Changeset: 1bef4498b2fd for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=1bef4498b2fd
Modified Files:
monetdb5/extras/rdf/rdf.h
monetdb5/extras/rdf/rdfalgebra.c
monetdb5/extras/rdf/rdfscan.c
monetdb5/extras/rdf/rdfscan.h
monetdb5/extras/rdf/rdfschema.c
Branch: rdf
Log Message:
Make persistent bat for tknzr-converted_oid mapping
diffs (truncated from 365 to 300 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
@@ -58,6 +58,12 @@ TKNZRrdf2str (bat *res, bat *bid, bat *m
rdf_export str
RDFpartialjoin (bat *res, bat *lmap, bat *rmap, bat *input);
+rdf_export str
+RDFtriplesubsort(BAT **sbat, BAT **pbat, BAT **obat);
+
+rdf_export str
+RDFbisubsort(BAT **lbat, BAT **rbat);
+
#define RDF_MIN_LITERAL (((oid) 1) << ((sizeof(oid)==8)?59:27))
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
@@ -167,6 +167,70 @@ RDFpartialjoin(bat *retid, bat *lid, bat
return MAL_SUCCEED;
}
+str RDFtriplesubsort(BAT **sbat, BAT **pbat, BAT **obat){
+
+ BAT *o1,*o2,*o3;
+ BAT *g1,*g2,*g3;
+ BAT *S = NULL, *P = NULL, *O = NULL;
+
+ S = *sbat;
+ P = *pbat;
+ O = *obat;
+ /* order SPO/SOP */
+ if (BATsubsort(sbat, &o1, &g1, S, NULL, NULL, 0, 0) == GDK_FAIL){
+ if (S != NULL) BBPreclaim(S);
+ throw(RDF, "rdf.triplesubsort", "Fail in sorting for S");
+ }
+
+ if (BATsubsort(pbat, &o2, &g2, P, o1, g1, 0, 0) == GDK_FAIL){
+ BBPreclaim(S);
+ if (P != NULL) BBPreclaim(P);
+ throw(RDF, "rdf.triplesubsort", "Fail in sub-sorting for P");
+ }
+ if (BATsubsort(obat, &o3, &g3, O, o2, g2, 0, 0) == GDK_FAIL){
+ BBPreclaim(S);
+ BBPreclaim(P);
+ if (O != NULL) BBPreclaim(O);
+ throw(RDF, "rdf.triplesubsort", "Fail in sub-sorting for O");
+ }
+
+ BBPunfix(o2->batCacheid);
+ BBPunfix(g2->batCacheid);
+ BBPunfix(o3->batCacheid);
+ BBPunfix(g3->batCacheid);
+
+ return MAL_SUCCEED;
+}
+
+/*
+ * Sort left bat and re-order right bat according to the lef bat
+ * */
+str RDFbisubsort(BAT **lbat, BAT **rbat){
+
+ BAT *o1,*o2;
+ BAT *g1,*g2;
+ BAT *L = NULL, *R = NULL;
+
+ L = *lbat;
+ R = *rbat;
+ if (BATsubsort(lbat, &o1, &g1, L, NULL, NULL, 0, 0) == GDK_FAIL){
+ if (L != NULL) BBPreclaim(L);
+ throw(RDF, "rdf.triplesubsort", "Fail in sorting for L");
+ }
+
+ if (BATsubsort(rbat, &o2, &g2, R, o1, g1, 0, 0) == GDK_FAIL){
+ BBPreclaim(L);
+ if (R != NULL) BBPreclaim(R);
+ throw(RDF, "rdf.triplesubsort", "Fail in sub-sorting for R");
+ }
+
+ BBPunfix(o1->batCacheid);
+ BBPunfix(g1->batCacheid);
+ BBPunfix(o2->batCacheid);
+ BBPunfix(g2->batCacheid);
+
+ return MAL_SUCCEED;
+}
str
TKNZRrdf2str(bat *res, bat *bid, bat *map)
diff --git a/monetdb5/extras/rdf/rdfscan.c b/monetdb5/extras/rdf/rdfscan.c
--- a/monetdb5/extras/rdf/rdfscan.c
+++ b/monetdb5/extras/rdf/rdfscan.c
@@ -37,18 +37,21 @@
#define MAX_PARAMS_NO 20
static
-void queryParser(RdfScanParams *rsParam, str query, str schema){
