Changeset: 8bb2096f7aca for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=8bb2096f7aca
Modified Files:
monetdb5/modules/kernel/arrays.c
monetdb5/modules/kernel/arrays.h
monetdb5/modules/kernel/arrays.mal
sql/backends/monet5/sql_arrays.mal
Branch: arrays
Log Message:
projecting the values of a non-dimensional column after a selection on
dimensional columns only
is the same as projecting any other column + Cleaning code
diffs (truncated from 628 to 300 lines):
diff --git a/monetdb5/modules/kernel/arrays.c b/monetdb5/modules/kernel/arrays.c
--- a/monetdb5/modules/kernel/arrays.c
+++ b/monetdb5/modules/kernel/arrays.c
@@ -4,13 +4,6 @@
#include <gdk_arrays.h>
-static gdk_dimension* getDimension(gdk_cells *dims, int dimNum) {
- dim_node *n;
-
- for(n=dims->h; n && n->data->dimNum<dimNum; n=n->next);
- return n->data;
-}
-
static int jumpSize(gdk_array *array, int dimNum) {
int i=0;
BUN skip = 1;
@@ -23,26 +16,13 @@ static int arrayCellsNum(gdk_array *arra
return jumpSize(array, array->dimsNum);
}
-static bool overlapingRanges(gdk_dimension *dim1, gdk_dimension *dim2) {
- if(*(oid*)dim1->max < *(oid*)dim2->min || *(oid*)dim2->max <
*(oid*)dim1->min) {
- //disjoint ranges. empty result
- return 0;
- }
- return 1;
+static gdk_dimension* getDimension(gdk_cells *dims, int dimNum) {
+ dim_node *n;
+
+ for(n=dims->h; n && n->data->dimNum<dimNum; n=n->next);
+ return n->data;
}
-static gdk_dimension* mergeCandidateDimensions(gdk_dimension *dim1,
gdk_dimension *dim2) {
- oid min, max, step;
-
- /*the biggest of the mins and the smallest of the maximums */
- min = *(oid*)dim1->min > *(oid*)dim2->min ? *(oid*)dim1->min :
*(oid*)dim2->min;
- max = *(oid*)dim1->max < *(oid*)dim2->max ? *(oid*)dim1->max :
*(oid*)dim2->max;
- step = *(oid*)dim1->step ; //step is always 1
-
- //the dimensions that are merged should have the same order
- //they also have the same number of initial elements because they came
from the same dimension
- return createDimension_oid(dim1->dimNum, dim1->initialElementsNum, min,
max, step); }
-
static BUN oidToIdx(oid oidVal, int dimNum, int currentDimNum, BUN skipCells,
gdk_array *dims) {
BUN oid = 0;
@@ -61,7 +41,6 @@ static BUN oidToIdx(oid oidVal, int dimN
return oid%skipCells;
}
-
static BUN* oidToIdx_bulk(oid* oidVals, int valsNum, int dimNum, int
currentDimNum, BUN skipCells, gdk_array *dims) {
BUN *oids = GDKmalloc(valsNum*sizeof(BUN));
int i;
@@ -102,129 +81,36 @@ static BUN* oidToIdx_bulk(oid* oidVals,
return oids;
}
-#define computeValues(TPE) \
-do { \
- TPE min = *(TPE*)dimension->min; \
- TPE step = *(TPE*)dimension->step; \
- TPE *vals; \
- oid *oids; \
- BUN *idxs; \
- BUN i; \
-\
- if(!(resBAT = BATnew(TYPE_void, TYPE_##TPE, resSize, TRANSIENT))) \
- throw(MAL, "algebra.dimensionLeftfetchjoin", "Problem allocating new
BAT"); \
- vals = (TPE*)Tloc(resBAT, BUNfirst(resBAT)); \
- oids = (oid*)Tloc(candsBAT, BUNfirst(candsBAT)); \
-\
- idxs = oidToIdx_bulk(oids, resSize, dimension->dimNum, 0, 1, array); \
- for(i=0; i<resSize; i++) { \
- *vals++ = min +idxs[i]*step; \
-/*fprintf(stderr, "%d - %d - %d\n", (int)oids[i], (int)idxs[i],
(int)vals[-1]); */\
- } \
-} while(0)
+static str readCands(gdk_cells** dimsRes, BAT** oidsRes, const ptr *dimsCand,
const bat *oidsCand, gdk_array* array) {
+ gdk_cells *dimensionsCandidates_in = NULL;
+ BAT *candidatesBAT_in = NULL;
+
+ if(oidsCand) {
+ if ((candidatesBAT_in = BATdescriptor(*oidsCand)) == NULL) {
+ throw(MAL, "algebra.dimensionSubselect",
RUNTIME_OBJECT_MISSING);
+ }
+ }
-str ALGdimensionLeftfetchjoin1(bat *result, const bat *cands, const ptr *dims,
const ptr *dim) {
- gdk_array *array = (gdk_array*)*dims;
