X-Git-Url: https://git.donarmstrong.com/?p=samtools.git;a=blobdiff_plain;f=bam_sort.c;h=7d00cd1a83321729c1384cf56f066d2c83f65548;hp=d38c7a11f8d22a74d30e5862b1d5440ba1a0e474;hb=307c147168f7154e3755712797078c513e0b242a;hpb=185e80bd42cead0e5af6ae0c54ebb6d0b66ca47c

diff --git a/bam_sort.c b/bam_sort.c
index d38c7a1..7d00cd1 100644
--- a/bam_sort.c
+++ b/bam_sort.c
@@ -89,7 +89,7 @@ static void swap_header_text(bam_header_t *h1, bam_header_t *h2)
   @discussion Padding information may NOT correctly maintained. This
   function is NOT thread safe.
  */
-int bam_merge_core2(int by_qname, const char *out, const char *headers, int n, char * const *fn, int flag, const char *reg, int n_threads)
+int bam_merge_core2(int by_qname, const char *out, const char *headers, int n, char * const *fn, int flag, const char *reg, int n_threads, int level)
 {
 	bamFile fpout, *fp;
 	heap1_t *heap;
@@ -97,7 +97,7 @@ int bam_merge_core2(int by_qname, const char *out, const char *headers, int n, c
 	bam_header_t *hheaders = NULL;
 	int i, j, *RG_len = 0;
 	uint64_t idx = 0;
-	char **RG = 0;
+	char **RG = 0, mode[8];
 	bam_iter_t *iter = 0;
 
 	if (headers) {
@@ -212,10 +212,11 @@ int bam_merge_core2(int by_qname, const char *out, const char *headers, int n, c
 		}
 		else h->pos = HEAP_EMPTY;
 	}
-	if (flag & MERGE_UNCOMP) fpout = strcmp(out, "-")? bam_open(out, "wu") : bam_dopen(fileno(stdout), "wu");
-	else if (flag & MERGE_LEVEL1) fpout = strcmp(out, "-")? bam_open(out, "w1") : bam_dopen(fileno(stdout), "w1");
-	else fpout = strcmp(out, "-")? bam_open(out, "w") : bam_dopen(fileno(stdout), "w");
-	if (fpout == 0) {
+	if (flag & MERGE_UNCOMP) level = 0;
+	else if (flag & MERGE_LEVEL1) level = 1;
+	strcpy(mode, "w");
+	if (level >= 0) sprintf(mode + 1, "%d", level < 9? level : 9);
+	if ((fpout = strcmp(out, "-")? bam_open(out, "w") : bam_dopen(fileno(stdout), "w")) == 0) {
 		fprintf(stderr, "[%s] fail to create the output file.\n", __func__);
 		return -1;
 	}
@@ -258,15 +259,15 @@ int bam_merge_core2(int by_qname, const char *out, const char *headers, int n, c
 
 int bam_merge_core(int by_qname, const char *out, const char *headers, int n, char * const *fn, int flag, const char *reg)
 {
-	return bam_merge_core2(by_qname, out, headers, n, fn, flag, reg, 0);
+	return bam_merge_core2(by_qname, out, headers, n, fn, flag, reg, 0, -1);
 }
 
 int bam_merge(int argc, char *argv[])
 {
-	int c, is_by_qname = 0, flag = 0, ret = 0, n_threads = 0;
+	int c, is_by_qname = 0, flag = 0, ret = 0, n_threads = 0, level = -1;
 	char *fn_headers = NULL, *reg = 0;
 
-	while ((c = getopt(argc, argv, "h:nru1R:f@:")) >= 0) {
+	while ((c = getopt(argc, argv, "h:nru1R:f@:l:")) >= 0) {
 		switch (c) {
 		case 'r': flag |= MERGE_RG; break;
 		case 'f': flag |= MERGE_FORCE; break;
@@ -275,6 +276,7 @@ int bam_merge(int argc, char *argv[])
 		case '1': flag |= MERGE_LEVEL1; break;
 		case 'u': flag |= MERGE_UNCOMP; break;
 		case 'R': reg = strdup(optarg); break;
+		case 'l': level = atoi(optarg); break;
 		case '@': n_threads = atoi(optarg); break;
 		}
 	}
@@ -286,6 +288,7 @@ int bam_merge(int argc, char *argv[])
 		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, "         -l INT   compression level, from 0 to 9 [-1]\n");
 		fprintf(stderr, "         -@ INT   number of BAM compression threads [0]\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");
@@ -302,7 +305,7 @@ int bam_merge(int argc, char *argv[])
 			return 1;
 		}
 	}
-	if (bam_merge_core2(is_by_qname, argv[optind], fn_headers, argc - optind - 1, argv + optind + 1, flag, reg, n_threads) < 0) ret = 1;
+	if (bam_merge_core2(is_by_qname, argv[optind], fn_headers, argc - optind - 1, argv + optind + 1, flag, reg, n_threads, level) < 0) ret = 1;
 	free(reg);
 	free(fn_headers);
 	return ret;
@@ -316,6 +319,40 @@ int bam_merge(int argc, char *argv[])
 
