Support arrays over domains.

Allowing arrays with a domain type as their element type was left un-done
in the original domain patch, but not for any very good reason.  This
omission leads to such surprising results as array_agg() not working on
a domain column, because the parser can't identify a suitable output type
for the polymorphic aggregate.

In order to fix this, first clean up the APIs of coerce_to_domain() and
some internal functions in parse_coerce.c so that we consistently pass
around a CoercionContext along with CoercionForm.  Previously, we sometimes
passed an "isExplicit" boolean flag instead, which is strictly less
information; and coerce_to_domain() didn't even get that, but instead had
to reverse-engineer isExplicit from CoercionForm.  That's contrary to the
documentation in primnodes.h that says that CoercionForm only affects
display and not semantics.  I don't think this change fixes any live bugs,
but it makes things more consistent.  The main reason for doing it though
is that now build_coercion_expression() receives ccontext, which it needs
in order to be able to recursively invoke coerce_to_target_type().

Next, reimplement ArrayCoerceExpr so that the node does not directly know
any details of what has to be done to the individual array elements while
performing the array coercion.  Instead, the per-element processing is
represented by a sub-expression whose input is a source array element and
whose output is a target array element.  This simplifies life in
parse_coerce.c, because it can build that sub-expression by a recursive
invocation of coerce_to_target_type().  The executor now handles the
per-element processing as a compiled expression instead of hard-wired code.
The main advantage of this is that we can use a single ArrayCoerceExpr to
handle as many as three successive steps per element: base type conversion,
typmod coercion, and domain constraint checking.  The old code used two
stacked ArrayCoerceExprs to handle type + typmod coercion, which was pretty
inefficient, and adding yet another array deconstruction to do domain
constraint checking seemed very unappetizing.

In the case where we just need a single, very simple coercion function,
doing this straightforwardly leads to a noticeable increase in the
per-array-element runtime cost.  Hence, add an additional shortcut evalfunc
in execExprInterp.c that skips unnecessary overhead for that specific form
of expression.  The runtime speed of simple cases is within 1% or so of
where it was before, while cases that previously required two levels of
array processing are significantly faster.

Finally, create an implicit array type for every domain type, as we do for
base types, enums, etc.  Everything except the array-coercion case seems
to just work without further effort.

Tom Lane, reviewed by Andrew Dunstan

Discussion: https://postgr.es/m/9852.1499791473@sss.pgh.pa.us

1 files changed, 28 insertions, 57 deletions

diff --git a/src/backend/utils/adt/arrayfuncs.c b/src/backend/utils/adt/arrayfuncs.c
index d1f2fe7d958..ca04b13e825 100644
--- a/src/backend/utils/adt/arrayfuncs.c
+++ b/src/backend/utils/adt/arrayfuncs.c
@@ -3092,21 +3092,18 @@ array_set(ArrayType *array, int nSubscripts, int *indx,
 /*
  * array_map()
  *
- * Map an array through an arbitrary function.  Return a new array with
- * same dimensions and each source element transformed by fn().  Each
- * source element is passed as the first argument to fn(); additional
- * arguments to be passed to fn() can be specified by the caller.
- * The output array can have a different element type than the input.
+ * Map an array through an arbitrary expression.  Return a new array with
+ * the same dimensions and each source element transformed by the given,
+ * already-compiled expression.  Each source element is placed in the
+ * innermost_caseval/innermost_casenull fields of the ExprState.
  *
  * Parameters are:
- * * fcinfo: a function-call data structure pre-constructed by the caller
- *	 to be ready to call the desired function, with everything except the
- *	 first argument position filled in.  In particular, flinfo identifies
- *	 the function fn(), and if nargs > 1 then argument positions after the
- *	 first must be preset to the additional values to be passed.  The
- *	 first argument position initially holds the input array value.
+ * * arrayd: Datum representing array argument.
+ * * exprstate: ExprState representing the per-element transformation.
+ * * econtext: context for expression evaluation.
  * * retType: OID of element type of output array.  This must be the same as,
- *	 or binary-compatible with, the result type of fn().
+ *	 or binary-compatible with, the result type of the expression.  It might
+ *	 be different from the input array's element type.
  * * amstate: workspace for array_map.  Must be zeroed by caller before
  *	 first call, and not touched after that.
  *
@@ -3116,11 +3113,14 @@ array_set(ArrayType *array, int nSubscripts, int *indx,
  *
  * NB: caller must assure that input array is not NULL.  NULL elements in
  * the array are OK however.
+ * NB: caller should be running in econtext's per-tuple memory context.
  */
 Datum
-array_map(FunctionCallInfo fcinfo, Oid retType, ArrayMapState *amstate)
+array_map(Datum arrayd,
+		  ExprState *exprstate, ExprContext *econtext,
+		  Oid retType, ArrayMapState *amstate)
 {
-	AnyArrayType *v;
+	AnyArrayType *v = DatumGetAnyArrayP(arrayd);
 	ArrayType  *result;
 	Datum	   *values;
 	bool	   *nulls;
@@ -3141,13 +3141,8 @@ array_map(FunctionCallInfo fcinfo, Oid retType, ArrayMapState *amstate)
 	array_iter	iter;
 	ArrayMetaState *inp_extra;
 	ArrayMetaState *ret_extra;
-
-	/* Get input array */
-	if (fcinfo->nargs < 1)
-		elog(ERROR, "invalid nargs: %d", fcinfo->nargs);
-	if (PG_ARGISNULL(0))
-		elog(ERROR, "null input array");
-	v = PG_GETARG_ANY_ARRAY_P(0);
+	Datum	   *transform_source = exprstate->innermost_caseval;
+	bool	   *transform_source_isnull = exprstate->innermost_casenull;
 
