Changeset: 1034ded84d4d for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=1034ded84d4d
Modified Files:
monetdb5/extras/rdf/rdf.h
monetdb5/extras/rdf/rdf_shredder.c
monetdb5/extras/rdf/rdf_shredder.mx
monetdb5/extras/rdf/rdfschema.c
monetdb5/extras/rdf/rdfschema.h
Branch: rdf
Log Message:
Change the implementation according to the new design.
- Using BAT for storing hash value of Characteristic set
- Store lists of P's corresponding to each CS in an BAT.
The offset for each list is stored in another BAT
diffs (truncated from 438 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
@@ -64,6 +64,9 @@ typedef enum {
#define STORE TRIPLE_STORE /* this should become a compile time option */
+#define batsz 10000000
+#define smallbatsz 100000
+
#if STORE == TRIPLE_STORE
typedef enum {
S_sort, P_sort, O_sort, /* sorted */
diff --git a/monetdb5/extras/rdf/rdf_shredder.c
b/monetdb5/extras/rdf/rdf_shredder.c
--- a/monetdb5/extras/rdf/rdf_shredder.c
+++ b/monetdb5/extras/rdf/rdf_shredder.c
@@ -37,7 +37,7 @@ typedef struct graphBATdef {
int tailType; /* type of right column */
} graphBATdef;
-static BUN batsz = 10000000;
+//static BUN batsz = 10000000;
/* this list should be kept alligned with the graphBATType enum */
#if STORE == TRIPLE_STORE
diff --git a/monetdb5/extras/rdf/rdf_shredder.mx
b/monetdb5/extras/rdf/rdf_shredder.mx
--- a/monetdb5/extras/rdf/rdf_shredder.mx
+++ b/monetdb5/extras/rdf/rdf_shredder.mx
@@ -40,8 +40,6 @@ typedef struct graphBATdef {
int tailType; /* type of right column */
} graphBATdef;
-static BUN batsz = 10000000;
-
/* this list should be kept alligned with the graphBATType enum */
#if STORE == TRIPLE_STORE
static graphBATdef graphdef[N_GRAPH_BAT] = {
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
@@ -26,6 +26,7 @@
#include <gdk.h>
#include <hashmap/hashmap.h>
+
str
RDFSchemaExplore(int *ret, str *tbname, str *clname)
{
@@ -107,6 +108,75 @@ CS* creatCS(int subId, int numP, int* bu
return cs;
}
+
+/*
+ * Hashing function for a set of values
+ * Rely on djb2 http://www.cse.yorku.ca/~oz/hash.html
+ *
+ */
+static unsigned int RDF_hash_intlist(int* key, int num){
+ unsigned int hashCode = 5381u;
+ int i;
+
+ for (i = 0; i < num; i++){
+ hashCode = ((hashCode << 5) + hashCode) + key[i];
+ }
+
+ // return 0x7fffffff & hashCode
+ return hashCode;
+}
+
+static
+void appendArrayToBat(BAT *b, int* inArray, int num){
+ int i;
+ BUN r = BUNlast(b);
+ if (r + num < b->batCapacity){
+ BATextend(b, smallbatsz);
+ }
+ for (i = 0; i < num; i++){
+ memcpy(Tloc(b, BUNlast(b)), inArray, sizeof(int) * num);
+ }
+ BATsetcount(b, (BUN) (b->batCount + num));
+
+}
+
+static
+void checkCSduplication(BAT* pOffsetBat, BAT* fullPBat, BUN pos, int* key, int
numK){
+ BUN *offset;
+ BUN *offset2;
+ int numP;
+ int i;
+ BUN *existvalue;
+
+ offset = (BUN *) Tloc(pOffsetBat, pos);
+ if ((pos + 1) < pOffsetBat->batCount){
+ offset2 = (BUN *)Tloc(pOffsetBat, pos + 1);
+ }
+ else{
+ offset2 = malloc(sizeof(BUN));
+ *offset2 = BUNlast(fullPBat);
+ }
+
+ numP = *offset2 - *offset;
+
+ // Check each value
+ if (numK != numP) {
+ printf("No duplication \n");
+ return;
+ }
+ else{
+ existvalue = (BUN *)Tloc(fullPBat, *offset);
+ for (i = 0; i < numP; i++){
+ if (key[i] != (int)*existvalue++) {
+ printf("No duplication \n");
+ return;
+ }
+ }
+ }
+
+ printf("There is duplication \n");
+ return;
+}
/*
* Put a CS to the hashmap.
