Changeset: e4bc64bf16d3 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=e4bc64bf16d3
Added Files:
gdk/gdk_project.c
Modified Files:
gdk/Makefile.ag
gdk/gdk_join.c
Branch: default
Log Message:
Moved BATproject into its own file gdk_project.c.
diffs (truncated from 857 to 300 lines):
diff --git a/gdk/Makefile.ag b/gdk/Makefile.ag
--- a/gdk/Makefile.ag
+++ b/gdk/Makefile.ag
@@ -27,7 +27,7 @@ lib_gdk = {
gdk_calc.c gdk_calc.h gdk_calc_compare.h gdk_calc_private.h \
gdk_aggr.c gdk_group.c \
gdk_imprints.c gdk_imprints.h \
- gdk_join.c \
+ gdk_join.c gdk_project.c \
gdk_unique.c \
gdk_firstn.c \
bat.feps bat1.feps bat2.feps \
diff --git a/gdk/gdk_join.c b/gdk/gdk_join.c
--- a/gdk/gdk_join.c
+++ b/gdk/gdk_join.c
@@ -12,12 +12,12 @@
#include "gdk_calc_private.h"
/*
- * All "sub" join variants produce some sort of join on two input
- * BATs, optionally subject to up to two candidate lists. Only values
- * in the input BATs that are mentioned in the associated candidate
- * list (if provided) are eligible. They all return two output BATs
- * in the first two arguments. The join operations differ in the way
- * in which tuples from the two inputs are matched.
+ * All join variants produce some sort of join on two input BATs,
+ * optionally subject to up to two candidate lists. Only values in
+ * the input BATs that are mentioned in the associated candidate list
+ * (if provided) are eligible. They all return two output BATs in the
+ * first two arguments. The join operations differ in the way in
+ * which tuples from the two inputs are matched.
*
* All inputs BATs must be dense headed, the output BATs will also be
* dense headed. The outputs consist of two aligned BATs (i.e. same
@@ -62,16 +62,12 @@
* right values; two extra Boolean parameters, li and hi,
* indicate whether equal values match
*
- * In addition to these functions, there are two more functions that
- * are closely related:
+ * In addition to these functions, there is one more functions that is
+ * closely related:
* BATdiff
* difference: return a candidate list compatible list of OIDs of
* tuples in the left input whose value does not occur in the
* right input
- * BATproject
- * projection: return a BAT aligned with the left input whose
- * values are the values from the right input that were referred
- * to by the OIDs in the tail of the left input
*/
/* Perform a bunch of sanity checks on the inputs to a join. */
@@ -3478,392 +3474,3 @@ BATrangejoin(BAT **r1p, BAT **r2p, BAT *
* it uses the imprints code there */
return rangejoin(r1, r2, l, rl, rh, sl, sr, li, hi, maxsize);
}
-
-#define project_loop(TYPE) \
-static gdk_return \
-project_##TYPE(BAT *bn, BAT *l, BAT *r, int nilcheck) \
-{ \
- oid lo, hi; \
- const TYPE *restrict rt; \
- TYPE *restrict bt; \
- TYPE v; \
- const oid *restrict o; \
- oid rseq, rend; \
- \
- o = (const oid *) Tloc(l, BUNfirst(l)); \
