A bug fix, "samtools view" is now working again.

[samtools.git] / bam_sort.c

diff --git a/bam_sort.c b/bam_sort.c
index a2d3d09473730715b78f38d6abb2b0123589daca..1817eb4e43f02e11b4310d7dc25355b6a3090dfc 100644 (file)
--- a/bam_sort.c
+++ b/bam_sort.c
@@ -93,7 +93,15 @@ void bam_merge_core(int by_qname, const char *out, const char *headers, int n, c
        for (i = 0; i != n; ++i) {
                heap1_t *h;
                bam_header_t *hin;
-               assert(fp[i] = bam_open(fn[i], "r"));
+               fp[i] = bam_open(fn[i], "r");
+               if (fp[i] == 0) {
+                       int j;
+                       fprintf(stderr, "[bam_merge_core] fail to open file %s\n", fn[i]);
+                       for (j = 0; j < i; ++j) bam_close(fp[j]);
+                       free(fp); free(heap);
+                       // FIXME: possible memory leak
+                       return;
+               }
                hin = bam_header_read(fp[i]);
                if (i == 0) { // the first SAM
                        hout = hin;
@@ -193,7 +201,7 @@ static inline int bam1_lt(const bam1_p a, const bam1_p b)
 }
 KSORT_INIT(sort, bam1_p, bam1_lt)
 
-static void sort_blocks(int n, int k, bam1_p *buf, const char *prefix, const bam_header_t *h)
+static void sort_blocks(int n, int k, bam1_p *buf, const char *prefix, const bam_header_t *h, int is_stdout)
 {
        char *name;
        int i;
@@ -202,7 +210,13 @@ static void sort_blocks(int n, int k, bam1_p *buf, const char *prefix, const bam
        name = (char*)calloc(strlen(prefix) + 20, 1);
        if (n >= 0) sprintf(name, "%s.%.4d.bam", prefix, n);
        else sprintf(name, "%s.bam", prefix);
-       assert(fp = bam_open(name, "w"));
+       fp = is_stdout? bam_dopen(fileno(stdout), "w") : bam_open(name, "w");
+       if (fp == 0) {
+               fprintf(stderr, "[sort_blocks] fail to create file %s.\n", name);
+               free(name);
+               // FIXME: possible memory leak
+               return;
+       }
        free(name);
        bam_header_write(fp, h);
        for (i = 0; i < k; ++i)
@@ -224,7 +238,7 @@ static void sort_blocks(int n, int k, bam1_p *buf, const char *prefix, const bam
   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)
+void bam_sort_core_ext(int is_by_qname, const char *fn, const char *prefix, size_t max_mem, int is_stdout)
 {
        int n, ret, k, i;
        size_t mem;
@@ -235,7 +249,10 @@ void bam_sort_core(int is_by_qname, const char *fn, const char *prefix, size_t m
        g_is_by_qname = is_by_qname;
        n = k = 0; mem = 0;
        fp = strcmp(fn, "-")? bam_open(fn, "r") : bam_dopen(fileno(stdin), "r");
-       assert(fp);
+       if (fp == 0) {
+               fprintf(stderr, "[bam_sort_core] fail to open file %s\n", fn);
+               return;
+       }
        header = bam_header_read(fp);
        buf = (bam1_t**)calloc(max_mem / BAM_CORE_SIZE, sizeof(bam1_t*));
        // write sub files
@@ -246,19 +263,20 @@ void bam_sort_core(int is_by_qname, const char *fn, const char *prefix, size_t m
                mem += ret;
                ++k;
                if (mem >= max_mem) {
-                       sort_blocks(n++, k, buf, prefix, header);
+                       sort_blocks(n++, k, buf, prefix, header, is_stdout);
                        mem = 0; k = 0;
                }
        }
        if (ret != -1)
                fprintf(stderr, "[bam_sort_core] truncated file. Continue anyway.\n");
-       if (n == 0) sort_blocks(-1, k, buf, prefix, header);
+       if (n == 0) sort_blocks(-1, k, buf, prefix, header, is_stdout);
        else { // then merge
                char **fns, *fnout;
                fprintf(stderr, "[bam_sort_core] merging from %d files...\n", n+1);
-               sort_blocks(n++, k, buf, prefix, header);
+               sort_blocks(n++, k, buf, prefix, header, is_stdout);
                fnout = (char*)calloc(strlen(prefix) + 20, 1);
-               sprintf(fnout, "%s.bam", prefix);
+               if (is_stdout) sprintf(fnout, "-");
+               else sprintf(fnout, "%s.bam", prefix);
                fns = (char**)calloc(n, sizeof(char*));
                for (i = 0; i < n; ++i) {
                        fns[i] = (char*)calloc(strlen(prefix) + 20, 1);
@@ -283,20 +301,26 @@ void bam_sort_core(int is_by_qname, const char *fn, const char *prefix, size_t m
        bam_close(fp);
 }
 
+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);
+}
+
 int bam_sort(int argc, char *argv[])
 {
        size_t max_mem = 500000000;
-       int c, is_by_qname = 0;
-       while ((c = getopt(argc, argv, "nm:")) >= 0) {
+       int c, is_by_qname = 0, is_stdout = 0;
+       while ((c = getopt(argc, argv, "nom:")) >= 0) {
                switch (c) {
+               case 'o': is_stdout = 1; break;
                case 'n': is_by_qname = 1; break;
                case 'm': max_mem = atol(optarg); break;
                }
        }
        if (optind + 2 > argc) {
-               fprintf(stderr, "Usage: samtools sort [-n] [-m <maxMem>] <in.bam> <out.prefix>\n");
+               fprintf(stderr, "Usage: samtools sort [-on] [-m <maxMem>] <in.bam> <out.prefix>\n");
                return 1;
        }
-       bam_sort_core(is_by_qname, argv[optind], argv[optind+1], max_mem);
+       bam_sort_core_ext(is_by_qname, argv[optind], argv[optind+1], max_mem, is_stdout);
        return 0;
 }