int gth_match_processor(GthMatchProcessorInfo *info, GthSeqCon *gen_seq_con,
GthSeqCon *ref_seq_con, GthMatch *match)
{
if (info->matchnumcounter) {
info->matchnumcounter[match->Storeseqnumreference]++;
if (info->maxnumofmatches > 0 &&
info->matchnumcounter[match->Storeseqnumreference] >
info->maxnumofmatches) {
/* discard matchA */
return 0;
}
}
if (!(info->refseqisindex && !info->online) &&
match->Storeseqnumreference != info->lastrefseqnum &&
gt_array_size(info->matches)) {
gt_assert(info->chain_collection && info->chaining_info);
/* chain all current matches */
calc_chains_from_matches(info->chain_collection, info->matches,
info->chaining_info, gen_seq_con, ref_seq_con,
info->rare, info->fragweightfactor,
info->jump_table_new, info->jump_table_new_reverse,
info->jump_table_delete);
/* and remove them afterwards */
gt_array_reset(info->matches);
}
/*...only if it does not equal the last one */
if (gt_array_size(info->matches) &&
gth_matches_are_equal(gt_array_get_last(info->matches), match)) {
return 0;
}
gt_array_add_elem(info->matches, match, sizeof *match);
/* update last reference sequence number */
info->lastrefseqnum = match->Storeseqnumreference;
return 0;
}
static int add_auto_sr_to_queue(GT_UNUSED void *key, void *value, void *data,
GT_UNUSED GtError *err)
{
AutomaticSequenceRegion *auto_sr = value;
GtQueue *genome_nodes = data;
GtGenomeNode *gf;
unsigned int i;
gt_error_check(err);
gt_assert(key && value && data);
if (gt_array_size(auto_sr->feature_nodes)) {
gt_queue_add(genome_nodes, auto_sr->sequence_region);
auto_sr->sequence_region = NULL;
for (i = 0; i < gt_array_size(auto_sr->feature_nodes); i++) {
gf = *(GtGenomeNode**) gt_array_get(auto_sr->feature_nodes, i);
gt_queue_add(genome_nodes, gf);
}
gt_array_reset(auto_sr->feature_nodes);
}
return 0;
}
static int gt_load_stream_next(GtNodeStream *ns, GtGenomeNode **gn,
GtError *err)
{
GtLoadStream *load_stream;
GtGenomeNode *node, *eofn;
int had_err = 0;
gt_error_check(err);
load_stream = gt_load_stream_cast(ns);
if (!load_stream->full) {
while (!(had_err = gt_node_stream_next(load_stream->in_stream, &node,
err)) && node) {
if ((eofn = gt_eof_node_try_cast(node)))
gt_genome_node_delete(eofn); /* get rid of EOF nodes */
else
gt_array_add(load_stream->nodes, node);
}
if (!had_err) {
load_stream->full = true;
}
}
if (!had_err) {
gt_assert(load_stream->full);
if (load_stream->idx < gt_array_size(load_stream->nodes)) {
*gn = *(GtGenomeNode**) gt_array_get(load_stream->nodes,
load_stream->idx);
load_stream->idx++;
return 0;
}
}
if (!had_err) {
gt_array_reset(load_stream->nodes);
*gn = NULL;
}
return had_err;
}
int gt_range_unit_test(GtError *err)
{
static GtRange ranges_in[] = { { 620432, 620536 }, { 620432, 620536 },
{ 620957, 621056 }, { 620957, 621056 },
{ 625234, 625253 }, { 625500, 625655 },
{ 625533, 625655 }, { 625533, 625655 },
{ 627618, 627729 }, { 627618, 627729 },
{ 627618, 627729 }, { 662083, 662194 },
{ 662083, 662194 }, { 662083, 662194 },
{ 663032, 663166 }, { 663032, 663166 },
{ 663032, 663166 }, { 664782, 664906 },
{ 664782, 664906 }, { 664782, 664906 },
{ 665748, 665823 }, { 665748, 665823 },
{ 665748, 665823 }, { 666825, 666881 },
{ 666825, 666881 }, { 667797, 667954 },
{ 667845, 667954 }, { 667845, 667954 },
