From: Heng Li <lh3@live.co.uk>
Date: Thu, 21 Apr 2011 18:50:12 +0000 (+0000)
Subject: fixed a bug in merge
X-Git-Url: https://git.donarmstrong.com/?a=commitdiff_plain;ds=sidebyside;h=d452c6f0d10693d0159a41fe128707e3366602a7;p=samtools.git

fixed a bug in merge
---

diff --git a/bam_sort.c b/bam_sort.c
index 94970b5..abf8d4f 100644
--- a/bam_sort.c
+++ b/bam_sort.c
@@ -71,6 +71,7 @@ static void swap_header_text(bam_header_t *h1, bam_header_t *h2)
 #define MERGE_RG     1
 #define MERGE_UNCOMP 2
 #define MERGE_LEVEL1 4
+#define MERGE_FORCE  8
 
 /*!
   @abstract    Merge multiple sorted BAM.
@@ -203,7 +204,7 @@ int bam_merge_core(int by_qname, const char *out, const char *headers, int n, ch
 		h->i = i;
 		h->b = (bam1_t*)calloc(1, sizeof(bam1_t));
 		if (bam_iter_read(fp[i], iter[i], h->b) >= 0) {
-			h->pos = ((uint64_t)h->b->core.tid<<32) | (uint32_t)h->b->core.pos<<1 | bam1_strand(h->b);
+			h->pos = ((uint64_t)h->b->core.tid<<32) | (uint32_t)((int32_t)h->b->core.pos+1)<<1 | bam1_strand(h->b);
 			h->idx = idx++;
 		}
 		else h->pos = HEAP_EMPTY;
@@ -228,7 +229,7 @@ int bam_merge_core(int by_qname, const char *out, const char *headers, int n, ch
 		}
 		bam_write1_core(fpout, &b->core, b->data_len, b->data);
 		if ((j = bam_iter_read(fp[heap->i], iter[heap->i], b)) >= 0) {
-			heap->pos = ((uint64_t)b->core.tid<<32) | (uint32_t)b->core.pos<<1 | bam1_strand(b);
+			heap->pos = ((uint64_t)b->core.tid<<32) | (uint32_t)((int)b->core.pos+1)<<1 | bam1_strand(b);
 			heap->idx = idx++;
 		} else if (j == -1) {
 			heap->pos = HEAP_EMPTY;
@@ -256,9 +257,10 @@ int bam_merge(int argc, char *argv[])
 	int c, is_by_qname = 0, flag = 0, ret = 0;
 	char *fn_headers = NULL, *reg = 0;
 
-	while ((c = getopt(argc, argv, "h:nru1R:")) >= 0) {
+	while ((c = getopt(argc, argv, "h:nru1R:f")) >= 0) {
 		switch (c) {
 		case 'r': flag |= MERGE_RG; break;
+		case 'f': flag |= MERGE_FORCE; break;
 		case 'h': fn_headers = strdup(optarg); break;
 		case 'n': is_by_qname = 1; break;
 		case '1': flag |= MERGE_LEVEL1; break;
@@ -272,6 +274,7 @@ int bam_merge(int argc, char *argv[])
 		fprintf(stderr, "Options: -n       sort by read names\n");
 		fprintf(stderr, "         -r       attach RG tag (inferred from file names)\n");
 		fprintf(stderr, "         -u       uncompressed BAM output\n");
+		fprintf(stderr, "         -f       overwrite the output BAM if exist\n");
 		fprintf(stderr, "         -1       compress level 1\n");
 		fprintf(stderr, "         -R STR   merge file in the specified region STR [all]\n");
 		fprintf(stderr, "         -h FILE  copy the header in FILE to <out.bam> [in1.bam]\n\n");
@@ -280,6 +283,14 @@ int bam_merge(int argc, char *argv[])
 		fprintf(stderr, "      the header dictionary in merging.\n\n");
 		return 1;
 	}
+	if (!(flag & MERGE_FORCE) && strcmp(argv[optind], "-")) {
+		FILE *fp = fopen(argv[optind], "rb");
+		if (fp != NULL) {
+			fclose(fp);
+			fprintf(stderr, "[%s] File '%s' exists. Please apply '-f' to overwrite. Abort.\n", __func__, argv[optind]);
+			return 1;
+		}
+	}
 	if (bam_merge_core(is_by_qname, argv[optind], fn_headers, argc - optind - 1, argv + optind + 1, flag, reg) < 0) ret = 1;
 	free(reg);
 	free(fn_headers);
@@ -292,8 +303,8 @@ static inline int bam1_lt(const bam1_p a, const bam1_p b)
 {
 	if (g_is_by_qname) {
 		int t = strnum_cmp(bam1_qname(a), bam1_qname(b));
-		return (t < 0 || (t == 0 && (((uint64_t)a->core.tid<<32|a->core.pos) < ((uint64_t)b->core.tid<<32|b->core.pos))));
-	} else return (((uint64_t)a->core.tid<<32|a->core.pos) < ((uint64_t)b->core.tid<<32|b->core.pos));
+		return (t < 0 || (t == 0 && (((uint64_t)a->core.tid<<32|(a->core.pos+1)) < ((uint64_t)b->core.tid<<32|(b->core.pos+1)))));
+	} else return (((uint64_t)a->core.tid<<32|(a->core.pos+1)) < ((uint64_t)b->core.tid<<32|(b->core.pos+1)));
 }
 KSORT_INIT(sort, bam1_p, bam1_lt)