Changeset: 50a2cd0713ec for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=50a2cd0713ec
Modified Files:
monetdb5/extras/rdf/hashmap/hashmap.c
monetdb5/extras/rdf/hashmap/hashmap.h
monetdb5/extras/rdf/rdf.h
monetdb5/extras/rdf/rdfschema.c
monetdb5/extras/rdf/rdfschema.h
monetdb5/extras/rdf/rdfschema.mal
sql/backends/monet5/sql.mx
Branch: rdf
Log Message:
Check the foreign key relationship from RDF triples.
diffs (truncated from 402 to 300 lines):
diff --git a/monetdb5/extras/rdf/hashmap/hashmap.c
b/monetdb5/extras/rdf/hashmap/hashmap.c
--- a/monetdb5/extras/rdf/hashmap/hashmap.c
+++ b/monetdb5/extras/rdf/hashmap/hashmap.c
@@ -358,6 +358,34 @@ int hashmap_statistic_CSbysupport(map_t
}
+/* Simply print all the elements in the hashmap */
+int hashmap_print(map_t in){
+
+ int i,j ;
+
+
+ /* Cast the hashmap */
+ hashmap_map* m = (hashmap_map*) in;
+
+ /* On empty hashmap, return immediately */
+ if (hashmap_length(m) <= 0)
+ return MAP_MISSING;
+
+ /* Linear probing */
+ for(i = 0; i< m->table_size; i++)
+ if(m->data[i].in_use != 0) {
+ /* Print each item */
+ printf("Key: ");
+ for (j = 0; j < m->data[i].num; j++){
+ printf(" %d ", m->data[i].key[j]);
+ }
+ printf("\n");
+ printf("Freq: %d \n", m->data[i].freq);
+ /* m->data[i].data */
+ }
+ return MAP_OK;
+}
+
/*
* Collect the number of CSs cummulatively for support values ranging
* from 1 to maxfreqthreshold
diff --git a/monetdb5/extras/rdf/hashmap/hashmap.h
b/monetdb5/extras/rdf/hashmap/hashmap.h
--- a/monetdb5/extras/rdf/hashmap/hashmap.h
+++ b/monetdb5/extras/rdf/hashmap/hashmap.h
@@ -37,7 +37,7 @@ typedef any_t map_t;
/* We need to keep keys and values */
typedef struct _hashmap_element{
int* key;
- int num;
+ int num; /* Number of integers for the key */
int freq;
char in_use;
any_t data;
@@ -139,6 +139,13 @@ extern int hashmap_remove(map_t in, int*
*/
extern int hashmap_get_one(map_t in, any_t *arg, int remove);
+/*
+ * Simply print all the elements in the hashmap
+ *
+ * */
+extern int hashmap_print(map_t in);
+
+
/*
* Free the hashmap
*/
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
@@ -52,9 +52,9 @@ rdf_export str
RDFleftfetchjoin_sorted(int *result, int* lid, int *rid);
typedef enum {
+ URI,
DATETIME,
NUMERIC,
- URI,
STRING
} ObjectType;
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
@@ -114,26 +114,26 @@ CS* creatCS(int subId, int numP, int* bu
* If yes, add that frequent CS to the freqCSset.