+str queryParser(RdfScanParams *rsParam, str query, str schema){
int paramNo;
str parts[MAX_PARAMS_NO];
int i = 0, j;
- int numRP = 0;
- int numOP = 0;
- int opIdx;
- int rpIdx;
+ int numRP = 0, numOP = 0;
+ int opIdx, rpIdx;
+ int ret;
(void) schema;
+ if (TKNZRopen (NULL, &schema) != MAL_SUCCEED) {
+ throw(RDF, "rdf.rdfschema", "could not open the tokenizer\n");
+ }
+
paramNo = TKNZRtokenize(query, parts, '|');
printf("Number of params from query %s is: %d \n", query, paramNo);
@@ -68,19 +71,19 @@ void queryParser(RdfScanParams *rsParam,
printf("Number of RPs is %d \n", numRP);
rsParam->numRP = numRP;
- rsParam->lstRPs = (char **)GDKmalloc(sizeof(char*) * rsParam->numRP);
+ rsParam->lstRPstr = (char **)GDKmalloc(sizeof(char*) * rsParam->numRP);
rsParam->lstLow = (char **)GDKmalloc(sizeof(char*) * rsParam->numRP);
rsParam->lstHi = (char **)GDKmalloc(sizeof(char*) * rsParam->numRP);
rsParam->numOP = numOP;
- rsParam->lstOPs = (char **)GDKmalloc(sizeof(char*) * rsParam->numOP);
+ rsParam->lstOPstr = (char **)GDKmalloc(sizeof(char*) * rsParam->numOP);
opIdx = 0;
rpIdx = 0;
for (i = 0; i < paramNo; i++){
if (parts[i][0] == '?'){ //optional param
parts[i]++;
- rsParam->lstOPs[opIdx] = GDKstrdup(parts[i]);
+ rsParam->lstOPstr[opIdx] = GDKstrdup(parts[i]);
opIdx++;
}
else{
@@ -113,11 +116,16 @@ void queryParser(RdfScanParams *rsParam,
}
}
- rsParam->lstRPs[rpIdx] = GDKstrdup(parts[i]);
+ rsParam->lstRPstr[rpIdx] = GDKstrdup(parts[i]);
rpIdx++;
}
}
+
+
+ TKNZRclose(&ret);
+
+ return MAL_SUCCEED;
}
static
@@ -125,12 +133,12 @@ void printParams(RdfScanParams *rsParam)
int i;
for (i = 0; i < rsParam->numRP; i++){
- printf("RP[%d] = %s\n", i, rsParam->lstRPs[i]);
+ printf("RP[%d] = %s\n", i, rsParam->lstRPstr[i]);
if (rsParam->lstLow[i] != NULL) printf(" Low
%s\n",rsParam->lstLow[i]);
if (rsParam->lstHi[i] != NULL) printf(" Hi
%s\n",rsParam->lstHi[i]);
}
for (i = 0; i < rsParam->numOP; i++){
- printf("OP[%d] = %s\n", i, rsParam->lstOPs[i]);
+ printf("OP[%d] = %s\n", i, rsParam->lstOPstr[i]);
}
}
@@ -139,18 +147,18 @@ static
void freeParams(RdfScanParams *rsParam){
int i;
for (i = 0; i < rsParam->numRP; i++){
- GDKfree(rsParam->lstRPs[i]);
+ GDKfree(rsParam->lstRPstr[i]);
if (rsParam->lstLow[i] != NULL) GDKfree(rsParam->lstLow[i]);
if (rsParam->lstHi[i] != NULL) GDKfree(rsParam->lstHi[i]);
}
- GDKfree(rsParam->lstRPs);
+ GDKfree(rsParam->lstRPstr);
GDKfree(rsParam->lstLow);
GDKfree(rsParam->lstHi);
for (i = 0; i < rsParam->numOP; i++){
- GDKfree(rsParam->lstOPs[i]);
+ GDKfree(rsParam->lstOPstr[i]);
}
- GDKfree(rsParam->lstOPs);
+ GDKfree(rsParam->lstOPstr);
GDKfree(rsParam);
diff --git a/monetdb5/extras/rdf/rdfscan.h b/monetdb5/extras/rdf/rdfscan.h
--- a/monetdb5/extras/rdf/rdfscan.h
+++ b/monetdb5/extras/rdf/rdfscan.h
@@ -31,11 +31,13 @@
typedef struct RdfScanParams {
char schema[20];
