Changeset: 775fcc0fb37d for MonetDB URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=775fcc0fb37d Modified Files: sql/server/rel_optimizer.c Branch: DVframework Log Message:
on the fly metadata derivation into partially materialized views (pmv):
required predicates collected and enumerated.
diffs (truncated from 888 to 300 lines):
diff --git a/sql/server/rel_optimizer.c b/sql/server/rel_optimizer.c
--- a/sql/server/rel_optimizer.c
+++ b/sql/server/rel_optimizer.c
@@ -30,9 +30,28 @@
#include "rel_planner.h"
#include "sql_env.h"
+#include "mtime.h"
+#include "mal_client.h"
+
#define new_func_list(sa) sa_list(sa)
#define new_col_list(sa) sa_list(sa)
+
+typedef struct table_pkeys
+{
+ str table_name;
+ struct list *pkey_column_names;
+} table_pkeys;
+
+typedef struct sel_predicate
+{
+ sht cmp_type; // cmp_gt = 0, cmp_gte = 1, cmp_lte = 2, cmp_lt = 3,
cmp_equal = 4, cmp_notequal = 5, cmp_filter = 6, cmp_or = 7, cmp_in = 8,
cmp_notin = 9, cmp_all = 10, cmp_project = 11 (up to here taken from
sql/rel.txt), cmp_range = 12, 13, 14, 15 (both not equal, first not equal
second equal, first equal second not equal, both equal, respectively).
+ sql_column* column; // column of the selection predicate
+ ValRecord** values; // array of values (valrecord pointers) compared.
Could be of different cardinality. E.g. 1 if cmp_equal, 2 if cmp_range,
anything if cmp_in, etc.
+ int num_values; // length of the values array.
+} sel_predicate;
+
+
typedef struct global_props {
int cnt[MAXOPS];
} global_props;
@@ -41,6 +60,20 @@ typedef sql_rel *(*rewrite_fptr)(int *ch
typedef int (*find_prop_fptr)(mvc *sql, sql_rel *rel);
static sql_subfunc *find_func( mvc *sql, char *name, list *exps );
+int is_table_in_list_table_pkeys(list *l, sql_table* st);
+int is_column_in_list_columns(list *l, sql_column* c);
+int is_column_of_table_in_list_table_pkeys(list *l, sql_table* st, sql_column*
c);
+list* extract_column_names_from_list_of_columns(list* list_of_columns);
+list* collect_PERPAD(mvc *sql, sql_rel *rel);
+lng get_enum_step_length(sql_column* c);
+sel_predicate** convert_all_into_in_clause_except_cmp_equal(list
*list_of_PERPAD);
+int enumerate_pkey_space(str** ret, sel_predicate** sps, int sps_enum_start,
int num_PERPAD, int* is_pkey_to_be_enumerated);
+int* enumerate_and_insert_into_temp_table(mvc *sql, sel_predicate** sps, int
num_PERPAD);
+str SQLstatementIntern(Client c, str *expr, str nme, int execute, bit output);
+str VAL2str(ValRecord* valp);
+void check_if_required_derived_metadata_is_already_available(list*
list_of_PERPAD, int* is_pkey_to_be_enumerated, int num_pkeys_to_be_enumerated);
+
+list *discovered_table_pkeys;
/* The important task of the relational optimizer is to optimize the
join order.
@@ -578,6 +611,792 @@ find_basetable( sql_rel *r)
}
}
+int is_table_in_list_table_pkeys(list *l, sql_table* st)
+{
+ node *n = NULL;
+
+ if (l && st) {
+ for (n = l->h; n; n = n->next)
+ {
+ table_pkeys *tp = n->data;
+ if (strcmp(tp->table_name, st->base.name) == 0)
+ return TRUE;
+ }
+ }
+ return FALSE;
+}
+
+int is_column_in_list_columns(list *l, sql_column* c)
+{
+ node *n = NULL;
+
+ if (l && c) {
+ for (n = l->h; n; n = n->next)
+ {
+ str s = n->data;
+ if (strcmp(s, c->base.name) == 0)
+ return TRUE;
+ }
+ }
+ return FALSE;
+}
+
+int is_column_of_table_in_list_table_pkeys(list *l, sql_table* st, sql_column*
c)
+{
+ node *n = NULL;
+
+ if (l && st && c) {
+ for (n = l->h; n; n = n->next)
+ {
+ table_pkeys *tp = n->data;
+ if (strcmp(tp->table_name, st->base.name) == 0)
+ {
+ return
is_column_in_list_columns(tp->pkey_column_names, c);
+ }
+ }
+ }
+ return FALSE;
+}
+
+
+list* extract_column_names_from_list_of_columns(list* list_of_sql_kc)
+{
+ node *n = NULL;
+ list* list_of_str = list_create(NULL);
+
+ if (list_of_sql_kc) {
+ for (n = list_of_sql_kc->h; n; n = n->next)
+ {
+ sql_kc* skc = n->data;
+ list_append(list_of_str, GDKstrdup(skc->c->base.name));
+ }
