X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bam_sort.c;h=402792aea44952a5962e340e9a0efbad75654494;hb=1a23a35867ee992dcffcf3d17df2b7d41732f33b;hp=c5ed5835b0c26cb4e3ed0a756af26ee8f6d76ef6;hpb=f93dae0d03856955f9424e8b2aaf261304ca647e;p=samtools.git

diff --git a/bam_sort.c b/bam_sort.c
index c5ed583..402792a 100644
--- a/bam_sort.c
+++ b/bam_sort.c
@@ -47,6 +47,16 @@ static inline int heap_lt(const heap1_t a, const heap1_t b)
 
 KSORT_INIT(heap, heap1_t, heap_lt)
 
+/*!
+  @abstract    Merge multiple sorted BAM.
+  @param  is_by_qname whether to sort by query name
+  @param  out  output BAM file name
+  @param  n    number of files to be merged
+  @param  fn   names of files to be merged
+
+  @discussion Padding information may NOT correctly maintained. This
+  function is NOT thread safe.
+ */
 void bam_merge_core(int by_qname, const char *out, int n, char * const *fn)
 {
 	bamFile fpout, *fp;
@@ -66,13 +76,17 @@ void bam_merge_core(int by_qname, const char *out, int n, char * const *fn)
 		else { // validate multiple baf
 			if (hout->n_targets != hin->n_targets) {
 				fprintf(stderr, "[bam_merge_core] file '%s' has different number of target sequences. Abort!\n", fn[i]);
-				abort();
+				exit(1);
 			}
 			for (j = 0; j < hout->n_targets; ++j) {
-				if (strcmp(hout->target_name[j], hin->target_name[j]) || hout->target_len[j] != hin->target_len[j]) {
-					fprintf(stderr, "[bam_merge_core] file '%s' has a different target sequence. Abort!\n", fn[i]);
-					abort();
+				if (strcmp(hout->target_name[j], hin->target_name[j])) {
+					fprintf(stderr, "[bam_merge_core] different target sequence name: '%s' != '%s' in file '%s'. Abort!\n",
+							hout->target_name[j], hin->target_name[j], fn[i]);
+					exit(1);
 				}
+				if (hout->target_len[j] != hin->target_len[j])
+					fprintf(stderr, "[bam_merge_core] different target sequence length: %d != %d in file '%s'. Continue.\n",
+							hout->target_len[j], hin->target_len[j], fn[i]);
 			}
 			bam_header_destroy(hin);
 		}
@@ -115,7 +129,7 @@ int bam_merge(int argc, char *argv[])
 		case 'n': is_by_qname = 1; break;
 		}
 	}
-	if (optind + 3 >= argc) {
+	if (optind + 2 >= argc) {
 		fprintf(stderr, "Usage: samtools merge [-n] <out.bam> <in1.bam> <in2.bam> [...]\n");
 		return 1;
 	}
@@ -151,6 +165,20 @@ static void sort_blocks(int n, int k, bam1_p *buf, const char *prefix, const bam
 	bam_close(fp);
 }
 
+/*!
+  @abstract Sort an unsorted BAM file based on the chromosome order
+  and the leftmost position of an alignment
+
+  @param  is_by_qname whether to sort by query name
+  @param  fn       name of the file to be sorted
+  @param  prefix   prefix of the output and the temporary files; upon
+	                   sucessess, prefix.bam will be written.
+  @param  max_mem  approxiate maximum memory (very inaccurate)
+
+  @discussion It may create multiple temporary subalignment files
+  and then merge them by calling bam_merge_core(). This function is
+  NOT thread safe.
+ */
 void bam_sort_core(int is_by_qname, const char *fn, const char *prefix, size_t max_mem)
 {
 	int n, ret, k, i;
@@ -191,7 +219,7 @@ void bam_sort_core(int is_by_qname, const char *fn, const char *prefix, size_t m
 			fns[i] = (char*)calloc(strlen(prefix) + 20, 1);
 			sprintf(fns[i], "%s.%.4d.bam", prefix, i);
 		}
-		bam_merge_core(0, fnout, n, fns);
+		bam_merge_core(is_by_qname, fnout, n, fns);
 		free(fnout);
 		for (i = 0; i < n; ++i) {
 			unlink(fns[i]);
@@ -221,7 +249,7 @@ int bam_sort(int argc, char *argv[])
 		}
 	}
 	if (optind + 2 > argc) {
-		fprintf(stderr, "Usage: samtools sort [-n] [-m <maxMem>] <in.baf> <out.prefix>\n");
+		fprintf(stderr, "Usage: samtools sort [-n] [-m <maxMem>] <in.bam> <out.prefix>\n");
 		return 1;
 	}
 	bam_sort_core(is_by_qname, argv[optind], argv[optind+1], max_mem);