/*
* Takes the command line options and turns them into something we can understand
*/
static opts_t* parse_args(int argc, char *argv[])
{
if (argc == 1) { usage(stdout); return NULL; }
const char* optstring = "i:vb:";
static const struct option lopts[] = {
{ "hash", 1, 0, 0 },
{ "input", 1, 0, 'i' },
{ "verbose", 0, 0, 'v' },
{ "input-fmt", 1, 0, 0 },
{ NULL, 0, NULL, 0 }
};
opts_t* opts = calloc(sizeof(opts_t), 1);
if (!opts) { perror("cannot allocate option parsing memory"); return NULL; }
opts->argv_list = stringify_argv(argc+1, argv-1);
if (opts->argv_list[strlen(opts->argv_list)-1] == ' ') opts->argv_list[strlen(opts->argv_list)-1] = 0;
// set defaults
opts->verbose = false;
opts->hash = DEFAULT_HASH_TYPE;
int opt;
int option_index = 0;
while ((opt = getopt_long(argc, argv, optstring, lopts, &option_index)) != -1) {
const char *arg;
switch (opt) {
case 'i': opts->input_name = strdup(optarg);
break;
case 'v': opts->verbose = true;
break;
case 0: arg = lopts[option_index].name;
if (strcmp(arg, "hash") == 0) opts->hash = decode_hash_name(optarg);
else if (strcmp(arg, "input-fmt") == 0) opts->input_fmt = strdup(optarg);
else {
printf("\nUnknown option: %s\n\n", arg);
usage(stdout); free_opts(opts);
return NULL;
}
break;
default: printf("Unknown option: '%c'\n", opt);
/* else fall-through */
case '?': usage(stdout); free_opts(opts); return NULL;
}
}
argc -= optind;
argv += optind;
if (argc > 0) opts->input_name = strdup(argv[0]);
optind = 0;
// some validation and tidying
// input defaults to stdin
if (!opts->input_name) opts->input_name = strdup("-");
// TODO: list valid hash types here
if (opts->hash == HASH_UNKNOWN) {
fprintf(stderr, "Unknown has type\n");
return NULL;
}
return opts;
}
int main(int argc, char **argv) {
cram_fd *fd;
bam_file_t *bfd;
bam_seq_t *bam = NULL;
char mode[4] = {'w', '\0', '\0', '\0'};
char *prefix = NULL;
int decode_md = 0;
int C;
int start, end;
char ref_name[1024] = {0}, *arg_list, *ref_fn = NULL;
int embed_ref = 0;
while ((C = getopt(argc, argv, "bu0123456789mp:hr:R:X")) != -1) {
switch (C) {
case 'b':
mode[1] = 'b';
break;
case 'u':
mode[2] = '0';
break;
case '0': case '1': case '2': case '3': case '4':
case '5': case '6': case '7': case '8': case '9':
mode[2] = C;
break;
case 'm':
decode_md = 1;
break;
case 'p':
prefix = optarg;
break;
case 'h':
usage(stdout);
return 0;
case 'r':
ref_fn = optarg;
break;
case 'X':
embed_ref = 1;
break;
case 'R': {
char *cp = strchr(optarg, ':');
if (cp) {
*cp = 0;
switch (sscanf(cp+1, "%d-%d", &start, &end)) {
case 1:
end = start;
break;
case 2:
break;
default:
fprintf(stderr, "Malformed range format\n");
return 1;
}
} else {
start = INT_MIN;
end = INT_MAX;
}
strncpy(ref_name, optarg, 1023);
break;
}
case '?':
fprintf(stderr, "Unrecognised option: -%c\n", optopt);
usage(stderr);
return 1;
}
}
if (argc - optind != 1 && argc - optind != 2) {
usage(stderr);
return 1;
}
if (argc - optind == 1) {
if (NULL == (bfd = bam_open("-", mode))) {
fprintf(stderr, "Failed to open SAM/BAM output\n.");
return 1;
}
} else {
