bgzip.c

   1 /* The MIT License
   2
   3    Copyright (c) 2008 Broad Institute / Massachusetts Institute of Technology
   4
   5    Permission is hereby granted, free of charge, to any person obtaining a copy
   6    of this software and associated documentation files (the "Software"), to deal
   7    in the Software without restriction, including without limitation the rights
   8    to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
   9    copies of the Software, and to permit persons to whom the Software is
  10    furnished to do so, subject to the following conditions:
  11
  12    The above copyright notice and this permission notice shall be included in
  13    all copies or substantial portions of the Software.
  14
  15    THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  16    IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  17    FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
  18    AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  19    LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
  20    OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
  21    THE SOFTWARE.
  22 */
  23
  24 #include <stdlib.h>
  25 #include <string.h>
  26 #include <stdio.h>
  27 #include <fcntl.h>
  28 #include <unistd.h>
  29 #include <errno.h>
  30 #include <sys/select.h>
  31 #include <sys/stat.h>
  32 #include "bgzf.h"
  33
  34 static const int WINDOW_SIZE = 64 * 1024;
  35
  36 static int bgzip_main_usage()
  37 {
  38         fprintf(stderr, "\n");
  39         fprintf(stderr, "Usage:   bgzip [options] [file] ...\n\n");
  40         fprintf(stderr, "Options: -c      write on standard output, keep original files unchanged\n");
  41         fprintf(stderr, "         -d      decompress\n");
  42         fprintf(stderr, "         -f      overwrite files without asking\n");
  43         fprintf(stderr, "         -b INT  decompress at virtual file pointer INT\n");
  44         fprintf(stderr, "         -s INT  decompress INT bytes in the uncompressed file\n");
  45         fprintf(stderr, "         -h      give this help\n");
  46         fprintf(stderr, "\n");
  47         return 1;
  48 }
  49
  50 static int write_open(const char *fn, int is_forced)
  51 {
  52         int fd = -1;
  53         char c;
  54         if (!is_forced) {
  55                 if ((fd = open(fn, O_WRONLY | O_CREAT | O_TRUNC | O_EXCL, 0666)) < 0 && errno == EEXIST) {
  56                         fprintf(stderr, "[bgzip] %s already exists; do you wish to overwrite (y or n)? ", fn);
  57                         scanf("%c", &c);
  58                         if (c != 'Y' && c != 'y') {
  59                                 fprintf(stderr, "[bgzip] not overwritten\n");
  60                                 exit(1);
  61                         }
  62                 }
  63         }
  64         if (fd < 0) {
  65                 if ((fd = open(fn, O_WRONLY | O_CREAT | O_TRUNC, 0666)) < 0) {
  66                         fprintf(stderr, "[bgzip] %s: Fail to write\n", fn);
  67                         exit(1);
  68                 }
  69         }
  70         return fd;
  71 }
  72
  73 static void fail(BGZF* fp)
  74 {
  75     fprintf(stderr, "Error: %s\n", fp->error);
  76     exit(1);
  77 }
  78
  79 int main(int argc, char **argv)
  80 {
  81         int c, compress, pstdout, is_forced;
  82         BGZF *fp;
  83         void *buffer;
  84         long start, end, size;
  85
  86         compress = 1; pstdout = 0; start = 0; size = -1; end = -1; is_forced = 0;
  87         while((c  = getopt(argc, argv, "cdhfb:s:")) >= 0){
  88                 switch(c){
  89                 case 'h': return bgzip_main_usage();
  90                 case 'd': compress = 0; break;
  91                 case 'c': pstdout = 1; break;
  92                 case 'b': start = atol(optarg); break;
  93                 case 's': size = atol(optarg); break;
  94                 case 'f': is_forced = 1; break;
  95                 }
  96         }
  97         if (size >= 0) end = start + size;
  98         if (end >= 0 && end < start) {
  99                 fprintf(stderr, "[bgzip] Illegal region: [%ld, %ld]\n", start, end);
 100                 return 1;
 101         }
 102         if (compress == 1) {
 103                 struct stat sbuf;
 104                 int f_src = fileno(stdin);
 105                 int f_dst = fileno(stdout);
 106
 107                 if ( argc>optind )
 108                 {
 109                         if ( stat(argv[optind],&sbuf)<0 )
 110                         {
 111                                 fprintf(stderr, "[bgzip] %s: %s\n", strerror(errno), argv[optind]);