{ 679175, 679280 }, { 679175, 679280 },
{ 679175, 679280 }, { 680427, 680540 },
{ 680427, 680540 }, { 680427, 680540 },
{ 684144, 684293 }, { 684144, 684293 },
{ 684144, 684293 }, { 724903, 724985 },
{ 724903, 724985 }, { 727099, 727325 },
{ 727099, 727325 }, { 732544, 732821 },
{ 732544, 732821 }, { 750016, 750280 },
{ 750016, 750280 }, { 769508, 769734 },
{ 769508, 769734 } },
ranges_out[] = { { 620432, 620536 }, { 620957, 621056 },
{ 625234, 625253 }, { 625500, 625655 },
{ 625533, 625655 }, { 627618, 627729 },
{ 662083, 662194 }, { 663032, 663166 },
{ 664782, 664906 }, { 665748, 665823 },
{ 666825, 666881 }, { 667797, 667954 },
{ 667845, 667954 }, { 679175, 679280 },
{ 680427, 680540 }, { 684144, 684293 },
{ 724903, 724985 }, { 727099, 727325 },
{ 732544, 732821 }, { 750016, 750280 },
{ 769508, 769734 }};
GtUword counts[] = { 2, 2, 1, 1, 2, 3, 3, 3, 3, 3, 2, 1, 2, 3, 3, 3, 2,
2, 2, 2, 2 };
GtArray *ranges, *tmp_ranges, *ctr;
GtUword i;
int had_err = 0;
gt_error_check(err);
gt_ensure(sizeof (ranges_out) / sizeof (ranges_out[0]) ==
sizeof (counts) / sizeof (counts[0]));
/* test gt_ranges_uniq() */
ranges = gt_array_new(sizeof (GtRange));
tmp_ranges = gt_array_new(sizeof (GtRange));
for (i = 0;
i < sizeof (ranges_in) / sizeof (ranges_in[0]) && !had_err;
i++)
gt_array_add(ranges, ranges_in[i]);
gt_ranges_uniq(tmp_ranges, ranges);
gt_ensure(gt_array_size(ranges) ==
sizeof (ranges_in) / sizeof (ranges_in[0]));
gt_ensure(gt_array_size(tmp_ranges) ==
sizeof (ranges_out) / sizeof (ranges_out[0]));
for (i = 0; i < gt_array_size(tmp_ranges) && !had_err; i++) {
gt_ensure(ranges_out[i].start ==
(*(GtRange*) gt_array_get(tmp_ranges, i)).start);
gt_ensure(ranges_out[i].end ==
(*(GtRange*) gt_array_get(tmp_ranges, i)).end);
}
/* test gt_ranges_uniq_in_place() */
gt_array_reset(tmp_ranges);
gt_array_add_array(tmp_ranges, ranges);
gt_ranges_uniq_in_place(tmp_ranges);
for (i = 0; i < gt_array_size(tmp_ranges) && !had_err; i++) {
gt_ensure(ranges_out[i].start ==
(*(GtRange*) gt_array_get(tmp_ranges, i)).start);
gt_ensure(ranges_out[i].end ==
(*(GtRange*) gt_array_get(tmp_ranges, i)).end);
}
/* test gt_ranges_uniq_count() */
gt_array_reset(tmp_ranges);
ctr = gt_ranges_uniq_count(tmp_ranges, ranges);
gt_ensure(gt_array_size(tmp_ranges) == gt_array_size(ctr));
gt_ensure(
gt_array_size(ctr) == sizeof (counts) / sizeof (counts[0]));
for (i = 0; i < gt_array_size(ctr) && !had_err; i++) {
gt_ensure(counts[i] == *(GtUword*) gt_array_get(ctr, i));
gt_ensure(ranges_out[i].start ==
(*(GtRange*) gt_array_get(tmp_ranges, i)).start);
gt_ensure(ranges_out[i].end ==
(*(GtRange*) gt_array_get(tmp_ranges, i)).end);
}
gt_array_delete(ctr);
/* test gt_ranges_uniq_in_place_count() */
ctr = gt_ranges_uniq_in_place_count(ranges);
gt_ensure(gt_array_size(ranges) == gt_array_size(ctr));
gt_ensure(
gt_array_size(ctr) == sizeof (counts) / sizeof (counts[0]));
for (i = 0; i < gt_array_size(ctr) && !had_err; i++) {
gt_ensure(counts[i] == *(GtUword*) gt_array_get(ctr, i));
gt_ensure(
ranges_out[i].start == (*(GtRange*)
gt_array_get(ranges, i)).start);
gt_ensure(
ranges_out[i].end == (*(GtRange*) gt_array_get(ranges, i)).end);
}
gt_array_delete(ctr);
/* test gt_range_reorder() */
if (!had_err) {
GtRange range = { 1, 100 };
range = gt_range_reorder(range);
