Build out the planner support function infrastructure.

Add support function requests for estimating the selectivity, cost,
and number of result rows (if a SRF) of the target function.

The lack of a way to estimate selectivity of a boolean-returning
function in WHERE has been a recognized deficiency of the planner
since Berkeley days.  This commit finally fixes it.

In addition, non-constant estimates of cost and number of output
rows are now possible.  We still fall back to looking at procost
and prorows if the support function doesn't service the request,
of course.

To make concrete use of the possibility of estimating output rowcount
for SRFs, this commit adds support functions for array_unnest(anyarray)
and the integer variants of generate_series; the lack of plausible
rowcount estimates for those, even when it's obvious to a human,
has been a repeated subject of complaints.  Obviously, much more
could now be done in this line, but I'm mostly just trying to get
the infrastructure in place.

Discussion: https://postgr.es/m/15193.1548028093@sss.pgh.pa.us

1 files changed, 74 insertions, 0 deletions

diff --git a/src/backend/utils/adt/int.c b/src/backend/utils/adt/int.c
index ad8e6d02ee4..04825fc77de 100644
--- a/src/backend/utils/adt/int.c
+++ b/src/backend/utils/adt/int.c
@@ -30,11 +30,15 @@
 
 #include <ctype.h>
 #include <limits.h>
+#include <math.h>
 
 #include "catalog/pg_type.h"
 #include "common/int.h"
 #include "funcapi.h"
 #include "libpq/pqformat.h"
+#include "nodes/nodeFuncs.h"
+#include "nodes/supportnodes.h"
+#include "optimizer/optimizer.h"
 #include "utils/array.h"
 #include "utils/builtins.h"
 
@@ -1427,3 +1431,73 @@ generate_series_step_int4(PG_FUNCTION_ARGS)
 		/* do when there is no more left */
 		SRF_RETURN_DONE(funcctx);
 }
+
+/*
+ * Planner support function for generate_series(int4, int4 [, int4])
+ */
+Datum
+generate_series_int4_support(PG_FUNCTION_ARGS)
+{
+	Node	   *rawreq = (Node *) PG_GETARG_POINTER(0);
+	Node	   *ret = NULL;
+
+	if (IsA(rawreq, SupportRequestRows))
+	{
+		/* Try to estimate the number of rows returned */
+		SupportRequestRows *req = (SupportRequestRows *) rawreq;
+
+		if (is_funcclause(req->node))	/* be paranoid */
+		{
+			List	   *args = ((FuncExpr *) req->node)->args;
+			Node	   *arg1,
+					   *arg2,
+					   *arg3;
+
+			/* We can use estimated argument values here */
+			arg1 = estimate_expression_value(req->root, linitial(args));
+			arg2 = estimate_expression_value(req->root, lsecond(args));
+			if (list_length(args) >= 3)
+				arg3 = estimate_expression_value(req->root, lthird(args));
+			else
+				arg3 = NULL;
+
+			/*
+			 * If any argument is constant NULL, we can safely assume that
+			 * zero rows are returned.  Otherwise, if they're all non-NULL
+			 * constants, we can calculate the number of rows that will be
+			 * returned.  Use double arithmetic to avoid overflow hazards.
+			 */
+			if ((IsA(arg1, Const) &&
+				 ((Const *) arg1)->constisnull) ||
+				(IsA(arg2, Const) &&
+				 ((Const *) arg2)->constisnull) ||
+				(arg3 != NULL && IsA(arg3, Const) &&
+				 ((Const *) arg3)->constisnull))
+			{
+				req->rows = 0;
+				ret = (Node *) req;
+			}
+			else if (IsA(arg1, Const) &&
+					 IsA(arg2, Const) &&
+					 (arg3 == NULL || IsA(arg3, Const)))
+			{
+				double		start,
+							finish,
+							step;
+
+				start = DatumGetInt32(((Const *) arg1)->constvalue);
+				finish = DatumGetInt32(((Const *) arg2)->constvalue);
+				step = arg3 ? DatumGetInt32(((Const *) arg3)->constvalue) : 1;
+
+				/* This equation works for either sign of step */
+				if (step != 0)
+				{
+					req->rows = floor((finish - start + step) / step);
+					ret = (Node *) req;
+				}
+			}
+		}
+	}
+
+	PG_RETURN_POINTER(ret);
+}