X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=misc%2Fsoap2sam.pl;h=b37135e811c6300d734488ae5a112f421bb920b5;hb=9f118264ea012adc21a46d7c03eaad4f9ce7d4d4;hp=0f99987940345887dd5b63979c55ee92c069d48a;hpb=38c2703b74ff9a045933206123c86c4e8f999d14;p=samtools.git

diff --git a/misc/soap2sam.pl b/misc/soap2sam.pl
index 0f99987..b37135e 100755
--- a/misc/soap2sam.pl
+++ b/misc/soap2sam.pl
@@ -1,7 +1,7 @@
 #!/usr/bin/perl -w
 
 # Contact: lh3
-# Version: 0.1.0
+# Version: 0.1.1
 
 use strict;
 use warnings;
@@ -36,14 +36,15 @@ sub mating {
 sub soap2sam {
   my %opts = ();
   getopts("p", \%opts);
-  die("Usage: soap2sam.pl [-p] <aln.soap>\n") if (@ARGV == 0);
+  die("Usage: soap2sam.pl [-p] <aln.soap>\n") if (@ARGV == 0 && -t STDIN);
   my $is_paired = defined($opts{p});
   # core loop
   my @s1 = ();
   my @s2 = ();
   my ($s_last, $s_curr) = (\@s1, \@s2);
   while (<>) {
-	&soap2sam_aux($_, $s_curr, $is_paired);
+	s/[\177-\377]|[\000-\010]|[\012-\040]//g;
+	next if (&soap2sam_aux($_, $s_curr, $is_paired) < 0);
 	if (@$s_last != 0 && $s_last->[0] eq $s_curr->[0]) {
 	  &mating($s_last, $s_curr);
 	  print join("\t", @$s_last), "\n";
@@ -60,8 +61,11 @@ sub soap2sam {
 sub soap2sam_aux {
   my ($line, $s, $is_paired) = @_;
   chomp($line);
-  my @t = split("\t", $line);
+  my @t = split(/\s+/, $line);
+  return -1 if (@t < 9 || $line =~ /^\s/ || !$t[0]);
   @$s = ();
+  # fix SOAP-2.1.x bugs
+  @t = @t[0..2,4..$#t] unless ($t[3] =~ /^\d+$/);
   # read name
   $s->[0] = $t[0];
   $s->[0] =~ s/\/[12]$//g;
@@ -70,7 +74,8 @@ sub soap2sam_aux {
   $s->[1] |= 1 | 1<<($t[4] eq 'a'? 6 : 7);
   $s->[1] |= 2 if ($is_paired);
   # read & quality
-  $s->[9] = $t[1]; $s->[10] = $t[2];
+  $s->[9] = $t[1];
+  $s->[10] = (length($t[2]) > length($t[1]))? substr($t[2], 0, length($t[1])) : $t[2];
   # cigar
   $s->[5] = length($s->[9]) . "M";
   # coor
@@ -100,4 +105,5 @@ sub soap2sam_aux {
 	$md = length($t[1]);
   }
   push(@$s, "MD:Z:$md");
+  return 0;
 }