From: Heng Li <lh3@live.co.uk>
Date: Fri, 27 Nov 2009 15:44:56 +0000 (+0000)
Subject:  * samtools-0.1.7-1 (r515)
X-Git-Url: https://git.donarmstrong.com/?a=commitdiff_plain;h=9215c0998a15d7ebc82baf78ded2443c55a1bc93;p=samtools.git

 * samtools-0.1.7-1 (r515)
 * report an error when .fai contains duplicated names, instead of segfault
---

diff --git a/bam_import.c b/bam_import.c
index 9d463d1..3a4a5cd 100644
--- a/bam_import.c
+++ b/bam_import.c
@@ -116,7 +116,7 @@ static bam_header_t *hash2header(const kh_ref_t *hash)
 bam_header_t *sam_header_read2(const char *fn)
 {
 	bam_header_t *header;
-	int c, dret, ret;
+	int c, dret, ret, error = 0;
 	gzFile fp;
 	kstream_t *ks;
 	kstring_t *str;
@@ -135,6 +135,10 @@ bam_header_t *sam_header_read2(const char *fn)
 		ks_getuntil(ks, 0, str, &dret);
 		len = atoi(str->s);
 		k = kh_put(ref, hash, s, &ret);
+		if (ret == 0) {
+			fprintf(stderr, "[sam_header_read2] duplicated sequence name: %s\n", s);
+			error = 1;
+		}
 		kh_value(hash, k) = (uint64_t)len<<32 | i;
 		if (dret != '\n')
 			while ((c = ks_getc(ks)) != '\n' && c != -1);
@@ -143,6 +147,7 @@ bam_header_t *sam_header_read2(const char *fn)
 	gzclose(fp);
 	free(str->s); free(str);
 	fprintf(stderr, "[sam_header_read2] %d sequences loaded.\n", kh_size(hash));
+	if (error) return 0;
 	header = hash2header(hash);
 	kh_destroy(ref, hash);
 	return header;
diff --git a/bamtk.c b/bamtk.c
index 48ac76b..2c7e4f0 100644
--- a/bamtk.c
+++ b/bamtk.c
@@ -9,7 +9,7 @@
 #endif
 
 #ifndef PACKAGE_VERSION
-#define PACKAGE_VERSION "0.1.7 (r510)"
+#define PACKAGE_VERSION "0.1.7-1 (r515)"
 #endif
 
 int bam_taf2baf(int argc, char *argv[]);
diff --git a/misc/wgsim_eval.pl b/misc/wgsim_eval.pl
index 01038f1..d3d9c61 100755
--- a/misc/wgsim_eval.pl
+++ b/misc/wgsim_eval.pl
@@ -12,9 +12,9 @@ exit;
 
 sub wgsim_eval {
   my %opts = (g=>5);
-  getopts('pcg:', \%opts);
-  die("Usage: wgsim_eval.pl [-pc] [-g $opts{g}] <in.sam>\n") if (@ARGV == 0 && -t STDIN);
-  my (@c0, @c1);
+  getopts('pcag:', \%opts);
+  die("Usage: wgsim_eval.pl [-pca] [-g $opts{g}] <in.sam>\n") if (@ARGV == 0 && -t STDIN);
+  my (@c0, @c1, %fnfp);
   my ($max_q, $flag) = (0, 0);
   my $gap = $opts{g};
   $flag |= 1 if (defined $opts{p});
@@ -66,14 +66,26 @@ sub wgsim_eval {
 	}
 	++$c0[$q];
 	++$c1[$q] unless ($is_correct);
+	@{$fnfp{$t[4]}} = (0, 0) unless (defined $fnfp{$t[4]});
+	++$fnfp{$t[4]}[0];
+	++$fnfp{$t[4]}[1] unless ($is_correct);
 	print STDERR $line if (($flag&1) && !$is_correct && $q > 0);
   }
   # print
   my ($cc0, $cc1) = (0, 0);
-  for (my $i = $max_q; $i >= 0; --$i) {
-	$c0[$i] = 0 unless (defined $c0[$i]);
-	$c1[$i] = 0 unless (defined $c1[$i]);
-	$cc0 += $c0[$i]; $cc1 += $c1[$i];
-	printf("%.2dx %12d / %-12d  %12d  %.3e\n", $i, $c1[$i], $c0[$i], $cc0, $cc1/$cc0);
+  if (!defined($opts{a})) {
+	for (my $i = $max_q; $i >= 0; --$i) {
+	  $c0[$i] = 0 unless (defined $c0[$i]);
+	  $c1[$i] = 0 unless (defined $c1[$i]);
+	  $cc0 += $c0[$i]; $cc1 += $c1[$i];
+	  printf("%.2dx %12d / %-12d  %12d  %.3e\n", $i, $c1[$i], $c0[$i], $cc0, $cc1/$cc0);
+	}
+  } else {
+	for (reverse(sort {$a<=>$b} (keys %fnfp))) {
+	  next if ($_ == 0);
+	  $cc0 += $fnfp{$_}[0];
+	  $cc1 += $fnfp{$_}[1];
+	  print join("\t", $_, $cc0, $cc1), "\n";
+	}
   }
 }
diff --git a/sam.c b/sam.c
index ad4325b..ecdee02 100644
--- a/sam.c
+++ b/sam.c
@@ -55,6 +55,7 @@ samfile_t *samopen(const char *fn, const char *mode, const void *aux)
 				if (aux) { // check if aux is present
 					bam_header_t *textheader = fp->header;
 					fp->header = sam_header_read2((const char*)aux);
+					if (fp->header == 0) goto open_err_ret;
 					append_header_text(fp->header, textheader->text, textheader->l_text);
 					bam_header_destroy(textheader);
 				}