 typedef bam1_t *bam1_p;
 
+static int change_SO(bam_header_t *h, const char *so)
+{
+	char *p, *q, *beg = 0, *end = 0, *newtext;
+	if (h->l_text > 3) {
+		if (strncmp(h->text, "@HD", 3) == 0) {
+			if ((p = strchr(h->text, '\n')) == 0) return -1;
+			*p = '\0';
+			if ((q = strstr(h->text, "\tSO:")) != 0) {
+				*p = '\n'; // change back
+				if (strncmp(q + 4, so, p - q - 4) != 0) {
+					beg = q;
+					for (q += 4; *q != '\n' && *q != '\t'; ++q);
+					end = q;
+				} else return 0; // no need to change
+			} else beg = end = p, *p = '\n';
+		}
+	}
+	if (beg == 0) { // no @HD
+		h->l_text += strlen(so) + 15;
+		newtext = malloc(h->l_text + 1);
+		sprintf(newtext, "@HD\tVN:1.3\tSO:%s\n", so);
+		strcat(newtext, h->text);
+	} else { // has @HD but different or no SO
+		h->l_text = (beg - h->text) + (4 + strlen(so)) + (h->text + h->l_text - end);
+		newtext = malloc(h->l_text + 1);
+		strncpy(newtext, h->text, beg - h->text);
+		sprintf(newtext + (beg - h->text), "\tSO:%s", so);
+		strcat(newtext, end);
+	}
+	free(h->text);
+	h->text = newtext;
+	return 0;
+}
+
 static inline int bam1_lt(const bam1_p a, const bam1_p b)
 {
 	if (g_is_by_qname) {
@@ -402,9 +439,9 @@ static int sort_blocks(int n_files, size_t k, bam1_p *buf, const char *prefix, c
   and then merge them by calling bam_merge_core(). This function is
   NOT thread safe.
  */
-void bam_sort_core_ext(int is_by_qname, const char *fn, const char *prefix, size_t _max_mem, int is_stdout, int n_threads)
+void bam_sort_core_ext(int is_by_qname, const char *fn, const char *prefix, size_t _max_mem, int is_stdout, int n_threads, int level)
 {
-	int ret, i, extra_mem, n_files = 0;
+	int ret, i, n_files = 0;
 	size_t mem, max_k, k, max_mem;
 	bam_header_t *header;
 	bamFile fp;
@@ -412,7 +449,6 @@ void bam_sort_core_ext(int is_by_qname, const char *fn, const char *prefix, size
 	char *fnout = 0;
 
