Changeset: 2b0ab4777950 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=2b0ab4777950
Modified Files:
monetdb5/extras/rdf/rdflabels.c
monetdb5/extras/rdf/rdflabels.h
monetdb5/extras/rdf/rdfschema.c
monetdb5/extras/rdf/rdfschema.h
Branch: rdf
Log Message:
Update labels when CS's are merged
Updates label, hierarchy and properties. Does not update candidates yet.
diffs (truncated from 851 to 300 lines):
diff --git a/monetdb5/extras/rdf/rdflabels.c b/monetdb5/extras/rdf/rdflabels.c
--- a/monetdb5/extras/rdf/rdflabels.c
+++ b/monetdb5/extras/rdf/rdflabels.c
@@ -1790,7 +1790,7 @@ oid* getOntoHierarchy(oid ontology, int*
// add 'ontology' to hierarchy
(*hierarchyCount) = 1;
- hierarchy = (oid *) malloc(sizeof(oid) * (*hierarchyCount));
+ hierarchy = (oid *) GDKmalloc(sizeof(oid) * (*hierarchyCount));
if (!hierarchy)
fprintf(stderr, "ERROR: Couldn't malloc memory!\n");
hierarchy[(*hierarchyCount) -1] = ontology;
@@ -1815,7 +1815,7 @@ oid* getOntoHierarchy(oid ontology, int*
// superclass
// add 'msuperstr' to hierarchy
(*hierarchyCount) += 1;
- hierarchy = realloc(hierarchy,
sizeof(oid) * (*hierarchyCount));
+ hierarchy = GDKrealloc(hierarchy,
sizeof(oid) * (*hierarchyCount));
if (!hierarchy)
fprintf(stderr, "ERROR:
Couldn't realloc memory!\n");
hierarchy[(*hierarchyCount) -1] =
msuper;
@@ -1843,7 +1843,7 @@ void getTableName(CSlabel* label, int cs
// --- ONTOLOGY ---
// add all ontology candidates to list of candidates
if (resultCount[csIdx] >= 1) {
- label->candidates = realloc(label->candidates, sizeof(oid) *
(label->candidatesCount + resultCount[csIdx]));
+ label->candidates = GDKrealloc(label->candidates, sizeof(oid) *
(label->candidatesCount + resultCount[csIdx]));
if (!label->candidates) fprintf(stderr, "ERROR: Couldn't
realloc memory!\n");
for (i = 0; i < resultCount[csIdx]; ++i) {
label->candidates[label->candidatesCount + i] =
result[csIdx][i];
@@ -1925,7 +1925,7 @@ void getTableName(CSlabel* label, int cs
// add all most frequent type values to list of candidates
if (tmpListCount >= 1) {
int counter = 0;
- label->candidates = realloc(label->candidates, sizeof(oid) *
(label->candidatesCount + tmpListCount));
+ label->candidates = GDKrealloc(label->candidates, sizeof(oid) *
(label->candidatesCount + tmpListCount));
if (!label->candidates) fprintf(stderr, "ERROR: Couldn't
realloc memory!\n");
for (i = 0; i < typeStatCount; ++i) {
for (j = 0; j < tmpListCount; ++j) {
@@ -1965,7 +1965,7 @@ void getTableName(CSlabel* label, int cs
// --- FK ---
// add top3 fk values to list of candidates
if (links[csIdx].num > 0) {
- label->candidates = realloc(label->candidates, sizeof(oid) *
(label->candidatesCount + MIN(3, links[csIdx].num)));
+ label->candidates = GDKrealloc(label->candidates, sizeof(oid) *
(label->candidatesCount + MIN(3, links[csIdx].num)));
if (!label->candidates) fprintf(stderr, "ERROR: Couldn't
realloc memory!\n");
for (i = 0; i < MIN(3, links[csIdx].num); ++i) {
label->candidates[label->candidatesCount + i] =
links[csIdx].fks[0].prop;
@@ -1983,7 +1983,7 @@ void getTableName(CSlabel* label, int cs
// --- NOTHING ---
if (label->candidatesCount == 0) {
- label->candidates = realloc(label->candidates, sizeof(oid));
+ label->candidates = GDKrealloc(label->candidates, sizeof(oid));
if (!label->candidates) fprintf(stderr, "ERROR: Couldn't
realloc memory!\n");
label->candidates[0] = BUN_NONE;
label->candidatesCount = 1;
@@ -2004,7 +2004,7 @@ CSlabel* initLabels(CSset *freqCSset) {
CSlabel *labels;
int i;
- labels = (CSlabel *) malloc(sizeof(CSlabel) * freqCSset->numCSadded);
+ labels = (CSlabel *) GDKmalloc(sizeof(CSlabel) * freqCSset->numCSadded);
