Changeset: 548ec8e2e127 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=548ec8e2e127
Modified Files:
monetdb5/extras/rdf/rdfontologyload.c
monetdb5/extras/rdf/rdfontologyload.h
monetdb5/extras/rdf/rdfschema.c
monetdb5/extras/rdf/rdfschema.h
monetdb5/extras/rdf/rdfschema.mal
Branch: rdf
Log Message:
Use the ontology labels if available for the table name
TODO: Use the ontology property labels if available for the column names in the
final schema
diffs (truncated from 376 to 300 lines):
diff --git a/monetdb5/extras/rdf/rdfontologyload.c
b/monetdb5/extras/rdf/rdfontologyload.c
--- a/monetdb5/extras/rdf/rdfontologyload.c
+++ b/monetdb5/extras/rdf/rdfontologyload.c
@@ -261,7 +261,7 @@ int compareProp (const void * a, const v
}
static
-str buildOntologyClassesInfo(oid **ontmetadat, int ontmetadataCount, oid
**ontattributes, int ontattributesCount){
+str buildOntologyClassesInfo(oid **ontmetadat, int ontmetadataCount, oid
**ontattributes, int ontattributesCount, str *tmpOntLabels){
int i;
oid classOid; //The class Oid comes from
@@ -281,7 +281,7 @@ str buildOntologyClassesInfo(oid **ontme
int tmpNumProp = 0;
oid* buffProps = NULL;
int maxNumPropPerOntology = 1000;
-
+ str *tmpLabelsShortlist = NULL;
//Read all ontmetadata and store them in the ontmetaBat
@@ -291,6 +291,11 @@ str buildOntologyClassesInfo(oid **ontme
if (!(ontmetaBat->T->hash)){
throw(RDF, "buildOntologyClassesInfo", "Cannot allocate the
hash for Bat");
}
+
+ tmpLabelsShortlist = (str *) malloc(sizeof(str) * ontmetadataCount);
+ for (i = 0; i < ontmetadataCount; i++) tmpLabelsShortlist[i] = NULL;
+
+ classIdx = 0;
for (i = 0; i < ontmetadataCount; i++){
classOid = ontmetadat[0][i];
assert(classOid != BUN_NONE);
@@ -299,8 +304,20 @@ str buildOntologyClassesInfo(oid **ontme
if (tmpBun == BUN_NONE){ //If it is a new class
if (BUNappend(ontmetaBat,&classOid, TRUE) == NULL)
throw(RDF, "buildOntologyClassesInfo", "Cannot
insert to ontmetaBat");
+
+ if (tmpOntLabels[i] != NULL){
+ tmpLabelsShortlist[classIdx] =
GDKstrdup(tmpOntLabels[i]);
+ }
+ else
+ tmpLabelsShortlist[classIdx] = NULL;
+
+ classIdx++;
}
+
+ }
+ //Also add super class to list of ontology classes
+ for (i = 0; i < ontmetadataCount; i++){
scOid = ontmetadat[1][i];
if (scOid != BUN_NONE){ //The superClass oid is there
@@ -308,6 +325,9 @@ str buildOntologyClassesInfo(oid **ontme
if (tmpBun == BUN_NONE){ //If it is a new class
if (BUNappend(ontmetaBat, &scOid, TRUE) ==
NULL)
throw(RDF, "buildOntologyClassesInfo",
"Cannot insert to ontmetaBat");
+
+ tmpLabelsShortlist[classIdx] = NULL;
+ classIdx++;
}
}
}
@@ -332,6 +352,11 @@ str buildOntologyClassesInfo(oid **ontme
tmpontclassSet[i].cOid = *tmpOid;
+ if (tmpLabelsShortlist[i] != NULL)
+ tmpontclassSet[i].label =
GDKstrdup(tmpLabelsShortlist[i]);
+ else
+ tmpontclassSet[i].label = NULL;
+
//Init other info
tmpontclassSet[i].scIdxes = (int *) malloc(sizeof(int) *
NUMSC_PER_ONTCLASS);
tmpontclassSet[i].numsc = 0;
@@ -343,6 +368,13 @@ str buildOntologyClassesInfo(oid **ontme
i++;
}
+ //Free
+ for (i = 0; i < numClass; i++){
+ if (tmpLabelsShortlist[i] != NULL)