gt_ensure(range.start == 1 && range.end == 100);
range.start = 100;
range.end = 1;
range = gt_range_reorder(range);
gt_ensure(range.start == 1 && range.end == 100);
}
/* free */
gt_array_delete(ranges);
gt_array_delete(tmp_ranges);
return had_err;
}
static int hcr_next_seq_qual(GtHcrSeqDecoder *seq_dec, char *seq, char *qual,
GtError *err)
{
enum state {
HCR_ERROR = -1,
END,
SUCCESS
};
unsigned char base;
GtUword i,
nearestsample,
*symbol;
size_t startofnearestsample = 0;
enum state status = END;
FastqFileInfo cur_read;
FastqFileInfo *fileinfo = NULL;
if (seq_dec->cur_read <= seq_dec->num_of_reads) {
status = SUCCESS;
if (seq_dec->symbols == NULL)
seq_dec->symbols = gt_array_new(sizeof (GtUword));
else
gt_array_reset(seq_dec->symbols);
cur_read.readnum = seq_dec->cur_read;
gt_log_log("cur_read: "GT_WU"",seq_dec->cur_read);
fileinfo = (FastqFileInfo *)gt_rbtree_next_key(seq_dec->file_info_rbt,
&cur_read,
hcr_cmp_FastqFileInfo,
NULL);
gt_assert(fileinfo);
/* reset huffman_decoder if next read is sampled */
if (gt_sampling_get_next_elementnum(seq_dec->sampling) ==
seq_dec->cur_read) {
gt_log_log("reset because sampled read is next");
(void) gt_sampling_get_next_sample(seq_dec->sampling,
&nearestsample,
&startofnearestsample);
reset_data_iterator_to_pos(seq_dec->data_iter, startofnearestsample);
(void) gt_huffman_decoder_get_new_mem_chunk(seq_dec->huff_dec, err);
if (gt_error_is_set(err))
status = HCR_ERROR;
}
if (status != HCR_ERROR) {
int ret;
ret = gt_huffman_decoder_next(seq_dec->huff_dec, seq_dec->symbols,
fileinfo->readlength, err);
if (ret != 1)
status = HCR_ERROR;
if (ret == 0)
gt_error_set(err, "reached end of file");
}
if (qual || seq) {
gt_log_log("set strings");
for (i = 0; i < gt_array_size(seq_dec->symbols); i++) {
symbol = (GtUword*) gt_array_get(seq_dec->symbols, i);
if (qual != NULL)
qual[i] = get_qual_from_symbol(seq_dec, *symbol);
if (seq != NULL) {
base = get_base_from_symbol(seq_dec, *symbol);
seq[i] = (char)toupper(gt_alphabet_decode(seq_dec->alpha,
(GtUchar) base));
}
}
if (qual != NULL)
qual[gt_array_size(seq_dec->symbols)] = '\0';
if (seq != NULL)
seq[gt_array_size(seq_dec->symbols)] = '\0';
}
seq_dec->cur_read++;
}
return (int) status;
}
static void compute_csas(ConsensusSA *csa)
{
unsigned long i, sa_i, sa_i_size = 0, sa_prime, sa_prime_size;
GtArray *splice_form;
GtBittab **C, **left, **right, **L, **R, *U_i, *SA_i, *SA_prime;
#ifndef NDEBUG
unsigned long u_i_size, u_i_minus_1_size;
gt_assert(csa && csa->set_of_sas);
#endif
/* init sets */
C = gt_malloc(sizeof (GtBittab*) * csa->number_of_sas);
left = gt_malloc(sizeof (GtBittab*) * csa->number_of_sas);
right = gt_malloc(sizeof (GtBittab*) * csa->number_of_sas);
L = gt_malloc(sizeof (GtBittab*) * csa->number_of_sas);
R = gt_malloc(sizeof (GtBittab*) * csa->number_of_sas);
for (i = 0; i < csa->number_of_sas; i++) {
C[i] = gt_bittab_new(csa->number_of_sas);
left[i] = gt_bittab_new(csa->number_of_sas);
right[i] = gt_bittab_new(csa->number_of_sas);
L[i] = gt_bittab_new(csa->number_of_sas);
R[i] = gt_bittab_new(csa->number_of_sas);
}
U_i = gt_bittab_new(csa->number_of_sas);
SA_i = gt_bittab_new(csa->number_of_sas);
SA_prime = gt_bittab_new(csa->number_of_sas);
splice_form = gt_array_new(sizeof (unsigned long));
/* compute sets */