if (NULL == (bfd = bam_open(argv[optind+1], mode))) {
fprintf(stderr, "Failed to open SAM/BAM output\n.");
perror(argv[optind+1]);
return 1;
}
}
if (NULL == (fd = cram_open(argv[optind], "rb"))) {
fprintf(stderr, "Error opening CRAM file '%s'.\n", argv[optind]);
return 1;
}
if (*ref_name != 0)
cram_index_load(fd, argv[optind]);
if (prefix)
cram_set_option(fd, CRAM_OPT_PREFIX, prefix);
if (decode_md)
cram_set_option(fd, CRAM_OPT_DECODE_MD, decode_md);
if (embed_ref)
cram_set_option(fd, CRAM_OPT_EMBED_REF, embed_ref);
/* Find and load reference */
cram_load_reference(fd, ref_fn);
if (!fd->refs && !embed_ref) {
fprintf(stderr, "Unable to find an appropriate reference.\n"
"Please specify a valid reference with -r ref.fa option.\n");
return 1;
}
bfd->header = fd->header;
if (*ref_name != 0) {
cram_range r;
int refid = sam_hdr_name2ref(fd->header, ref_name);
if (refid == -1 && *ref_name != '*') {
fprintf(stderr, "Unknown reference name '%s'\n", ref_name);
return 1;
}
r.refid = refid;
r.start = start;
r.end = end;
cram_set_option(fd, CRAM_OPT_RANGE, &r);
}
/* SAM Header */
if (!(arg_list = stringify_argv(argc, argv)))
return 1;
sam_hdr_add_PG(bfd->header, "cram_to_sam",
"VN", PACKAGE_VERSION,
"CL", arg_list, NULL);
free(arg_list);
bam_write_header(bfd);
while (cram_get_bam_seq(fd, &bam) == 0) {
bam_put_seq(bfd, bam);
}
if (!cram_eof(fd)) {
fprintf(stderr, "Error while reading file\n");
return 1;
}
cram_close(fd);
bfd->header = NULL;
bam_close(bfd);
free(bam);
return 0;
}
int main(int argc, char **argv) {
cram_fd *out;
bam_file_t *in;
bam_seq_t *s = NULL;
char *out_fn;
int level = '\0'; // nul terminate string => auto level
char out_mode[4];
int c, verbose = 0;
int s_opt = 0, S_opt = 0, embed_ref = 0;
char *arg_list, *ref_fn = NULL;
while ((c = getopt(argc, argv, "u0123456789hvs:S:V:r:X")) != -1) {
switch (c) {
case '0': case '1': case '2': case '3': case '4':
case '5': case '6': case '7': case '8': case '9':
level = c;
break;
case 'u':
level = '0';
break;
case 'h':
usage(stdout);
return 0;
case 'v':
verbose++;
break;
case 's':
s_opt = atoi(optarg);
break;
case 'S':
S_opt = atoi(optarg);
break;
case 'V':
cram_set_option(NULL, CRAM_OPT_VERSION, optarg);
break;
case 'r':
ref_fn = optarg;
break;
case 'X':
embed_ref = 1;
break;
case '?':
fprintf(stderr, "Unrecognised option: -%c\n", optopt);
usage(stderr);
return 1;
}
}
if (argc - optind != 1 && argc - optind != 2) {
usage(stderr);
return 1;
}
/* opening */
if (NULL == (in = bam_open(argv[optind], "rb"))) {
perror(argv[optind]);
return 1;
}
out_fn = argc - optind == 2 ? argv[optind+1] : "-";
sprintf(out_mode, "wb%c", level);
if (NULL == (out = cram_open(out_fn, out_mode))) {
fprintf(stderr, "Error opening CRAM file '%s'.\n", out_fn);
return 1;
}
/* SAM Header */
if (!(arg_list = stringify_argv(argc, argv)))
return 1;
sam_hdr_add_PG(in->header, "sam_to_cram",
"VN", PACKAGE_VERSION,
"CL", arg_list, NULL);
free(arg_list);
/* Find and load reference */
if (!ref_fn) {
SAM_hdr_type *ty = sam_hdr_find(in->header, "SQ", NULL, NULL);
if (ty) {
