Datum
cosine(PG_FUNCTION_ARGS)
{
char *a, *b;
TokenList *s, *t;
int atok, btok, comtok, alltok;
float4 res;
a = DatumGetPointer(DirectFunctionCall1(textout, PointerGetDatum(PG_GETARG_TEXT_P(0))));
b = DatumGetPointer(DirectFunctionCall1(textout, PointerGetDatum(PG_GETARG_TEXT_P(1))));
if (strlen(a) > PGS_MAX_STR_LEN || strlen(b) > PGS_MAX_STR_LEN)
ereport(ERROR,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("argument exceeds the maximum length of %d bytes",
PGS_MAX_STR_LEN)));
/* sets */
s = initTokenList(1);
t = initTokenList(1);
switch (pgs_cosine_tokenizer)
{
case PGS_UNIT_WORD:
tokenizeBySpace(s, a);
tokenizeBySpace(t, b);
break;
case PGS_UNIT_GRAM:
tokenizeByGram(s, a);
tokenizeByGram(t, b);
break;
case PGS_UNIT_CAMELCASE:
tokenizeByCamelCase(s, a);
tokenizeByCamelCase(t, b);
break;
case PGS_UNIT_ALNUM: /* default */
default:
tokenizeByNonAlnum(s, a);
tokenizeByNonAlnum(t, b);
break;
}
elog(DEBUG3, "Token List A");
printToken(s);
elog(DEBUG3, "Token List B");
printToken(t);
atok = s->size;
btok = t->size;
/* combine the sets */
switch (pgs_cosine_tokenizer)
{
case PGS_UNIT_WORD:
tokenizeBySpace(s, b);
break;
case PGS_UNIT_GRAM:
tokenizeByGram(s, b);
break;
case PGS_UNIT_CAMELCASE:
tokenizeByCamelCase(s, b);
break;
case PGS_UNIT_ALNUM: /* default */
default:
tokenizeByNonAlnum(s, b);
break;
}
elog(DEBUG3, "All Token List");
printToken(s);
alltok = s->size;
destroyTokenList(s);
destroyTokenList(t);
comtok = atok + btok - alltok;
elog(DEBUG1, "is normalized: %d", pgs_cosine_is_normalized);
elog(DEBUG1, "token list A size: %d", atok);
elog(DEBUG1, "token list B size: %d", btok);
elog(DEBUG1, "all tokens size: %d", alltok);
elog(DEBUG1, "common tokens size: %d", comtok);
/* normalized and unnormalized version are the same */
res = (float) comtok / (sqrt(atok) * sqrt(btok));
PG_RETURN_FLOAT4(res);
}
Datum
mongeelkan(PG_FUNCTION_ARGS)
{
char *a, *b;
TokenList *s, *t;
Token *p, *q;
double summatches;
double maxvalue;
float8 res;
a = DatumGetPointer(DirectFunctionCall1(textout,
PointerGetDatum(PG_GETARG_TEXT_P(0))));
b = DatumGetPointer(DirectFunctionCall1(textout,
PointerGetDatum(PG_GETARG_TEXT_P(1))));
if (strlen(a) > PGS_MAX_STR_LEN || strlen(b) > PGS_MAX_STR_LEN)
ereport(ERROR,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("argument exceeds the maximum length of %d bytes",
PGS_MAX_STR_LEN)));
/* lists */
s = initTokenList(0);
t = initTokenList(0);
switch (pgs_mongeelkan_tokenizer)
{
case PGS_UNIT_WORD:
tokenizeBySpace(s, a);
tokenizeBySpace(t, b);
break;
case PGS_UNIT_GRAM:
tokenizeByGram(s, a);
tokenizeByGram(t, b);
break;
case PGS_UNIT_CAMELCASE:
tokenizeByCamelCase(s, a);
tokenizeByCamelCase(t, b);
break;
case PGS_UNIT_ALNUM:
default:
tokenizeByNonAlnum(s, a);
tokenizeByNonAlnum(t, b);
break;
}
summatches = 0.0;
p = s->head;
while (p != NULL)
{
maxvalue = 0.0;
q = t->head;
while (q != NULL)
{
double val = _mongeelkan(p->data, q->data);
elog(DEBUG3, "p: %s; q: %s", p->data, q->data);
if (val > maxvalue)
maxvalue = val;
q = q->next;
}
summatches += maxvalue;
p = p->next;
}
/* normalized and unnormalized version are the same */
res = summatches / s->size;
elog(DEBUG1, "is normalized: %d", pgs_mongeelkan_is_normalized);
elog(DEBUG1, "sum matches: %.3f", summatches);
elog(DEBUG1, "s size: %d", s->size);
elog(DEBUG1, "medistance(%s, %s) = %.3f", a, b, res);
destroyTokenList(s);
destroyTokenList(t);
PG_RETURN_FLOAT8(res);
}
