Changeset: 660366c5d68d for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=660366c5d68d
Modified Files:
monetdb5/extras/rdf/hashmap/hashmap.c
monetdb5/extras/rdf/hashmap/hashmap.h
monetdb5/extras/rdf/rdf_shredder.mx
monetdb5/extras/rdf/rdfschema.c
Branch: rdf
Log Message:
Modify the log handler for Raptor parser
diffs (217 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
@@ -13,7 +13,8 @@
/* We need to keep keys and values */
typedef struct _hashmap_element{
int* key;
- int num;
+ int num;
+ int freq;
int in_use;
any_t data;
} hashmap_element;
@@ -171,6 +172,7 @@ int hashmap_put(map_t in, int* key, int
m->data[index].key = key;
m->data[index].num = num;
m->data[index].in_use = 1;
+ m->data[index].freq = 1;
m->size++;
return MAP_OK;
@@ -179,7 +181,7 @@ int hashmap_put(map_t in, int* key, int
/*
* Get your pointer out of the hashmap with a key
*/
-int hashmap_get(map_t in, int* key, int num, any_t *arg){
+int hashmap_get(map_t in, int* key, int num, any_t *arg, char
isUpdateFrequency, int *retfreq){
int curr;
int i;
hashmap_map* m;
@@ -193,14 +195,17 @@ int hashmap_get(map_t in, int* key, int
/* Linear probing, if necessary */
for(i = 0; i<MAX_CHAIN_LENGTH; i++){
- int in_use = m->data[curr].in_use;
- if (in_use == 1){
- if ((m->data[curr].num == num) &&
(intsetcmp(m->data[curr].key,key,num)==0)){
- *arg = (m->data[curr].data);
- return MAP_OK;
- }
+ int in_use = m->data[curr].in_use;
+ if (in_use == 1){
+ if ((m->data[curr].num == num) &&
(intsetcmp(m->data[curr].key,key,num)==0)){
+ *arg = (m->data[curr].data);
+ if (isUpdateFrequency == 1){
+ m->data[curr].freq++;
+ *retfreq = m->data[curr].freq;
+ }
+ return MAP_OK;
+ }
}
-
curr = (curr + 1) % m->table_size;
}
@@ -210,6 +215,8 @@ int hashmap_get(map_t in, int* key, int
return MAP_MISSING;
}
+
+
/*
* Iterate the function parameter over each element in the hashmap. The
* additional any_t argument is passed to the function as its first
@@ -239,6 +246,37 @@ int hashmap_iterate(map_t in, PFany f, a
}
/*
+ * This function is used for filtering the CSs only:
+ *
+ * Iterate over the hashmap and show the values from
+ * all the hash element that has the frequent value > freqthreshold
+ *
+ * */
+
+int hashmap_iterate_threshold(map_t in, int freqthreshold){
+
+ int i;
+
+ /* 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) {
+ if (m->data[i].freq > freqthreshold){
+ //any_t data = (any_t) (m->data[i].data);
+
+ }
+ }
+
+ return MAP_OK;
+}
+
+/*
* Remove an element with that key from the map
*/
int hashmap_remove(map_t in, int* key, int num){
@@ -261,6 +299,7 @@ int hashmap_remove(map_t in, int* key, i
/* Blank out the fields */
m->data[curr].in_use = 0;
m->data[curr].num = 0;
+ m->data[curr].freq = 1;
m->data[curr].data = NULL;
m->data[curr].key = NULL;
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
@@ -51,6 +51,15 @@ extern map_t hashmap_new(void);
extern int hashmap_iterate(map_t in, PFany f, any_t item);
/*
+ * This function is used for filtering the CSs only:
+ *
+ * Iterate over the hashmap and show the values from
+ * all the hash element that has the frequent value > freqthreshold
+ *
+ * */
+extern int hashmap_iterate_threshold(map_t in, int freqthreshold);
+
+/*
* Add an element to the hashmap. Return MAP_OK or MAP_OMEM.
*/
extern int hashmap_put(map_t in, int* key, int num, any_t value);
@@ -58,7 +67,7 @@ extern int hashmap_put(map_t in, int* ke
/*
* Get an element from the hashmap. Return MAP_OK or MAP_MISSING.
*/
-extern int hashmap_get(map_t in, int* key, int num, any_t *arg);
+extern int hashmap_get(map_t in, int* key, int num, any_t *arg, char
isUpdateFreq, int *retfreq);
/*
* Remove an element from the hashmap. Return MAP_OK or MAP_MISSING.
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
@@ -108,20 +108,20 @@ raptor_parser_parse_abort (@1->rparser);
@= rdf_parser_handler
static void
-@1Handler (void *user_data, raptor_locator* locator,
- const char *message)
+@1Handler (void *user_data, raptor_log_message* message)
{
- parserData *pdata = (parserData *) user_data;
- pdata->@1Msg = GDKstrdup(message);
+ parserData *pdata = ((parserData *) user_data);
+ pdata->@1Msg = GDKstrdup(message->text);
mnstr_printf(GDKout, "rdflib: @1:%s\n", pdata->@1Msg);
pdata->@1++;
/* check for a valid locator object and only then use it */
- if (locator != NULL) {
- pdata->line = locator->line;
- pdata->column = locator->column;
- } else {
- }
+ if (message->locator != NULL) {
+ pdata->line = message->locator->line;
+ pdata->column = message->locator->column;
+ mnstr_printf(GDKout, "rdflib: @1: at line %d column %d\n",
pdata->line, pdata->column);
+ }
+
}
@
@@ -563,6 +563,7 @@ RDFParser (BAT **graph, str *location, s
@:set_handlers(world, rparser, pdata)@
//raptor_parser_set_option(rparser, 0); //MDPHAM: CHECK FOR THIS SETTING
+ raptor_parser_set_option(rparser, RAPTOR_OPTION_SCANNING);
/* Parse URI or local file. */
ret = URLisaURL(&isURI, location);
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
@@ -44,11 +44,12 @@ static void putCStoHash(map_t csmap, int
oid *getCSoid;
oid *putCSoid;
int err;
- int* cs;
+ int* cs;
+ int freq;
cs = (int*) malloc(sizeof(int) * num);
copyIntSet(cs, buff, num);
- if (hashmap_get(csmap, cs, num,(void**)(&getCSoid)) != MAP_OK){
+ if (hashmap_get(csmap, cs, num,(void**)(&getCSoid),1, &freq) != MAP_OK){
putCSoid = malloc(sizeof(oid));
*putCSoid = *csoid;
@@ -60,12 +61,18 @@ static void putCStoHash(map_t csmap, int
(*csoid)++;
}
else{
- printf("The key %d exists in the hashmap \n", (int) *getCSoid);
+ printf("The key %d exists in the hashmap with freq %d \n",
(int) *getCSoid, freq);
free(cs);
}
}
+/*
+static void getTopFreqCSs(map_t csmap){
+
+}
+*/
+
str
RDFextractCS(int *ret, bat *sbatid, bat *pbatid){
BUN p, q;
_______________________________________________
checkin-list mailing list
[email protected]
http://mail.monetdb.org/mailman/listinfo/checkin-list
