Implement the FILTER clause for aggregate function calls.

This is SQL-standard with a few extensions, namely support for
subqueries and outer references in clause expressions.

catversion bump due to change in Aggref and WindowFunc.

David Fetter, reviewed by Dean Rasheed.

Author: nmisch

contrib/pg_stat_statements/pg_stat_statements.c

@@ -1546,6 +1546,7 @@ JumbleExpr(pgssJumbleState *jstate, Node *node)
 				JumbleExpr(jstate, (Node *) expr->args);
 				JumbleExpr(jstate, (Node *) expr->aggorder);
 				JumbleExpr(jstate, (Node *) expr->aggdistinct);
+				JumbleExpr(jstate, (Node *) expr->aggfilter);
 			}
 			break;
 		case T_WindowFunc:
@@ -1555,6 +1556,7 @@ JumbleExpr(pgssJumbleState *jstate, Node *node)
 				APP_JUMB(expr->winfnoid);
 				APP_JUMB(expr->winref);
 				JumbleExpr(jstate, (Node *) expr->args);
+				JumbleExpr(jstate, (Node *) expr->aggfilter);
 			}
 			break;
 		case T_ArrayRef:

doc/src/sgml/keywords.sgml

@@ -1786,7 +1786,7 @@
    </row>
    <row>
     <entry><token>FILTER</token></entry>
-    <entry></entry>
+    <entry>non-reserved</entry>
     <entry>reserved</entry>
     <entry>reserved</entry>
     <entry></entry>

doc/src/sgml/ref/select.sgml

@@ -598,6 +598,11 @@ GROUP BY <replaceable class="parameter">expression</replaceable> [, ...]
     making up each group, producing a separate value for each group
     (whereas without <literal>GROUP BY</literal>, an aggregate
     produces a single value computed across all the selected rows).
+    The set of rows fed to the aggregate function can be further filtered by
+    attaching a <literal>FILTER</literal> clause to the aggregate function
+    call; see <xref linkend="syntax-aggregates"> for more information.  When
+    a <literal>FILTER</literal> clause is present, only those rows matching it
+    are included.
     When <literal>GROUP BY</literal> is present, it is not valid for
     the <command>SELECT</command> list expressions to refer to
     ungrouped columns except within aggregate functions or if the

doc/src/sgml/syntax.sgml

@@ -1554,6 +1554,10 @@ sqrt(2)
     <secondary>invocation</secondary>
    </indexterm>
 
+   <indexterm zone="syntax-aggregates">
+    <primary>filter</primary>
+   </indexterm>
+
    <para>
     An <firstterm>aggregate expression</firstterm> represents the
     application of an aggregate function across the rows selected by a
@@ -1562,19 +1566,19 @@ sqrt(2)
     syntax of an aggregate expression is one of the following:
 
 <synopsis>
-<replaceable>aggregate_name</replaceable> (<replaceable>expression</replaceable> [ , ... ] [ <replaceable>order_by_clause</replaceable> ] )
-<replaceable>aggregate_name</replaceable> (ALL <replaceable>expression</replaceable> [ , ... ] [ <replaceable>order_by_clause</replaceable> ] )
-<replaceable>aggregate_name</replaceable> (DISTINCT <replaceable>expression</replaceable> [ , ... ] [ <replaceable>order_by_clause</replaceable> ] )
-<replaceable>aggregate_name</replaceable> ( * )
+<replaceable>aggregate_name</replaceable> (<replaceable>expression</replaceable> [ , ... ] [ <replaceable>order_by_clause</replaceable> ] ) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ]
+<replaceable>aggregate_name</replaceable> (ALL <replaceable>expression</replaceable> [ , ... ] [ <replaceable>order_by_clause</replaceable> ] ) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ]
+<replaceable>aggregate_name</replaceable> (DISTINCT <replaceable>expression</replaceable> [ , ... ] [ <replaceable>order_by_clause</replaceable> ] ) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ]
+<replaceable>aggregate_name</replaceable> ( * ) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ]
 </synopsis>
 
     where <replaceable>aggregate_name</replaceable> is a previously
-    defined aggregate (possibly qualified with a schema name),
+    defined aggregate (possibly qualified with a schema name) and
     <replaceable>expression</replaceable> is
     any value expression that does not itself contain an aggregate
-    expression or a window function call, and
-    <replaceable>order_by_clause</replaceable> is a optional
-    <literal>ORDER BY</> clause as described below.
+    expression or a window function call.  The optional
+    <replaceable>order_by_clause</replaceable> and
+    <replaceable>filter_clause</replaceable> are described below.
    </para>
 
