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
hemdist(TSQuerySign a, TSQuerySign b)
{
TSQuerySign res = a ^ b;
return sizebitvec(res);
}
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);
}
Datum
gsfp_consistent(PG_FUNCTION_ARGS)
{
GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0);
StrategyNumber strategy = (StrategyNumber) PG_GETARG_UINT16(2);
bool *recheck = (bool *) PG_GETARG_POINTER(4);
bytea *key = (bytea*)DatumGetPointer(entry->key);
bytea *query;
MolSparseFingerPrint data;
fcinfo->flinfo->fn_extra = SearchSparseFPCache(
fcinfo->flinfo->fn_extra,
fcinfo->flinfo->fn_mcxt,
PG_GETARG_DATUM(1),
NULL, &data, &query);
*recheck = true; /* we use signature, so it's needed to recheck */
if (ISALLTRUE(key) && !GIST_LEAF(entry))
{
PG_RETURN_BOOL(true);
}
else
{
int sum,
overlapSum,
overlapN;
countOverlapValues(
(ISALLTRUE(key)) ? NULL : key, data, NUMBITS,
&sum, &overlapSum, &overlapN
);
PG_RETURN_BOOL(
calcConsistency(
GIST_LEAF(entry), strategy,
overlapSum, /* nCommonUp */
overlapN, /* nCommonDown */
(ISALLTRUE(key)) ? NUMBITS : sizebitvec(key), /* nKey */
sum /* nQuery */
)
);
}
}
Datum
gtsvectorout(PG_FUNCTION_ARGS)
{
SignTSVector *key = (SignTSVector *) DatumGetPointer(PG_DETOAST_DATUM(PG_GETARG_POINTER(0)));
char *outbuf;
if (outbuf_maxlen == 0)
outbuf_maxlen = 2 * EXTRALEN + Max(strlen(SINGOUTSTR), strlen(ARROUTSTR)) + 1;
outbuf = palloc(outbuf_maxlen);
if (ISARRKEY(key))
sprintf(outbuf, ARROUTSTR, (int) ARRNELEM(key));
else
{
int cnttrue = (ISALLTRUE(key)) ? SIGLENBIT : sizebitvec(GETSIGN(key));
sprintf(outbuf, SINGOUTSTR, cnttrue, (int) SIGLENBIT - cnttrue);
}
PG_FREE_IF_COPY(key, 0);
PG_RETURN_POINTER(outbuf);
}
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);
}
