Changeset: 054157433545 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=054157433545
Modified Files:
monetdb5/extras/rdf/rdfschema.c
monetdb5/extras/rdf/rdfschema.h
sql/backends/monet5/sql_rdf.c
Branch: rdf
Log Message:
Create function for generating string from both URI-based oid and string-based
oid.
This function is used for table name since the name's oid can appears in
tokenizer or mapbat
diffs (truncated from 521 to 300 lines):
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
@@ -120,6 +120,13 @@ static void initcsIdFreqIdxMap(int* inpu
}
+char
+getObjType(oid objOid){
+ char objType = (char) (objOid >> (sizeof(BUN)*8 - 4)) & 7 ;
+
+ return objType;
+
+}
str printTKNZStringFromOid(oid id){
int ret;
@@ -140,6 +147,50 @@ str printTKNZStringFromOid(oid id){
return MAL_SUCCEED;
}
+//Get the string for
+static
+char getStringName(oid objOid, str *objStr, BATiter mapi, BAT *mapbat, char
isTblName){
+
+ char objType = getObjType(objOid);
+ 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);
+
+ if (isTblName){
+ s = *objStr;
+ len = strlen(s);
+ //Replace all non-alphabet character by XXX
+ for (i = 0; i < len; i++)
+ {
+ //printf("i = %d: %c \n",i, s[i]);
+ if (!isalpha(*s)){
+ *s = 'X';
+ }
+ s++;
+
+ }
+ }
+
+ }
+
+
+ return objType;
+}
+
char isCSTable(CS item){
if (item.parentFreqIdx != -1) return 0;
@@ -620,13 +671,7 @@ void printSubCSInformation(SubCSSet *sub
#endif /*NO_OUTPUTFILE*/
#endif /* NEEDSUBCS */
-char
-getObjType(oid objOid){
- char objType = (char) (objOid >> (sizeof(BUN)*8 - 4)) & 7 ;
-
- return objType;
-
-}
+
@@ -1881,8 +1926,6 @@ str printMergedFreqCSSet(CSset *freqCSse
str subStr;
str objStr;
oid objOid;
- char objType;
- BUN bun;
#endif
int ret;
char* schema = "rdf";
@@ -1929,12 +1972,12 @@ str printMergedFreqCSSet(CSset *freqCSse
if (i < freqCSset->numOrigFreqCS){
if (cs.subject != BUN_NONE){
takeOid(cs.subject, &subStr);
-
if (labels[i].name == BUN_NONE) {
fprintf(fout,"CS " BUNFMT " -
FreqId %d - Name: %s (Freq: %d) | Subject: %s | FreqParentIdx %d \n",
cs.csId, i, "DUMMY", freq, subStr, cs.parentFreqIdx);
} else {
str labelStr;
- takeOid(labels[i].name,
&labelStr);
+ //takeOid(labels[i].name,
&labelStr);
+ getStringName(labels[i].name,
&labelStr, mapi, mapbat, 1);
fprintf(fout,"CS " BUNFMT " -
FreqId %d - Name: %s (Freq: %d) | Subject: %s | FreqParentIdx %d \n",
cs.csId, i, labelStr, freq, subStr, cs.parentFreqIdx);
GDKfree(labelStr);
}
@@ -1946,7 +1989,8 @@ str printMergedFreqCSSet(CSset *freqCSse
fprintf(fout,"CS " BUNFMT " -
FreqId %d - Name: %s (Freq: %d) | FreqParentIdx %d \n", cs.csId, i, "DUMMY",
freq, cs.parentFreqIdx);
} else {
str labelStr;
- takeOid(labels[i].name,
&labelStr);
+ //takeOid(labels[i].name,
&labelStr);
+ getStringName(labels[i].name,
&labelStr, mapi, mapbat, 1);
fprintf(fout,"CS " BUNFMT " -
FreqId %d - Name: %s (Freq: %d) | FreqParentIdx %d \n", cs.csId, i, labelStr,
freq, cs.parentFreqIdx);
GDKfree(labelStr);
}