* While putting CS to the hashmap, update the support (frequency) value
@@ -114,39 +184,44 @@ CS* creatCS(int subId, int numP, int* bu
* If yes, add that frequent CS to the 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* csKey;
+static
+void putaCStoHash(BAT* hsKeyBat, BAT* pOffsetBat, BAT* fullPBat, oid subjId,
int* key, int num,
+ oid *csoid, char isStoreFreqCS, int freqThreshold, CSset
**freqCSset){
+ int csKey;
int freq = 0;
CS *freqCS;
+ BUN bun;
+ BUN offset;
- csKey = (int*) malloc(sizeof(int) * num);
- if (csKey==NULL){
- printf("Malloc failed. at %d", num);
- exit(-1);
- }
+ csKey = RDF_hash_intlist(key, num);
+ bun = BUNfnd(BATmirror(hsKeyBat),(ptr) &csKey);
+ if (bun == BUN_NONE) {
+ if (hsKeyBat->T->hash && BATcount(hsKeyBat) > 4 *
hsKeyBat->T->hash->mask) {
+ HASHdestroy(hsKeyBat);
+ BAThash(BATmirror(hsKeyBat), 2*BATcount(hsKeyBat));
+ }
+ hsKeyBat = BUNappend(hsKeyBat, (ptr) &csKey, TRUE);
- copyIntSet(csKey, key, num);
- if (hashmap_get(csmap, csKey, num,(void**)(&getCSoid),1, &freq) !=
MAP_OK){
- putCSoid = malloc(sizeof(oid));
- *putCSoid = *csoid;
+ (*csoid)++;
- err = hashmap_put(csmap, csKey, num, 1, putCSoid);
- assert(err == MAP_OK);
+ offset = BUNlast(fullPBat);
+ /* Add list of p to fullPBat and pOffsetBat*/
+ BUNappend(pOffsetBat, &offset , TRUE);
+ appendArrayToBat(fullPBat, key, num);
- (*csoid)++;
}
else{
+ printf("This CS exists \n");
+ /* Check whether it is really an duplication (same hashvalue
but different list of */
+ checkCSduplication(pOffsetBat, fullPBat, bun, key, num );
+
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, key);
+ freqCS = creatCS(subjId, num, key);
addCStoSet(*freqCSset, *freqCS);
}
}
- free(csKey);
}
}
@@ -334,128 +409,29 @@ static void getStatisticCSsBySupports(ma
free(statCS);
}
-/* Extract CS from SPO triples table */
-str
-RDFextractCS(int *ret, bat *sbatid, bat *pbatid, int *freqThreshold){
- BUN p, q;
- BAT *sbat = NULL, *pbat = NULL;
- BATiter si, pi; /*iterator for BAT of s,p columns in spo table
*/
- oid *bt, *pbt;
- 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 */
-
-
- buff = (int *) malloc (sizeof(int) * INIT_PROPERTY_NUM);
-
- if ((sbat = BATdescriptor(*sbatid)) == NULL) {
- throw(MAL, "rdf.RDFextractCS", RUNTIME_OBJECT_MISSING);
- }
- if ((pbat = BATdescriptor(*pbatid)) == NULL) {
- throw(MAL, "rdf.RDFextractCS", RUNTIME_OBJECT_MISSING);
- }
-
- si = bat_iterator(sbat);
- pi = bat_iterator(pbat);
-
- /* Init a hashmap */
- csMap = hashmap_new();
- freqCSset = initCSset();
-
- numP = 0;
- curP = 0;
-
- printf("freqThreshold = %d \n", *freqThreshold);
- BATloop(sbat, p, q){
- bt = (oid *) BUNtloc(si, p);
- if (*bt != curS){
- if (p != 0){ /* Not the first S */
- putaCStoHash(csMap, buff, numP, &CSoid, 1,
*freqThreshold, &freqCSset);
-
- if (numP > maxNumProp)
- maxNumProp = numP;
- }
- curS = *bt;
- 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;
- }
- //printf("Travel sbat at %d value: %d , for pbat: %d \n", (int)
p, (int) *bt, (int) *pbt);
- }
-
- /*put the last CS */
- putaCStoHash(csMap, buff, numP, &CSoid, 1, *freqThreshold, &freqCSset
);
-
- if (numP > maxNumProp)
- maxNumProp = numP;
-
- printf("Number of frequent CSs is: %d \n", freqCSset->numCSadded);
-
- /*get the statistic */
-
- getTopFreqCSs(csMap,*freqThreshold);
-
- getMaximumFreqCSs(freqCSset);
-
- //getStatisticCSsBySize(csMap,maxNumProp);
-
- getStatisticCSsBySupports(csMap, 5000, 1, 0);
-
- BBPreclaim(sbat);
- BBPreclaim(pbat);
-
- free (buff);
-
- freeCSset(freqCSset);
-
- hashmap_free(csMap);
-
- *ret = 1;
- 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){
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
http://mail.monetdb.org/mailman/listinfo/checkin-list