+ GDKfree(tmpLabelsShortlist[i]);
+ }
+ free(tmpLabelsShortlist);
+
//Add sc
for (i = 0; i < ontmetadataCount; i++){
@@ -425,14 +457,15 @@ str buildOntologyClassesInfo(oid **ontme
}
str
-RDFloadsqlontologies(int *ret, bat *auriid, bat *aattrid, bat *muriid, bat
*msuperid){
+RDFloadsqlontologies(int *ret, bat *auriid, bat *aattrid, bat *muriid, bat
*msuperid, bat *mlabelid){
BUN p, q;
- BAT *auri = NULL, *aattr = NULL, *muri = NULL,
*msuper = NULL;
- BATiter aurii, aattri, murii, msuperi;
- BUN bun, bun2, bun3, bun4;
+ BAT *auri = NULL, *aattr = NULL, *muri = NULL,
*msuper = NULL, *mlabel = NULL;
+ BATiter aurii, aattri, murii, msuperi, mlabeli;
+ BUN bun, bun2, bun3, bun4, bun5;
BUN auriCount, muriCount;
int i;
str schema = "rdf";
+ str *tmpOntLabels = NULL; //Set of ontology labels
TKNZRopen (NULL, &schema);
@@ -462,10 +495,19 @@ RDFloadsqlontologies(int *ret, bat *auri
throw(MAL, "rdf.RDFloadsqlontologies", RUNTIME_OBJECT_MISSING);
}
+ if ((mlabel = BATdescriptor(*mlabelid)) == NULL) {
+ BBPreleaseref(auri->batCacheid);
+ BBPreleaseref(aattr->batCacheid);
+ BBPreleaseref(muri->batCacheid);
+ BBPreleaseref(msuper->batCacheid);
+ throw(MAL, "rdf.RDFloadsqlontologies", RUNTIME_OBJECT_MISSING);
+ }
+
aurii = bat_iterator(auri);
aattri = bat_iterator(aattr);
murii = bat_iterator(muri);
msuperi = bat_iterator(msuper);
+ mlabeli = bat_iterator(mlabel);
// load ontattributes
i = 0;
@@ -517,18 +559,23 @@ RDFloadsqlontologies(int *ret, bat *auri
i = 0;
bun3 = BUNfirst(muri);
bun4 = BUNfirst(msuper);
-
+ bun5 = BUNfirst(mlabel);
+
muriCount = BATcount(muri);
ontmetadata = (oid**) malloc(sizeof(oid *) * 2);
if (!ontmetadata) fprintf(stderr, "ERROR: Couldn't malloc memory!\n");
- ontmetadata[0] = malloc(sizeof(str) * muriCount); // uri
- ontmetadata[1] = malloc(sizeof(str) * muriCount); // superclass
+ ontmetadata[0] = malloc(sizeof(oid) * muriCount); // uri
+ ontmetadata[1] = malloc(sizeof(oid) * muriCount); // superclass
if (!ontmetadata[0] || !ontmetadata[1]) fprintf(stderr, "ERROR:
Couldn't malloc memory!\n");
+ tmpOntLabels = (str*)malloc(sizeof(str) * muriCount); //labels of
ontology classes
+ if (!tmpOntLabels) fprintf(stderr, "ERROR: Couldn't malloc memory!\n");
+
BATloop(muri, p, q){
str muristr = (str) BUNtail(murii, bun3 + i);
str msuperstr = (str) BUNtail(msuperi, bun4 + i);
+ str mlabelstr = (str) BUNtail(mlabeli, bun5 + i);
oid murioid, msuperoid;
@@ -554,6 +601,14 @@ RDFloadsqlontologies(int *ret, bat *auri
} else {
ontmetadata[1][ontmetadataCount] = msuperoid;
}
+
+ if (strcmp(mlabelstr, "\x80") == 0) {
+ tmpOntLabels[ontmetadataCount] = NULL;
+ } else {
+ tmpOntLabels[ontmetadataCount] = GDKstrdup(mlabelstr);
+ }
+
+
ontmetadataCount += 1;
++i;
@@ -562,8 +617,14 @@ RDFloadsqlontologies(int *ret, bat *auri
GDKfree(msuperstr2);
}
- buildOntologyClassesInfo(ontmetadata, ontmetadataCount, ontattributes,
ontattributesCount);
+ buildOntologyClassesInfo(ontmetadata, ontmetadataCount, ontattributes,
ontattributesCount, tmpOntLabels);
+ for (i = 0; i < ontmetadataCount; i++){
+ if (tmpOntLabels[i] != NULL)
