From a655cbef86f24619675da97eae99ee76bd929029 Mon Sep 17 00:00:00 2001
From: Heng Li <lh3@me.com>
Date: Sun, 15 Jul 2018 11:03:18 -0400
Subject: [PATCH] print SAM header; remove tmp files

---
 main.c     |  2 +-
 map.c      |  5 ++++-
 mmpriv.h   |  1 +
 splitidx.c | 16 ++++++++++++++--
 4 files changed, 20 insertions(+), 4 deletions(-)

diff --git a/main.c b/main.c
index d32fff0..064e2fc 100644
--- a/main.c
+++ b/main.c
@@ -329,7 +329,7 @@ int main(int argc, char *argv[])
 				mm_write_sam_hdr(mi, rg, MM_VERSION, argc, argv);
 			} else {
 				mm_write_sam_hdr(0, rg, MM_VERSION, argc, argv);
-				if (mm_verbose >= 2)
+				if (opt.split_prefix == 0 && mm_verbose >= 2)
 					fprintf(stderr, "[WARNING]\033[1;31m For a multi-part index, no @SQ lines will be outputted.\033[0m\n");
 			}
 		}
diff --git a/map.c b/map.c
index be4ae74..1a6ff1a 100644
--- a/map.c
+++ b/map.c
@@ -659,7 +659,6 @@ int mm_split_merge(int n_segs, const char **fn, const mm_mapopt_t *opt, int n_sp
 {
 	int i;
 	pipeline_t pl;
-	mm_idx_t *mi;
 	if (n_segs < 1 || n_split_idx < 1) return -1;
 	memset(&pl, 0, sizeof(pipeline_t));
 	pl.n_fp = n_segs;
@@ -681,6 +680,9 @@ int mm_split_merge(int n_segs, const char **fn, const mm_mapopt_t *opt, int n_sp
 		pl.rid_shift[i] = pl.rid_shift[i - 1];
 	for (pl.rid_shift[0] = 0, i = 1; i < n_split_idx; ++i)
 		pl.rid_shift[i] += pl.rid_shift[i - 1];
+	if (opt->flag & MM_F_OUT_SAM)
+		for (i = 0; i < pl.mi->n_seq; ++i)
+			printf("@SQ\tSN:%s\tLN:%d\n", pl.mi->seq[i].name, pl.mi->seq[i].len);
 
 	kt_pipeline(2, worker_pipeline, &pl, 3);
 
@@ -692,5 +694,6 @@ int mm_split_merge(int n_segs, const char **fn, const mm_mapopt_t *opt, int n_sp
 	for (i = 0; i < pl.n_fp; ++i)
 		mm_bseq_close(pl.fp[i]);
 	free(pl.fp);
+	mm_split_rm_tmp(opt->split_prefix, n_split_idx);
 	return 0;
 }
diff --git a/mmpriv.h b/mmpriv.h
index a02a1b7..f847445 100644
--- a/mmpriv.h
+++ b/mmpriv.h
@@ -92,6 +92,7 @@ void mm_pair(void *km, int max_gap_ref, int dp_bonus, int sub_diff, int match_sc
 FILE *mm_split_init(const char *prefix, const mm_idx_t *mi);
 mm_idx_t *mm_split_merge_prep(const char *prefix, int n_splits, FILE **fp, uint32_t *n_seq_part);
 int mm_split_merge(int n_segs, const char **fn, const mm_mapopt_t *opt, int n_split_idx);
+void mm_split_rm_tmp(const char *prefix, int n_splits);
 
 void mm_err_puts(const char *str);
 void mm_err_fwrite(const void *p, size_t size, size_t nitems, FILE *fp);
diff --git a/splitidx.c b/splitidx.c
index 7d8627d..6d12071 100644
--- a/splitidx.c
+++ b/splitidx.c
@@ -33,7 +33,7 @@ mm_idx_t *mm_split_merge_prep(const char *prefix, int n_splits, FILE **fp, uint3
 	int i, j;
 
 	if (n_splits < 1) return 0;
-	fn = (char*)calloc(strlen(prefix) + 10, 1);
+	fn = CALLOC(char, strlen(prefix) + 10);
 	for (i = 0; i < n_splits; ++i) {
 		sprintf(fn, "%s.%.4d.tmp", prefix, i);
 		if ((fp[i] = fopen(fn, "rb")) == 0) {
@@ -47,7 +47,7 @@ mm_idx_t *mm_split_merge_prep(const char *prefix, int n_splits, FILE **fp, uint3
 	}
 	free(fn);
 
-	mi = (mm_idx_t*)calloc(1, sizeof(mm_idx_t));
+	mi = CALLOC(mm_idx_t, 1);
 	for (i = 0; i < n_splits; ++i) {
 		mm_err_fread(&mi->k, 4, 1, fp[i]); // TODO: check if k is all the same
 		mm_err_fread(&n_seq_part[i], 4, 1, fp[i]);
@@ -66,3 +66,15 @@ mm_idx_t *mm_split_merge_prep(const char *prefix, int n_splits, FILE **fp, uint3
 	}
 	return mi;
 }
+
+void mm_split_rm_tmp(const char *prefix, int n_splits)
+{
+	int i;
+	char *fn;
+	fn = CALLOC(char, strlen(prefix) + 10);
+	for (i = 0; i < n_splits; ++i) {
+		sprintf(fn, "%s.%.4d.tmp", prefix, i);
+		remove(fn);
+	}
+	free(fn);
+}