* samtools-0.1.4-21 (r368)

[samtools.git] / sam_view.c

diff --git a/sam_view.c b/sam_view.c
index cefac2f4835e84c4d6846862b85dc59b9032badf..02aee3c034cbc668b51295039d92a6adb1e74d51 100644 (file)
--- a/sam_view.c
+++ b/sam_view.c
@@ -5,14 +5,29 @@
 #include "sam.h"
 
 static int g_min_mapQ = 0, g_flag_on = 0, g_flag_off = 0;
+static char *g_library, *g_rg;
+
+static inline int __g_skip_aln(const bam_header_t *h, const bam1_t *b)
+{
+       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_library || g_rg) {
+               uint8_t *s = bam_aux_get(b, "RG");
+               if (s) {
+                       if (g_rg && strcmp(g_rg, (char*)(s + 1)) == 0) return 0;
+                       if (g_library) {
+                               const char *p = bam_strmap_get(h->rg2lib, (char*)(s + 1));
+                               return (p && strcmp(p, g_library) == 0)? 0 : 1;
+                       } return 1;
+               } else return 1;
+       } else return 0;
+}
 
 // callback function for bam_fetch()
 static int view_func(const bam1_t *b, void *data)
 {
-       if (b->core.qual < g_min_mapQ) return 0;
-       if (g_flag_on && (b->core.flag & g_flag_on) == 0) return 0;
-       if (b->core.flag & g_flag_off) return 0;
-       samwrite((samfile_t*)data, b);
+       if (!__g_skip_aln(((samfile_t*)data)->header, b))
+               samwrite((samfile_t*)data, b);
        return 0;
 }
 
@@ -20,28 +35,35 @@ static int usage(void);
 
 int main_samview(int argc, char *argv[])
 {
-       int c, is_header = 0, is_header_only = 0, is_bamin = 1, ret = 0;
+       int c, is_header = 0, is_header_only = 0, is_bamin = 1, ret = 0, is_uncompressed = 0, is_bamout = 0;
        samfile_t *in = 0, *out = 0;
-       char in_mode[4], out_mode[4], *fn_out = 0, *fn_list = 0;
+       char in_mode[5], out_mode[5], *fn_out = 0, *fn_list = 0;
 
        /* parse command-line options */
        strcpy(in_mode, "r"); strcpy(out_mode, "w");
-       while ((c = getopt(argc, argv, "bt:hHo:q:f:F:")) >= 0) {
+       while ((c = getopt(argc, argv, "Sbt:hHo:q:f:F:ul:r:")) >= 0) {
                switch (c) {
-               case 'b': strcat(out_mode, "b"); break;
+               case 'S': is_bamin = 0; break;
+               case 'b': is_bamout = 1; break;
                case 't': fn_list = strdup(optarg); is_bamin = 0; break;
                case 'h': is_header = 1; break;
                case 'H': is_header_only = 1; break;
                case 'o': fn_out = strdup(optarg); break;
                case 'f': g_flag_on = strtol(optarg, 0, 0); break;
-               case 'F': g_flag_on = strtol(optarg, 0, 0); break;
+               case 'F': g_flag_off = strtol(optarg, 0, 0); break;
                case 'q': g_min_mapQ = atoi(optarg); break;
+               case 'u': is_uncompressed = 1; break;
+               case 'l': g_library = strdup(optarg); break;
+               case 'r': g_rg = strdup(optarg); break;
                default: return usage();
                }
        }
+       if (is_uncompressed) is_bamout = 1;
        if (is_header_only) is_header = 1;
+       if (is_bamout) strcat(out_mode, "b");
        if (is_bamin) strcat(in_mode, "b");
        if (is_header) strcat(out_mode, "h");
+       if (is_uncompressed) strcat(out_mode, "u");
        if (argc == optind) return usage();
 
        // open file handlers
@@ -59,7 +81,8 @@ int main_samview(int argc, char *argv[])
                bam1_t *b = bam_init1();
                int r;
                while ((r = samread(in, b)) >= 0) // read one alignment from `in'
-                       samwrite(out, b); // write the alignment to `out'
+                       if (!__g_skip_aln(in->header, b))
+                               samwrite(out, b); // write the alignment to `out'
                if (r < -1) fprintf(stderr, "[main_samview] truncated file.\n");
                bam_destroy1(b);
        } else { // retrieve alignments in specified regions
@@ -85,7 +108,7 @@ int main_samview(int argc, char *argv[])
 
 view_end:
        // close files, free and return
-       free(fn_list); free(fn_out);
+       free(fn_list); free(fn_out); free(g_library); free(g_rg);
        samclose(in);
        samclose(out);
        return ret;
@@ -98,11 +121,15 @@ static int usage()
        fprintf(stderr, "Options: -b       output BAM\n");
        fprintf(stderr, "         -h       print header for the SAM output\n");
        fprintf(stderr, "         -H       print header only (no alignments)\n");
-       fprintf(stderr, "         -t FILE  list of reference names and lengths, assuming SAM input [null]\n");
+       fprintf(stderr, "         -S       input is SAM\n");
+       fprintf(stderr, "         -u       uncompressed BAM output (force -b)\n");
+       fprintf(stderr, "         -t FILE  list of reference names and lengths (force -S) [null]\n");
        fprintf(stderr, "         -o FILE  output file name [stdout]\n");
        fprintf(stderr, "         -f INT   required flag, 0 for unset [0]\n");
        fprintf(stderr, "         -F INT   filtering flag, 0 for unset [0]\n");
        fprintf(stderr, "         -q INT   minimum mapping quality [0]\n");
+       fprintf(stderr, "         -l STR   only output reads in library STR [null]\n");
+       fprintf(stderr, "         -r STR   only output reads in read group STR [null]\n");
        fprintf(stderr, "\n\
 Notes:\n\
 \n\
@@ -121,6 +148,9 @@ Notes:\n\
   4. A region should be presented in one of the following formats:\n\
      `chr1', `chr2:1,000' and `chr3:1000-2,000'. When a region is\n\
      specified, the input alignment file must be an indexed BAM file.\n\
+\n\
+  5. Option `-u' is preferred over `-b' when the output is piped to\n\
+     another samtools command.\n\
 \n");
        return 1;
 }