SAM_hdr_tag *tag;
if ((tag = sam_hdr_find_key(in->header, ty, "UR", NULL))) {
ref_fn = tag->str + 3;
if (strncmp(ref_fn, "file:", 5) == 0)
ref_fn += 5;
}
}
}
out->header = in->header;
if (ref_fn)
cram_load_reference(out, ref_fn);
if (!out->refs) {
fprintf(stderr, "Unable to open reference.\n"
"Please specify a valid reference with -r ref.fa option.\n");
return 1;
}
refs2id(out->refs, out->header);
if (-1 == cram_write_SAM_hdr(out, in->header))
return 1;
cram_set_option(out, CRAM_OPT_VERBOSITY, verbose);
if (s_opt)
cram_set_option(out, CRAM_OPT_SEQS_PER_SLICE, s_opt);
if (S_opt)
cram_set_option(out, CRAM_OPT_SLICES_PER_CONTAINER, S_opt);
if (embed_ref)
cram_set_option(out, CRAM_OPT_EMBED_REF, embed_ref);
/* Sequence iterators */
while (bam_get_seq(in, &s) > 0) {
if (-1 == cram_put_bam_seq(out, s)) {
fprintf(stderr, "Failed in cram_put_bam_seq()\n");
return 1;
}
}
bam_close(in);
out->header = NULL; // freed by bam_close()
if (-1 == cram_close(out)) {
fprintf(stderr, "Failed in cram_close()\n");
return 1;
}
if (s)
free(s);
return 0;
}
int main(int argc, char **argv) {
scram_fd *in, *out;
bam_seq_t *s;
char imode[10], *in_f = "", omode[10], *out_f = "";
int level = '\0'; // nul terminate string => auto level
int c, verbose = 0;
int s_opt = 0, S_opt = 0, embed_ref = 0, ignore_md5 = 0, decode_md = 0;
char *ref_fn = NULL;
int start, end, multi_seq = -1, no_ref = 0;
int use_bz2 = 0, use_arith = 0, use_lzma = 0;
char ref_name[1024] = {0};
refs_t *refs;
int nthreads = 1;
t_pool *p = NULL;
int max_reads = -1;
enum quality_binning binning = BINNING_NONE;
/* Parse command line arguments */
while ((c = getopt(argc, argv, "u0123456789hvs:S:V:r:xXeI:O:R:!MmjJZt:BN:")) != -1) {
switch (c) {
case '0': case '1': case '2': case '3': case '4':
case '5': case '6': case '7': case '8': case '9':
level = c;
break;
case 'u':
level = '0';
break;
case 'h':
usage(stdout);
return 0;
case 'v':
verbose++;
break;
case 's':
s_opt = atoi(optarg);
break;
case 'S':
S_opt = atoi(optarg);
break;
case 'm':
decode_md = 1;
break;
case 'V':
if (cram_set_option(NULL, CRAM_OPT_VERSION, optarg))
return 1;
break;
case 'r':
ref_fn = optarg;
break;
case 'X':
fprintf(stderr, "-X is deprecated in favour of -e.\n");
case 'e':
embed_ref = 1;
break;
case 'x':
no_ref = 1;
break;
case 'I':
in_f = parse_format(optarg);
break;
case 'O':
out_f = parse_format(optarg);
break;
case 'R': {
char *cp = strchr(optarg, ':');
if (cp) {
*cp = 0;
switch (sscanf(cp+1, "%d-%d", &start, &end)) {
case 1:
end = start;
break;
case 2:
break;
default:
fprintf(stderr, "Malformed range format\n");
return 1;
}
} else {
start = INT_MIN;
end = INT_MAX;
}
strncpy(ref_name, optarg, 1023);
break;
}
case '!':
ignore_md5 = 1;
break;
case 'M':
multi_seq = 1;
break;
case 'j':
#ifdef HAVE_LIBBZ2
use_bz2 = 1;
#else
fprintf(stderr, "Warning: bzip2 support is not compiled into this"
" version.\nPlease recompile.\n");
#endif
break;
case 'J':
use_arith = 1;
break;
case 'Z':
#ifdef HAVE_LIBLZMA
use_lzma = 1;
#else
fprintf(stderr, "Warning: lzma support is not compiled into this"
