int fai_build(const char *fn)
{
char *str;
RAZF *rz;
FILE *fp;
faidx_t *fai;
str = (char*)calloc(strlen(fn) + 5, 1);
sprintf(str, "%s.fai", fn);
rz = razf_open(fn, "r");
if (rz == 0) {
fprintf(stderr, "[fai_build] fail to open the FASTA file %s\n",fn);
free(str);
return -1;
}
fai = fai_build_core(rz);
razf_close(rz);
fp = fopen(str, "wb");
if (fp == 0) {
fprintf(stderr, "[fai_build] fail to write FASTA index %s\n",str);
fai_destroy(fai); free(str);
return -1;
}
fai_save(fai, fp);
fclose(fp);
free(str);
fai_destroy(fai);
return 0;
}
faidx_t *fai_load(const char *fn)
{
char *str;
FILE *fp;
faidx_t *fai;
str = (char*)calloc(strlen(fn) + 5, 1);
sprintf(str, "%s.fai", fn);
#ifdef _USE_KNETFILE
if (strstr(fn, "ftp://") == fn || strstr(fn, "http://") == fn)
{
fp = download_and_open(str);
if ( !fp )
{
fprintf(stderr, "[fai_load] failed to open remote FASTA index %s\n", str);
free(str);
return 0;
}
}
else
#endif
fp = fopen(str, "rb");
if (fp == 0) {
fprintf(stderr, "[fai_load] build FASTA index.\n");
fai_build(fn);
fp = fopen(str, "rb");
if (fp == 0) {
fprintf(stderr, "[fai_load] fail to open FASTA index.\n");
free(str);
return 0;
}
}
fai = fai_read(fp);
fclose(fp);
fai->rz = razf_open(fn, "rb");
free(str);
if (fai->rz == 0) {
fprintf(stderr, "[fai_load] fail to open FASTA file.\n");
return 0;
}
return fai;
}
int main(int argc, char **argv)
{
int c, compress, pstdout, is_forced;
RAZF *rz;
void *buffer;
long start, end, size;
compress = 1; pstdout = 0; start = 0; size = -1; end = -1; is_forced = 0;
while((c = getopt(argc, argv, "cdlhfb:s:")) >= 0){
switch(c){
case 'h': return razf_main_usage();
case 'd': compress = 0; break;
case 'c': pstdout = 1; break;
case 'l': compress = 2; break;
case 'b': start = atol(optarg); break;
case 's': size = atol(optarg); break;
case 'f': is_forced = 1; break;
}
}
if (size >= 0) end = start + size;
if(end >= 0 && end < start){
fprintf(stderr, " -- Illegal region: [%ld, %ld] --\n", start, end);
return 1;
}
if(compress == 1){
int f_src, f_dst = -1;
if(argc > optind){
if((f_src = open(argv[optind], O_RDONLY)) < 0){
fprintf(stderr, " -- Cannot open file: %s --\n", argv[optind]);
return 1;
}
if(pstdout){
f_dst = fileno(stdout);
} else {
char *name = malloc(sizeof(strlen(argv[optind]) + 5));
strcpy(name, argv[optind]);
strcat(name, ".rz");
f_dst = write_open(name, is_forced);
if (f_dst < 0) return 1;
free(name);
}
} else if(pstdout){
f_src = fileno(stdin);
f_dst = fileno(stdout);
} else return razf_main_usage();
rz = razf_dopen(f_dst, "w");
buffer = malloc(WINDOW_SIZE);
while((c = read(f_src, buffer, WINDOW_SIZE)) > 0) razf_write(rz, buffer, c);
razf_close(rz); // f_dst will be closed here
if (argc > optind && !pstdout) unlink(argv[optind]);
free(buffer);
close(f_src);
return 0;
} else {
if(argc <= optind) return razf_main_usage();
if(compress == 2){
rz = razf_open(argv[optind], "r");
if(rz->file_type == FILE_TYPE_RZ) {
printf("%20s%20s%7s %s\n", "compressed", "uncompressed", "ratio", "name");
printf("%20lld%20lld%6.1f%% %s\n", (long long)rz->end, (long long)rz->src_end, rz->end * 100.0f / rz->src_end,
argv[optind]);
} else fprintf(stdout, "%s is not a regular rz file\n", argv[optind]);
} else {
int f_dst;
if (argc > optind && !pstdout) {
char *name;
if (strstr(argv[optind], ".rz") - argv[optind] != strlen(argv[optind]) - 3) {
printf("razip: %s: unknown suffix -- ignored\n", argv[optind]);
return 1;
}
name = strdup(argv[optind]);
name[strlen(name) - 3] = '\0';
f_dst = write_open(name, is_forced);
free(name);
} else f_dst = fileno(stdout);
rz = razf_open(argv[optind], "r");
buffer = malloc(WINDOW_SIZE);
razf_seek(rz, start, SEEK_SET);
while(1){
if(end < 0) c = razf_read(rz, buffer, WINDOW_SIZE);
else c = razf_read(rz, buffer, (end - start > WINDOW_SIZE)? WINDOW_SIZE:(end - start));
if(c <= 0) break;
start += c;
write(f_dst, buffer, c);
if(end >= 0 && start >= end) break;
}
free(buffer);
if (!pstdout) unlink(argv[optind]);
}
razf_close(rz);
return 0;
}
}