    <para>
@@ -1606,6 +1610,23 @@ sqrt(2)
     distinct non-null values of <literal>f1</literal>.
    </para>
 
+   <para>
+    If <literal>FILTER</literal> is specified, then only the input
+    rows for which the <replaceable>filter_clause</replaceable>
+    evaluates to true are fed to the aggregate function; other rows
+    are discarded.  For example:
+<programlisting>
+SELECT
+    count(*) AS unfiltered,
+    count(*) FILTER (WHERE i < 5) AS filtered
+FROM generate_series(1,10) AS s(i);
+ unfiltered | filtered 
+------------+----------
+         10 |        4
+(1 row)
+</programlisting>
+   </para>
+
    <para>
     Ordinarily, the input rows are fed to the aggregate function in an
     unspecified order.  In many cases this does not matter; for example,
@@ -1709,10 +1730,10 @@ SELECT string_agg(a ORDER BY a, ',') FROM table;  -- incorrect
     The syntax of a window function call is one of the following:
 
 <synopsis>
-<replaceable>function_name</replaceable> (<optional><replaceable>expression</replaceable> <optional>, <replaceable>expression</replaceable> ... </optional></optional>) OVER ( <replaceable class="parameter">window_definition</replaceable> )
-<replaceable>function_name</replaceable> (<optional><replaceable>expression</replaceable> <optional>, <replaceable>expression</replaceable> ... </optional></optional>) OVER <replaceable>window_name</replaceable>
-<replaceable>function_name</replaceable> ( * ) OVER ( <replaceable class="parameter">window_definition</replaceable> )
-<replaceable>function_name</replaceable> ( * ) OVER <replaceable>window_name</replaceable>
+<replaceable>function_name</replaceable> (<optional><replaceable>expression</replaceable> <optional>, <replaceable>expression</replaceable> ... </optional></optional>) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ] OVER ( <replaceable class="parameter">window_definition</replaceable> )
+<replaceable>function_name</replaceable> (<optional><replaceable>expression</replaceable> <optional>, <replaceable>expression</replaceable> ... </optional></optional>) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ] OVER <replaceable>window_name</replaceable>
+<replaceable>function_name</replaceable> ( * ) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ] OVER ( <replaceable class="parameter">window_definition</replaceable> )
+<replaceable>function_name</replaceable> ( * ) [ FILTER ( WHERE <replaceable>filter_clause</replaceable> ) ] OVER <replaceable>window_name</replaceable>
 </synopsis>
     where <replaceable class="parameter">window_definition</replaceable>
     has the syntax
@@ -1836,7 +1857,8 @@ UNBOUNDED FOLLOWING
     The built-in window functions are described in <xref
     linkend="functions-window-table">.  Other window functions can be added by
     the user.  Also, any built-in or user-defined aggregate function can be
-    used as a window function.
+    used as a window function.  Only aggregate window functions accept
+    a <literal>FILTER</literal> clause.
    </para>
 
    <para>

src/backend/executor/execQual.c

@@ -4410,6 +4410,8 @@ ExecInitExpr(Expr *node, PlanState *parent)
 
 					astate->args = (List *) ExecInitExpr((Expr *) aggref->args,
 														 parent);
+					astate->aggfilter = ExecInitExpr(aggref->aggfilter,
+													 parent);
 
 					/*
 					 * Complain if the aggregate's arguments contain any
@@ -4448,6 +4450,8 @@ ExecInitExpr(Expr *node, PlanState *parent)
 
 					wfstate->args = (List *) ExecInitExpr((Expr *) wfunc->args,
 														  parent);
+					wfstate->aggfilter = ExecInitExpr(wfunc->aggfilter,
+													  parent);
 