- rt = (const TYPE *) Tloc(r, BUNfirst(r)); \
- bt = (TYPE *) Tloc(bn, BUNfirst(bn)); \
- rseq = r->hseqbase; \
- rend = rseq + BATcount(r); \
- lo = 0; \
- hi = lo + BATcount(l); \
- if (nilcheck) { \
- for (; lo < hi; lo++) { \
- if (o[lo] < rseq || o[lo] >= rend) { \
- if (o[lo] == oid_nil) { \
- bt[lo] = TYPE##_nil; \
- bn->T->nonil = 0; \
- bn->T->nil = 1; \
- bn->tsorted = 0; \
- bn->trevsorted = 0; \
- bn->tkey = 0; \
- lo++; \
- break; \
- } else { \
- GDKerror("BATproject: does not match
always\n"); \
- return GDK_FAIL; \
- } \
- } else { \
- v = rt[o[lo] - rseq]; \
- bt[lo] = v; \
- if (v == TYPE##_nil && bn->T->nonil) { \
- bn->T->nonil = 0; \
- bn->T->nil = 1; \
- lo++; \
- break; \
- } \
- } \
- } \
- } \
- for (; lo < hi; lo++) { \
- if (o[lo] < rseq || o[lo] >= rend) { \
- if (o[lo] == oid_nil) { \
- bt[lo] = TYPE##_nil; \
- bn->T->nonil = 0; \
- bn->T->nil = 1; \
- bn->tsorted = 0; \
- bn->trevsorted = 0; \
- bn->tkey = 0; \
- } else { \
- GDKerror("BATproject: does not match
always\n"); \
- return GDK_FAIL; \
- } \
- } else { \
- v = rt[o[lo] - rseq]; \
- bt[lo] = v; \
- } \
- } \
- assert((BUN) lo == BATcount(l)); \
- BATsetcount(bn, (BUN) lo); \
- return GDK_SUCCEED; \
-}
-
-
-/* project type switch */
-project_loop(bte)
-project_loop(sht)
-project_loop(int)
-project_loop(flt)
-project_loop(dbl)
-project_loop(lng)
-#ifdef HAVE_HGE
-project_loop(hge)
-#endif
-
-static gdk_return
-project_void(BAT *bn, BAT *l, BAT *r)
-{
- oid lo, hi;
- oid *restrict bt;
- const oid *o;
- oid rseq, rend;
-
- assert(r->tseqbase != oid_nil);
- o = (const oid *) Tloc(l, BUNfirst(l));
- bt = (oid *) Tloc(bn, BUNfirst(bn));
- bn->tsorted = l->tsorted;
- bn->trevsorted = l->trevsorted;
- bn->tkey = l->tkey & 1;
- bn->T->nonil = 1;
- bn->T->nil = 0;
- rseq = r->hseqbase;
- rend = rseq + BATcount(r);
- for (lo = 0, hi = lo + BATcount(l); lo < hi; lo++) {
- if (o[lo] < rseq || o[lo] >= rend) {
- if (o[lo] == oid_nil) {
- bt[lo] = oid_nil;
- bn->T->nonil = 0;
- bn->T->nil = 1;
- bn->tsorted = 0;
- bn->trevsorted = 0;
- bn->tkey = 0;
- } else {
- GDKerror("BATproject: does not match always\n");
- return GDK_FAIL;
- }
- } else {
- bt[lo] = o[lo] - rseq + r->tseqbase;
- }
- }
- assert((BUN) lo == BATcount(l));
- BATsetcount(bn, (BUN) lo);
- return GDK_SUCCEED;
-}
-
-static gdk_return
-project_any(BAT *bn, BAT *l, BAT *r, int nilcheck)
-{
- BUN n;
- oid lo, hi;
- BATiter ri;
- int (*cmp)(const void *, const void *) = ATOMcompare(r->ttype);
- const void *nil = ATOMnilptr(r->ttype);
- const void *v;
- const oid *o;
- oid rseq, rend;
-
- o = (const oid *) Tloc(l, BUNfirst(l));
- n = BUNfirst(bn);
- ri = bat_iterator(r);
- rseq = r->hseqbase;
- rend = rseq + BATcount(r);
- for (lo = 0, hi = lo + BATcount(l); lo < hi; lo++, n++) {
- if (o[lo] < rseq || o[lo] >= rend) {
- if (o[lo] == oid_nil) {