if (!labels) fprintf(stderr, "ERROR: Couldn't malloc memory!\n");
for (i = 0; i < freqCSset->numCSadded; ++i) {
labels[i].candidates = NULL;
@@ -2031,7 +2031,7 @@ void getAllLabels(CSlabel* labels, CSset
// copy attribute oids (names)
labels[i].numProp = cs.numProp;
- labels[i].lstProp = (oid *) malloc(sizeof(oid) * cs.numProp);
+ labels[i].lstProp = (oid *) GDKmalloc(sizeof(oid) * cs.numProp);
if (!labels[i].lstProp) fprintf(stderr, "ERROR: Couldn't malloc
memory!\n");
for (j = 0; j < cs.numProp; ++j) {
labels[i].lstProp[j] = cs.lstProp[j];
@@ -2169,7 +2169,6 @@ void addToOntoUsageTree(OntoUsageNode* t
addToOntoUsageTree(leaf, hierarchy, hierarchyCount, numTuples);
}
-
static
void printTree(OntoUsageNode* tree, int level) {
int i;
@@ -2253,7 +2252,7 @@ void createOntoUsageTree(OntoUsageNode**
// search class in tree and add CS to statistics
addToOntoUsageTree(*tree, hierarchy, hierarchyCount,
freqCSset->items[i].support);
- free(hierarchy);
+ GDKfree(hierarchy);
// numTuples += freqCSset->items[i].support; // update total
number of tuples in dataset // TODO cs.support not yet available
numTuples += 1;
}
@@ -2353,60 +2352,6 @@ void freeOntologyLookupResult(oid** onto
free(ontologyLookupResult);
}
-/* Returns the indexes of all CS a CS consists of. */
-static
-int* getSubCS(CSset* freqCSset, int csIdx, int* csListLength) {
- int *csList = NULL;
- int i, j;
-
- CS cs = freqCSset->items[csIdx];
-
- (*csListLength) = 0;
-
- if (cs.type == MAXCS) {
-//printf("MAXCS ");
- // get itself & all childen
- (*csListLength)++;
- csList = (int *) malloc(sizeof(int) * *csListLength);
- csList[*csListLength - 1] = csIdx;
- for (i = 0; i < freqCSset->numCSadded; ++i) {
- if (freqCSset->items[i].parentFreqIdx != csIdx)
continue;
- (*csListLength)++;
- csList = (int *) realloc(csList, sizeof(int) *
*csListLength);
- if (!csList) fprintf(stderr, "ERROR: Couldn't realloc
memory!\n");
- csList[*csListLength - 1] = i;
-//printf(BUNFMT" ", freqCSset->items[i].csId);
- }
- } else if (cs.type == MERGECS) {
-//printf("MERGECS ( ");
- // get all children
- for (i = 0; i < cs.numConsistsOf; ++i) {
- int length;
- int *list = getSubCS(freqCSset, cs.lstConsistsOf[i],
&length); // recursive call
-
- // merge into existing list
- csList = realloc(csList, sizeof(int) * (*csListLength +
length));
- if (!csList) fprintf(stderr, "ERROR: Couldn't realloc
memory!\n");
- for (j = 0; j < length; ++j) {
- csList[*csListLength + j] = list[j];
- }
- (*csListLength) += length;
- free(list);
- }
-//printf(")\n");
- } else { // FREQCS
-//printf("FREQCS ");
- // copy entry
- csList = (int *) malloc(sizeof(int) * 1);
- *csListLength = 1;
- csList[0] = csIdx;
-//printf(BUNFMT" ", freqCSset->items[csIdx].csId);
- return csList;
- }
-
- return csList;
-}
-
/* Creates labels for all CS (without a parent). */
CSlabel* createLabels(CSset* freqCSset, CSrel* csrelSet, int num, BAT *sbat,
BATiter si, BATiter pi, BATiter oi, oid *subjCSMap, int *csIdFreqIdxMap, oid**
ontattributes, int ontattributesCount, oid** ontmetadata, int ontmetadataCount,
OntoUsageNode** ontoUsageTree) {
#if USE_TYPE_NAMES
@@ -2505,155 +2450,180 @@ CSlabel* createLabels(CSset* freqCSset,
return labels;
}
-str updateLabel(int ruleNumber, CSlabel *labels, int mergeCSFreqId, int
freqCS1, int freqCS2){
- (void) ruleNumber;
- (void) labels;
- (void) mergeCSFreqId;
- (void) freqCS1;
- (void) freqCS2;
+/* Create labels for merged CS's. Uses rules S1 to S5 (new names!).
+ * If no MERGECS is created (subset-superset relation), mergeCSFreqId contains
the Id of the superset class.
