Teach genbki.pl to auto-generate pg_type entries for array types.

This eliminates some more tedium in adding new catalog entries,
specifically the need to set up an array type when adding a new
built-in data type.  Now it's sufficient to assign an OID for the
array type and write it in an "array_type_oid" metadata field.
You don't have to fill the base type's typarray link explicitly, either.

No catversion bump since the contents of pg_type aren't changed.
(Well, their order might be different, but that doesn't matter.)

John Naylor, reviewed and whacked around a bit by
Dagfinn Ilmari Mannsåker, and some more by me.

Discussion: https://postgr.es/m/CAJVSVGVTb6m9pJF49b3SuA8J+T-THO9c0hxOmoyv-yGKh-FbNg@mail.gmail.com

doc/src/sgml/bki.sgml

@@ -217,12 +217,14 @@
       <replaceable>value</replaceable> pairs.  The
       allowed <replaceable>key</replaceable>s are the names of the catalog's
       columns, plus the metadata keys <literal>oid</literal>,
-      <literal>oid_symbol</literal>, and <literal>descr</literal>.
+      <literal>oid_symbol</literal>,
+      <literal>array_type_oid</literal>, and <literal>descr</literal>.
       (The use of <literal>oid</literal> and <literal>oid_symbol</literal>
-      is described in <xref linkend="system-catalog-oid-assignment"/>
-      below.  <literal>descr</literal> supplies a description string for
-      the object, which will be inserted
-      into <structname>pg_description</structname>
+      is described in <xref linkend="system-catalog-oid-assignment"/> below,
+      while <literal>array_type_oid</literal> is described in
+      <xref linkend="system-catalog-auto-array-types"/>.
+      <literal>descr</literal> supplies a description string for the object,
+      which will be inserted into <structname>pg_description</structname>
       or <structname>pg_shdescription</structname> as appropriate.)
       While the metadata keys are optional, the catalog's defined columns
       must all be provided, except when the catalog's <literal>.h</literal>
@@ -282,8 +284,9 @@
         <para>
          Within each pair of curly braces, the metadata
          fields <literal>oid</literal>, <literal>oid_symbol</literal>,
-         and <literal>descr</literal> (if present) come first, in that
-         order, then the catalog's own fields appear in their defined order.
+         <literal>array_type_oid</literal>, and <literal>descr</literal>
+         (if present) come first, in that order, then the catalog's own
+         fields appear in their defined order.
         </para>
        </listitem>
 
@@ -498,6 +501,41 @@
    </para>
   </sect2>
 
+  <sect2 id="system-catalog-auto-array-types">
+   <title>Automatic Creation of Array Types</title>
+
+   <para>
+    Most scalar data types should have a corresponding array type (that is,
+    a standard varlena array type whose element type is the scalar type, and
+    which is referenced by the <structfield>typarray</structfield> field of
+    the scalar type's <structname>pg_type</structname>
+    entry).  <filename>genbki.pl</filename> is able to generate
+    the <structname>pg_type</structname> entry for the array type
+    automatically in most cases.
+   </para>
+
+   <para>
+    To use this facility, just write an <literal>array_type_oid
+    =&gt; <replaceable>nnnn</replaceable></literal> metadata field in the
+    scalar type's <structname>pg_type</structname> entry, specifying the OID
+    to use for the array type.  You may then omit
+    the <structfield>typarray</structfield> field, since it will be filled
+    automatically with that OID.
+   </para>
+
+   <para>
+    The generated array type's name is the scalar type's name with an
+    underscore prepended.  The array entry's other fields are filled from
+    <literal>BKI_ARRAY_DEFAULT(<replaceable>value</replaceable>)</literal>
+    annotations in <filename>pg_type.h</filename>, or if there isn't one,
+    copied from the scalar type.  (There's also a special case
+    for <structfield>typalign</structfield>.)  Then
+    the <structfield>typelem</structfield>
+    and <structfield>typarray</structfield> fields of the two entries are
+    set to cross-reference each other.
+   </para>
+  </sect2>
+
   <sect2 id="system-catalog-recipes">
    <title>Recipes for Editing Data Files</title>
 

src/backend/catalog/Catalog.pm

@@ -191,6 +191,11 @@ sub ParseHeader
 					{
 						$column{default} = $1;
 					}
+					elsif (
+						$attopt =~ /BKI_ARRAY_DEFAULT\(['"]?([^'"]+)['"]?\)/)
+					{
+						$column{array_default} = $1;
+					}
 					elsif ($attopt =~ /BKI_LOOKUP\((\w+)\)/)
 					{
 						$column{lookup} = $1;
@@ -277,12 +282,16 @@ sub ParseData
 			}
 		}
 
-		# If we found a hash reference, keep it.
-		# Only keep non-data strings if we
-		# are told to preserve formatting.
+		# If we found a hash reference, keep it, unless it is marked as
+		# autogenerated; in that case it'd duplicate an entry we'll
+		# autogenerate below.  (This makes it safe for reformat_dat_file.pl
+		# with --full-tuples to print autogenerated entries, which seems like
+		# useful behavior for debugging.)
+		#
+		# Only keep non-data strings if we are told to preserve formatting.
 		if (defined $hash_ref)
 		{
-			push @$data, $hash_ref;
+			push @$data, $hash_ref if !$hash_ref->{autogenerated};
 		}
 		elsif ($preserve_formatting)
 		{
@@ -290,6 +299,10 @@ sub ParseData
 		}
 	}
 	close $ifd;
+
+	# If this is pg_type, auto-generate array types too.
+	GenerateArrayTypes($schema, $data) if $catname eq 'pg_type';
+
 	return $data;
 }
 
