Changeset: 650aa35e15c4 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=650aa35e15c4
Modified Files:
monetdb5/extras/rdf/rdf.h
monetdb5/extras/rdf/rdf_shredder.mx
monetdb5/modules/mal/tokenizer.h
sql/backends/monet5/sql.mx
Branch: rdf
Log Message:
Change two bits in the oid for the RDF triple object so that
the oid can specify the type of that object
diffs (188 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
@@ -51,7 +51,12 @@ RDFleftfetchjoin_sortedestimate(int *res
rdf_export str
RDFleftfetchjoin_sorted(int *result, int* lid, int *rid);
-
+typedef enum {
+ DATETIME,
+ NUMERIC,
+ URI,
+ STRING
+} ObjectType;
#define IS_DUPLICATE_FREE 0 /* 0: Duplications have not been removed,
otherwise 1 */
#define TRIPLE_STORE 1
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
@@ -150,8 +150,29 @@ if (bun == BUN_NONE) {
HASHdestroy(@1);
BAThash(BATmirror(@1), 2*BATcount(@1));
}
- bun = (BUN) ((@1)->hseqbase + (@1)->batCount);
- @1 = BUNappend(@1, (ptr)@2, TRUE);
+ //bun = (BUN) ((@1)->hseqbase + (@1)->batCount);
+ bun = (BUN) (RDF_MIN_LITERAL + (@1)->batCount);
+
+ /* Add the type here */
+ if (@3 == DATETIME){
+ printf("Datetime appears here \n Before: " BUNFMT "\n", bun);
+ bun |= (BUN)1 << (sizeof(BUN)*8 - 2);
+ printf("After: " BUNFMT "\n", bun);
+ }
+ else if (@3 == NUMERIC){
+ printf("Numeric value appears here \n Before: " BUNFMT "\n",
bun);
+ bun |= (BUN)2 << (sizeof(BUN)*8 - 2);
+ printf("After: " BUNFMT "\n", bun);
+ }
+ else { /* @3 == STRING */
+ printf("String value appears here \n Before: " BUNFMT "\n",
bun);
+ bun |= (BUN)3 << (sizeof(BUN)*8 - 2);
+ printf("After: " BUNFMT "\n", bun);
+ }
+
+ //@1 = BUNappend(@1, (ptr)@2, TRUE);
+ @1 = BUNins(@1, (ptr) &bun, (ptr)@2, TRUE);
+
if (@1 == NULL) {
@:raptor_exception(pdata, "could not append in@1")@
}
@@ -187,6 +208,39 @@ if (@1 == NULL) {
@
@c
+
+
+/*
+* Get the specific type of the object value in an RDF triple
+* The URI object can be recoginized by raptor parser.
+* If the object value is not an URI ==> it is a literal, and
+* specifically, a numeric, a dateTime or a string.
+* This function will find the specific type of Object value
+*/
+
+static ObjectType
+getObjectType(unsigned char* objStr){
+ ObjectType obType;
+ if (strstr((const char*) objStr, "XMLSchema#date") != NULL){
+ obType = DATETIME;
+ printf("%s: DateTime \n", objStr);
+ }
+ else if (strstr((const char*) objStr, "XMLSchema#float") != NULL
+ || strstr((const char*) objStr, "XMLSchema#integer") != NULL
+ )
+ {
+ obType = NUMERIC;
+ printf("%s: Numeric \n", objStr);
+ }
+ else {
+ obType = STRING;
+ printf("%s: String \n", objStr);
+ }
+
+ return obType;
+}
+
+
static void
tripleHandler(void* user_data, const raptor_statement* triple)
{
@@ -227,8 +281,10 @@ tripleHandler(void* user_data, const rap
free(objStr);
} else if (triple->object->type == RAPTOR_TERM_TYPE_LITERAL) {
