Datum
bfp_tanimoto_sml_op(PG_FUNCTION_ARGS) {
MolBitmapFingerPrint abfp,
bbfp;
double res;
fcinfo->flinfo->fn_extra = SearchBitmapFPCache(
fcinfo->flinfo->fn_extra,
fcinfo->flinfo->fn_mcxt,
PG_GETARG_DATUM(0),
NULL, &abfp, NULL);
fcinfo->flinfo->fn_extra = SearchBitmapFPCache(
fcinfo->flinfo->fn_extra,
fcinfo->flinfo->fn_mcxt,
PG_GETARG_DATUM(1),
NULL, &bbfp, NULL);
res = calcBitmapTanimotoSml(abfp, bbfp);
PG_RETURN_BOOL( res >= getTanimotoLimit() );
}
bool
calcConsistency(bool isLeaf, uint16 strategy,
double nCommonUp, double nCommonDown,
double nKey, double nQuery)
{
bool res = false;
/*
* We don't wish to use RDKit's functions to compute similarity
* for two reasons:
* - too expensive
* - on inner page it will give wrong result, because inner keys
* are not a real fingerprints, they are OR-ed all subsequent
* fingerprints
*
* For inner key we should compule maximum possible numerator and
* minimum possible denominator to find an upper bound of similarity.
*
* nCommon and nKey could not become greater on child - so they are
* an upper bound of number of common bits and number of set bit
* correspondingly. Lower bound of nKey is nCommon.
*/
switch(strategy) {
case RDKitTanimotoStrategy:
/*
* Nsame / (Na + Nb - Nsame)
*/
if ( isLeaf ) {
if ( nCommonUp / (nKey + nQuery - nCommonUp) >= getTanimotoLimit() ) {
res = true;
}
}
else {
if ( nCommonUp / nQuery >= getTanimotoLimit() ) {
res = true;
}
}
break;
case RDKitDiceStrategy:
/*
* 2 * Nsame / (Na + Nb)
*/
if ( isLeaf ) {
if ( 2.0 * nCommonUp / (nKey + nQuery) >= getDiceLimit() ) {
res = true;
}
}
else {
if ( 2.0 * nCommonUp / (nCommonDown + nQuery) >= getDiceLimit() ) {
res = true;
}
}
break;
default:
elog(ERROR,"Unknown strategy: %d", strategy);
}
return res;
}