+ GDKfree(tmpOntLabels[i]);
+ }
+ free(tmpOntLabels);
+
BBPreclaim(auri);
BBPreclaim(aattr);
BBPreclaim(muri);
diff --git a/monetdb5/extras/rdf/rdfontologyload.h
b/monetdb5/extras/rdf/rdfontologyload.h
--- a/monetdb5/extras/rdf/rdfontologyload.h
+++ b/monetdb5/extras/rdf/rdfontologyload.h
@@ -24,6 +24,7 @@
typedef struct OntClass {
oid cOid; /*class Oid*/
+ char* label; /*ontology label (if available) */
int* scIdxes; /*Idx of super classes*/
int numsc; /*Number of super classes*/
int numAllocation;
@@ -36,7 +37,7 @@ rdf_export str
RDFOntologyParser(int *ret, str *location, str *schema);
rdf_export str
-RDFloadsqlontologies(int *ret, bat *auri, bat *aattr, bat *muri, bat *msuper);
+RDFloadsqlontologies(int *ret, bat *auri, bat *aattr, bat *muri, bat *msuper,
bat *mlabel);
/*
rdf_export str
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
@@ -158,6 +158,15 @@ str printTKNZStringFromOid(oid id){
return MAL_SUCCEED;
}
+
+static
+char isOntologyName(oid valueOid, BUN *ontClassPos){
+ *ontClassPos = BUN_NONE;
+ *ontClassPos = BUNfnd(BATmirror(ontmetaBat), &valueOid);
+ if (*ontClassPos == BUN_NONE) return 0;
+ else return 1;
+}
+
//Get the string for
static
char getStringName(oid objOid, str *objStr, BATiter mapi, BAT *mapbat, char
isTblName){
@@ -166,38 +175,57 @@ char getStringName(oid objOid, str *objS
oid realObjOid;
BUN bun;
int i = 0;
-
- if (objType == URI || objType == BLANKNODE){
- realObjOid = objOid - ((oid)objType << (sizeof(BUN)*8 - 4));
- takeOid(realObjOid, objStr);
- }
- else{
- str tmpObjStr;
- str s;
- int len;
- realObjOid = objOid - (objType*2 + 1) * RDF_MIN_LITERAL; /*
Get the real objOid from Map or Tokenizer */
- bun = BUNfirst(mapbat);
- tmpObjStr = (str) BUNtail(mapi, bun + realObjOid);
-
- *objStr = GDKstrdup(tmpObjStr);
+ char hasOntologyLabel = 0;
+
+ #if USE_ONTLABEL_FOR_NAME
+ if (isTblName){
+ char isOntName = 0;
+ BUN tmpontClassPos = BUN_NONE;
+
+ isOntName = isOntologyName(objOid, &tmpontClassPos);
+
+ if (isOntName == 1){
+ //Check if label is availabel
+ if (ontclassSet[tmpontClassPos].label != NULL){ //Use
this label
+ *objStr =
GDKstrdup(ontclassSet[tmpontClassPos].label);
+ hasOntologyLabel = 1;
+ }
+ }
+ }
+ #endif
+
+ if (hasOntologyLabel == 0){
+ if (objType == URI || objType == BLANKNODE){
+ realObjOid = objOid - ((oid)objType << (sizeof(BUN)*8 -
4));
+ takeOid(realObjOid, objStr);
+ }
+ else{
+ str tmpObjStr;
+ str s;
+ int len;
+ realObjOid = objOid - (objType*2 + 1) *
RDF_MIN_LITERAL; /* Get the real objOid from Map or Tokenizer */
+ bun = BUNfirst(mapbat);
+ tmpObjStr = (str) BUNtail(mapi, bun + realObjOid);
- if (isTblName){
- s = *objStr;
- len = strlen(s);
- //Replace all non-alphabet character by ___
- for (i = 0; i < len; i++)
- {
- //printf("i = %d: %c \n",i, s[i]);
- if (!isalpha(*s)){
- *s = '_';
- }
- s++;
-
- }
- }
-
- }
-
+ *objStr = GDKstrdup(tmpObjStr);
+
+ if (isTblName){
+ s = *objStr;
+ len = strlen(s);
+ //Replace all non-alphabet character by ___
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
https://www.monetdb.org/mailman/listinfo/checkin-list