*
* */
-static void putaCStoHash(map_t csmap, int* buff, int num, oid *csoid, char
isStoreFreqCS, int freqThreshold, CSset **freqCSset){
+static void putaCStoHash(map_t csmap, int* key, int num, oid *csoid, char
isStoreFreqCS, int freqThreshold, CSset **freqCSset){
oid *getCSoid;
oid *putCSoid;
int err;
- int* cs;
+ int* csKey;
int freq = 0;
CS *freqCS;
- cs = (int*) malloc(sizeof(int) * num);
- if (cs==NULL){
+ csKey = (int*) malloc(sizeof(int) * num);
+ if (csKey==NULL){
printf("Malloc failed. at %d", num);
exit(-1);
}
- copyIntSet(cs, buff, num);
- if (hashmap_get(csmap, cs, num,(void**)(&getCSoid),1, &freq) != MAP_OK){
+ copyIntSet(csKey, key, num);
+ if (hashmap_get(csmap, csKey, num,(void**)(&getCSoid),1, &freq) !=
MAP_OK){
putCSoid = malloc(sizeof(oid));
*putCSoid = *csoid;
- err = hashmap_put(csmap, cs, num, 1, putCSoid);
+ err = hashmap_put(csmap, csKey, num, 1, putCSoid);
assert(err == MAP_OK);
(*csoid)++;
@@ -142,11 +142,11 @@ static void putaCStoHash(map_t csmap, in
if (isStoreFreqCS == 1){ /* Store the frequent CS to the
CSset*/
//printf("FreqCS: Support = %d, Threshold %d \n ",
freq, freqThreshold);
if (freq == freqThreshold){
- freqCS = creatCS(*getCSoid, num, buff);
+ freqCS = creatCS(*getCSoid, num, key);
addCStoSet(*freqCSset, *freqCS);
}
}
- free(cs);
+ free(csKey);
}
}
@@ -234,7 +234,7 @@ void getMaximumFreqCSs(CSset *freqCSset)
-static void putPtoHash(map_t pmap, int value, oid *poid, int support){
+static void putPtoHash(map_t pmap, int key, oid *poid, int support){
oid *getPoid;
oid *putPoid;
int err;
@@ -242,7 +242,7 @@ static void putPtoHash(map_t pmap, int v
pkey = (int*) malloc(sizeof(int));
- *pkey = value;
+ *pkey = key;
if (hashmap_get_forP(pmap, pkey,(void**)(&getPoid)) != MAP_OK){
putPoid = malloc(sizeof(oid));
@@ -351,6 +351,7 @@ RDFextractCS(int *ret, bat *sbatid, bat
int maxNumProp = 0;
CSset *freqCSset; /* Set of frequent CSs */
+
buff = (int *) malloc (sizeof(int) * INIT_PROPERTY_NUM);
if ((sbat = BATdescriptor(*sbatid)) == NULL) {
@@ -431,10 +432,179 @@ RDFextractCS(int *ret, bat *sbatid, bat
return MAL_SUCCEED;
}
+/*
+ * Get the refer CS
+ * Input: oid of a URI object
+ * Return the id of the CS
+ * */
+static
+str getReferCS(BAT *sbat, BAT *pbat, oid *obt){
+
+ /* For detecting foreign key relationships */
+ BAT *tmpbat; /* Get the result of searching objectURI from
sbat */
+ BATiter ti;
+ oid *tbt;
+ BUN pt, qt;
+ oid *s_t, *p_t;
+ //int *tmpbuff;
+
+ //tmpbuff = (int *) malloc (sizeof(int) * INIT_PROPERTY_NUM);
+
+ /* BATsubselect(inputbat, <dont know yet>, lowValue, Highvalue,
isIncludeLowValue, isIncludeHigh, <anti> */
+ printf("Checking for object " BUNFMT "\n", *obt);
+ tmpbat = BATsubselect(sbat, NULL, obt, obt, 1, 1, 0);
+ /* tmpbat tail contain head oid of sbat for matching elements */
+ if (tmpbat != NULL){
+ printf("Matching: " BUNFMT "\n", BATcount(tmpbat));
+ BATprint(tmpbat);
+
+ if (BATcount(tmpbat) > 0){