 					/*
 					 * Complain if the windowfunc's arguments contain any

src/backend/executor/execUtils.c

@@ -649,9 +649,9 @@ get_last_attnums(Node *node, ProjectionInfo *projInfo)
 	}
 
 	/*
-	 * Don't examine the arguments of Aggrefs or WindowFuncs, because those do
-	 * not represent expressions to be evaluated within the overall
-	 * targetlist's econtext.
+	 * Don't examine the arguments or filters of Aggrefs or WindowFuncs,
+	 * because those do not represent expressions to be evaluated within the
+	 * overall targetlist's econtext.
 	 */
 	if (IsA(node, Aggref))
 		return false;

src/backend/executor/functions.c

@@ -380,7 +380,7 @@ sql_fn_post_column_ref(ParseState *pstate, ColumnRef *cref, Node *var)
 		param = ParseFuncOrColumn(pstate,
 								  list_make1(subfield),
 								  list_make1(param),
-								  NIL, false, false, false,
+								  NIL, NULL, false, false, false,
 								  NULL, true, cref->location);
 	}
 

src/backend/executor/nodeAgg.c

@@ -484,10 +484,23 @@ advance_aggregates(AggState *aggstate, AggStatePerGroup pergroup)
 	{
 		AggStatePerAgg peraggstate = &aggstate->peragg[aggno];
 		AggStatePerGroup pergroupstate = &pergroup[aggno];
+		ExprState  *filter = peraggstate->aggrefstate->aggfilter;
 		int			nargs = peraggstate->numArguments;
 		int			i;
 		TupleTableSlot *slot;
 
+		/* Skip anything FILTERed out */
+		if (filter)
+		{
+			bool		isnull;
+			Datum		res;
+
+			res = ExecEvalExprSwitchContext(filter, aggstate->tmpcontext,
+											&isnull, NULL);
+			if (isnull || !DatumGetBool(res))
+				continue;
+		}
+
 		/* Evaluate the current input expressions for this aggregate */
 		slot = ExecProject(peraggstate->evalproj, NULL);
 

src/backend/executor/nodeWindowAgg.c

@@ -227,9 +227,23 @@ advance_windowaggregate(WindowAggState *winstate,
 	int			i;
 	MemoryContext oldContext;
 	ExprContext *econtext = winstate->tmpcontext;
+	ExprState  *filter = wfuncstate->aggfilter;
 
 	oldContext = MemoryContextSwitchTo(econtext->ecxt_per_tuple_memory);
 
+	/* Skip anything FILTERed out */
+	if (filter)
+	{
+		bool		isnull;
+		Datum		res = ExecEvalExpr(filter, econtext, &isnull, NULL);
+
+		if (isnull || !DatumGetBool(res))
+		{
+			MemoryContextSwitchTo(oldContext);
+			return;
+		}
+	}
+
 	/* We start from 1, since the 0th arg will be the transition value */
 	i = 1;
 	foreach(arg, wfuncstate->args)

src/backend/nodes/copyfuncs.c

@@ -1137,6 +1137,7 @@ _copyAggref(const Aggref *from)
 	COPY_NODE_FIELD(args);
 	COPY_NODE_FIELD(aggorder);
 	COPY_NODE_FIELD(aggdistinct);
+	COPY_NODE_FIELD(aggfilter);
 	COPY_SCALAR_FIELD(aggstar);
 	COPY_SCALAR_FIELD(agglevelsup);
 	COPY_LOCATION_FIELD(location);
@@ -1157,6 +1158,7 @@ _copyWindowFunc(const WindowFunc *from)
 	COPY_SCALAR_FIELD(wincollid);
 	COPY_SCALAR_FIELD(inputcollid);
 	COPY_NODE_FIELD(args);
+	COPY_NODE_FIELD(aggfilter);
 	COPY_SCALAR_FIELD(winref);
 	COPY_SCALAR_FIELD(winstar);
 	COPY_SCALAR_FIELD(winagg);
@@ -2152,6 +2154,7 @@ _copyFuncCall(const FuncCall *from)
 	COPY_NODE_FIELD(funcname);
 	COPY_NODE_FIELD(args);
 	COPY_NODE_FIELD(agg_order);
+	COPY_NODE_FIELD(agg_filter);
 	COPY_SCALAR_FIELD(agg_star);
 	COPY_SCALAR_FIELD(agg_distinct);
 	COPY_SCALAR_FIELD(func_variadic);