+ * For S1 and S2, parameter 'name' is used to avoid recomputation of CS names
+ */
+str updateLabel(int ruleNumber, CSset *freqCSset, CSlabel **labels, int newCS,
int mergeCSFreqId, int freqCS1, int freqCS2, oid name, oid **ontmetadata, int
ontmetadataCount){
+ int i;
+ int freqCS1Counter;
+ CSlabel *big;
+ CSlabel *label;
+ CS cs;
- return MAL_SUCCEED;
+ if (newCS) {
+ // realloc labels
+ *labels = GDKrealloc(*labels, sizeof(CSlabel) *
freqCSset->numCSadded);
+ if (!(*labels)) fprintf(stderr, "ERROR: Couldn't realloc
memory!\n");
+ (*labels)[mergeCSFreqId].candidates = NULL;
+ (*labels)[mergeCSFreqId].candidatesCount = 0;
+ (*labels)[mergeCSFreqId].hierarchy = NULL;
+ (*labels)[mergeCSFreqId].hierarchyCount = 0;
+ (*labels)[mergeCSFreqId].numProp = 0;
+ (*labels)[mergeCSFreqId].lstProp = NULL;
+ }
+ label = &(*labels)[mergeCSFreqId];
+ cs = freqCSset->items[mergeCSFreqId];
+
+ // copy properties
+ if (ruleNumber != S3) {
+ if (label->numProp > 0) GDKfree(label->lstProp);
+ label->numProp = cs.numProp;
+ label->lstProp = (oid *) GDKmalloc(sizeof(oid) *
label->numProp);
+ if (!label->lstProp) fprintf(stderr, "ERROR: Couldn't malloc
memory!\n");
+ for (i = 0; i < label->numProp; ++i) {
+ label->lstProp[i] = cs.lstProp[i];
+ }
+ }
+
+ switch (ruleNumber) {
+ case S1: // was: (S1 or S2), now combined
+ // use common name
+ label->name = name;
+
+ // TODO candidates
+ //label->candidates = ;
+ //label->candidatesCount = ;
+
+ // hierarchy
+ if ((*labels)[freqCS1].name == label->name) {
+ // copy hierarchy from CS freqCS1
+ label->hierarchyCount =
(*labels)[freqCS1].hierarchyCount;
+ if (label->hierarchyCount > 0) {
+ label->hierarchy = (oid *)
GDKmalloc(sizeof(oid) * label->hierarchyCount);
+ if (!label->hierarchy) fprintf(stderr, "ERROR:
Couldn't malloc memory!\n");
+ for (i = 0; i < label->hierarchyCount; ++i) {
+ label->hierarchy[i] =
(*labels)[freqCS1].hierarchy[i];
+ }
+ }
+ } else if ((*labels)[freqCS2].name == label->name) {
+ // copy hierarchy from CS freqCS2
+ label->hierarchyCount =
(*labels)[freqCS2].hierarchyCount;
+ if (label->hierarchyCount > 0) {
+ label->hierarchy = (oid *)
GDKmalloc(sizeof(oid) * label->hierarchyCount);
+ if (!label->hierarchy) fprintf(stderr, "ERROR:
Couldn't malloc memory!\n");
+ for (i = 0; i < label->hierarchyCount; ++i) {
+ label->hierarchy[i] =
(*labels)[freqCS2].hierarchy[i];
+ }
+ }
+ } else {
+ // no top 1 name, no hierarchy available
+ label->hierarchy = getOntoHierarchy(name,
&(label->hierarchyCount), ontmetadata, ontmetadataCount);
+ }
+
+ break;
+
+ case S2:
+ // use common ancestor
+ label->name = name;
+
+ // TODO candidates
+ //label->candidates = ;
+ //label->candidatesCount = ;
+
+ // hierarchy
+ freqCS1Counter = (*labels)[freqCS1].hierarchyCount - 1;
+ while (TRUE) {
+ if ((*labels)[freqCS1].hierarchy[freqCS1Counter] ==
label->name)
+ break;
+ freqCS1Counter--;
+ }
+ label->hierarchyCount = (*labels)[freqCS1].hierarchyCount -
freqCS1Counter;
+ if (label->hierarchyCount > 0) {
+ label->hierarchy = (oid *) GDKmalloc(sizeof(oid) *
label->hierarchyCount);
+ if (!label->hierarchy) fprintf(stderr, "ERROR: Couldn't
malloc memory!\n");
+ for (i = 0; i < label->hierarchyCount; ++i) {
+ label->hierarchy[i] =
(*labels)[freqCS1].hierarchy[freqCS1Counter + i];
+ }
+ }
+
+ break;
+
+ case S3:
+ // subset-superset relation
+ // candidates already set
+ // hierarchy already set
+ // properties already set
+
+ break;
+
+ case S4: // FALLTHROUGH
+ case S5:
+ // use label of biggest CS (higher coverage value)
+ if (freqCSset->items[freqCS1].coverage >
freqCSset->items[freqCS2].coverage) {
+ big = &(*labels)[freqCS1];
+ } else {
+ big = &(*labels)[freqCS2];
+ }
+
+ label->name = big->name;
+
+ // TODO candidates
+ //label->candidatesCount = ;
+ //label->candidates = ;
+
+ // hierarchy
+ label->hierarchyCount = big->hierarchyCount;
+ if (label->hierarchyCount > 0) {
+ label->hierarchy = (oid *) GDKmalloc(sizeof(oid) *
label->hierarchyCount);
+ if (!label->hierarchy) fprintf(stderr, "ERROR: Couldn't
malloc memory!\n");
+ for (i = 0; i < label->hierarchyCount; ++i) {
+ label->hierarchy[i] = big->hierarchy[i];
+ }
+ }
+
+ break;
+
+ default:
+ // error
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
http://mail.monetdb.org/mailman/listinfo/checkin-list