- tfastins_nocheck(bn, n, nil, Tsize(bn));
- bn->T->nonil = 0;
- bn->T->nil = 1;
- bn->tsorted = 0;
- bn->trevsorted = 0;
- bn->tkey = 0;
- } else {
- GDKerror("BATproject: does not match always\n");
- goto bunins_failed;
- }
- } else {
- v = BUNtail(ri, o[lo] - rseq + BUNfirst(r));
- tfastins_nocheck(bn, n, v, Tsize(bn));
- if (nilcheck && bn->T->nonil && cmp(v, nil) == 0) {
- bn->T->nonil = 0;
- bn->T->nil = 1;
- }
- }
- }
- assert(n == BATcount(l));
- BATsetcount(bn, n);
- return GDK_SUCCEED;
-bunins_failed:
- return GDK_FAIL;
-}
-
-BAT *
-BATproject(BAT *l, BAT *r)
-{
- BAT *bn;
- oid lo, hi;
- gdk_return res;
- int tpe = ATOMtype(r->ttype), nilcheck = 1, stringtrick = 0;
- BUN lcount = BATcount(l), rcount = BATcount(r);
- lng t0 = GDKusec();
-
- ALGODEBUG fprintf(stderr, "#BATproject(l=%s#" BUNFMT "%s%s%s,"
- "r=%s#" BUNFMT "[%s]%s%s%s)\n",
- BATgetId(l), BATcount(l),
- l->tsorted ? "-sorted" : "",
- l->trevsorted ? "-revsorted" : "",
- l->tkey & 1 ? "-key" : "",
- BATgetId(r), BATcount(r), ATOMname(r->ttype),
- r->tsorted ? "-sorted" : "",
- r->trevsorted ? "-revsorted" : "",
- r->tkey & 1 ? "-key" : "");
-
- assert(BAThdense(l));
- assert(BAThdense(r));
- assert(ATOMtype(l->ttype) == TYPE_oid);
-
- if (BATtdense(l) && BATcount(l) > 0) {
- lo = l->tseqbase;
- hi = l->tseqbase + BATcount(l);
- if (lo < r->hseqbase || hi > r->hseqbase + BATcount(r)) {
- GDKerror("BATproject: does not match always\n");
- return NULL;
- }
- bn = BATslice(r, lo - r->hseqbase, hi - r->hseqbase);
- if (bn == NULL)
- return NULL;
- BATseqbase(bn, l->hseqbase + (lo - l->tseqbase));
- ALGODEBUG fprintf(stderr,
"#BATproject(l=%s,r=%s)=%s#"BUNFMT"%s%s%s\n",
- BATgetId(l), BATgetId(r), BATgetId(bn),
BATcount(bn),
- bn->tsorted ? "-sorted" : "",
- bn->trevsorted ? "-revsorted" : "",
- bn->tkey & 1 ? "-key" : "");
- assert(bn->htype == TYPE_void);
- return bn;
- }
- if (l->ttype == TYPE_void || BATcount(l) == 0 ||
- (r->ttype == TYPE_void && r->tseqbase == oid_nil)) {
- /* trivial: all values are nil */
- const void *nil = ATOMnilptr(r->ttype);
-
- bn = BATconstant(r->ttype == TYPE_oid ? TYPE_void : r->ttype,
- nil, BATcount(l), TRANSIENT);
- if (bn == NULL)
- return NULL;
- BATseqbase(bn, l->hseqbase);
- if (ATOMtype(bn->ttype) == TYPE_oid &&
- BATcount(bn) == 0) {
- bn->tdense = 1;
- BATseqbase(BATmirror(bn), 0);
- }
- ALGODEBUG fprintf(stderr,
"#BATproject(l=%s,r=%s)=%s#"BUNFMT"%s%s%s\n",
- BATgetId(l), BATgetId(r),
- BATgetId(bn), BATcount(bn),
- bn->tsorted ? "-sorted" : "",
- bn->trevsorted ? "-revsorted" : "",
- bn->tkey & 1 ? "-key" : "");
- return bn;
- }
- assert(l->ttype == TYPE_oid);
-
- if (ATOMstorage(tpe) == TYPE_str &&
- l->T->nonil &&
- (rcount == 0 ||
- lcount > (rcount >> 3) ||
- r->batRestricted == BAT_READ)) {
- /* insert strings as ints, we need to copy the string
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
