Complain when BAM cannot be open. Severe bug fixed in -m haploid calling.

[samtools.git] / sam_view.c

diff --git a/sam_view.c b/sam_view.c
index 402e280606f7551dfdcc1667dcb289dc78c41a4e..528309897bbf7c2a35bb4f2fa4727f522afb889e 100644 (file)
--- a/sam_view.c
+++ b/sam_view.c
@@ -14,14 +14,14 @@ KHASH_SET_INIT_STR(rg)
 // data passed to the bam_fetch callback is encapsulated in this struct.
 typedef struct {
        bam_header_t *header;
-       int *count;
+       int64_t *count;  // int does overflow for very big BAMs
 } count_func_data_t;
 
 typedef khash_t(rg) *rghash_t;
 
 // FIXME: we'd better use no global variables...
 static rghash_t g_rghash = 0;
-static int g_min_mapQ = 0, g_flag_on = 0, g_flag_off = 0, g_qual_scale = 0;
+static int g_min_mapQ = 0, g_flag_on = 0, g_flag_off = 0, g_qual_scale = 0, g_min_qlen = 0;
 static uint32_t g_subsam_seed = 0;
 static double g_subsam_frac = -1.;
 static char *g_library, *g_rg;
@@ -41,6 +41,14 @@ static int process_aln(const bam_header_t *h, bam1_t *b)
                        qual[i] = c < 93? c : 93;
                }
        }
+       if (g_min_qlen > 0) {
+               int k, qlen = 0;
+               uint32_t *cigar = bam1_cigar(b);
+               for (k = 0; k < b->core.n_cigar; ++k)
+                       if ((bam_cigar_type(bam_cigar_op(cigar[k]))&1) || bam_cigar_op(cigar[k]) == BAM_CHARD_CLIP)
+                               qlen += bam_cigar_oplen(cigar[k]);
+               if (qlen < g_min_qlen) return 1;
+       }
        if (b->core.qual < g_min_mapQ || ((b->core.flag & g_flag_on) != g_flag_on) || (b->core.flag & g_flag_off))
                return 1;
        if (g_bed && b->core.tid >= 0 && !bed_overlap(g_bed, h->target_name[b->core.tid], b->core.pos, bam_calend(&b->core, bam1_cigar(b))))
@@ -120,13 +128,13 @@ int main_samview(int argc, char *argv[])
 {
        int c, is_header = 0, is_header_only = 0, is_bamin = 1, ret = 0, compress_level = -1, is_bamout = 0, is_count = 0;
        int of_type = BAM_OFDEC, is_long_help = 0, n_threads = 0;
-       int count = 0;
+       int64_t count = 0;
        samfile_t *in = 0, *out = 0;
        char in_mode[5], out_mode[5], *fn_out = 0, *fn_list = 0, *fn_ref = 0, *fn_rg = 0, *q;
 
        /* parse command-line options */
        strcpy(in_mode, "r"); strcpy(out_mode, "w");
-       while ((c = getopt(argc, argv, "SbBct:h1Ho:q:f:F:ul:r:xX?T:R:L:s:Q:@:")) >= 0) {
+       while ((c = getopt(argc, argv, "SbBct:h1Ho:q:f:F:ul:r:xX?T:R:L:s:Q:@:m:")) >= 0) {
                switch (c) {
                case 's':
                        if ((g_subsam_seed = strtol(optarg, &q, 10)) != 0) {
@@ -135,6 +143,7 @@ int main_samview(int argc, char *argv[])
                        }
                        g_subsam_frac = strtod(q, &q);
                        break;
+               case 'm': g_min_qlen = atoi(optarg); break;
                case 'c': is_count = 1; break;
                case 'S': is_bamin = 0; break;
                case 'b': is_bamout = 1; break;
@@ -265,7 +274,7 @@ int main_samview(int argc, char *argv[])
 
 view_end:
        if (is_count && ret == 0) {
-               printf("%d\n", count);
+               printf("%ld\n", count); // compilers on some platforms may complain about printing int64_t with %ld
        }
        // close files, free and return
        free(fn_list); free(fn_ref); free(fn_out); free(g_library); free(g_rg); free(fn_rg);