" version.\nPlease recompile.\n");
#endif
break;
case 't':
nthreads = atoi(optarg);
if (nthreads < 1) {
fprintf(stderr, "Number of threads needs to be >= 1\n");
return 1;
}
break;
case 'B':
binning = BINNING_ILLUMINA;
break;
case 'N': // For debugging
max_reads = atoi(optarg);
break;
case '?':
fprintf(stderr, "Unrecognised option: -%c\n", optopt);
usage(stderr);
return 1;
}
}
if (argc - optind > 2) {
fprintf(stderr, "Usage: scramble [input_file [output_file]]\n");
return 1;
}
/* Open up input and output files */
sprintf(imode, "r%s%c", in_f, level);
if (argc - optind > 0) {
if (*in_f == 0)
sprintf(imode, "r%s%c", detect_format(argv[optind]), level);
if (!(in = scram_open(argv[optind], imode))) {
fprintf(stderr, "Failed to open file %s\n", argv[optind]);
return 1;
}
} else {
if (!(in = scram_open("-", imode))) {
fprintf(stderr, "Failed to open file %s\n", argv[optind]);
return 1;
}
}
if (!in->is_bam && ref_fn) {
cram_load_reference(in->c, ref_fn);
if (!in->c->refs && !embed_ref) {
fprintf(stderr, "Unable to find an appropriate reference.\n"
"Please specify a valid reference with "
"-r ref.fa option.\n");
return 1;
}
}
sprintf(omode, "w%s%c", out_f, level);
if (argc - optind > 1) {
if (*out_f == 0)
sprintf(omode, "w%s%c", detect_format(argv[optind+1]), level);
if (!(out = scram_open(argv[optind+1], omode))) {
fprintf(stderr, "Failed to open file %s\n", argv[optind+1]);
return 1;
}
} else {
if (!(out = scram_open("-", omode))) {
fprintf(stderr, "Failed to open file %s\n", argv[optind+1]);
return 1;
}
}
/* Set any format specific options */
scram_set_refs(out, refs = scram_get_refs(in));
scram_set_option(out, CRAM_OPT_VERBOSITY, verbose);
if (s_opt)
if (scram_set_option(out, CRAM_OPT_SEQS_PER_SLICE, s_opt))
return 1;
if (S_opt)
if (scram_set_option(out, CRAM_OPT_SLICES_PER_CONTAINER, S_opt))
return 1;
if (embed_ref)
if (scram_set_option(out, CRAM_OPT_EMBED_REF, embed_ref))
return 1;
if (use_bz2)
if (scram_set_option(out, CRAM_OPT_USE_BZIP2, use_bz2))
return 1;
if (use_arith)
if (scram_set_option(out, CRAM_OPT_USE_ARITH, use_arith))
return 1;
if (use_lzma)
if (scram_set_option(out, CRAM_OPT_USE_LZMA, use_lzma))
return 1;
if (binning != BINNING_NONE)
if (scram_set_option(out, CRAM_OPT_BINNING, binning))
return 1;
if (no_ref)
if (scram_set_option(out, CRAM_OPT_NO_REF, no_ref))
return 1;
if (multi_seq)
if (scram_set_option(out, CRAM_OPT_MULTI_SEQ_PER_SLICE, multi_seq))
return 1;
if (decode_md) {
if (no_ref) {
fprintf(stderr, "Cannot use -m in conjunction with -x.\n");
return 1;
}
if (scram_set_option(in, CRAM_OPT_DECODE_MD, decode_md))
return 1;
}
if (nthreads > 1) {
if (NULL == (p = t_pool_init(nthreads*2, nthreads)))
return 1;
if (scram_set_option(in, CRAM_OPT_THREAD_POOL, p))
return 1;
if (scram_set_option(out, CRAM_OPT_THREAD_POOL, p))
return 1;
}
if (ignore_md5)
if (scram_set_option(in, CRAM_OPT_IGNORE_MD5, ignore_md5))
return 1;
/* Copy header and refs from in to out, for writing purposes */
scram_set_header(out, scram_get_header(in));
// Needs doing after loading the header.