compute_C(C, csa);
compute_left(left, csa);
compute_right(right, csa);
compute_L(L, C, left, csa->number_of_sas);
compute_R(R, C, right, csa->number_of_sas);
/* U_0 = SA */
for (i = 0; i < csa->number_of_sas; i++)
gt_bittab_set_bit(U_i, i);
#ifndef NDEBUG
/* preparation for assertion below */
u_i_minus_1_size = gt_bittab_count_set_bits(U_i);
#endif
while (gt_bittab_is_true(U_i)) {
sa_i = GT_UNDEF_ULONG;
for (sa_prime = gt_bittab_get_first_bitnum(U_i);
sa_prime != gt_bittab_get_last_bitnum(U_i);
sa_prime = gt_bittab_get_next_bitnum(U_i, sa_prime)) {
if (sa_i == GT_UNDEF_ULONG) {
sa_i = sa_prime;
gt_bittab_or(SA_i, L[sa_i], R[sa_i]);
sa_i_size = gt_bittab_count_set_bits(SA_i);
}
else {
gt_bittab_or(SA_prime, L[sa_prime], R[sa_prime]);
sa_prime_size = gt_bittab_count_set_bits(SA_prime);
if (sa_prime_size > sa_i_size) {
sa_i = sa_prime;
sa_i_size = sa_prime_size;
gt_bittab_equal(SA_i, SA_prime);
}
}
}
/* make sure the computed splice form is maximal w.r.t. to compatibility */
gt_assert(splice_form_is_valid(SA_i, csa));
/* process splice form */
if (csa->process_splice_form) {
gt_array_reset(splice_form);
gt_bittab_get_all_bitnums(SA_i, splice_form);
csa->process_splice_form(splice_form, csa->set_of_sas, csa->number_of_sas,
csa->size_of_sa, csa->userdata);
}
/* U_i = U_i-1 \ SA_i */
gt_bittab_nand(U_i, U_i, SA_i);
#ifndef NDEBUG
/* ensure that |U_i| < |U_i-1| */
u_i_size = gt_bittab_count_set_bits(U_i);
gt_assert(u_i_size < u_i_minus_1_size);
u_i_minus_1_size = u_i_size;
#endif
}
/* free sets */
for (i = 0; i < csa->number_of_sas; i++) {
gt_bittab_delete(C[i]);
gt_bittab_delete(left[i]);
gt_bittab_delete(right[i]);
gt_bittab_delete(L[i]);
gt_bittab_delete(R[i]);
}
gt_free(C);
gt_free(left);
gt_free(right);
gt_free(L);
gt_free(R);
gt_bittab_delete(U_i);
gt_bittab_delete(SA_i);
gt_bittab_delete(SA_prime);
gt_array_delete(splice_form);
}
int gt_ovlfind_kmp_unit_test(GtError *err)
{
int had_err = 0;
GtArray *a;
struct GtOvlfindKmpResult *r;
GtContfind retval;
gt_kmp_t *u_pi, *v_pi;
/*@i1@*/ gt_error_check(err);
had_err = gt_kmp_preproc_unit_test(err);
if (had_err != 0)
return had_err;
a = gt_array_new(sizeof (struct GtOvlfindKmpResult));
/* u suffix == v prefix */
if (!had_err)
{
gt_array_reset(a);
u_pi = gt_kmp_preproc("aacgcacctg", 10UL);
v_pi = gt_kmp_preproc("acctgatttc", 10UL);
retval = gt_ovlfind_kmp("aacgcacctg", 10UL, u_pi, "acctgatttc", 10UL, v_pi,
GT_OVLFIND_PROPER_SPM, 1UL, false, ovlfind_kmp_test_save, a);
gt_ensure(had_err, retval == GT_CONTFIND_NO);
gt_ensure(had_err, gt_array_size(a) == 1UL);
GT_OVLFIND_KMP_EXPECT_RESULT(0UL, true, 5UL);
gt_free(u_pi);
gt_free(v_pi);
}
/* v suffix == u prefix */
if (!had_err)
{
gt_array_reset(a);
u_pi = gt_kmp_preproc("atccgtgacgtg", 12UL);
v_pi = gt_kmp_preproc("aagaagaatccg", 12UL);
retval = gt_ovlfind_kmp("atccgtgacgtg", 12UL, u_pi, "aagaagaatccg", 12UL,
v_pi, GT_OVLFIND_ALL, 1UL, false, ovlfind_kmp_test_save, a);
gt_ensure(had_err, retval == GT_CONTFIND_NO);
gt_ensure(had_err, gt_array_size(a) == 1UL);
GT_OVLFIND_KMP_EXPECT_RESULT(0UL, false, 5UL);
gt_free(u_pi);
gt_free(v_pi);
}
/* no overlap */
if (!had_err)
{
gt_array_reset(a);
u_pi = gt_kmp_preproc("aac", 3UL);
v_pi = gt_kmp_preproc("tgc", 3UL);
retval = gt_ovlfind_kmp("aac", 3UL, u_pi, "tgc", 3UL, v_pi,
