char* gt_shredder_shred(GtShredder *shredder, unsigned long *fragment_length,
GtStr *desc)
{
char *frag;
gt_assert(shredder && fragment_length);
while ((frag = generate_fragment(shredder, fragment_length, desc))) {
if (shredder->sample_probability == 1.0 ||
gt_rand_0_to_1() <= shredder->sample_probability) {
return frag;
} else gt_free(frag);
}
return NULL;
}
static int gt_seqfilter_runner(int argc, const char **argv, int parsed_args,
void *tool_arguments, GtError *err)
{
SeqFilterArguments *arguments = tool_arguments;
GtBioseqIterator *bsi;
GtBioseq *bioseq;
GtUint64 passed = 0, filtered = 0, num_of_sequences = 0, steps = 0;
int had_err = 0;
gt_error_check(err);
gt_assert(tool_arguments);
bsi = gt_bioseq_iterator_new(argc - parsed_args, argv + parsed_args);
while (!(had_err = gt_bioseq_iterator_next(bsi, &bioseq, err)) &&
bioseq != NULL) {
GtUword i;
GtUint64 current_num = gt_bioseq_number_of_sequences(bioseq);
for (i = 0;
i < current_num &&
(arguments->maxseqnum == GT_UNDEF_UWORD ||
passed + 1 <= arguments->maxseqnum);
i++) {
char *seq;
if ((arguments->step == 1 ||
steps + 1 == arguments->step) &&
(arguments->sample_prob == 1.0 ||
gt_rand_0_to_1() <= arguments->sample_prob) &&
(arguments->minlength == GT_UNDEF_UWORD ||
gt_bioseq_get_sequence_length(bioseq, i) >= arguments->minlength) &&
(arguments->maxlength == GT_UNDEF_UWORD ||
gt_bioseq_get_sequence_length(bioseq, i) <= arguments->maxlength)) {
seq = gt_bioseq_get_sequence(bioseq, i);
gt_fasta_show_entry(gt_bioseq_get_description(bioseq, i),
seq,
gt_bioseq_get_sequence_length(bioseq, i),
arguments->width, arguments->outfp);
gt_free(seq);
passed++;
}
else {
filtered++;
}
steps = (steps + 1 == arguments->step) ? 0 : steps + 1;
}
filtered += current_num - i;
num_of_sequences += current_num;
gt_bioseq_delete(bioseq);
}
/* show statistics */
if (!had_err) {
gt_assert(passed + filtered == num_of_sequences);
fprintf(stderr, "# " GT_LLU " out of " GT_LLU
" sequences have been removed (%.3f%%)\n",
filtered, num_of_sequences,
((double) filtered / num_of_sequences) * 100.0);
}
gt_bioseq_iterator_delete(bsi);
return had_err;
}