Changeset: 4ef5e5da0ac0 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=4ef5e5da0ac0
Modified Files:
monetdb5/extras/rdf/rdfschema.c
monetdb5/extras/rdf/rdfschema.h
Branch: rdf
Log Message:
Nil values to each missing property of a CS (in comparing to its maxCS)
diffs (125 lines):
diff --git a/monetdb5/extras/rdf/rdfschema.c b/monetdb5/extras/rdf/rdfschema.c
--- a/monetdb5/extras/rdf/rdfschema.c
+++ b/monetdb5/extras/rdf/rdfschema.c
@@ -3219,7 +3219,21 @@ void updateTblIdxPropIdxMap(int* tblIdxP
}
-str RDFdistTriplesToCSs(int *ret, bat *sbatid, bat *pbatid, bat *obatid,
PropStat* propStat, CStableStat *cstablestat){
+static
+void fillMissingvalues(BAT* curBat, oid lastSubjId){
+ oid k;
+ BUN bun;
+ //Insert nil values to the last column if it does not have the same
+ //size as the table
+ if (curBat != NULL){
+ for(k = BATcount(curBat) -1; k < lastSubjId; k++){
+ bun = oid_nil;
+ BUNappend(curBat,&bun , TRUE);
+ }
+ }
+}
+
+str RDFdistTriplesToCSs(int *ret, bat *sbatid, bat *pbatid, bat *obatid,
PropStat* propStat, CStableStat *cstablestat, oid* lastSubjId){
BAT *sbat = NULL, *pbat = NULL, *obat = NULL;
BATiter si,pi,oi;
BUN p,q;
@@ -3233,9 +3247,11 @@ str RDFdistTriplesToCSs(int *ret, bat *s
// list of that table's properties
Postinglist tmpPtl;
int tmpColIdx = -1;
+ int lasttblIdx = -1;
+ int lastColIdx = -1;
BUN bun;
int i,j;
- BAT *curBat;
+ BAT *curBat = NULL;
oid tmplastInsertedS;
oid k;
@@ -3277,6 +3293,8 @@ str RDFdistTriplesToCSs(int *ret, bat *s
if (*pbt != lastP){
+ fillMissingvalues(curBat, lastSubjId[tblIdx]);
+
//Get number of BATs for this p
ppos = BUNfnd(BATmirror(propStat->pBat),pbt);
if (ppos == BUN_NONE)
@@ -3308,8 +3326,11 @@ str RDFdistTriplesToCSs(int *ret, bat *s
tmpColIdx = tmpTblIdxPropIdxMap[tblIdx];
- printf(BUNFMT": Table %d | column %d for prop " BUNFMT " | sub
" BUNFMT " | obj " BUNFMT "\n",p, tblIdx,
- tmpColIdx, *pbt,
tmpSoid, *obt);
+ if (tmpColIdx != lastColIdx || lasttblIdx != tblIdx){
+ fillMissingvalues(curBat, lastSubjId[lasttblIdx]);
+ lastColIdx = tmpColIdx;
+ lasttblIdx = tblIdx;
+ }
curBat = cstablestat->lstcstable[tblIdx].colBats[tmpColIdx];
tmplastInsertedS =
cstablestat->lastInsertedS[tblIdx][tmpColIdx];
@@ -3317,18 +3338,23 @@ str RDFdistTriplesToCSs(int *ret, bat *s
//TODO: Check last subjectId for this prop. If the subjectId is
not continuous, insert NIL
if (tmpSoid > (tmplastInsertedS + 1)){
for (k = tmplastInsertedS; k < tmpSoid-1; k++){
- bun = BUN_NONE;
+ printf(" Add nil value \n");
+ bun = oid_nil;
BUNappend(curBat,&bun , TRUE);
}
}
BUNappend(curBat, obt, TRUE);
+ printf(BUNFMT": Table %d | column %d for prop " BUNFMT " | sub
" BUNFMT " | obj " BUNFMT "\n",p, tblIdx,
+ tmpColIdx, *pbt,
tmpSoid, *obt);
//Update last inserted S
cstablestat->lastInsertedS[tblIdx][tmpColIdx] = tmpSoid;
}
- //Keep the batCacheId
+ fillMissingvalues(curBat, lastSubjId[tblIdx]);
+
+ // Keep the batCacheId
for (i = 0; i < cstablestat->numTables; i++){
for (j = 0; j < cstablestat->numPropPerTable[i];j++){
cstablestat->lstbatid[i][j] =
cstablestat->lstcstable[i].colBats[j]->batCacheid;
@@ -3454,7 +3480,7 @@ RDFreorganize(int *ret, CStableStat *cst
lmap = BUNappend(lmap, &l, TRUE);
rmap = BUNappend(rmap, &r, TRUE);
- lastSubjId[tblIdx] = newId;
+ lastSubjId[tblIdx]++;
}
}
@@ -3509,13 +3535,14 @@ RDFreorganize(int *ret, CStableStat *cst
printPropStat(propStat);
- if (RDFdistTriplesToCSs(ret, &sNewBat->batCacheid,
&pNewBat->batCacheid, &oNewBat->batCacheid, propStat, cstablestat) !=
MAL_SUCCEED){
+ if (RDFdistTriplesToCSs(ret, &sNewBat->batCacheid,
&pNewBat->batCacheid, &oNewBat->batCacheid, propStat, cstablestat,lastSubjId)
!= MAL_SUCCEED){
throw(RDF, "rdf.RDFreorganize", "Problem in distributing
triples to BATs using CSs");
}
freeCSset(freqCSset);
free(subjCSMap);
free(csTblIdxMapping);
+ free(lastSubjId);
//freeCStableStat(cstablestat);
BBPreclaim(lmap);
diff --git a/monetdb5/extras/rdf/rdfschema.h b/monetdb5/extras/rdf/rdfschema.h
--- a/monetdb5/extras/rdf/rdfschema.h
+++ b/monetdb5/extras/rdf/rdfschema.h
@@ -195,7 +195,7 @@ typedef struct CStableStat {
rdf_export str
-RDFdistTriplesToCSs(int *ret, bat *sbatid, bat *pbatid, bat *obatid, PropStat*
propStat, CStableStat *cstablestat);
+RDFdistTriplesToCSs(int *ret, bat *sbatid, bat *pbatid, bat *obatid, PropStat*
propStat, CStableStat *cstablestat, oid* lastSubjId);
rdf_export str
RDFreorganize(int *ret, CStableStat *cstablestat, bat *sbatid, bat *pbatid,
bat *obatid, bat *mapbatid, int *freqThreshold);
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
http://mail.monetdb.org/mailman/listinfo/checkin-list
