/* Third pass: extract all sequences and their execution counts. We
* walk the toplevel and then recur throughout each tree; if the count
* at one level is less than the count at the previous level then one
* sequence ended at the previous level but continues with the lower
* count at this level.
*
* Discard sequences of length 1 (duh) and anything below the
* user-settable cutoff.
*
* Every sequence that is a proper suffix of another sequence with the
* same execution count is discarded too (subsumed by the longer
* sequence).
*/
void extract_superwords()
{
int i, j, ki, kj;
for ( i=0 ; i < INSTRCOUNT ; i++ )
descend(&toplevel[i], 0);
qsort(seqs, nextseq, sizeof(seq_t), seq_cmp);
for ( i=0 ; i < nextseq ; i++ ) {
for ( j=i+1 ; j < nextseq && seqs[i].count == seqs[j].count ; j++ ) {
if (seqs[j].length < seqs[i].length) {
for ( ki=seqs[i].length-1, kj=seqs[j].length-1 ; kj >= 0 && seqs[i].opcode[ki] == seqs[j].opcode[kj] ; ki--, kj-- )
;
if (kj < 0)
seqs[j].suffix = 1;
}
}
}
/* Remove suffix sequences */
j=0;
for ( i=0 ; i < nextseq ; i++ )
if (!seqs[i].suffix)
seqs[j++] = seqs[i];
nextseq = j;
/* Flat */
if (flat)
for ( i=0 ; i < nextseq ; i++ )
printseq(i, 0);
/* Hierarchical. Every sequence that is a proper prefix of
another sequence will necessarily have a higher execution
count, and will be printed with the longer sequence as a
child.
Note this messes with the suffix flag.
*/
if (hierarchical) {
if (flat)
printf("\n\n----------\n\n");
for ( i=0 ; i < nextseq ; i++ ) {
if (!seqs[i].suffix) {
printseq(i, 0);
printhseq(i, 1);
}
}
}
}
void
main(void)
{
FILE *fin, *fout;
int a, i, impossible;
fin = fopen("lamps.in", "r");
fout = fopen("lamps.out", "w");
assert(fin != NULL && fout != NULL);
fscanf(fin, "%d %d", &nlamp, &nswitch);
for(;;) {
fscanf(fin, "%d", &a);
if(a == -1)
break;
a = MAXLAMP-1 - (a-1) % MAXLAMP;
ison |= 1<<a;
known |= 1<<a;
}
for(;;) {
fscanf(fin, "%d", &a);
if(a == -1)
break;
a = MAXLAMP-1 - (a-1) % MAXLAMP;
assert((ison & (1<<a)) == 0);
known |= 1<<a;
}
if(nswitch > 4)
if(nswitch%2 == 0)
nswitch = 4;
else
nswitch = 3;
for(; nswitch >= 0; nswitch -= 2)
search(LAMPMASK, 0, nswitch);
impossible = 1;
for(i=0; i<(1<<MAXLAMP); i++) {
if(poss[i]) {
printseq(fout, i);
impossible = 0;
}
}
if(impossible)
fprintf(fout, "IMPOSSIBLE\n");
exit(0);
}
void printhseq(int i, int level)
{
int j, k, l;
l=seqs[i].length;
for ( j=i+1 ; j < nextseq ; j++ ) {
if (seqs[j].suffix || seqs[j].length < l)
goto next_j;
for ( k=0 ; k < l ; k++ )
if (seqs[i].opcode[k] != seqs[j].opcode[k])
goto next_j;
printseq(j, level);
seqs[j].suffix = 1;
printhseq(j, level+1);
next_j:
;
}
}
int pairs_join(int argc, char *argv[]) {
gzFile fp[2];
kseq_t *ks[2];
int c, i, tag=0, strict=0, l[] = {0, 0};
while ((c = getopt(argc, argv, "ts")) >= 0) {
switch (c) {
case 't': tag = 1; break;
case 's': strict = 1; break;
default: return 1;
}
}
if (optind == argc) return join_usage();
for (i = 0; i < 2; ++i) {