GT_OVLFIND_PROPER_SPM, 1UL, false, ovlfind_kmp_test_save, a);
gt_ensure(had_err, retval == GT_CONTFIND_NO);
gt_ensure(had_err, gt_array_size(a) == 0UL);
gt_free(u_pi);
gt_free(v_pi);
}
/* u suffix of v */
if (!had_err)
{
gt_array_reset(a);
u_pi = gt_kmp_preproc("acagc", 5UL);
v_pi = gt_kmp_preproc("gtacagc", 7UL);
retval = gt_ovlfind_kmp("acagc", 5UL, u_pi, "gtacagc", 7UL, v_pi,
GT_OVLFIND_SPM, 1UL, false, ovlfind_kmp_test_save, a);
gt_ensure(had_err, gt_array_size(a) == 1UL);
gt_ensure(had_err, retval == GT_CONTFIND_OFF);
GT_OVLFIND_KMP_EXPECT_RESULT(0UL, false, 5UL);
gt_array_reset(a);
retval = gt_ovlfind_kmp("acagc", 5UL, u_pi, "gtacagc", 7UL, v_pi,
GT_OVLFIND_PROPER_SPM, 1UL, false, ovlfind_kmp_test_save, a);
gt_ensure(had_err, retval == GT_CONTFIND_U);
gt_ensure(had_err, gt_array_size(a) == 0UL);
gt_array_reset(a);
retval = gt_ovlfind_kmp("acagc", 5UL, u_pi, "gtacagc", 7UL, v_pi,
GT_OVLFIND_CNT, 1UL, false, ovlfind_kmp_test_save, a);
gt_ensure(had_err, gt_array_size(a) == 0UL);
gt_ensure(had_err, retval == GT_CONTFIND_U);
gt_array_reset(a);
retval = gt_ovlfind_kmp("acagc", 5UL, u_pi, "gtacagc", 7UL, v_pi,
GT_OVLFIND_ALL, 1UL, false, ovlfind_kmp_test_save, a);
gt_ensure(had_err, gt_array_size(a) == 1UL);
gt_ensure(had_err, retval == GT_CONTFIND_U);
GT_OVLFIND_KMP_EXPECT_RESULT(0UL, false, 5UL);
gt_free(u_pi);
gt_free(v_pi);
}
/* v suffix of u */
if (!had_err)
{
gt_array_reset(a);
u_pi = gt_kmp_preproc("gtacagc", 7UL);
v_pi = gt_kmp_preproc("acagc", 5UL);
retval = gt_ovlfind_kmp("gtacagc", 7UL, u_pi, "acagc", 5UL, v_pi,
GT_OVLFIND_SPM, 1UL, false, ovlfind_kmp_test_save, a);
gt_ensure(had_err, retval == GT_CONTFIND_OFF);
gt_ensure(had_err, gt_array_size(a) == 1UL);
GT_OVLFIND_KMP_EXPECT_RESULT(0UL, true, 5UL);
gt_array_reset(a);
retval = gt_ovlfind_kmp("gtacagc", 7UL, u_pi, "acagc", 5UL, v_pi,
GT_OVLFIND_PROPER_SPM, 1UL, false, ovlfind_kmp_test_save, a);
gt_ensure(had_err, retval == GT_CONTFIND_V);
gt_ensure(had_err, gt_array_size(a) == 0UL);
gt_free(u_pi);
gt_free(v_pi);
}
/* u prefix of v */
if (!had_err)
{
gt_array_reset(a);
u_pi = gt_kmp_preproc("ctat", 4UL);
v_pi = gt_kmp_preproc("ctatacagg", 9UL);
retval = gt_ovlfind_kmp("ctat", 4UL, u_pi, "ctatacagg", 9UL, v_pi,
GT_OVLFIND_SPM, 1UL, false, ovlfind_kmp_test_save, a);
gt_ensure(had_err, retval == GT_CONTFIND_OFF);
gt_ensure(had_err, gt_array_size(a) == 1UL);
GT_OVLFIND_KMP_EXPECT_RESULT(0UL, true, 4UL);
gt_array_reset(a);
retval = gt_ovlfind_kmp("ctat", 4UL, u_pi, "ctatacagg", 9UL, v_pi,
GT_OVLFIND_PROPER_SPM, 1UL, false, ovlfind_kmp_test_save, a);
gt_ensure(had_err, retval == GT_CONTFIND_U);
gt_ensure(had_err, gt_array_size(a) == 0UL);
gt_free(u_pi);
gt_free(v_pi);
}
/* v prefix of u */
if (!had_err)
{
gt_array_reset(a);
u_pi = gt_kmp_preproc("ctatacagg", 9UL);
v_pi = gt_kmp_preproc("ctat", 4UL);
retval = gt_ovlfind_kmp("ctatacagg", 9UL, u_pi, "ctat", 4UL, v_pi,
GT_OVLFIND_SPM, 1UL, false, ovlfind_kmp_test_save, a);
gt_ensure(had_err, retval == GT_CONTFIND_OFF);
gt_ensure(had_err, gt_array_size(a) == 1UL);
GT_OVLFIND_KMP_EXPECT_RESULT(0UL, false, 4UL);
gt_array_reset(a);
retval = gt_ovlfind_kmp("ctatacagg", 9UL, u_pi, "ctat", 4UL, v_pi,
GT_OVLFIND_PROPER_SPM, 1UL, false, ovlfind_kmp_test_save, a);
gt_ensure(had_err, retval == GT_CONTFIND_V);