if (ref_fn) {
if (scram_set_option(out, CRAM_OPT_REFERENCE, ref_fn))
return 1;
} else {
// Attempt to fill out a cram->refs[] array from @SQ headers
scram_set_option(out, CRAM_OPT_REFERENCE, NULL);
}
if (scram_get_header(out)) {
char *arg_list = stringify_argv(argc, argv);
if (!arg_list)
return 1;
if (sam_hdr_add_PG(scram_get_header(out), "scramble",
"VN", PACKAGE_VERSION,
"CL", arg_list, NULL))
return 1;
if (scram_write_header(out))
return 1;
free(arg_list);
}
/* Support for sub-range queries, currently implemented for CRAM only */
if (*ref_name != 0) {
cram_range r;
int refid;
if (in->is_bam) {
fprintf(stderr, "Currently the -R option is only implemented for CRAM indices\n");
return 1;
}
cram_index_load(in->c, argv[optind]);
refid = sam_hdr_name2ref(in->c->header, ref_name);
if (refid == -1 && *ref_name != '*') {
fprintf(stderr, "Unknown reference name '%s'\n", ref_name);
return 1;
}
r.refid = refid;
r.start = start;
r.end = end;
if (scram_set_option(in, CRAM_OPT_RANGE, &r))
return 1;
}
/* Do the actual file format conversion */
s = NULL;
while (scram_get_seq(in, &s) >= 0) {
if (-1 == scram_put_seq(out, s)) {
fprintf(stderr, "Failed to encode sequence\n");
return 1;
}
if (max_reads >= 0)
if (--max_reads == 0)
break;
}
if (max_reads == -1) {
switch(scram_eof(in)) {
case 0:
fprintf(stderr, "Failed to decode sequence\n");
return 1;
case 2:
fprintf(stderr, "Warning: no end-of-file block identified. "
"File may be truncated.\n");
break;
case 1: default:
// expected case
break;
}
}
/* Finally tidy up and close files */
if (scram_close(in))
return 1;
if (scram_close(out))
return 1;
if (p)
t_pool_destroy(p, 0);
if (s)
free(s);
return 0;
}
int main_reheader(int argc, char *argv[])
{
int inplace = 0, r, add_PG = 1, c;
bam_hdr_t *h;
samFile *in;
char *arg_list = stringify_argv(argc+1, argv-1);
static const struct option lopts[] = {
{"help", no_argument, NULL, 'h'},
{"in-place", no_argument, NULL, 'i'},
{"no-PG", no_argument, NULL, 'P'},
{NULL, 0, NULL, 0}
};
while ((c = getopt_long(argc, argv, "hiP", lopts, NULL)) >= 0) {
switch (c) {
case 'P': add_PG = 0; break;
case 'i': inplace = 1; break;
case 'h': usage(stdout, 0); break;
default:
fprintf(stderr, "Invalid option '%c'\n", c);
usage(stderr, 1);
}
}
if (argc - optind != 2)
usage(stderr, 1);
{ // read the header
samFile *fph = sam_open(argv[optind], "r");
if (fph == 0) {
fprintf(stderr, "[%s] fail to read the header from %s.\n", __func__, argv[optind]);
return 1;
}
h = sam_hdr_read(fph);
sam_close(fph);
if (h == NULL) {
fprintf(stderr, "[%s] failed to read the header for '%s'.\n",
__func__, argv[1]);
return 1;
}
}
in = sam_open(argv[optind+1], inplace?"r+":"r");
if (in == 0) {
fprintf(stderr, "[%s] fail to open file %s.\n", __func__, argv[optind+1]);
return 1;
}
if (hts_get_format(in)->format == bam) {
r = bam_reheader(in->fp.bgzf, h, fileno(stdout), arg_list, add_PG);
} else {
if (inplace)
r = cram_reheader_inplace(in->fp.cram, h, arg_list, add_PG);
else
r = cram_reheader(in->fp.cram, h, arg_list, add_PG);
}
if (sam_close(in) != 0)
r = -1;
bam_hdr_destroy(h);
if (arg_list)
free(arg_list);
return -r;
}
