Changeset: cd42437881d1 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=cd42437881d1
Modified Files:
monetdb5/modules/mal/txtsim.c
monetdb5/modules/mal/txtsim.h
monetdb5/modules/mal/txtsim.mal
Branch: default
Log Message:
Split return BAT for qgram operation
diffs (78 lines):
diff --git a/monetdb5/modules/mal/txtsim.c b/monetdb5/modules/mal/txtsim.c
--- a/monetdb5/modules/mal/txtsim.c
+++ b/monetdb5/modules/mal/txtsim.c
@@ -875,11 +875,11 @@ fstrcmp0_impl(dbl *ret, str *string1, st
/* ============ Q-GRAM SELF JOIN ============== */
str
-CMDqgramselfjoin(BAT **res, BAT *qgram, BAT *id, BAT *pos, BAT *len, flt *c,
int *k)
+CMDqgramselfjoin(BAT **res, BAT **res2, BAT *qgram, BAT *id, BAT *pos, BAT
*len, flt *c, int *k)
{
BUN n = BATcount(qgram);
BUN i, j;
- BAT *bn;
+ BAT *bn, *bn2;
oid *qbuf = (oid *) Tloc(qgram, BUNfirst(qgram));
int *ibuf = (int *) Tloc(id, BUNfirst(id));
@@ -904,12 +904,19 @@ CMDqgramselfjoin(BAT **res, BAT *qgram,
ERRORcheck((Tsize(pos) != ATOMsize(pos->ttype)), "CMDqgramselfjoin: pos
is not a true void bat");
ERRORcheck((Tsize(len) != ATOMsize(len->ttype)), "CMDqgramselfjoin: len
is not a true void bat");
- *res = bn = BATnew(TYPE_int, TYPE_int, n);
+ *res = bn = BATnew(TYPE_void, TYPE_int, n);
+ *res2 = bn2 = BATnew(TYPE_void, TYPE_int, n);
+ if (bn == NULL || bn2 == NULL){
+ if (bn) BBPreclaim(bn);
+ if (bn2) BBPreclaim(bn2);
+ throw(MAL, "txtsim.qgramselfjoin", MAL_MALLOC_FAIL);
+ }
for (i = 0; i < n - 1; i++) {
for (j = i + 1; (j < n && qbuf[j] == qbuf[i] && pbuf[j] <=
(pbuf[i] + (*k + *c * MYMIN(lbuf[i], lbuf[j])))); j++) {
if (ibuf[i] != ibuf[j] && abs(lbuf[i] - lbuf[j]) <= (*k
+ *c * MYMIN(lbuf[i], lbuf[j]))) {
- bunfastins(bn, ibuf + i, ibuf + j);
+ BUNappend(bn, ibuf + i, FALSE);
+ BUNappend(bn2, ibuf + j, FALSE);
}
}
}
@@ -918,10 +925,11 @@ CMDqgramselfjoin(BAT **res, BAT *qgram,
bn->hrevsorted = bn->trevsorted = 0;
bn->H->nonil = bn->T->nonil = 0;
+ bn2->hsorted = bn2->tsorted = 0;
+ bn2->hrevsorted = bn2->trevsorted = 0;
+ bn2->H->nonil = bn2->T->nonil = 0;
+
return MAL_SUCCEED;
- bunins_failed:
- BBPreclaim(bn);
- throw(MAL, "txtsim.qgramselfjoin", MAL_MALLOC_FAIL);
}
/* copy up to utf8len UTF-8 encoded characters from src to buf
diff --git a/monetdb5/modules/mal/txtsim.h b/monetdb5/modules/mal/txtsim.h
--- a/monetdb5/modules/mal/txtsim.h
+++ b/monetdb5/modules/mal/txtsim.h
@@ -51,7 +51,7 @@ txtsim_export str fstrcmp0_impl(dbl *ret
txtsim_export str soundex_impl(str *res, str *Name);
txtsim_export str stringdiff_impl(int *res, str *s1, str*s2);
txtsim_export str CMDqgramnormalize(str *res, str *input);
-txtsim_export str CMDqgramselfjoin(BAT **res, BAT *qgram, BAT *id, BAT *pos,
BAT *len, flt *c, int *k);
+txtsim_export str CMDqgramselfjoin(BAT **res, BAT **res2, BAT *qgram, BAT *id,
BAT *pos, BAT *len, flt *c, int *k);
txtsim_export str CMDstr2qgrams(int *ret, str *val);
#endif /*_TXTSIM_H*/
diff --git a/monetdb5/modules/mal/txtsim.mal b/monetdb5/modules/mal/txtsim.mal
--- a/monetdb5/modules/mal/txtsim.mal
+++ b/monetdb5/modules/mal/txtsim.mal
@@ -53,7 +53,7 @@ command qgramnormalize(input:str): str
address CMDqgramnormalize
comment "'Normalizes' strings (eg. toUpper and replaces non-alphanumerics with
one space";
-command qgramselfjoin(qgram:bat[:oid, :oid], id:bat[:oid, :oid], pos:bat[:oid,
:int], len:bat[:oid, :int], c:flt, k:int) :bat[:int, :int]
+command qgramselfjoin(qgram:bat[:oid, :oid], id:bat[:oid, :oid], pos:bat[:oid,
:int], len:bat[:oid, :int], c:flt, k:int) (:bat[:oid,:int],:bat[:oid, :int])
address CMDqgramselfjoin
comment "QGram self-join on ordered(!) qgram tables and sub-ordered q-gram
positions";
_______________________________________________
Checkin-list mailing list
[email protected]
http://mail.monetdb.org/mailman/listinfo/checkin-list