 	inpType = AARR_ELEMTYPE(v);
 	ndim = AARR_NDIM(v);
@@ -3158,7 +3153,7 @@ array_map(FunctionCallInfo fcinfo, Oid retType, ArrayMapState *amstate)
 	if (nitems <= 0)
 	{
 		/* Return empty array */
-		PG_RETURN_ARRAYTYPE_P(construct_empty_array(retType));
+		return PointerGetDatum(construct_empty_array(retType));
 	}
 
 	/*
@@ -3203,39 +3198,15 @@ array_map(FunctionCallInfo fcinfo, Oid retType, ArrayMapState *amstate)
 
 	for (i = 0; i < nitems; i++)
 	{
-		bool		callit = true;
-
 		/* Get source element, checking for NULL */
-		fcinfo->arg[0] = array_iter_next(&iter, &fcinfo->argnull[0], i,
-										 inp_typlen, inp_typbyval, inp_typalign);
-
-		/*
-		 * Apply the given function to source elt and extra args.
-		 */
-		if (fcinfo->flinfo->fn_strict)
-		{
-			int			j;
+		*transform_source =
+			array_iter_next(&iter, transform_source_isnull, i,
+							inp_typlen, inp_typbyval, inp_typalign);
 
-			for (j = 0; j < fcinfo->nargs; j++)
-			{
-				if (fcinfo->argnull[j])
-				{
-					callit = false;
-					break;
-				}
-			}
-		}
+		/* Apply the given expression to source element */
+		values[i] = ExecEvalExpr(exprstate, econtext, &nulls[i]);
 
-		if (callit)
-		{
-			fcinfo->isnull = false;
-			values[i] = FunctionCallInvoke(fcinfo);
-		}
-		else
-			fcinfo->isnull = true;
-
-		nulls[i] = fcinfo->isnull;
-		if (fcinfo->isnull)
+		if (nulls[i])
 			hasnulls = true;
 		else
 		{
@@ -3254,7 +3225,7 @@ array_map(FunctionCallInfo fcinfo, Oid retType, ArrayMapState *amstate)
 		}
 	}
 
-	/* Allocate and initialize the result array */
+	/* Allocate and fill the result array */
 	if (hasnulls)
 	{
 		dataoffset = ARR_OVERHEAD_WITHNULLS(ndim, nitems);
@@ -3273,18 +3244,18 @@ array_map(FunctionCallInfo fcinfo, Oid retType, ArrayMapState *amstate)
 	memcpy(ARR_DIMS(result), AARR_DIMS(v), ndim * sizeof(int));
 	memcpy(ARR_LBOUND(result), AARR_LBOUND(v), ndim * sizeof(int));
 
-	/*
-	 * Note: do not risk trying to pfree the results of the called function
-	 */
 	CopyArrayEls(result,
 				 values, nulls, nitems,
 				 typlen, typbyval, typalign,
 				 false);
 
+	/*
+	 * Note: do not risk trying to pfree the results of the called expression
+	 */
 	pfree(values);
 	pfree(nulls);
 
-	PG_RETURN_ARRAYTYPE_P(result);
+	return PointerGetDatum(result);
 }
 
 /*