- gdk_dimension *dimension = (gdk_dimension*)*dim;
- BAT *candsBAT = NULL, *resBAT = NULL;
- BUN resSize = 0;
+ if(dimsCand)
+ dimensionsCandidates_in = (gdk_cells*)*dimsCand;
-//TODO: Make sure that the cabdidates form an MBR
+ //if there are no candidates then everything is a candidate
+ if(!dimsCand && !oidsCand) {
+ dimensionsCandidates_in = arrayToCells(array);
+ //create an empy candidates BAT
+ if((candidatesBAT_in = BATnew(TYPE_void, TYPE_oid, 0,
TRANSIENT)) == NULL)
+ throw(MAL, "algebra.dimensionSubselect", GDK_EXCEPTION);
+ BATsetcount(candidatesBAT_in, 0);
+ BATseqbase(candidatesBAT_in, 0);
+ BATderiveProps(candidatesBAT_in, FALSE);
+ }
- if ((candsBAT = BATdescriptor(*cands)) == NULL) {
- throw(MAL, "algebra.dimensionLeftfetchjoin", RUNTIME_OBJECT_MISSING);
- }
- resSize = BATcount(candsBAT);
- /*for each oid in the candsBAT find the real value of the dimension */
- switch(dimension->type) { \
- case TYPE_bte: \
- computeValues(bte);
- break;
- case TYPE_sht:
- computeValues(sht);
- break;
- case TYPE_int:
- computeValues(int);
- break;
- case TYPE_wrd:
- computeValues(wrd);
- break;
- case TYPE_oid:
- computeValues(oid);
- break;
- case TYPE_lng:
- computeValues(lng);
- break;
- case TYPE_dbl:
- computeValues(dbl);
- break;
- case TYPE_flt:
- computeValues(flt);
- break;
- default:
- throw(MAL, "algebra.dimensionLeftfetchjoin", "Dimension
type not supported\n");
- }
-
- BATsetcount(resBAT, resSize);
- BATseqbase(resBAT, 0);
- BATderiveProps(resBAT, FALSE);
-
- *result = resBAT->batCacheid;
- BBPkeepref(*result);
-
- //free the space occupied by the dimension
- freeDimension(dimension);
- return MAL_SUCCEED;
-
-}
-
-//str ALGnonDimensionLeftfetchjoin1(bat* result, const ptr* dimsCand, const
bat *candBat, const bat *valsBat) {
-str ALGnonDimensionLeftfetchjoin1(bat* result, const bat* cands, const bat
*vals, const ptr *dims) {
- (void)*result;
- (void)*cands;
- (void)*vals;
- (void)*dims;
+ *dimsRes = dimensionsCandidates_in;
+ *oidsRes = candidatesBAT_in;
return MAL_SUCCEED;
}
-str ALGnonDimensionLeftfetchjoin2(bat* result, ptr *dimsRes, const bat *tids,
const bat *vals, const ptr *dims) {
- BAT *materialisedBAT = NULL;
- BAT *nonDimensionalBAT = NULL;
- BUN totalCellsNum, neededCellsNum;
-
- gdk_array *array = (gdk_array*)*dims;
-
- if ((nonDimensionalBAT = BATdescriptor(*vals)) == NULL) {
- throw(MAL, "algebra.leftfecthjoin", RUNTIME_OBJECT_MISSING);
- }
- (void)*tids; //ignore the tids
-
- totalCellsNum = arrayCellsNum(array);
- neededCellsNum = totalCellsNum - BATcount(nonDimensionalBAT);
-
- /*TODO: fix this so that I can have the real default value of the
column */
- materialisedBAT =
materialise_nonDimensional_column(ATOMtype(BATttype(nonDimensionalBAT)),
neededCellsNum, NULL);
- if(!materialisedBAT) {
- BBPunfix(nonDimensionalBAT->batCacheid);
- throw(MAL, "algebra.leftfetchjoin", "Problem materialising
non-dimensional column");
- }
-
-
- /*append the missing values to the BAT */
- BATappend(nonDimensionalBAT, materialisedBAT, TRUE);
- BATsetcount(nonDimensionalBAT, totalCellsNum);
-
- BBPunfix(materialisedBAT->batCacheid);
- BBPkeepref(*result = nonDimensionalBAT->batCacheid);
-
- //sent this to the output so that we do not lose it afterwards
- *dimsRes = *dims;
-
- return MAL_SUCCEED;
-
-}
-
static str emptyCandidateResults(ptr *candsRes_dims, bat* candsRes_bid) {
BAT *candidatesBAT = NULL;
@@ -240,6 +126,27 @@ static str emptyCandidateResults(ptr *ca
return MAL_SUCCEED;
}
+static bool overlapingRanges(gdk_dimension *dim1, gdk_dimension *dim2) {
+ if(*(oid*)dim1->max < *(oid*)dim2->min || *(oid*)dim2->max <
*(oid*)dim1->min) {
+ //disjoint ranges. empty result
+ return 0;
+ }
+ return 1;