fp[i] = gzopen(argv[optind + i], "r");
ks[i] = kseq_init(fp[i]);
}
for (;;) {
for (i = 0; i < 2; ++i) l[i] = kseq_read(ks[i]);
if (l[0] < 0 || l[1] < 0)
break;
if (!is_interleaved_pair(ks[0]->name.s, ks[1]->name.s)) {
fprintf(stderr, "[%s] warning: different sequence names: %s != %s\n", __func__, ks[0]->name.s, ks[1]->name.s);
if (strict) return 1;
}
for (i = 0; i < 2; ++i) printseq(stdout, ks[i], ks[i]->seq.l, tag ? i+1 : 0);
}
if (l[0] > 0 || l[1] > 0) {
fprintf(stderr, "[%s] error: paired end files have differing numbers of reads.\n", __func__);
exit(1);
}
for (i = 0; i < 2; ++i) {
kseq_destroy(ks[i]);
gzclose(fp[i]);
}
return 0;
}
int pairs_split(int argc, char *argv[]) {
kseq_t **seq, *tmp;
FILE *fpout[] = {NULL, NULL, NULL};
gzFile fp;
int c, l, i, strict=1, min_length=0;
unsigned total[]={0, 0}, removed[]={0, 0}, both_removed = 0;
unsigned is_empty[]={0, 0};
char *p, *tags[] = {"/1", "/2"};
while ((c = getopt(argc, argv, "1:2:u:n")) >= 0) {
switch (c) {
case '1':
fpout[0] = fopen(optarg, "w");
break;
case '2':
fpout[1] = fopen(optarg, "w");
break;
case 'u':
fpout[2] = fopen(optarg, "w");
break;
case 'm':
min_length = atoi(optarg);
if (min_length < 1) fprintf(stderr, "[%s] error: minimum length must be >= 1\n", __func__);
return 1;
break;
case 'n': strict = 0; break;
default: return 1;
}
}
if (optind == argc) return split_usage();
for (i = 0; i < 3; ++i) {
if (!fpout[i]) {
fprintf(stderr, "[%s] error: arguments -1, -2, and -u are required.", __func__);
return 1;
}
}
fp = (strcmp(argv[optind], "-") == 0) ? gzdopen(fileno(stdin), "r") : gzopen(argv[optind], "r");
seq = calloc(2, sizeof(kseq_t*));
for (i = 0; i < 2; ++i) seq[i] = malloc(sizeof(kseq_t));
tmp = kseq_init(fp);
while ((l=kseq_read(tmp)) >= 0) {
/* always read in chunks of two FASTX entries */
cpy_kseq(seq[0], tmp);
l = kseq_read(tmp);
if (l < 0) break;
cpy_kseq(seq[1], tmp);
for (i = 0; i < 2; ++i) {
/* remove /1 and /2 tags */
p = strstr(seq[i]->name.s, tags[i]);
if (p) {
strncpy(p, "\0", 1);
}
}
if (strcmp(seq[0]->name.s, seq[1]->name.s) != 0) {
fprintf(stderr, "[%s] warning: interleaved reads names differ '%s' != '%s'\n", __func__, seq[0]->name.s, seq[1]->name.s);
if (strict) return 1;
}
/* deal with unpaired cases (either no seq or single 'N') */
for (i = 0; i < 2; ++i) {
is_empty[i] = seq[i]->seq.l <= min_length || strcmp(seq[i]->seq.s, "N") == 0;
removed[i] += is_empty[i];
total[i] += 1;
}
if (!is_empty[0] && !is_empty[1]) {
for (i = 0; i < 2; i++)
printseq(fpout[i], seq[i], seq[i]->seq.l, 0);
} else if (is_empty[0] && is_empty[1]) {
both_removed += 1;
continue;
} else {
i = is_empty[0] ? 1 : 0;
printseq(fpout[2], seq[i], seq[i]->seq.l, 0);
}
}
if (total[0] != total[1]) {
fprintf(stderr, "[%s] error: mismatched totals of interleaved pairs! %u != %u\n", __func__, total[0], total[1]);
return 1;
}
fprintf(stderr, "totals: %u %u\nremoved: %u %u\n", total[0], total[1], removed[0], removed[1]);
return 0;
}
