kword_t * Kmerizer::canonicalize(kword_t *packed, kword_t *rcpack, size_t *bin) {
// copy the words (reverse their order)
for (size_t i=0;i<nwords;i++)
rcpack[i] = packed[nwords-1-i];
// shift bits from word to word if necessary
if (lshift) {
for (size_t i=0;i<nwords-1;i++) {
rcpack[i] |= rcpack[i+1] << lshift;
rcpack[i+1] >>= rshift;
}
}
size_t rcbin = revcomp_8(rcpack[0]);
if (rcbin > *bin)
return packed;
// shift each word by 8 bits
rcpack[0] >>= 8;
for (size_t i=0;i<nwords-1;i++) {
rcpack[i] |= rcpack[i+1] << 56;
rcpack[i+1] >>= 8;
}
if (lshift >= 8)
rcpack[nwords-1] |= *bin << (lshift-8);
else if (lshift > 0) {
rcpack[nwords-2] |= *bin << 56+lshift;
rcpack[nwords-1] |= *bin >> lshift;
}
inline static uint64_t revcomp_block(uint64_t x) {
uint64_t output;
unsigned char * p = (unsigned char *) &x;
unsigned char * q = (unsigned char *) &output;
q[7] = revcomp_8(p[0]);
q[6] = revcomp_8(p[1]);
q[5] = revcomp_8(p[2]);
q[4] = revcomp_8(p[3]);
q[3] = revcomp_8(p[4]);
q[2] = revcomp_8(p[5]);
q[1] = revcomp_8(p[6]);
q[0] = revcomp_8(p[7]);
return output;
}