gt_ensure(had_err, gt_array_size(a) == 0UL);
gt_free(u_pi);
gt_free(v_pi);
}
/* identical sequences */
if (!had_err)
{
gt_array_reset(a);
u_pi = gt_kmp_preproc("acagc", 5UL);
retval = gt_ovlfind_kmp("acagc", 5UL, u_pi, "acagc", 5UL, u_pi,
GT_OVLFIND_SPM, 1UL, false, ovlfind_kmp_test_save, a);
gt_ensure(had_err, retval == GT_CONTFIND_OFF);
gt_ensure(had_err, gt_array_size(a) == 2UL);
GT_OVLFIND_KMP_EXPECT_RESULT(0UL, true, 5UL);
GT_OVLFIND_KMP_EXPECT_RESULT(1UL, false, 5UL);
gt_array_reset(a);
retval = gt_ovlfind_kmp("acagc", 5UL, u_pi, "acagc", 5UL, u_pi,
GT_OVLFIND_PROPER_SPM, 1UL, false, ovlfind_kmp_test_save, a);
gt_ensure(had_err, retval == GT_CONTFIND_EQ);
gt_ensure(had_err, gt_array_size(a) == 0UL);
gt_free(u_pi);
}
/* find_nonmaximal */
if (!had_err)
{
gt_array_reset(a);
u_pi = gt_kmp_preproc("aacagtagtagt", 12UL);
v_pi = gt_kmp_preproc("agtagtagttaa", 12UL);
retval = gt_ovlfind_kmp("aacagtagtagt", 12UL, u_pi, "agtagtagttaa", 12UL,
v_pi, GT_OVLFIND_SPM, 1UL, false, ovlfind_kmp_test_save, a);
gt_ensure(had_err, retval == GT_CONTFIND_OFF);
gt_ensure(had_err, gt_array_size(a) == 2UL);
GT_OVLFIND_KMP_EXPECT_RESULT(0UL, true, 9UL);
GT_OVLFIND_KMP_EXPECT_RESULT(1UL, false, 2UL);
gt_array_reset(a);
retval = gt_ovlfind_kmp("aacagtagtagt", 12UL, u_pi, "agtagtagttaa", 12UL,
v_pi, GT_OVLFIND_SPM, 1UL, true, ovlfind_kmp_test_save, a);
gt_ensure(had_err, gt_array_size(a) == 5UL);
gt_ensure(had_err, retval == GT_CONTFIND_OFF);
gt_free(u_pi);
gt_free(v_pi);
}
/* min_length */
if (!had_err)
{
gt_array_reset(a);
u_pi = gt_kmp_preproc("aggaccagtagt", 12UL);
v_pi = gt_kmp_preproc("agtagttactac", 12UL);
retval = gt_ovlfind_kmp("aggaccagtagt", 12UL, u_pi, "agtagttactac", 12UL,
v_pi, GT_OVLFIND_SPM, 1UL, true, ovlfind_kmp_test_save, a);
gt_ensure(had_err, retval == GT_CONTFIND_OFF);
gt_ensure(had_err, gt_array_size(a) == 2UL);
gt_array_reset(a);
retval = gt_ovlfind_kmp("aggaccagtagt", 12UL, u_pi, "agtagttactac", 12UL,
v_pi, GT_OVLFIND_SPM, 4UL, true, ovlfind_kmp_test_save, a);
gt_ensure(had_err, gt_array_size(a) == 1UL);
gt_ensure(had_err, retval == GT_CONTFIND_OFF);
gt_free(u_pi);
gt_free(v_pi);
}
gt_array_delete(a);
return had_err;
}
int gt_string_matching_unit_test(GtError *err)
{
char s[STRING_MATCHING_MAX_STRING_LENGTH+1],
p[STRING_MATCHING_MAX_PATTERN_LENGTH+1], *text = "foo";
GtArray *brute_force_matches,
*bmh_matches,
*kmp_matches,
*shift_and_matches;
unsigned long i, brute_force_match, bmh_match, kmp_match, shift_and_match;
int had_err = 0;
gt_error_check(err);
brute_force_matches = gt_array_new(sizeof (unsigned long));
bmh_matches = gt_array_new(sizeof (unsigned long));
kmp_matches = gt_array_new(sizeof (unsigned long));
shift_and_matches = gt_array_new(sizeof (unsigned long));
/* match the empty pattern */
gt_string_matching_brute_force(text, strlen(text), "", 0, store_match,
brute_force_matches);
gt_string_matching_bmh(text, strlen(text), "", 0, store_match, bmh_matches);
gt_string_matching_kmp(text, strlen(text), "", 0, store_match, kmp_matches);
gt_string_matching_shift_and(text, strlen(text), "", 0, store_match,
shift_and_matches);
ensure(had_err, !gt_array_size(brute_force_matches));
ensure(had_err, !gt_array_size(bmh_matches));
ensure(had_err, !gt_array_size(kmp_matches));
