
        }

	sub _get_file_text {
		my ( $self, $arg_ref ) = @_;
		my $file = defined $arg_ref->{file} ? $arg_ref->{file} : '';
		$self->_debug( "INFILE: $file" );
		my $lines;

		if ( $file ) {
			open( INFILE, '<', $file );
			while ( my $line =  ) { $lines .= $line; }
			close( INFILE );
		}

                return $lines;
        }

	sub _get_url {
		my ( $self, $url ) = @_;
		my $response;
		eval { $response = get( $url ); };
		if( $@ ) {
		  $self->warn("Can't query website: $@");
		  return;
		}
		$self->debug( "resp is $response\n");
		return $response;
	}

	sub _debug {
		my ( $self, $message ) = @_;
		if ( $self->debug() ) { print( $message . "\n" ); }
		return $self;
	}

	sub __exception {
		my ( $data ) = @_;
		print "\n#######################################\n";
		print "# Exception                           #\n";
		print "#######################################\n";
		print Dumper( $data );
		print "#######################################\n\n";
	}

}

1; # Magic true value required at end of module
__END__