/**
* returns 0 on success, -1 on failure.
*/
int writeToFile(scram_fd* fp) {
int r1 = scram_put_seq(fp, read1);
if (r1 == 0 and isPaired()) {
return scram_put_seq(fp, read2);
} else {
return r1;
}
}
int main(int argc, char **argv) {
scram_fd **in, *out;
int n_input, i;
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;
char *ref_fn = NULL;
int start, end;
char ref_name[1024] = {0};
refs_t *refs = NULL;
/* Parse command line arguments */
while ((c = getopt(argc, argv, "u0123456789hvs:S:V:r:XI:O:R:")) != -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 '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 '?':
fprintf(stderr, "Unrecognised option: -%c\n", optopt);
usage(stderr);
return 1;
}
}
/* Open output file */
sprintf(omode, "w%s%c", out_f, level);
if (!(out = scram_open("-", omode))) {
fprintf(stderr, "Failed to open bam file %s\n", argv[optind+1]);
return 1;
}
/* Open multiple input files */
sprintf(imode, "r%s%c", in_f, level);
n_input = argc - optind;
if (!n_input) {
fprintf(stderr, "No input files specified.\n");
return 1;
}
if (!(in = malloc(n_input * sizeof(*in))))
return 1;
if (!(s = malloc(n_input * sizeof(*s))))
return 1;
for (i = 0; i < n_input; i++, optind++) {
s[i] = NULL;
if (*in_f == 0)
sprintf(imode, "r%s%c", detect_format(argv[optind]), level);
if (!(in[i] = scram_open(argv[optind], imode))) {
fprintf(stderr, "Failed to open bam file %s\n", argv[optind]);
return 1;
}
if (i && !hdr_compare(scram_get_header(in[0]),
scram_get_header(in[i]))) {
fprintf(stderr, "Incompatible reference sequence list.\n");
fprintf(stderr, "Currently the @SQ lines need to be identical"
" in all files.\n");
return 1;
}
if (!refs && scram_get_refs(in[i]))
refs = scram_get_refs(in[i]);
if (refs && scram_set_option(in[i], CRAM_OPT_SHARED_REF, refs))
return 1;
}
/* Set any format specific options */
if (refs)
scram_set_option(out, CRAM_OPT_SHARED_REF, refs);
if (scram_set_option(out, CRAM_OPT_VERBOSITY, verbose))
return 1;
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;
/* Copy header and refs from in to out, for writing purposes */
// FIXME: do proper merging of @PG lines
// FIXME: track mapping of old PG aux name to new PG aux name per seq
scram_set_header(out, sam_hdr_dup(scram_get_header(in[0])));
// Needs doing after loading the header.
if (ref_fn)
if (scram_set_option(out, CRAM_OPT_REFERENCE, ref_fn))
return 1;
if (scram_get_header(in[0])) {
if (scram_write_header(out))
return 1;
}
/* Do the actual file format conversion */
fprintf(stderr, "Opening and loading initial seqs\n");
for (i = 0; i < n_input; i++) {
if (scram_get_seq(in[i], &s[i]) < 0) {
if (scram_close(in[i]))
return 1;
in[i] = NULL;
free(s[i]);
continue;
}
}
fprintf(stderr, "Merging...\n");
for (;;) {
int64_t best_val = INT64_MAX;
int best_j = 0, j;
for (j = 0; j < n_input; j++) {
bam_seq_t *b = s[j];
uint64_t x;
if (!in[j])
continue;
x = (((uint64_t)bam_ref(b))<<33)
| (bam_pos(b)<<2)
| (bam_strand(b)<<1)
| !(bam_flag(b) & BAM_FREAD1);
if (best_val > x) {
best_val = x;
best_j = j;
}
}
if (best_val == INT64_MAX) { // all closed
break;
}
if (-1 == scram_put_seq(out, s[best_j]))
return 1;
if (scram_get_seq(in[best_j], &s[best_j]) < 0) {
if (scram_close(in[best_j]))
return 1;
in[best_j] = NULL;
free(s[best_j]);
}
}
for (i = 0; i < n_input; i++) {
if (!in[i])
continue;
scram_close(in[i]);
if (s[i])
free(s[i]);
}
/* Finally tidy up and close files */
if (scram_close(out))
return 1;
free(in);
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;
}
// return 0 on success, -1 on failure
int writeToFile(scram_fd* fp) {
return scram_put_seq(fp, read);
}
