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 "0.2.3 (r876)"
  10
  11 int main(int argc, char *argv[])
  12 {
  13         int c, skip = -1, meta = -1, list_chrms = 0, force = 0, print_header = 0;
  14         ti_conf_t conf = ti_conf_gff;
  15         while ((c = getopt(argc, argv, "p:s:b:e:0S:c:lhf")) >= 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 || strcmp(optarg, "vcf4") == 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 'h': print_header = 1; break;
  36                 case 'f': force = 1; break;
  37                 }
  38         }
  39         if (skip >= 0) conf.line_skip = skip;
  40         if (meta >= 0) conf.meta_char = meta;
  41         if (optind == argc) {
  42                 fprintf(stderr, "\n");
  43                 fprintf(stderr, "Program: tabix (TAB-delimited file InderXer)\n");
  44                 fprintf(stderr, "Version: %s\n\n", PACKAGE_VERSION);
  45                 fprintf(stderr, "Usage:   tabix <in.tab.bgz> [region1 [region2 [...]]]\n\n");
  46                 fprintf(stderr, "Options: -p STR     preset: gff, bed, sam, vcf, psltbl [gff]\n");
  47                 fprintf(stderr, "         -s INT     sequence name column [1]\n");
  48                 fprintf(stderr, "         -b INT     start column [4]\n");
  49                 fprintf(stderr, "         -e INT     end column [5]\n");
  50                 fprintf(stderr, "         -S INT     skip first INT lines [0]\n");
  51                 fprintf(stderr, "         -c CHAR    symbol for comment/meta lines [#]\n");
  52                 fprintf(stderr, "         -0         zero-based coordinate\n");
  53                 fprintf(stderr, "         -h         print the VCF header\n");
  54                 fprintf(stderr, "         -l         list chromosome names\n");
  55                 fprintf(stderr, "         -f         force to overwrite the index\n");
  56                 fprintf(stderr, "\n");
  57                 return 1;
  58         }
  59     if (list_chrms) {
  60                 ti_index_t *idx;
  61                 int i, n;
  62                 const char **names;
  63                 idx = ti_index_load(argv[optind]);
  64                 if (idx == 0) {
  65                         fprintf(stderr, "[main] fail to load the index file.\n");
  66                         return 1;
  67                 }
  68                 names = ti_seqname(idx, &n);
  69                 for (i = 0; i < n; ++i) printf("%s\n", names[i]);
  70                 free(names);
  71                 ti_index_destroy(idx);
  72                 return 0;
  73         }
  74         if (optind + 1 == argc) {
  75                 if (force == 0) {
  76                         struct stat buf;
  77                         char *fnidx = calloc(strlen(argv[optind]) + 5, 1);
  78                         strcat(strcpy(fnidx, argv[optind]), ".tbi");
  79                         if (stat(fnidx, &buf) == 0) {
  80                                 fprintf(stderr, "[tabix] the index file exists. Please use '-f' to overwrite.\n");
  81                                 free(fnidx);
  82                                 return 1;
  83                         }
  84                         free(fnidx);
  85                 }
  86         if ( is_bgzipped(argv[optind])!=1 )
  87         {
  88             fprintf(stderr,"[tabix] was bgzip used to compress this file? %s\n", argv[optind]);
  89             return 1;
  90         }
  91                 return ti_index_build(argv[optind], &conf);
  92         }
  93         { // retrieve
  94                 tabix_t *t;
  95                 if ((t = ti_open(argv[optind], 0)) == 0) {
  96                         fprintf(stderr, "[main] fail to open the data file.\n");
  97                         return 1;
  98                 }
  99                 if (strcmp(argv[optind+1], ".") == 0) { // retrieve all
 100                         ti_iter_t iter;
 101                         const char *s;
 102                         int len;
 103                         iter = ti_query(t, 0, 0, 0);
 104                         while ((s = ti_read(t, iter, &len)) != 0) {
 105                                 fputs(s, stdout); fputc('\n', stdout);
 106                         }
 107                         ti_iter_destroy(iter);
 108                 } else { // retrieve from specified regions
 109                         int i;
 110                         if ( ti_lazy_index_load(t) )
 111             {
 112                 fprintf(stderr,"[tabix] failed to load the index file.\n");
 113                 return 1;
 114             }
 115
 116             ti_iter_t iter;
 117             const char *s;
 118             int len;
 119             if ( print_header )
 120             {
 121                 // If requested, print the header lines here
 122                 iter = ti_query(t, 0, 0, 0);
 123                 while ((s = ti_read(t, iter, &len)) != 0) {
 124                     if ( *s != '#' ) break;
 125                     fputs(s, stdout); fputc('\n', stdout);
 126                 }
 127                 ti_iter_destroy(iter);
 128             }
 129                         for (i = optind + 1; i < argc; ++i) {
 130                                 int tid, beg, end;
 131                                 if (ti_parse_region(t->idx, argv[i], &tid, &beg, &end) == 0) {
 132                                         iter = ti_queryi(t, tid, beg, end);
 133                                         while ((s = ti_read(t, iter, &len)) != 0) {
 134                                                 fputs(s, stdout); fputc('\n', stdout);
 135                                         }
 136                                         ti_iter_destroy(iter);
 137                                 }
 138                 // else fprintf(stderr, "[main] invalid region: unknown target name or minus interval.\n");
 139                         }
 140                 }
 141                 ti_close(t);
 142         }
 143         return 0;
 144 }