- char** lstRPs; /* List of required params */
- char** lstLow; /* Low bound of RP*/
- char** lstHi; /* High bound of RP*/
+ char **lstRPstr; /* List of required params (string format) */
+ oid *lstRP; /* List of required params' oids */
+ char **lstLow; /* Low bound of RP*/
+ char **lstHi; /* High bound of RP*/
int numRP; /* Number of required params */
- char** lstOPs; /* List of optional params */
+ char **lstOPstr; /* List of optional params (string format) */
+ oid *lstOP; /* List of optional params' oids */
int numOP;
} RdfScanParams;
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
@@ -10091,41 +10091,6 @@ str getOrigPbt(oid *pbt, oid *origPbt, B
return MAL_SUCCEED;
}
-static
-str triplesubsort(BAT **sbat, BAT **pbat, BAT **obat){
-
- BAT *o1,*o2,*o3;
- BAT *g1,*g2,*g3;
- BAT *S = NULL, *P = NULL, *O = NULL;
-
- S = *sbat;
- P = *pbat;
- O = *obat;
- /* order SPO/SOP */
- if (BATsubsort(sbat, &o1, &g1, S, NULL, NULL, 0, 0) == GDK_FAIL){
- if (S != NULL) BBPreclaim(S);
- throw(RDF, "rdf.triplesubsort", "Fail in sorting for S");
- }
-
- if (BATsubsort(pbat, &o2, &g2, P, o1, g1, 0, 0) == GDK_FAIL){
- BBPreclaim(S);
- if (P != NULL) BBPreclaim(P);
- throw(RDF, "rdf.triplesubsort", "Fail in sub-sorting for P");
- }
- if (BATsubsort(obat, &o3, &g3, O, o2, g2, 0, 0) == GDK_FAIL){
- BBPreclaim(S);
- BBPreclaim(P);
- if (O != NULL) BBPreclaim(O);
- throw(RDF, "rdf.triplesubsort", "Fail in sub-sorting for O");
- }
-
- BBPunfix(o2->batCacheid);
- BBPunfix(g2->batCacheid);
- BBPunfix(o3->batCacheid);
- BBPunfix(g3->batCacheid);
-
- return MAL_SUCCEED;
-}
static
BAT* BATnewPropSet(int ht, int tt, BUN cap){
@@ -11254,6 +11219,9 @@ str RDFdistTriplesToCSs(int *ret, bat *s
*
* Input String --> Original TKNZR oid --> New Oid (TKNRZ_to_new_MapBAT)
* New oid --> Original TKNZR oid --> Input String (New_to_TKNZR_MapBat)
+ *
+ * To convert from mapId to tknz Id, we use the lmap, rmap bats.
+ *
* */
static
str buildTKNZRMappingBat(BAT *lmap, BAT *rmap){
@@ -11263,9 +11231,30 @@ str buildTKNZRMappingBat(BAT *lmap, BAT
int ret;
int num = 0;
bat mapBatId;
- BAT *mapBat;
- str batname = NULL;
-
+ BAT *tmpmapBat = NULL, *pMapBat = NULL;
+ str bname = NULL, bnamelBat = NULL, bnamerBat = NULL;
+ bat *lstCommits = NULL;
+
+ /* Check if the bat has already built */
+ bname = (str) GDKmalloc(50 * sizeof(char));
+ snprintf(bname, 50, "tknzr_to_map");
+
+ bnamelBat = (str) GDKmalloc(50 * sizeof(char));
+ snprintf(bnamelBat, 50, "map_to_tknz_left");
+
+ bnamerBat = (str) GDKmalloc(50 * sizeof(char));
+ snprintf(bnamerBat, 50, "map_to_tknz_right");
+
+
+ mapBatId = BBPindex(bname);
+ if (mapBatId != 0){
+ printf("The tokenizer-mapping-bat %s has been built \n",
bname);
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
https://www.monetdb.org/mailman/listinfo/checkin-list