+ ti = bat_iterator(tmpbat);
+ BATloop(tmpbat, pt, qt){
+ tbt = (oid *) BUNtail(ti, pt);
+ s_t = (oid *) Tloc(sbat, *tbt);
+ p_t = (oid *) Tloc(pbat, *tbt);
+ printf("s_t: " BUNFMT "\n", (*s_t));
+ printf("p_t: " BUNFMT "\n", (*p_t));
+ /* Check which CS is referred */
+
+ }
+ }
+ }
+ else
+ throw(MAL, "rdf.RDFextractCSwithTypes", "Null Bat returned for
BATsubselect");
+
+
+
+ /* temporarily use */
+ if (tmpbat)
+ BBPunfix(tmpbat->batCacheid);
+
+ return MAL_SUCCEED;
+}
+
+/* Extract CS from SPO triples table */
+str
+RDFextractCSwithTypes(int *ret, bat *sbatid, bat *pbatid, bat *obatid, int
*freqThreshold){
+ BUN p, q;
+ BAT *sbat = NULL, *pbat = NULL, *obat = NULL;
+ BATiter si, pi, oi; /*iterator for BAT of s,p,o columns in spo
table */
+ oid *sbt, *pbt, *obt;
+ oid curS; /* current Subject oid */
+ oid curP; /* current Property oid */
+ oid CSoid = 0; /* Characteristic set oid */
+ int numP; /* Number of properties for current S */
+ map_t csMap;
+ int* buff;
+ int INIT_PROPERTY_NUM = 5000;
+ int maxNumProp = 0;
+ CSset *freqCSset; /* Set of frequent CSs */
+ oid objType;
+
+
+
+ buff = (int *) malloc (sizeof(int) * INIT_PROPERTY_NUM);
+
+ if ((sbat = BATdescriptor(*sbatid)) == NULL) {
+ throw(MAL, "rdf.RDFextractCSwithTypes", RUNTIME_OBJECT_MISSING);
+ }
+ if (!(sbat->tsorted)){
+ throw(MAL, "rdf.RDFextractCSwithTypes", "sbat is not sorted");
+ }
+
+ if ((pbat = BATdescriptor(*pbatid)) == NULL) {
+ throw(MAL, "rdf.RDFextractCSwithTypes", RUNTIME_OBJECT_MISSING);
+ }
+ if ((obat = BATdescriptor(*obatid)) == NULL) {
+ throw(MAL, "rdf.RDFextractCSwithTypes", RUNTIME_OBJECT_MISSING);
+ }
+
+ si = bat_iterator(sbat);
+ pi = bat_iterator(pbat);
+ oi = bat_iterator(obat);
+
+ /* Init a hashmap */
+ csMap = hashmap_new();
+ freqCSset = initCSset();
+
+ numP = 0;
+ curP = 0;
+
+ printf("freqThreshold = %d \n", *freqThreshold);
+ BATloop(sbat, p, q){
+ sbt = (oid *) BUNtloc(si, p);
+ if (*sbt != curS){
+ if (p != 0){ /* Not the first S */
+ putaCStoHash(csMap, buff, numP, &CSoid, 1,
*freqThreshold, &freqCSset);
+
+ if (numP > maxNumProp)
+ maxNumProp = numP;
+ }
+ curS = *sbt;
+ curP = 0;
+ numP = 0;
+ }
+
+ pbt = (oid *) BUNtloc(pi, p);
+
+ if (numP > INIT_PROPERTY_NUM){
+ throw(MAL, "rdf.RDFextractCS", "# of properties is
greater than INIT_PROPERTY_NUM");
+ exit(-1);
+ }
+
+ if (curP != *pbt){ /* Multi values property */
+ buff[numP] = *pbt;
+ numP++;
+ curP = *pbt;
+ }
+
+ obt = (oid *) BUNtloc(oi, p);
+ /* Check type of object */
+ objType = ((*obt) >> (sizeof(BUN)*8 - 3)) & 3 ; /* Get
two bits 63th, 62nd from object oid */
+
+ //printf("object type: " BUNFMT "\n", objType);
+
+ /* Look at sbat*/
+ if (objType == URI){
+
+ getReferCS(sbat, pbat, obt);
+ }
+ }
+
+ /*put the last CS */
+ putaCStoHash(csMap, buff, numP, &CSoid, 1, *freqThreshold, &freqCSset
);
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
http://mail.monetdb.org/mailman/listinfo/checkin-list
