view release on metacpan or  search on metacpan

lib/App/Fasops/Command/stat.pm  view on Meta::CPAN

package App::Fasops::Command::stat;
use strict;
use warnings;
use autodie;

use Text::CSV_XS;

use App::Fasops -command;
use App::Fasops::Common;

sub abstract {
    return 'basic statistics on alignments';
}

sub opt_spec {
    return (
        [ "outfile|o=s", "output filename. [stdout] for screen" ],
        [ "length|l=i", "the threshold of alignment length", { default => 1 } ],
        [ 'outgroup',   'alignments have an outgroup', ],
        { show_defaults => 1, }
    );
}

sub usage_desc {
    return "fasops stat [options] <infile>";
}

sub description {
    my $desc;
    $desc .= ucfirst(abstract) . ".\n";
    $desc .= <<'MARKDOWN';

* <infiles> are paths to axt files, .axt.gz is supported
* infile == stdin means reading from STDIN

MARKDOWN

    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( @{$args} != 1 ) {
        my $message = "This command need one input file.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        next if lc $_ eq "stdin";
        if ( !Path::Tiny::path($_)->is_file ) {
            $self->usage_error("The input file [$_] doesn't exist.");
        }
    }

    if ( !exists $opt->{outfile} ) {
        $opt->{outfile} = Path::Tiny::path( $args->[0] )->absolute . ".csv";
    }
}

sub execute {
    my ( $self, $opt, $args ) = @_;

    #@type IO::Handle
    my $in_fh;
    if ( lc $args->[0] eq "stdin" ) {
        $in_fh = *STDIN{IO};
    }
    else {
        $in_fh = IO::Zlib->new( $args->[0], "rb" );
    }

    my $out_fh;
    if ( lc( $opt->{outfile} ) eq "stdout" ) {
        $out_fh = *STDOUT{IO};
    }
    else {
        open $out_fh, ">", $opt->{outfile};
    }

    # csv object
    my $csv = Text::CSV_XS->new( { eol => $/, } );

    # headers
    my @headers = qw{
        first legnth comparables identities differences gaps ns errors D indel
    };
    $csv->print( $out_fh, \@headers );

    my $content = '';    # content of one block
    while (1) {
        last if $in_fh->eof and $content eq '';
        my $line = '';
        if ( !$in_fh->eof ) {
            $line = $in_fh->getline;
        }
        next if substr( $line, 0, 1 ) eq "#";

        if ( ( $line eq '' or $line =~ /^\s+$/ ) and $content ne '' ) {
            my $info_of = App::Fasops::Common::parse_block( $content, 1 );