+ }
+ return list_of_str;
+}
+
+
+list* collect_PERPAD(mvc *sql, sql_rel *rel)
+{
+ node *n = NULL;
+ int i;
+ list *res = NULL;
+ list *first = NULL, *second = NULL;
+
+ if(rel == NULL)
+ return NULL;
+
+ printf("=====enter: collect_PERPAD\n");
+
+ switch (rel->op)
+ {
+ case op_basetable:
+ case op_table:
+ return NULL;
+ case op_join:
+ case op_left:
+ case op_right:
+ case op_full:
+
+ case op_semi:
+ case op_anti:
+
+ case op_union:
+ case op_inter:
+ case op_except:
+ first = collect_PERPAD(sql, rel->l);
+ second = collect_PERPAD(sql, rel->r);
+ if(first == NULL && second == NULL)
+ return NULL;
+ else if(first == NULL)
+ return list_merge(second, first, (fdup)NULL);
+ else
+ return list_merge(first, second, (fdup)NULL);
+ case op_project:
+ return collect_PERPAD(sql, rel->l);
+ case op_select:
+ printf("op_select!\n");
+ if(rel->exps == NULL)
+ break;
+
+ if(rel->exps->h == NULL)
+ break;
+
+
+ res = list_create(NULL);
+
+ for (n = rel->exps->h, i = 0; n; n = n->next, i++)
+ {
+ sql_exp *e = n->data;
+
+ printf("i: %d\n", i);
+ if(e == NULL)
+ continue;
+
+ if(e->type == e_cmp)
+ {
+ sql_exp *el = (sql_exp*) e->l;
+ sql_exp *er = (sql_exp*) e->r;
+ sql_exp *ef = NULL;
+ sql_exp *pivot = NULL;
+
+ if(e->f)
+ {
+ ef = (sql_exp*) e->f;
+ }
+
+ if(el->type == e_column || er->type ==
e_column)
+ {
+ sql_column *c;
+ atom *a;
+
+ if(el->type != e_column)
+ {
+ pivot = el;
+ el = er;
+ er = pivot;
+ }
+
+ c = exp_find_column(rel, el,
-1);
+ if(c == NULL)
+ continue;
+ printf("column: %s & table: %s
& er_type: %d & atom: %s & atom_r: %s\n", c->base.name, c->t->base.name,
er->type, er->name, er->rname);
+
+ if(find_prop(el->p,
PROP_HASHCOL))
+ {
+ sql_ukey *su =
(sql_ukey*) ((prop*)el->p)->value;
+
if(!is_table_in_list_table_pkeys(discovered_table_pkeys, c->t))
+ {
+ // TODO: add
the table only if it is a partially materialized view
+ table_pkeys *tp
= (table_pkeys*) GDKmalloc(sizeof(table_pkeys));
+ tp->table_name
= GDKstrdup(c->t->base.name);
+
tp->pkey_column_names =
extract_column_names_from_list_of_columns(su->k.columns);
+
list_append(discovered_table_pkeys, tp);
+ }
+ }
+
+
if(is_column_of_table_in_list_table_pkeys(discovered_table_pkeys, c->t, c))
+ {
+
+ sel_predicate* sp =
(sel_predicate*) GDKmalloc(sizeof(sel_predicate));
+
+ sp->cmp_type = e->flag;
+ sp->column = c;
+ sp->values =
(ValRecord**) GDKmalloc(sizeof(ValRecord*));
+
+ printf("pkey!\n");
+
+ /* if it is a range */
+ if(ef)
+ {
+ if(ef->type ==
e_convert)
+ ef =
ef->l;
+ if
(is_atom(ef->type) && (a = exp_value(ef, sql->args, sql->argc)) != NULL)
+ {
+
ValRecord** vr = NULL;
+
if(a->isnull)
+
printf("ERROR: range boundary is given as NULL\n");
+
+ /*
TODO: how to understand which type of range it is */
+
sp->cmp_type = 15;
+
sp->num_values = 2;
+ vr =
(ValRecord**) GDKrealloc(sp->values, 2*sizeof(ValRecord*));
+ if(vr
== NULL)
+ {
+
printf("ERROR: can not allocate memory\n");
+ }
+
sp->values = vr;
+ /*
TODO: check if ef is always the upper boundary of the range */
+
sp->values[1] = &(a->data);
+
printf("atom2sql: %s\n", atom2sql(a));
+ }
+ else
+ {
+
printf("ERROR: NOT atom or NO value in atom!\n");
+ }
+ }
+
+ if(er->type ==
e_convert)
+ er = er->l;
+
+
+
+ if (is_atom(er->type)
&& (a = exp_value(er, sql->args, sql->argc)) != NULL)
+ {
+
+ if(a->isnull)
+
printf("ERROR: primary key should not be NULL\n");
+
+
+
switch(sp->cmp_type)
+ {
+ case 0:
+ case 1:
+ case 2:
+ case 3:
+ case 4:
+ case 5:
+
sp->num_values = 1;
+
sp->values[0] = &(a->data);
+
break;
+ case 6:
+ case 7:
+
/* not handled (yet) */
+
printf("ERROR: case not handled yet!\n");
+
break;
+ case 8:
+ case 9:
+
/* TODO: handle IN and NOT IN */
+
printf("ERROR: case not handled yet!\n");
+
break;
+ case 10:
+ case 11:
+
/* not handled (yet) */
+
printf("ERROR: case not handled yet!\n");
+
break;
+ case 12:
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
https://www.monetdb.org/mailman/listinfo/checkin-list
