Changeset: d1665931f0f6 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=d1665931f0f6
Modified Files:
monetdb5/extras/rdf/rdf.h
monetdb5/extras/rdf/rdfschema.c
Branch: rdf
Log Message:
Fix a bug in extracting type-specific CS (i.e., subCS).
This bug generates a lot of subCSs. After fixing, one CS has a few subCSs which
is reasonably good.
diffs (113 lines):
diff --git a/monetdb5/extras/rdf/rdf.h b/monetdb5/extras/rdf/rdf.h
--- a/monetdb5/extras/rdf/rdf.h
+++ b/monetdb5/extras/rdf/rdf.h
@@ -60,7 +60,8 @@ typedef enum {
DATETIME,
NUMERIC,
STRING,
- BLANKNODE
+ BLANKNODE,
+ MULTIVALUES // For the multi-value property
} ObjectType;
#define IS_DUPLICATE_FREE 0 /* 0: Duplications have not been
removed, otherwise 1 */
@@ -68,6 +69,8 @@ typedef enum {
#define TRIPLE_STORE 1
#define MLA_STORE 2
#define NOT_IGNORE_ERROR_TRIPLE 0
+#define USE_MULTIPLICITY 1 /* Properties having >= 2 values are
being considered as having
+ the same type, i.e., MULTIVALUES */
#define STORE TRIPLE_STORE /* this should become a compile time option */
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
@@ -596,7 +596,7 @@ void printSubCSInformation(SubCSSet *sub
if (subcsset[i].numSubCS != 0){
freq = (int *) Tloc(freqBat, i);
fprintf(fout, "CS " BUNFMT ": ",
subcsset[i].csId);
-
+
if (*freq > freqThreshold){
fprintf(foutfreq, BUNFMT " ",
subcsset[i].csId);
fprintf(foutfreqfilter, BUNFMT " ",
subcsset[i].csId);
@@ -2391,6 +2391,7 @@ str RDFrelationships(int *ret, BAT *sbat
char* buffTypes;
oid realObjOid;
char isBlankNode;
+ oid curP;
if (BATcount(sbat) == 0) {
throw(RDF, "rdf.RDFrelationships", "sbat must not be empty");
@@ -2402,39 +2403,54 @@ str RDFrelationships(int *ret, BAT *sbat
numPwithDup = 0;
curS = 0;
+ curP = 0;
BATloop(sbat, p, q){
sbt = (oid *) BUNtloc(si, p);
if (*sbt != curS){
if (p != 0){ /* Not the first S */
- returnSubCSid = addSubCS(buffTypes,
numPwithDup, subjCSMap[*sbt], csSubCSMap);
+ returnSubCSid = addSubCS(buffTypes,
numPwithDup, subjCSMap[curS], csSubCSMap);
+
//Get the subCSId
subjSubCSMap[*sbt] = returnSubCSid;
}
curS = *sbt;
numPwithDup = 0;
+ curP = 0;
}
obt = (oid *) BUNtloc(oi, p);
/* Check type of object */
objType = (char) ((*obt) >> (sizeof(BUN)*8 - 4)) & 7 ;
/* Get two bits 63th, 62nd from object oid */
-
- buffTypes[numPwithDup] = objType;
- numPwithDup++;
-
+
+ pbt = (oid *) BUNtloc(pi, p);
+
/* Look at the referenced CS Id using subjCSMap */
isBlankNode = 0;
if (objType == URI || objType == BLANKNODE){
- pbt = (oid *) BUNtloc(pi, p);
realObjOid = (*obt) - ((oid) objType << (sizeof(BUN)*8
- 4));
if (realObjOid <= maxSoid && subjCSMap[realObjOid] !=
BUN_NONE){
if (objType == BLANKNODE) isBlankNode = 1;
- ////printf(" Subject " BUNFMT " refer to CS "
BUNFMT " \n",*sbt, subjCSMap[*obt]);
addReltoCSRel(subjCSMap[*sbt],
subjCSMap[realObjOid], *pbt, &csrelSet[subjCSMap[*sbt]], isBlankNode);
}
}
+
+ if (curP == *pbt){
+ #if USE_MULTIPLICITY == 1
+ // Update the object type for this P as MULTIVALUES
+ buffTypes[numPwithDup-1] = MULTIVALUES;
+ #else
+ buffTypes[numPwithDup] = objType;
+ numPwithDup++;
+ #endif
+ }
+ else{
+ buffTypes[numPwithDup] = objType;
+ numPwithDup++;
+ curP = *pbt;
+ }
}
/* Check for the last CS */
@@ -2443,8 +2459,6 @@ str RDFrelationships(int *ret, BAT *sbat
free (buffTypes);
-
-
*ret = 1;
return MAL_SUCCEED;
_______________________________________________
checkin-list mailing list
[email protected]
http://mail.monetdb.org/mailman/listinfo/checkin-list