@@ -1958,7 +2002,8 @@ str printMergedFreqCSSet(CSset *freqCSse
fprintf(fout,"CS " BUNFMT " - FreqId %d
- Name: %s (Freq: %d) | Subject: <Not available> | FreqParentIdx %d \n",
cs.csId, i, "DUMMY", freq, cs.parentFreqIdx);
} else {
str labelStr;
- takeOid(labels[i].name, &labelStr);
+ //takeOid(labels[i].name, &labelStr);
+ getStringName(labels[i].name,
&labelStr, mapi, mapbat, 1);
fprintf(fout,"CS " BUNFMT " - FreqId %d
- Name: %s (Freq: %d) | Subject: <Not available> | FreqParentIdx %d \n",
cs.csId, i, labelStr, freq, cs.parentFreqIdx);
GDKfree(labelStr);
}
@@ -1972,7 +2017,8 @@ str printMergedFreqCSSet(CSset *freqCSse
else{
str labelStr = NULL;
str labelShortStr = NULL;
-
takeOid(labels[tmpParentFreqId].name, &labelStr);
+
//takeOid(labels[tmpParentFreqId].name, &labelStr);
+
getStringName(labels[tmpParentFreqId].name, &labelStr, mapi, mapbat, 1);
getPropNameShort(&labelShortStr,labelStr);
fprintf(fout, "[%s]
",labelShortStr);
GDKfree(labelShortStr);
@@ -1999,24 +2045,9 @@ str printMergedFreqCSSet(CSset *freqCSse
}
if (cs.lstObj != NULL){
objOid = cs.lstObj[j];
-
- objType = getObjType(objOid);
-
- if (objType == URI || objType ==
BLANKNODE){
- objOid = objOid - ((oid)objType
<< (sizeof(BUN)*8 - 4));
- takeOid(objOid, &objStr);
- }
- else{
- objOid = objOid - (objType*2 +
1) * RDF_MIN_LITERAL; /* Get the real objOid from Map or Tokenizer */
- bun = BUNfirst(mapbat);
- objStr = (str) BUNtail(mapi,
bun + objOid);
- }
-
+ getStringName(objOid, &objStr, mapi,
mapbat, 0);
fprintf(fout, " O: %s \n", objStr);
-
- if (objType == URI || objType ==
BLANKNODE){
- GDKfree(objStr);
- }
+ GDKfree(objStr);
}
else{
fprintf(fout, " <No Object value> \n");
@@ -2140,7 +2171,7 @@ str printmergeCSSet(CSset *freqCSset, in
#if NO_OUTPUTFILE == 0
static
-str printsubsetFromCSset(CSset *freqCSset, BAT* subsetIdxBat, int num, int*
mergeCSFreqCSMap, CSlabel *label, int sampleVersion){
+str printsubsetFromCSset(CSset *freqCSset, BAT* subsetIdxBat, BAT *mbat, int
num, int* mergeCSFreqCSMap, CSlabel *label, int sampleVersion){
int i,j;
FILE *fout;
@@ -2155,8 +2186,9 @@ str printsubsetFromCSset(CSset *freqCSse
CS cs;
int tmpNumcand;
str canStr;
-
-
+ BATiter mapi;
+
+ mapi = bat_iterator(mbat);
if (TKNZRopen (NULL, &schema) != MAL_SUCCEED) {
throw(RDF, "rdf.rdfschema",
"could not open the tokenizer\n");
@@ -2184,7 +2216,9 @@ str printsubsetFromCSset(CSset *freqCSse
#if USE_SHORT_NAMES
str canStrShort = NULL;
#endif
- takeOid(label[freqIdx].candidates[j], &canStr);
+ //takeOid(label[freqIdx].candidates[j],
&canStr);
+ getStringName(label[freqIdx].candidates[j],
&canStr, mapi, mbat, 1);
+
#if USE_SHORT_NAMES
getPropNameShort(&canStrShort, canStr);
fprintf(fout," %s ::: ", canStrShort);
@@ -3445,7 +3479,7 @@ void doMerge(CSset *freqCSset, int ruleN
}
static
-str mergeMaxFreqCSByS1(CSset *freqCSset, CSlabel** labels, oid *mergecsId,
oid** ontmetadata, int ontmetadataCount){
+str mergeMaxFreqCSByS1(CSset *freqCSset, CSlabel** labels, oid *mergecsId,
