database/postgres
1
0
Fork 0
mirror of https://github.com/postgres/postgres.git synced 2025-06-27 23:21:58 +03:00
Code Activity

Allow an alias to be attached to a JOIN ... USING

Browse Source 
This allows something like

    SELECT ... FROM t1 JOIN t2 USING (a, b, c) AS x

where x has the columns a, b, c and unlike a regular alias it does not
hide the range variables of the tables being joined t1 and t2.

Per SQL:2016 feature F404 "Range variable for common column names".

Reviewed-by: Vik Fearing <vik.fearing@2ndquadrant.com>
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://www.postgresql.org/message-id/flat/454638cf-d563-ab76-a585-2564428062af@2ndquadrant.com
This commit is contained in:
Peter Eisentraut
2021-03-31 17:09:24 +02:00
parent 27e1f14563
commit 055fee7eb4
22 changed files with 315 additions and 29 deletions
Download Patch File Download Diff File Expand all files Collapse all files

65
src/backend/parser/parse_expr.c
View File

@ -2512,26 +2512,61 @@ static Node *
transformWholeRowRef(ParseState *pstate, ParseNamespaceItem *nsitem,
int sublevels_up, int location)
{
Var *result;
/*
* Build the appropriate referencing node. Note that if the RTE is a
* function returning scalar, we create just a plain reference to the
* function value, not a composite containing a single column. This is
* pretty inconsistent at first sight, but it's what we've done
* historically. One argument for it is that "rel" and "rel.*" mean the
* same thing for composite relations, so why not for scalar functions...
* Build the appropriate referencing node. Normally this can be a
* whole-row Var, but if the nsitem is a JOIN USING alias then it contains
* only a subset of the columns of the underlying join RTE, so that will
* not work. Instead we immediately expand the reference into a RowExpr.
* Since the JOIN USING's common columns are fully determined at this
* point, there seems no harm in expanding it now rather than during
* planning.
*
* Note that if the RTE is a function returning scalar, we create just a
* plain reference to the function value, not a composite containing a
* single column. This is pretty inconsistent at first sight, but it's
* what we've done historically. One argument for it is that "rel" and
* "rel.*" mean the same thing for composite relations, so why not for
* scalar functions...
*/
result = makeWholeRowVar(nsitem->p_rte, nsitem->p_rtindex,
sublevels_up, true);
if (nsitem->p_names == nsitem->p_rte->eref)
{
Var *result;
/* location is not filled in by makeWholeRowVar */
result->location = location;
result = makeWholeRowVar(nsitem->p_rte, nsitem->p_rtindex,
sublevels_up, true);
/* mark relation as requiring whole-row SELECT access */
markVarForSelectPriv(pstate, result);
/* location is not filled in by makeWholeRowVar */
result->location = location;
return (Node *) result;
/* mark relation as requiring whole-row SELECT access */
markVarForSelectPriv(pstate, result);
return (Node *) result;
}
else
{
RowExpr *rowexpr;
List *fields;
/*
* We want only as many columns as are listed in p_names->colnames,
* and we should use those names not whatever possibly-aliased names
* are in the RTE. We needn't worry about marking the RTE for SELECT
* access, as the common columns are surely so marked already.
*/
expandRTE(nsitem->p_rte, nsitem->p_rtindex,
sublevels_up, location, false,
NULL, &fields);
rowexpr = makeNode(RowExpr);
rowexpr->args = list_truncate(fields,
list_length(nsitem->p_names->colnames));
rowexpr->row_typeid = RECORDOID;
rowexpr->row_format = COERCE_IMPLICIT_CAST;
rowexpr->colnames = copyObject(nsitem->p_names->colnames);
rowexpr->location = location;
return (Node *) rowexpr;
}
}
/*