File Coverage

blib/lib/Proch/Seqfu.pm

Criterion	Covered	Total	%
statement	40	69	57.9
branch	6	40	15.0
condition			n/a
subroutine	11	15	73.3
pod	8	8	100.0
total	65	132	49.2

package Proch::Seqfu;
#ABSTRACT: SeqFu utilities

use strict;
use warnings;
use 5.014;
use Carp qw(confess);
use Data::Dumper;
use Term::ANSIColor qw(:constants);
use base 'Exporter';

# Version and configuration
our $VERSION     = '1.7.0';
our $fu_linesize = 0;
our $fu_verbose  = 0;




# Explicitly declare exports
our @EXPORT = qw(
    rc
    fu_printfasta
    fu_printfastq
    verbose
    has_seqfu
    seqfu_version
    is_seq
);

our @EXPORT_OK = qw($fu_linesize $fu_verbose);

# Function to check SeqFu version
sub seqfu_version {
    my $cmd = '';

    eval {
        my $path = $ENV{PATH};
        $path =~ /^(.+)$/;  # Untaint PATH by capturing
        local $ENV{PATH} = $1;
        $cmd = qx(seqfu version 2>/dev/null);
        chomp($cmd);
    };

    return -2 if $@;
    return $cmd if $cmd =~ /^(\d+)\.(\d+)(?:\.(\d+))?$/;
    return "-$cmd";
}

# Function to check if SeqFu is available
sub has_seqfu {
    my $ver = seqfu_version();
    return 0 if $ver =~ /^-/;        # Not installed/error
    return 1 if length($ver) > 0;     # Valid version found
    return undef;                     # Unknown state
}

# Validate sequence string
sub is_seq {
    my $string = $_[0];
    return 0 unless defined $string;
    return $string !~ /[^ACGTRYSWKMBDHVNU]/i;
}

# Get reverse complement
sub rc {
    my $sequence = reverse($_[0]);
    return unless is_seq($sequence);

    if ($sequence =~ /U/i) {
        $sequence =~ tr/ACGURYSWKMBDHVacguryswkmbdhv/UGCAYRSWMKVHDBugcayrswmkvhdb/;
    } else {
        $sequence =~ tr/ACGTRYSWKMBDHVacgtryswkmbdhv/TGCAYRSWMKVHDBtgcayrswmkvhdb/;
    }
    return $sequence;
}

# Print verbose messages
sub verbose {
    say STDERR " - ", $_[0] if $fu_verbose;
}

# Print FASTA format
sub fu_printfasta {
    my ($name, $comment, $seq) = @_;
    confess "Error: Name parameter required" unless defined $name;
    confess "Error: Sequence parameter required" unless defined $seq;
    confess "Error: Invalid sequence characters detected" unless is_seq($seq);

    my $print_comment = defined $comment ? ' ' . $comment : '';
    say '>', $name, $print_comment;
    print split_string($seq);
}

# Print FASTQ format
sub fu_printfastq {
    my ($name, $comment, $seq, $qual) = @_;
    confess "Error: Name parameter required" unless defined $name;
    confess "Error: Sequence parameter required" unless defined $seq;
    confess "Error: Quality string required" unless defined $qual;
    confess "Error: Invalid sequence characters detected" unless is_seq($seq);
    confess "Error: Sequence and quality length mismatch"
        unless length($seq) == length($qual);

    my $print_comment = defined $comment ? ' ' . $comment : '';
    say '@', $name, $print_comment;
    print split_string($seq), "+\n", split_string($qual);
}

# Split string into lines
sub split_string {
    my $input_string = $_[0];
    return unless defined $input_string;

    my $formatted = '';
    my $line_width = $fu_linesize;

    return $input_string . "\n" unless $line_width;

    for (my $i = 0; $i < length($input_string); $i += $line_width) {
        my $frag = substr($input_string, $i, $line_width);
        $formatted .= $frag . "\n";
    }
    return $formatted;
}

1;

__END__