main.c

   1 #include <string.h>
   2 #include <unistd.h>
   3 #include <stdlib.h>
   4 #include <stdio.h>
   5 #include <sys/stat.h>
   6 #include "bgzf.h"
   7 #include "tabix.h"
   8
   9 #define PACKAGE_VERSION "0.1.6 (r565)"
  10
  11 int main(int argc, char *argv[])
  12 {
  13         int c, skip = -1, meta = -1, list_chrms = 0, force = 0;
  14         ti_conf_t conf = ti_conf_gff;
  15         while ((c = getopt(argc, argv, "p:s:b:e:0S:c:lf")) >= 0) {
  16                 switch (c) {
  17                 case '0': conf.preset |= TI_FLAG_UCSC; break;
  18                 case 'S': skip = atoi(optarg); break;
  19                 case 'c': meta = optarg[0]; break;
  20                 case 'p':
  21                         if (strcmp(optarg, "gff") == 0) conf = ti_conf_gff;
  22                         else if (strcmp(optarg, "bed") == 0) conf = ti_conf_bed;
  23                         else if (strcmp(optarg, "sam") == 0) conf = ti_conf_sam;
  24                         else if (strcmp(optarg, "vcf") == 0) conf = ti_conf_vcf;
  25                         else if (strcmp(optarg, "psltbl") == 0) conf = ti_conf_psltbl;
  26                         else {
  27                                 fprintf(stderr, "[main] unrecognized preset '%s'\n", optarg);
  28                                 return 1;
  29                         }
  30                         break;
  31                 case 's': conf.sc = atoi(optarg); break;
  32                 case 'b': conf.bc = atoi(optarg); break;
  33                 case 'e': conf.ec = atoi(optarg); break;
  34         case 'l': list_chrms = 1; break;
  35                 case 'f': force = 1; break;
  36                 }
  37         }
  38         if (skip >= 0) conf.line_skip = skip;
  39         if (meta >= 0) conf.meta_char = meta;
  40         if (optind == argc) {
  41                 fprintf(stderr, "\n");
  42                 fprintf(stderr, "Program: tabix (TAB-delimited file InderXer)\n");
  43                 fprintf(stderr, "Version: %s\n\n", PACKAGE_VERSION);
  44                 fprintf(stderr, "Usage:   tabix <in.tab.bgz> [region1 [region2 [...]]]\n\n");
  45                 fprintf(stderr, "Options: -p STR     preset: gff, bed, sam, vcf, psltbl [gff]\n");
  46                 fprintf(stderr, "         -s INT     sequence name column [1]\n");
  47                 fprintf(stderr, "         -b INT     start column [4]\n");
  48                 fprintf(stderr, "         -e INT     end column [5]\n");
  49                 fprintf(stderr, "         -S INT     skip first INT lines [0]\n");
  50                 fprintf(stderr, "         -c CHAR    symbol for comment/meta lines [#]\n");
  51                 fprintf(stderr, "         -0         zero-based coordinate\n");
  52                 fprintf(stderr, "         -l         list chromosome names\n");
  53                 fprintf(stderr, "         -f         force to overwrite the index\n");
  54                 fprintf(stderr, "\n");
  55                 return 1;
  56         }
  57     if (list_chrms) {
  58                 ti_index_t *idx;
  59                 int i, n;
  60                 const char **names;
  61                 idx = ti_index_load(argv[optind]);
  62                 if (idx == 0) {
  63                         fprintf(stderr, "[main] fail to load the index file.\n");
  64                         return 1;
  65                 }
  66                 names = ti_seqname(idx, &n);
  67                 for (i = 0; i < n; ++i) printf("%s\n", names[i]);
  68                 free(names);
  69                 ti_index_destroy(idx);
  70                 return 0;
  71         }
  72         if (optind + 1 == argc) {
  73                 if (force == 0) {
  74                         struct stat buf;
  75                         char *fnidx = calloc(strlen(argv[optind]) + 5, 1);
  76                         strcat(strcpy(fnidx, argv[optind]), ".tbi");
  77                         if (stat(fnidx, &buf) == 0) {
  78                                 fprintf(stderr, "[tabix] the index file exists. Please use '-f' to overwrite.\n");
  79                                 free(fnidx);
  80                                 return 1;
  81                         }
  82                         free(fnidx);
  83                 }
  84                 return ti_index_build(argv[optind], &conf);
  85         }
  86         { // retrieve
  87                 tabix_t *t;
  88                 if ((t = ti_open(argv[optind], 0)) == 0) {
  89                         fprintf(stderr, "[main] fail to open the data file.\n");
  90                         return 1;
  91                 }
  92                 if (strcmp(argv[optind+1], ".") == 0) { // retrieve all
  93                         ti_iter_t iter;
  94                         const char *s;
  95                         int len;
  96                         iter = ti_query(t, 0, 0, 0);
  97                         while ((s = ti_read(t, iter, &len)) != 0) {
  98                                 fputs(s, stdout); fputc('\n', stdout);
  99                         }
 100                         ti_iter_destroy(iter);
 101                 } else { // retrieve from specified regions
 102                         int i;
 103                         ti_lazy_index_load(t);
 104                         for (i = optind + 1; i < argc; ++i) {
 105                                 int tid, beg, end;
 106                                 if (ti_parse_region(t->idx, argv[i], &tid, &beg, &end) == 0) {
 107                                         ti_iter_t iter;
 108                                         const char *s;
 109                                         int len;
 110                                         iter = ti_queryi(t, tid, beg, end);
 111                                         while ((s = ti_read(t, iter, &len)) != 0) {
 112                                                 fputs(s, stdout); fputc('\n', stdout);
 113                                         }
 114                                         ti_iter_destroy(iter);
 115                                 } else fprintf(stderr, "[main] invalid region: unknown target name or minus interval.\n");
 116                         }
 117                 }
 118                 ti_close(t);
 119         }
 120         return 0;
 121 }