oid** ontmetadata, int ontmetadataCount,bat *mapbatid){
int i;
#if !USE_MULTIWAY_MERGING
@@ -3467,6 +3501,8 @@ str mergeMaxFreqCSByS1(CSset *freqCSset,
char* schema = "rdf";
int ret = 0;
str tmpLabel;
+ BAT *mbat = NULL;
+ BATiter mapi;
#if USE_SHORT_NAMES
str canStrShort = NULL;
@@ -3479,12 +3515,19 @@ str mergeMaxFreqCSByS1(CSset *freqCSset,
(void) cs1;
(void) cs2;
#endif
+ (void) mapbatid;
+
labelStat = initLabelStat();
buildLabelStat(labelStat, (*labels), freqCSset, TOPK);
printf("Num FreqCSadded before using S1 = %d \n",
freqCSset->numCSadded);
#if OUTPUT_FREQID_PER_LABEL
+ if ((mbat = BATdescriptor(*mapbatid)) == NULL) {
+ throw(MAL, "rdf.RDFreorganize", RUNTIME_OBJECT_MISSING);
+ }
+ mapi = bat_iterator(mbat);
+
if (TKNZRopen (NULL, &schema) != MAL_SUCCEED) {
throw(RDF, "rdf.rdfschema",
"could not open the tokenizer\n");
@@ -3627,7 +3670,9 @@ str mergeMaxFreqCSByS1(CSset *freqCSset,
#endif /* USE_MULTIWAY_MERGING */
#if OUTPUT_FREQID_PER_LABEL
- takeOid(*name, &tmpLabel);
+ //takeOid(*name, &tmpLabel);
+ getStringName(*name, &tmpLabel, mapi, mbat, 1);
+
#if USE_SHORT_NAMES
getPropNameShort(&canStrShort, tmpLabel);
fprintf(fout,"Label %d: %s \n", i, canStrShort);
@@ -5659,7 +5704,7 @@ void getTblName(char *name, oid nameId){
}
*/
-void getTblName(str *name, oid nameId){
+void getTblName(str *name, oid nameId, BATiter mapi, BAT *mbat){
str canStr = NULL;
str canStrShort = NULL;
char *pch;
@@ -5668,7 +5713,8 @@ void getTblName(str *name, oid nameId){
int i;
if (nameId != BUN_NONE){
- takeOid(nameId, &canStr);
+ //takeOid(nameId, &canStr);
+ getStringName(nameId, &canStr, mapi, mbat, 1);
getPropNameShort(&canStrShort, canStr);
if (strstr (canStrShort,".") != NULL ||
@@ -5788,7 +5834,8 @@ str printSampleData(CSSample *csSample,
#if USE_SHORT_NAMES
str canStrShort = NULL;
#endif
- takeOid(sample.candidates[j], &canStr);
+ //takeOid(sample.candidates[j], &canStr);
+ getStringName(sample.candidates[j], &canStr,
mapi, mbat, 1);
#if USE_SHORT_NAMES
getPropNameShort(&canStrShort, canStr);
fprintf(fout,";%s", canStrShort);
@@ -5805,7 +5852,8 @@ str printSampleData(CSSample *csSample,
if (sample.name != BUN_NONE){
str canStrShort = NULL;
- takeOid(sample.name, &canStr);
+ //takeOid(sample.name, &canStr);
+ getStringName(sample.name, &canStr, mapi, mbat, 1);
getPropNameShort(&canStrShort, canStr);
if (strstr (canStrShort,".") != NULL ||
@@ -5971,7 +6019,9 @@ str printSampleData(CSSample *csSample,
if (sample.name != BUN_NONE){
str canStrShort = NULL;
- takeOid(sample.name, &canStr);
+ //takeOid(sample.name, &canStr);
+ getStringName(sample.name, &canStr, mapi, mbat, 1);
+
getPropNameShort(&canStrShort, canStr);
if (strstr (canStrShort,".") != NULL ||
@@ -6011,7 +6061,7 @@ str printSampleData(CSSample *csSample,
#if NO_OUTPUTFILE == 0
static
-str printFullSampleData(CSSampleExtend *csSampleEx, int num){
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
