Changeset: db631d067a69 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=db631d067a69
Modified Files:
monetdb5/extras/rdf/rdfalgebra.c
monetdb5/extras/rdf/rdfschema.c
monetdb5/extras/rdf/rdfschema.h
monetdb5/modules/mal/tokenizer.c
monetdb5/modules/mal/tokenizer.h
Branch: rdf
Log Message:
Build BAT mapping original oid to converted oid.
BAT: original oid created from Tokenizer maps to converted oid.
diffs (238 lines):
diff --git a/monetdb5/extras/rdf/rdfalgebra.c b/monetdb5/extras/rdf/rdfalgebra.c
--- a/monetdb5/extras/rdf/rdfalgebra.c
+++ b/monetdb5/extras/rdf/rdfalgebra.c
@@ -137,6 +137,7 @@ RDFpartialjoin(bat *retid, bat *lid, bat
result = BATproject(result2, right);
result->T->nil = 0;
+
BBPreleaseref(left->batCacheid);
BBPreleaseref(right->batCacheid);
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
@@ -37,6 +37,7 @@
#include <mtime.h>
#include <rdfgraph.h>
#include <rdfparams.h>
+#include "bat5.h"
#define SHOWPROPERTYNAME 1
@@ -11243,6 +11244,74 @@ str RDFdistTriplesToCSs(int *ret, bat *s
return MAL_SUCCEED;
}
+#if BUILDTOKENZIER_TO_MAPID
+/*
+ * Since the of s,p,o oids are converted to table-based oids,
+ * a bat that maps original tokenizer oids to the converted oid
+ * need to be built. Then, when it comes a string, it can first look
+ * at the tokenizer for its original oid. After that, it can
+ * use this TKNZRMappingBat for receiving the converted Id.
+ *
+ * Input String --> Original TKNZR oid --> New Oid (TKNRZ_to_new_MapBAT)
+ * New oid --> Original TKNZR oid --> Input String (New_to_TKNZR_MapBat)
+ * */
+static
+str buildTKNZRMappingBat(BAT *lmap, BAT *rmap){
+
+ BAT *tmpBat = NULL;
+ char* schema = "rdf";
+ int ret;
+ int num = 0;
+ bat mapBatId;
+ BAT *mapBat;
+ str batname = NULL;
+
+ if (TKNZRopen (NULL, &schema) != MAL_SUCCEED) {
+ throw(RDF, "rdf.rdfschema",
+ "could not open the tokenizer\n");
+ }
+
+ TKNZRgetTotalCount(&num);
+ //create a tmpBat from 0 to number of TKNZR items
+ tmpBat = BATnew(TYPE_void, TYPE_oid , num + 1, TRANSIENT);
+ BATsetcount(tmpBat,num);
+ BATseqbase(tmpBat, 0);
+ BATseqbase(BATmirror(tmpBat), 0);
+
+ tmpBat->T->nonil = 1;
+ tmpBat->tkey = 1;
+ tmpBat->tsorted = 1;
+ tmpBat->trevsorted = 0;
+ tmpBat->tdense = 1;
+
+
+ if (RDFpartialjoin(&mapBatId, &lmap->batCacheid, &rmap->batCacheid,
&tmpBat->batCacheid) != MAL_SUCCEED){
+ throw(RDF, "rdf.RDFreorganize", "Problem in using
RDFpartialjoin for tokenizer map bat");
+ }
+
+ if ((mapBat = BATdescriptor(mapBatId)) == NULL) {
+ throw(MAL, "rdf.RDFreorganize", RUNTIME_OBJECT_MISSING);
+ }
+
+ batname = (str) GDKmalloc(50 * sizeof(char));
+ snprintf(batname, 50, "tknzr_to_map");
+
+ if (BKCsetName(&ret, (int *) &(mapBat->batCacheid), (str *) &batname)
!= MAL_SUCCEED)
+ throw(MAL, "tokenizer.open", OPERATION_FAILED);
+ /*
+ if (BKCsetPersistent(&ret, (int *) &(mapBat->batCacheid) != MAL_SUCCEED)
+ throw(MAL, "tokenizer.open", OPERATION_FAILED);
+ */
+
+ GDKfree(batname);
+
+ TKNZRclose(&ret);
+
+ return MAL_SUCCEED;
+}
+
+#endif /* BUILDTOKENZIER_TO_MAPID*/
+
str
RDFreorganize(int *ret, CStableStat *cstablestat, CSPropTypes **csPropTypes,
bat *sbatid, bat *pbatid, bat *obatid, bat *mapbatid, bat *ontbatid, int
*freqThreshold, int *mode){
@@ -11561,6 +11630,10 @@ RDFreorganize(int *ret, CStableStat *cst
#endif
//BATprint(VIEWcreate(BATmirror(lmap),rmap));
+ #if BUILDTOKENZIER_TO_MAPID
+ buildTKNZRMappingBat(lmap, rmap);
+ #endif
+
origobat = getOriginalUriOBat(obat); //Return obat without
type-specific information for URI & BLANKNODE
//This is to get the same oid
as the subject oid for a same URI, BLANKNODE
//--> There will be no
BLANKNODE indication in this obat object oid.