unsigned char* objStr;
+ ObjectType objType;
objStr = raptor_term_to_string(triple->object);
- @:rdf_BUNappend_unq_1(graph[MAP_LEX], (str)objStr)@
+ objType = getObjectType(objStr);
+ @:rdf_BUNappend_unq_1(graph[MAP_LEX], (str)objStr, objType)@
@:rdf_BUNappend(graph[O_sort], &bun)@
bun = BUN_NONE;
free(objStr);
@@ -323,9 +379,14 @@ parserData_create (str location, BAT** g
pdata->graph[MAP_LEX]->T->nokey[0] = 0;
pdata->graph[MAP_LEX]->T->nokey[1] = 0;
+ /* Reset the dense property of graph[MAP_LEX] */
+ pdata->graph[MAP_LEX]->hdense = FALSE;
+
return pdata;
}
+
+
/*
* @-
* After the RDF document has been shredded into 3 bats and a lexical value
@@ -392,11 +453,15 @@ post_processing (parserData *pdata)
BAT *ctref= NULL;
#endif
#ifdef _TKNZR_H
- BATiter bi, mi;
- BUN p, d, r;
- oid *bt;
+
+ //BATiter bi, mi;
+ //BUN p, d, r;
+ //oid *bt;
/* order MAP_LEX */
+
+ /* Do not order the MAP_LEX BAT */
+ #ifdef ORDER_MAPLEX
BATorder(BATmirror(graph[MAP_LEX]));
map_oid = BATmark(graph[MAP_LEX], RDF_MIN_LITERAL); /* BATmark will
create a copy */
BATorder(map_oid);
@@ -418,6 +483,8 @@ post_processing (parserData *pdata)
}
BBPreclaim(map_oid);
+ #endif
+
S = graph[S_sort];
P = graph[P_sort];
O = graph[O_sort];
@@ -583,7 +650,7 @@ RDFParser (BAT **graph, str *location, s
char *buf = (char*) GDKmalloc(RDF_CHUNK_SIZE);
if (buf == NULL) {
throw(RDF, "rdf.rdfShred",
- "could not allocate a %dMB file buffer\n",
(int) (RDF_CHUNK_SIZE>>20));
+ "could not allocate a %dMB file buffer\n", (int)
(RDF_CHUNK_SIZE>>20));
}
uri =
raptor_new_uri(raptor_uri_filename_to_uri_string(pdata->location));
iret = raptor_start_parse(rparser, uri);
diff --git a/monetdb5/modules/mal/tokenizer.h b/monetdb5/modules/mal/tokenizer.h
--- a/monetdb5/modules/mal/tokenizer.h
+++ b/monetdb5/modules/mal/tokenizer.h
@@ -22,7 +22,7 @@
#include "mal_client.h"
#include "mal_interpreter.h"
-#define RDF_MIN_LITERAL (((oid) 1) << ((sizeof(oid)==8)?62:30))
+#define RDF_MIN_LITERAL (((oid) 1) << ((sizeof(oid)==8)?60:28))
#ifdef WIN32
#if !defined(LIBMAL) && !defined(LIBATOMS) && !defined(LIBKERNEL) &&
!defined(LIBMAL) && !defined(LIBOPTIMIZER) && !defined(LIBSCHEDULER) &&
!defined(LIBMONETDB5)
diff --git a/sql/backends/monet5/sql.mx b/sql/backends/monet5/sql.mx
--- a/sql/backends/monet5/sql.mx
+++ b/sql/backends/monet5/sql.mx
@@ -7298,13 +7298,13 @@ SQLrdfShred(Client cntxt, MalBlkPtr mb,
store_funcs.append_col(m->session->tr,
mvc_bind_column(m,
spo_tbl, "subject"),
- sbt, TYPE_int);
+ sbt, TYPE_oid);
store_funcs.append_col(m->session->tr,
mvc_bind_column(m,
spo_tbl, "property"),
- pbt, TYPE_int);
+ pbt, TYPE_oid);
store_funcs.append_col(m->session->tr,
mvc_bind_column(m,
spo_tbl, "object"),
- obt, TYPE_int);
+ obt, TYPE_oid);
/* Update current value */
curS = *sbt;
curP = *pbt;
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
http://mail.monetdb.org/mailman/listinfo/checkin-list