ensure(had_err, !gt_array_size(shift_and_matches));
for (i = 0; !had_err && i < STRING_MATCHING_NUM_OF_TESTS; i++) {
unsigned long j, n, m;
/* generate random string and pattern */
n = gt_rand_max(STRING_MATCHING_MAX_STRING_LENGTH);
m = gt_rand_max(STRING_MATCHING_MAX_PATTERN_LENGTH);
for (j = 0; j < n; j++)
s[j] = gt_rand_char();
s[n] = '\0';
for (j = 0; j < m; j++)
p[j] = gt_rand_char();
p[m] = '\0';
/* matching (first match) */
brute_force_match = GT_UNDEF_ULONG;
bmh_match = GT_UNDEF_ULONG;
kmp_match = GT_UNDEF_ULONG;
shift_and_match = GT_UNDEF_ULONG;
gt_string_matching_brute_force(s, n, p, m, store_first_match,
&brute_force_match);
gt_string_matching_bmh(s, n, p, m, store_first_match, &bmh_match);
gt_string_matching_kmp(s, n, p, m, store_first_match, &kmp_match);
gt_string_matching_shift_and(s, n, p, m, store_first_match,
&shift_and_match);
/* comparing (first match) */
ensure(had_err, brute_force_match == bmh_match);
ensure(had_err, brute_force_match == kmp_match);
ensure(had_err, brute_force_match == shift_and_match);
/* matching (all matches) */
gt_string_matching_brute_force(s, n, p, m, store_match,
brute_force_matches);
gt_string_matching_bmh(s, n, p, m, store_match, bmh_matches);
gt_string_matching_kmp(s, n, p, m, store_match, kmp_matches);
gt_string_matching_shift_and(s, n, p, m, store_match, shift_and_matches);
/* comparing (all matches) */
ensure(had_err, gt_array_size(brute_force_matches) ==
gt_array_size(bmh_matches));
ensure(had_err, gt_array_size(brute_force_matches) ==
gt_array_size(kmp_matches));
ensure(had_err, gt_array_size(brute_force_matches) ==
gt_array_size(shift_and_matches));
ensure(had_err, !gt_array_cmp(brute_force_matches, bmh_matches));
ensure(had_err, !gt_array_cmp(brute_force_matches, kmp_matches));
ensure(had_err, !gt_array_cmp(brute_force_matches, shift_and_matches));
/* reset */
gt_array_reset(brute_force_matches);
gt_array_reset(bmh_matches);
gt_array_reset(kmp_matches);
gt_array_reset(shift_and_matches);
}
gt_array_delete(shift_and_matches);
gt_array_delete(bmh_matches);
gt_array_delete(kmp_matches);
gt_array_delete(brute_force_matches);
return had_err;
}
int gt_interval_tree_unit_test(GT_UNUSED GtError *err)
{
GtIntervalTree *it = NULL;
GtIntervalTreeNode *res = NULL;
unsigned long i = 0;
int had_err = 0, num_testranges = 3000,
num_samples = 300000, num_find_all_samples = 10000,
gt_range_max_basepos = 90000, width = 700,
query_width = 5000;
GtRange *res_rng = NULL, qrange;
GtArray *arr = NULL, *narr = NULL;
arr = gt_array_new(sizeof (GtRange*));
/* generate test ranges */
for (i = 0;i<num_testranges;i++)
{
unsigned long start;
GtRange *rng;
rng = gt_calloc(1, sizeof (GtRange));
start = gt_rand_max(gt_range_max_basepos);
rng->start = start;
rng->end = start + gt_rand_max(width);
gt_array_add(arr, rng);
}
it = gt_interval_tree_new(gt_free_func);
/* insert ranges */
for (i = 0; i < num_testranges && !had_err; i++)
{
GtIntervalTreeNode *new_node;
GtRange *rng;
rng = *(GtRange**) gt_array_get(arr, i);
new_node = gt_interval_tree_node_new(rng, rng->start, rng->end);
gt_interval_tree_insert(it, new_node);
}
gt_ensure(had_err, gt_interval_tree_size(it) == num_testranges);
/* perform test queries */
for (i = 0; i < num_samples && !had_err; i++)
{
unsigned long start = gt_rand_max(gt_range_max_basepos);