@@ -11582,6 +11655,7 @@ RDFreorganize(int *ret, CStableStat *cst
else
throw(RDF, "rdf.RDFreorganize", "Problem in using
RDFpartialjoin for obat");
+
printf("Done! \n");
printf("Sort triple table according to P, S, O order ... ");
if (triplesubsort(&pNewBat, &sNewBat, &oNewBat) != MAL_SUCCEED){
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
@@ -298,6 +298,9 @@ typedef struct SubCSSet{
// However, for creating the foreign
key relationship, we add this column and
// markt it as a primary key
+#define BUILDTOKENZIER_TO_MAPID 1 /* Build the BAT storing the
mapping between the
+ original tokenizer oid of a URI and
converted oid */
+
typedef struct CSset{
CS* items;
diff --git a/monetdb5/modules/mal/tokenizer.c b/monetdb5/modules/mal/tokenizer.c
--- a/monetdb5/modules/mal/tokenizer.c
+++ b/monetdb5/modules/mal/tokenizer.c
@@ -458,6 +458,26 @@ TKNZRdepositFile(int *r, str *fnme)
str
TKNZRlocate(Client cntxt, MalBlkPtr mb, MalStkPtr stk, InstrPtr pci)
+{
+ str ret;
+ oid id;
+ (void) cntxt;
+ (void) mb;
+
+ if (TRANS == NULL)
+ throw(MAL, "tokenizer", "no tokenizer store open");
+
+ ret = TKNRstringToOid(&id, (str *) getArgReference(stk, pci, 1));
+ if (ret == MAL_SUCCEED) {
+ VALset(getArgReference(stk, pci, 0), TYPE_oid, &id);
+ }
+ return MAL_SUCCEED;
+}
+
+/*
+ * This is similar to the TKNRlocate function */
+
+str TKNRstringToOid(oid *id, str *s)
{
oid pos;
str url;
@@ -466,19 +486,11 @@ TKNZRlocate(Client cntxt, MalBlkPtr mb,
BUN p;
oid prv = 0;
oid comp;
- (void) cntxt;
- (void) mb;
- if (TRANS == NULL)
- throw(MAL, "tokenizer", "no tokenizer store open");
- url = (str) GDKmalloc(sizeof(char) *
- (strlen(*(str *) getArgReference(stk, pci, 1)) + 1));
- if (url == NULL) {
- throw(MAL, "tokenizer.locate", MAL_MALLOC_FAIL);
+ if ((url = GDKstrdup(*s)) == NULL) {
+ throw(MAL, "tokenizer.append", OPERATION_FAILED
MAL_MALLOC_FAIL);
}
- strcpy(url, *(str *) getArgReference(stk, pci, 1));
-
depth = TKNZRtokenize(url, parts, '/');
@@ -509,11 +521,11 @@ TKNZRlocate(Client cntxt, MalBlkPtr mb,
}
}
- VALset(getArgReference(stk, pci, 0), TYPE_oid, &pos);
+ *id = pos;
+
GDKfree(url);
return MAL_SUCCEED;
}
-
str
takeOid(oid id, str *val)
{
@@ -651,6 +663,7 @@ TKNZRgetLevel(int *r, int *level)
return MAL_SUCCEED;
}
+
str
TKNZRgetCount(int *r)
{
@@ -675,8 +688,21 @@ TKNZRgetCount(int *r)
BBPkeepref(*r);
return MAL_SUCCEED;
}
+/*
+ * Get total number of the strings in the tokenizer
+ * */
str
+TKNZRgetTotalCount(int *r){
+
+ if (TRANS == NULL)
+ throw(MAL, "tokenizer", "no tokenizer store open");
+
+ *r = BATcount(tokenBAT[INDEX].val);
+
+ return MAL_SUCCEED;
+}
+str
TKNZRgetCardinality(int *r)
{
BAT *b, *en;
diff --git a/monetdb5/modules/mal/tokenizer.h b/monetdb5/modules/mal/tokenizer.h
--- a/monetdb5/modules/mal/tokenizer.h
+++ b/monetdb5/modules/mal/tokenizer.h
@@ -45,6 +45,8 @@ tokenizer_export str TKNZRgetCount
tokenizer_export str TKNZRgetCardinality (int *r);
tokenizer_export str takeOid (oid id, str *val);
tokenizer_export str takeOid2 (oid id, str *val);
-tokenizer_export int TKNZRtokenize (str in, str *parts, char tkn);
+tokenizer_export int TKNZRtokenize (str in, str *parts, char tkn);
+tokenizer_export str TKNRstringToOid (oid *id, str *s);
+tokenizer_export str TKNZRgetTotalCount (int *r);
#endif /* _TKNZR_H */
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
