static int
hemdist(bytea *a, bytea *b) {
if (ISALLTRUE(a)) {
if (ISALLTRUE(b))
return 0;
else
return SIGLENBIT(b) - bitstringWeight(SIGLEN(b), (uint8 *)VARDATA(b));
}
else if (ISALLTRUE(b)) {
return SIGLENBIT(a) - bitstringWeight(SIGLEN(a), (uint8 *)VARDATA(a));
}
return hemdistsign(a, b);
}
static int
soergeldist(bytea *a, bytea *b) {
if (ISALLTRUE(a)) {
if (ISALLTRUE(b))
return 0;
else
return SIGLENBIT(b) - sizebitvec(b);
}
else if (ISALLTRUE(b)) {
return SIGLENBIT(a) - sizebitvec(a);
}
return soergeldistsign(a, b);
}
static int
soergeldist(bytea *a, bytea *b) {
double d;
if (ISALLTRUE(a)) {
if (ISALLTRUE(b))
return 0;
else
// FIXME shouldn't it be double(sizebitvec(b))/SIGLENBIT(b); ?
return SIGLENBIT(b) - bitstringWeight(SIGLEN(b), (uint8 *)VARDATA(b));
}
else if (ISALLTRUE(b)) {
// FIXME shouldn't it be double(sizebitvec(a))/SIGLENBIT(a); ?
return SIGLENBIT(a) - bitstringWeight(SIGLEN(a), (uint8 *)VARDATA(a));
}
return (int) floor(10000*soergeldistsign(a, b));
}
static bool
rdkit_consistent(GISTENTRY *entry, StrategyNumber strategy, bytea *key, bytea *query)
{
double nCommon, nQuery, nKey = 0.0;
if (ISALLTRUE(query))
elog(ERROR, "Query malformed");
/*
* Counts basic numbers, but don't count nKey on inner
* page (see comments below)
*/
nQuery = (double)sizebitvec(query);
if (ISALLTRUE(key)) {
if (GIST_LEAF(entry))
nKey = (double)SIGLENBIT(query);
nCommon = nQuery;
} else {
int i, cnt = 0;
unsigned char *pk = (unsigned char*)VARDATA(key);
unsigned char *pq = (unsigned char*)VARDATA(query);
if (SIGLEN(key) != SIGLEN(query))
elog(ERROR, "All fingerprints should be the same length");
#ifndef USE_BUILTIN_POPCOUNT
for(i=0;i<SIGLEN(key);i++)
cnt += number_of_ones[ pk[i] & pq[i] ];
#else
unsigned eidx=SIGLEN(key)/sizeof(unsigned int);
for(i=0;i<eidx;++i){
cnt += __builtin_popcount(((unsigned int *)pk)[i] & ((unsigned int *)pq)[i]);
}
for(i=eidx*sizeof(unsigned);i<SIGLEN(key);++i){
cnt += number_of_ones[ pk[i] & pq[i] ];
}
#endif
nCommon = (double)cnt;
if (GIST_LEAF(entry))
nKey = (double)sizebitvec(key);
}
return calcConsistency(GIST_LEAF(entry), strategy, nCommon, nCommon, nKey, nQuery);
}
static bool
rdkit_consistent(GISTENTRY *entry, StrategyNumber strategy, bytea *key, bytea *query)
{
double nCommon, nQuery, nKey = 0.0;
if (ISALLTRUE(query)) {
elog(ERROR, "Query malformed");
}
/*
* Counts basic numbers, but don't count nKey on inner
* page (see comments below)
*/
int siglen = SIGLEN(query);
uint8 *q = (uint8 *)VARDATA(query);
nQuery = (double)bitstringWeight(siglen, q);
if (ISALLTRUE(key)) {
if (GIST_LEAF(entry)) {
nKey = (double)SIGLENBIT(query);
}
nCommon = nQuery;
}
else {
if (siglen != SIGLEN(key)) {
elog(ERROR, "All fingerprints should be the same length");
}
uint8 *k = (uint8 *)VARDATA(key);
nCommon = bitstringIntersectionWeight(siglen, k, q);
if (GIST_LEAF(entry)) {
nKey = (double)bitstringWeight(siglen, k);
}
}
return calcConsistency(GIST_LEAF(entry), strategy, nCommon, nCommon, nKey, nQuery);
}
Datum
gbfp_distance(PG_FUNCTION_ARGS)
{
GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0);
// bytea *query = PG_GETARG_DATA_TYPE_P(1);
StrategyNumber strategy = (StrategyNumber) PG_GETARG_UINT16(2);
bytea *key = (bytea*)DatumGetPointer(entry->key);
bytea *query;
double nCommon, nCommonUp, nCommonDown, nQuery, distance;
double nKey = 0.0;
fcinfo->flinfo->fn_extra = SearchBitmapFPCache(
fcinfo->flinfo->fn_extra,
fcinfo->flinfo->fn_mcxt,
PG_GETARG_DATUM(1),
NULL, NULL,&query);
if (ISALLTRUE(query))
elog(ERROR, "Query malformed");
/*
* Counts basic numbers, but don't count nKey on inner
* page (see comments below)
*/
nQuery = (double)sizebitvec(query);
if (ISALLTRUE(key))
{
if (GIST_LEAF(entry)) nKey = (double)SIGLENBIT(query);
nCommon = nQuery;
}
else
{
int i, cnt = 0;
unsigned char *pk = (unsigned char*)VARDATA(key),
*pq = (unsigned char*)VARDATA(query);
if (SIGLEN(key) != SIGLEN(query))
elog(ERROR, "All fingerprints should be the same length");
#ifndef USE_BUILTIN_POPCOUNT
for(i=0;i<SIGLEN(key);i++)
cnt += number_of_ones[ pk[i] & pq[i] ];
#else
unsigned eidx=SIGLEN(key)/sizeof(unsigned int);
for(i=0;i<SIGLEN(key)/sizeof(unsigned int);++i){
cnt += __builtin_popcount(((unsigned int *)pk)[i] & ((unsigned int *)pq)[i]);
}
for(i=eidx*sizeof(unsigned);i<SIGLEN(key);++i){
cnt += number_of_ones[ pk[i] & pq[i] ];
}
#endif
nCommon = (double)cnt;
if (GIST_LEAF(entry))
nKey = (double)sizebitvec(key);
}
nCommonUp = nCommon;
nCommonDown = nCommon;
switch(strategy)
{
case RDKitOrderByTanimotoStrategy:
/*
* Nsame / (Na + Nb - Nsame)
*/
if (GIST_LEAF(entry))
{
distance = nCommonUp / (nKey + nQuery - nCommonUp);
}
else
{
distance = nCommonUp / nQuery;
}
break;
case RDKitOrderByDiceStrategy:
/*
* 2 * Nsame / (Na + Nb)
*/
if (GIST_LEAF(entry))
{
distance = 2.0 * nCommonUp / (nKey + nQuery);
}
else
{
distance = 2.0 * nCommonUp / (nCommonDown + nQuery);
}
break;
default:
elog(ERROR,"Unknown strategy: %d", strategy);
}
PG_RETURN_FLOAT8(1.0 - distance);
}
