/*!
* get the approximate count of objects with value arg
* \param sketch a countmin sketch
* \param arg the Datum we want to find the count of
* \param funcOid the Postgres function that converts arg to a string
*/
int64 cmsketch_count_c(countmin sketch, Datum arg, Oid funcOid, Oid typOid)
{
bytea *nhash;
/* get the md5 hash of the argument. */
nhash = sketch_md5_bytea(arg, typOid);
return(cmsketch_count_md5_datum(sketch, nhash, funcOid));
}
/*!
* transition function to maintain a CountMin sketch with
* Most-Frequent Values
*/
Datum __mfvsketch_trans(PG_FUNCTION_ARGS)
{
bytea * transblob = PG_GETARG_BYTEA_P(0);
Datum newdatum = PG_GETARG_DATUM(1);
int max_mfvs = PG_GETARG_INT32(2);
mfvtransval *transval;
uint64 tmpcnt;
int i;
Datum md5_datum;
/*
* This function makes destructive updates to its arguments.
* Make sure it's being called in an agg context.
*/
if (!(fcinfo->context &&
(IsA(fcinfo->context, AggState)
#ifdef NOTGP
|| IsA(fcinfo->context, WindowAggState)
#endif
)))
elog(ERROR,
"destructive pass by reference outside agg");
/* initialize if this is first call */
if (VARSIZE(transblob) <= sizeof(MFV_TRANSVAL_SZ(0))) {
Oid typOid = get_fn_expr_argtype(fcinfo->flinfo, 1);
transblob = mfv_init_transval(max_mfvs, typOid);
}
else {
check_mfvtransval(transblob);
}
/* ignore NULL inputs */
if (PG_ARGISNULL(1) || PG_ARGISNULL(2))
PG_RETURN_DATUM(PointerGetDatum(transblob));
transval = (mfvtransval *)VARDATA(transblob);
if (transval->typOid != get_fn_expr_argtype(fcinfo->flinfo, 1)) {
elog(ERROR, "cannot aggregate on elements with different types");
}
/* insert into the countmin sketch */
md5_datum = countmin_trans_c(transval->sketch,
newdatum,
transval->outFuncOid,
transval->typOid);
tmpcnt = cmsketch_count_md5_datum(transval->sketch,
(bytea *)DatumGetPointer(md5_datum),
transval->outFuncOid);
i = mfv_find(transblob, newdatum);
if (i > -1) {
transval->mfvs[i].cnt = tmpcnt;
}
else {
/* try to insert as either a new or replacement entry */
for (i = 0; i < (int)transval->max_mfvs; i++) {
if ((i == (int)transval->next_mfv)) {
/* room for new */
transblob = mfv_transval_append(transblob, newdatum);
transval = (mfvtransval *)VARDATA(transblob);
transval->mfvs[i].cnt = tmpcnt;
break;
}
else if (transval->mfvs[i].cnt < tmpcnt) {
/* arg beats this mfv */
transblob = mfv_transval_replace(transblob, newdatum, i);
transval = (mfvtransval *)VARDATA(transblob);
transval->mfvs[i].cnt = tmpcnt;
break;
}
/* else this is not a frequent value */
}
}
PG_RETURN_DATUM(PointerGetDatum(transblob));
}