+}
+
+static gdk_dimension* mergeCandidateDimensions(gdk_dimension *dim1,
gdk_dimension *dim2) {
+ oid min, max, step;
+
+ /*the biggest of the mins and the smallest of the maximums */
+ min = *(oid*)dim1->min > *(oid*)dim2->min ? *(oid*)dim1->min :
*(oid*)dim2->min;
+ max = *(oid*)dim1->max < *(oid*)dim2->max ? *(oid*)dim1->max :
*(oid*)dim2->max;
+ step = *(oid*)dim1->step ; //step is always 1
+
+ //the dimensions that are merged should have the same order
+ //they also have the same number of initial elements because they came
from the same dimension
+ return createDimension_oid(dim1->dimNum, dim1->initialElementsNum, min,
max, step);
+}
+
static BAT* joinBATs(BAT *candsBAT, BAT* dimBAT, gdk_array *array, int dimNum)
{
oid *candsOIDs, *dimOIDs, *mergedOIDs;
BAT* mergedBAT;
@@ -461,37 +368,7 @@ static bool updateCandidateResults(gdk_a
return 1;
}
-static str readCands(gdk_cells** dimsRes, BAT** oidsRes, const ptr *dimsCand,
const bat *oidsCand, gdk_array* array) {
- gdk_cells *dimensionsCandidates_in = NULL;
- BAT *candidatesBAT_in = NULL;
-
- if(oidsCand) {
- if ((candidatesBAT_in = BATdescriptor(*oidsCand)) == NULL) {
- throw(MAL, "algebra.dimensionSubselect",
RUNTIME_OBJECT_MISSING);
- }
- }
-
- if(dimsCand)
- dimensionsCandidates_in = (gdk_cells*)*dimsCand;
-
- //if there are no candidates then everything is a candidate
- if(!dimsCand && !oidsCand) {
- dimensionsCandidates_in = arrayToCells(array);
- //create an empy candidates BAT
- if((candidatesBAT_in = BATnew(TYPE_void, TYPE_oid, 0,
TRANSIENT)) == NULL)
- throw(MAL, "algebra.dimensionSubselect", GDK_EXCEPTION);
- BATsetcount(candidatesBAT_in, 0);
- BATseqbase(candidatesBAT_in, 0);
- BATderiveProps(candidatesBAT_in, FALSE);
- }
-
- *dimsRes = dimensionsCandidates_in;
- *oidsRes = candidatesBAT_in;
-
- return MAL_SUCCEED;
-}
-
-str ALGdimensionSubselect2(ptr *dimsRes, bat* oidsRes, const ptr *dims, const
ptr* dim, const ptr *dimsCand, const bat* oidsCand,
+str ALGdimensionSubselect2(ptr *dimsRes, bat* oidsRes, const ptr *dim, const
ptr* dims, const ptr *dimsCand, const bat* oidsCand,
const void *low, const
void *high, const bit *li, const bit *hi, const bit *anti) {
gdk_array *array = (gdk_array*)*dims; //the sizes of all the dimensions
(I treat them as indices and I do not care about the exact values)
gdk_dimension *dimension = (gdk_dimension*)*dim;
@@ -646,12 +523,12 @@ str ALGdimensionSubselect2(ptr *dimsRes,
return MAL_SUCCEED;
}
-str ALGdimensionSubselect1(ptr *dimsRes, bat* oidsRes, const ptr *dims, const
ptr* dim,
+str ALGdimensionSubselect1(ptr *dimsRes, bat* oidsRes, const ptr *dim, const
ptr* dims,
const void *low, const
void *high, const bit *li, const bit *hi, const bit *anti) {
- return ALGdimensionSubselect2(dimsRes, oidsRes, dims, dim, NULL, NULL,
low, high, li, hi, anti);
+ return ALGdimensionSubselect2(dimsRes, oidsRes, dim, dims, NULL, NULL,
low, high, li, hi, anti);
}
-str ALGdimensionThetasubselect2(ptr *dimsRes, bat* oidsRes, const ptr *dims,
const ptr* dim, const ptr *dimsCand, const bat* oidsCand, const void *val,
const char **opp) {
+str ALGdimensionThetasubselect2(ptr *dimsRes, bat* oidsRes, const ptr *dim,
const ptr* dims, const ptr *dimsCand, const bat* oidsCand, const void *val,
const char **opp) {
bit li = 0;
bit hi = 0;
bit anti = 0;
@@ -663,52 +540,190 @@ str ALGdimensionThetasubselect2(ptr *dim
/* "=" or "==" */
li = hi = 1;
anti = 0;
- return ALGdimensionSubselect2(dimsRes, oidsRes, dims, dim, dimsCand,
oidsCand, val, nil, &li, &hi, &anti);
+ return ALGdimensionSubselect2(dimsRes, oidsRes, dim, dims, dimsCand,
oidsCand, val, nil, &li, &hi, &anti);
}
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