 112                                 return 1;
 113                         }
 114
 115                         if ((f_src = open(argv[optind], O_RDONLY)) < 0) {
 116                                 fprintf(stderr, "[bgzip] %s: %s\n", strerror(errno), argv[optind]);
 117                                 return 1;
 118                         }
 119
 120                         if (pstdout)
 121                                 f_dst = fileno(stdout);
 122                         else
 123                         {
 124                                 char *name = malloc(strlen(argv[optind]) + 5);
 125                                 strcpy(name, argv[optind]);
 126                                 strcat(name, ".gz");
 127                                 f_dst = write_open(name, is_forced);
 128                                 if (f_dst < 0) return 1;
 129                                 free(name);
 130                         }
 131                 }
 132                 else if (!pstdout && isatty(fileno((FILE *)stdout)) )
 133                         return bgzip_main_usage();
 134
 135                 fp = bgzf_fdopen(f_dst, "w");
 136                 buffer = malloc(WINDOW_SIZE);
 137                 while ((c = read(f_src, buffer, WINDOW_SIZE)) > 0)
 138                         if (bgzf_write(fp, buffer, c) < 0) fail(fp);
 139                 // f_dst will be closed here
 140                 if (bgzf_close(fp) < 0) fail(fp);
 141                 if (argc > optind && !pstdout) unlink(argv[optind]);
 142                 free(buffer);
 143                 close(f_src);
 144                 return 0;
 145         } else {
 146                 struct stat sbuf;
 147                 int f_dst;
 148
 149                 if ( argc>optind )
 150                 {
 151                         if ( stat(argv[optind],&sbuf)<0 )
 152                         {
 153                                 fprintf(stderr, "[bgzip] %s: %s\n", strerror(errno), argv[optind]);
 154                                 return 1;
 155                         }
 156                         char *name;
 157                         int len = strlen(argv[optind]);
 158                         if ( strcmp(argv[optind]+len-3,".gz") )
 159                         {
 160                                 fprintf(stderr, "[bgzip] %s: unknown suffix -- ignored\n", argv[optind]);
 161                                 return 1;
 162                         }
 163                         fp = bgzf_open(argv[optind], "r");
 164                         if (fp == NULL) {
 165                                 fprintf(stderr, "[bgzip] Could not open file: %s\n", argv[optind]);
 166                                 return 1;
 167                         }
 168
 169                         if (pstdout) {
 170                                 f_dst = fileno(stdout);
 171                         }
 172                         else {
 173                                 name = strdup(argv[optind]);
 174                                 name[strlen(name) - 3] = '\0';
 175                                 f_dst = write_open(name, is_forced);
 176                                 free(name);
 177                         }
 178                 }
 179                 else if (!pstdout && isatty(fileno((FILE *)stdin)) )
 180                         return bgzip_main_usage();
 181                 else
 182                 {
 183                         f_dst = fileno(stdout);
 184                         fp = bgzf_fdopen(fileno(stdin), "r");
 185                         if (fp == NULL) {
 186                                 fprintf(stderr, "[bgzip] Could not read from stdin: %s\n", strerror(errno));
 187                                 return 1;
 188                         }
 189                 }
 190                 buffer = malloc(WINDOW_SIZE);
 191                 if (bgzf_seek(fp, start, SEEK_SET) < 0) fail(fp);
 192                 while (1) {
 193                         if (end < 0) c = bgzf_read(fp, buffer, WINDOW_SIZE);
 194                         else c = bgzf_read(fp, buffer, (end - start > WINDOW_SIZE)? WINDOW_SIZE:(end - start));
 195                         if (c == 0) break;
 196                         if (c < 0) fail(fp);
 197                         start += c;
 198                         write(f_dst, buffer, c);
 199                         if (end >= 0 && start >= end) break;
 200                 }
 201                 free(buffer);
 202                 if (bgzf_close(fp) < 0) fail(fp);
 203                 if (!pstdout) unlink(argv[optind]);
 204                 return 0;
 205         }
 206 }