qrange.start = start;
qrange.end = start + gt_rand_max(width);
res = gt_interval_tree_find_first_overlapping(it, qrange.start, qrange.end);
if (res)
{
/* we have a hit, check if really overlapping */
res_rng = (GtRange*) gt_interval_tree_node_get_data(res);
gt_ensure(had_err, gt_range_overlap(&qrange, res_rng));
} else {
/* no hit, check whether there really is no overlapping
interval in tree */
GtRange *this_rng;
unsigned long j;
bool found = false;
for (j = 0; j < gt_array_size(arr); j++)
{
this_rng = *(GtRange**) gt_array_get(arr, j);
if (gt_range_overlap(this_rng, &qrange))
{
found = true;
break;
}
}
gt_ensure(had_err, !found);
}
}
/* test searching for all overlapping intervals */
for (i = 0; i < num_find_all_samples && !had_err; i++)
{
unsigned long start = gt_rand_max(gt_range_max_basepos);
qrange.start = start;
qrange.end = start + gt_rand_max(query_width);
GtArray *res = gt_array_new(sizeof (GtRange*));
gt_interval_tree_find_all_overlapping(it, qrange.start, qrange.end, res);
if (res)
{
/* generate reference overlapping interval list by linear search */
GtArray *ref;
unsigned long j;
ref = gt_array_new(sizeof (GtRange*));
for (j = 0; j < gt_array_size(arr); j++)
{
GtRange *this_rng;
this_rng = *(GtRange**) gt_array_get(arr, j);
if (gt_range_overlap(this_rng, &qrange))
{
gt_array_add(ref, this_rng);
}
}
/* compare reference with interval tree query result */
gt_array_sort_stable(ref, range_ptr_compare);
gt_array_sort_stable(res, range_ptr_compare);
/* must be equal */
gt_ensure(had_err, gt_array_cmp(ref, res)==0);
gt_array_delete(ref);
}
gt_array_delete(res);
}
gt_interval_tree_delete(it);
it = gt_interval_tree_new(NULL);
gt_array_reset(arr);
/* generate test ranges */
for (i = 0;i<num_testranges && !had_err;i++)
{
unsigned long start;
GtIntervalTreeNode *new_node;
start = gt_rand_max(gt_range_max_basepos);
new_node = gt_interval_tree_node_new((void*) i, start,
start + gt_rand_max(width));
gt_interval_tree_insert(it, new_node);
}
gt_ensure(had_err, gt_interval_tree_size(it) == num_testranges);
narr = gt_array_new(sizeof (GtIntervalTreeNode*));
for (i = 0; i < num_testranges && !had_err; i++) {
unsigned long idx, n, val;
GtIntervalTreeNode *node = NULL;
/* get all nodes referenced by the interval tree */
interval_tree_find_all_internal(it, it->root, itree_test_get_node, 0,
gt_range_max_basepos+width, narr);
/* remove a random node */
idx = gt_rand_max(gt_array_size(narr)-1);
node = *(GtIntervalTreeNode**) gt_array_get(narr, idx);
gt_ensure(had_err, node != NULL);
val = (unsigned long) gt_interval_tree_node_get_data(node);
gt_interval_tree_remove(it, node);
gt_array_reset(narr);
/* make sure that the node has disappeared */
gt_ensure(had_err, gt_interval_tree_size(it) == num_testranges - (i+1));
interval_tree_find_all_internal(it, it->root, itree_test_get_node, 0,
gt_range_max_basepos+width, narr);
gt_ensure(had_err, gt_array_size(narr) == num_testranges - (i+1));
for (n = 0; !had_err && n < gt_array_size(narr); n++) {
GtIntervalTreeNode *onode = *(GtIntervalTreeNode**) gt_array_get(narr, n);
gt_ensure(had_err, (unsigned long) gt_interval_tree_node_get_data(onode)
!= val);
}
}
gt_array_delete(arr);
gt_array_delete(narr);
gt_interval_tree_delete(it);
return had_err;
}