File Coverage

blib/lib/CracTools/SAMReader.pm

Criterion	Covered	Total	%
statement	93	106	87.7
branch	23	46	50.0
condition	2	3	66.6
subroutine	16	17	94.1
pod	11	11	100.0
total	145	183	79.2

package CracTools::SAMReader;

{
  $CracTools::SAMReader::DIST = 'CracTools';
}
# ABSTRACT: An easy to use tool to read files in SAM format.
$CracTools::SAMReader::VERSION = '1.251';

use strict;
use warnings;
use Carp;
use CracTools::SAMReader::SAMline;


sub new {
  my $class = shift;
  my ($sam_file,$sam_type) = @_;

  my $self = bless{
      sam_file => $sam_file,
      sam_type => $sam_type,
  }, $class;

  $self->init();

  return $self;
}


sub iterator {
  my $self = shift;
  my $f_it = $self->iteratorFile("IGNORE_HEADERS");

  return sub {
    my ($line) = $f_it->();
    my $sam_line;
    if(defined $line) {
      $sam_line = CracTools::SAMReader::SAMline->new($line);
    }
    return $sam_line;
  };
}


sub iteratorFile {
  my $self = shift;
  my $option = shift;
  my $sam_file = $self->{sam_file};

  if($sam_file =~ /\.sam$/) {
    open(SAM,"< $sam_file") or die ("Cannot open $sam_file");
  } elsif($self->{sam_file} =~ /\.sam.gz$/) {
    open(SAM,"gunzip -c $sam_file |") or die ("Cannot open $sam_file");
  } elsif($self->{sam_file} =~ /\.bam$/) {
    open(SAM, "-|", "samtools view -h $sam_file" )or die "Cannot open $sam_file, check if samtools are installed.";
  } else {
    open(SAM,"< $sam_file") or die ("Cannot open $sam_file");
    warn "Unknown file format. We assume this is SAM (uncompressed).";
  }

  my $next_line;
  my $line_number = 0;

  if(defined $option && $option eq "IGNORE_HEADERS") {
    while(my $line = ) {
      if(!($line =~ /^@/)) {
        $next_line = $line;
        $line_number++;
        last;
      }
    }
  } else {
    $next_line = ;
  }

  return sub {
    my $sam_line = $next_line;
    $next_line = ;
    $line_number++;
    if($sam_line) {
      return $sam_line, $line_number;
    } else {
      close(SAM) or warn $! ? "Error closing samtools pipe: $!" : "Exit status $? from samtools";
      return ();
    }
  };
}



sub header {
  my $self = shift;
  return $self->{header};
}


sub refSeqLength {
  my $self = shift;
  my $ref_seq = shift;
  croak("Missing reference sequence name in arguement") unless defined $ref_seq;
  my $refseq_lengths = $self->allRefSeqLengths();
  return $refseq_lengths->{$ref_seq};
}


sub allRefSeqLengths {
  my $self = shift;
  my @header_lines = split('\n',$self->header);
  my %refseq_lengths;
  foreach (@header_lines) {
    if ($_ =~/\@SQ.*SN:/) {
      my ($name,$length) = $_ =~/\@SQ.*SN:(\S+)\s+LN:(\d+)+/;
      $refseq_lengths{$name} = $length;
    }
  }
  return \%refseq_lengths;
}


sub commandLine {
  my $self = shift;
  if(defined $self->header) {
    my @header_lines = split('\n',$self->header);
    my $command_line;
    foreach (@header_lines) {
      if ($_ =~/\@PG.*PN:crac/) {
        ($command_line) = $_ =~ /CL:([^\t]+)/;
      }
    }
    return $command_line;
  } else {
    return undef;
  }
}


sub getCracArgumentValue {
  my $self = shift;
  my $argument = shift;
  my $command_line = $self->commandLine;
  if(defined $command_line) {
    my ($value) = $command_line =~ /--$argument\s+(\S+)/;
    return $value;
  } else {
    return undef;
  }
}


sub hasCracOption {
  my $self = shift;
  my $option = shift;
  croak("Missing argument") unless defined $option;
  if(defined $self->commandLine) {
    return $self->commandLine =~ /--$option/;
  } else {
    return 0;
  }
}


sub getCracVersionNumber {
  my $self = shift;
  if(defined $self->header) {
    my @header_lines = split('\n',$self->header);
    my $version_number;
    foreach (@header_lines) {
      if ($_ =~/\@PG.*PN:crac/) {
        ($version_number) = $_ =~ /VN:([^\t]+)/;
        last;
      }
    }
    return $version_number;
  } else {
    return undef;
  }
}


sub init {
  my $self = shift;
  my $f_it = $self->iteratorFile;
  my $header;
  while(my ($line) = $f_it->()) {
    if($line =~ /^@/) {
      $header .= $line;
    } else {
      last;
    }
  }
  $self->{header} = $header;
}

1;

__END__