 	if (n_threads < 2) n_threads = 1;
-	extra_mem = sizeof(void*) + sizeof(void*) + (sizeof(bam1_t) - sizeof(bam1_core_t));
 	g_is_by_qname = is_by_qname;
 	max_k = k = 0; mem = 0;
 	max_mem = _max_mem * n_threads;
@@ -423,6 +459,8 @@ void bam_sort_core_ext(int is_by_qname, const char *fn, const char *prefix, size
 		return;
 	}
 	header = bam_header_read(fp);
+	if (is_by_qname) change_SO(header, "queryname");
+	else change_SO(header, "coordinate");
 	// write sub files
 	for (;;) {
 		if (k == max_k) {
@@ -434,7 +472,12 @@ void bam_sort_core_ext(int is_by_qname, const char *fn, const char *prefix, size
 		if (buf[k] == 0) buf[k] = (bam1_t*)calloc(1, sizeof(bam1_t));
 		b = buf[k];
 		if ((ret = bam_read1(fp, b)) < 0) break;
-		mem += ret + extra_mem;
+		if (b->data_len < b->m_data>>2) { // shrink
+			b->m_data = b->data_len;
+			kroundup32(b->m_data);
+			b->data = realloc(b->data, b->m_data);
+		}
+		mem += sizeof(bam1_t) + b->m_data + sizeof(void*) + sizeof(void*); // two sizeof(void*) for the data allocated to pointer arrays
 		++k;
 		if (mem >= max_mem) {
 			n_files = sort_blocks(n_files, k, buf, prefix, header, n_threads);
@@ -449,8 +492,11 @@ void bam_sort_core_ext(int is_by_qname, const char *fn, const char *prefix, size
 	else sprintf(fnout, "%s.bam", prefix);
 	// write the final output
 	if (n_files == 0) { // a single block
+		char mode[8];
+		strcpy(mode, "w");
+		if (level >= 0) sprintf(mode + 1, "%d", level < 9? level : 9);
 		ks_mergesort(sort, k, buf, 0);
-		write_buffer(fnout, "w", k, buf, header, n_threads);
+		write_buffer(fnout, mode, k, buf, header, n_threads);
 	} else { // then merge
 		char **fns;
 		n_files = sort_blocks(n_files, k, buf, prefix, header, n_threads);
@@ -460,7 +506,7 @@ void bam_sort_core_ext(int is_by_qname, const char *fn, const char *prefix, size
 			fns[i] = (char*)calloc(strlen(prefix) + 20, 1);
 			sprintf(fns[i], "%s.%.4d.bam", prefix, i);
 		}
-		bam_merge_core2(is_by_qname, fnout, 0, n_files, fns, 0, 0, n_threads);
+		bam_merge_core2(is_by_qname, fnout, 0, n_files, fns, 0, 0, n_threads, level);
 		for (i = 0; i < n_files; ++i) {
 			unlink(fns[i]);
 			free(fns[i]);
@@ -481,14 +527,14 @@ void bam_sort_core_ext(int is_by_qname, const char *fn, const char *prefix, size
 
 void bam_sort_core(int is_by_qname, const char *fn, const char *prefix, size_t max_mem)
 {
-	bam_sort_core_ext(is_by_qname, fn, prefix, max_mem, 0, 0);
+	bam_sort_core_ext(is_by_qname, fn, prefix, max_mem, 0, 0, -1);
 }
 
 int bam_sort(int argc, char *argv[])
 {
 	size_t max_mem = 768<<20; // 512MB
-	int c, is_by_qname = 0, is_stdout = 0, n_threads = 0;
-	while ((c = getopt(argc, argv, "nom:@:")) >= 0) {
+	int c, is_by_qname = 0, is_stdout = 0, n_threads = 0, level = -1;
+	while ((c = getopt(argc, argv, "nom:@:l:")) >= 0) {
 		switch (c) {
 		case 'o': is_stdout = 1; break;
 		case 'n': is_by_qname = 1; break;
@@ -501,12 +547,20 @@ int bam_sort(int argc, char *argv[])
 				break;
 			}
 		case '@': n_threads = atoi(optarg); break;
+		case 'l': level = atoi(optarg); break;
 		}
 	}
 	if (optind + 2 > argc) {
-		fprintf(stderr, "Usage: samtools sort [-on] [-m maxMem=1G] <in.bam> <out.prefix>\n");
+		fprintf(stderr, "\n");
+		fprintf(stderr, "Usage:   samtools sort [options] <in.bam> <out.prefix>\n\n");
+		fprintf(stderr, "Options: -n        sort by read name\n");
+		fprintf(stderr, "         -o        final output to stdout\n");
+		fprintf(stderr, "         -l INT    compression level, from 0 to 9 [-1]\n");
+		fprintf(stderr, "         -@ INT    number of sorting and compression threads [1]\n");
+		fprintf(stderr, "         -m INT    max memory per thread; suffix K/M/G recognized [768M]\n");
+		fprintf(stderr, "\n");
 		return 1;
 	}
-	bam_sort_core_ext(is_by_qname, argv[optind], argv[optind+1], max_mem, is_stdout, n_threads);
+	bam_sort_core_ext(is_by_qname, argv[optind], argv[optind+1], max_mem, is_stdout, n_threads, level);
 	return 0;
 }