@@ -343,6 +356,63 @@ sub AddDefaultValues
 	}
 }
 
+# If a pg_type entry has an array_type_oid metadata field,
+# auto-generate an entry for its array type.
+sub GenerateArrayTypes
+{
+	my $pgtype_schema = shift;
+	my $types         = shift;
+	my @array_types;
+
+	foreach my $elem_type (@$types)
+	{
+		next if !(ref $elem_type eq 'HASH');
+		next if !defined($elem_type->{array_type_oid});
+
+		my %array_type;
+
+		# Set up metadata fields for array type.
+		$array_type{oid}           = $elem_type->{array_type_oid};
+		$array_type{autogenerated} = 1;
+		$array_type{line_number}   = $elem_type->{line_number};
+
+		# Set up column values derived from the element type.
+		$array_type{typname} = '_' . $elem_type->{typname};
+		$array_type{typelem} = $elem_type->{typname};
+
+		# Arrays require INT alignment, unless the element type requires
+		# DOUBLE alignment.
+		$array_type{typalign} = $elem_type->{typalign} eq 'd' ? 'd' : 'i';
+
+		# Fill in the rest of the array entry's fields.
+		foreach my $column (@$pgtype_schema)
+		{
+			my $attname = $column->{name};
+
+			# Skip if we already set it above.
+			next if defined $array_type{$attname};
+
+			# Apply the BKI_ARRAY_DEFAULT setting if there is one,
+			# otherwise copy the field from the element type.
+			if (defined $column->{array_default})
+			{
+				$array_type{$attname} = $column->{array_default};
+			}
+			else
+			{
+				$array_type{$attname} = $elem_type->{$attname};
+			}
+		}
+
+		# Lastly, cross-link the array to the element type.
+		$elem_type->{typarray} = $array_type{typname};
+
+		push @array_types, \%array_type;
+	}
+
+	push @$types, @array_types;
+}
+
 # Rename temporary files to final names.
 # Call this function with the final file name and the .tmp extension.
 #

src/backend/catalog/genbki.pl

@@ -394,7 +394,9 @@ EOM
 			next
 			  if $key eq "oid"
 			  || $key eq "oid_symbol"
+			  || $key eq "array_type_oid"
 			  || $key eq "descr"
+			  || $key eq "autogenerated"
 			  || $key eq "line_number";
 			die sprintf "unrecognized field name \"%s\" in %s.dat line %s\n",
 			  $key, $catname, $bki_values{line_number}

src/include/catalog/genbki.h

@@ -34,6 +34,8 @@
 #define BKI_FORCE_NOT_NULL
 /* Specifies a default value for a catalog field */
 #define BKI_DEFAULT(value)
+/* Specifies a default value for auto-generated array types */
+#define BKI_ARRAY_DEFAULT(value)
 /* Indicates how to perform name lookups for an OID or OID-array field */
 #define BKI_LOOKUP(catalog)
 

src/include/catalog/pg_type.h

@@ -52,7 +52,7 @@ CATALOG(pg_type,1247,TypeRelationId) BKI_BOOTSTRAP BKI_ROWTYPE_OID(71,TypeRelati
 	 * "varlena" type (one that has a length word), -2 to indicate a
 	 * null-terminated C string.
 	 */
-	int16		typlen;
+	int16		typlen BKI_ARRAY_DEFAULT(-1);
 
 	/*
 	 * typbyval determines whether internal Postgres routines pass a value of
@@ -62,7 +62,7 @@ CATALOG(pg_type,1247,TypeRelationId) BKI_BOOTSTRAP BKI_ROWTYPE_OID(71,TypeRelati
 	 * typbyval can be false even if the length would allow pass-by-value; for
 	 * example, type macaddr8 is pass-by-ref even when Datum is 8 bytes.
 	 */
-	bool		typbyval;
+	bool		typbyval BKI_ARRAY_DEFAULT(f);
 
 	/*
 	 * typtype is 'b' for a base type, 'c' for a composite type (e.g., a
@@ -71,7 +71,7 @@ CATALOG(pg_type,1247,TypeRelationId) BKI_BOOTSTRAP BKI_ROWTYPE_OID(71,TypeRelati
 	 *
 	 * If typtype is 'c', typrelid is the OID of the class' entry in pg_class.
 	 */
-	char		typtype BKI_DEFAULT(b);
+	char		typtype BKI_DEFAULT(b) BKI_ARRAY_DEFAULT(b);
 
 	/*
 	 * typcategory and typispreferred help the parser distinguish preferred
@@ -81,10 +81,10 @@ CATALOG(pg_type,1247,TypeRelationId) BKI_BOOTSTRAP BKI_ROWTYPE_OID(71,TypeRelati
 	 */
 
 	/* arbitrary type classification */
-	char		typcategory;
+	char		typcategory BKI_ARRAY_DEFAULT(A);
 
 	/* is type "preferred" within its category? */
-	bool		typispreferred BKI_DEFAULT(f);
+	bool		typispreferred BKI_DEFAULT(f) BKI_ARRAY_DEFAULT(f);
 
 	/*
 	 * If typisdefined is false, the entry is only a placeholder (forward
@@ -97,7 +97,7 @@ CATALOG(pg_type,1247,TypeRelationId) BKI_BOOTSTRAP BKI_ROWTYPE_OID(71,TypeRelati
 	char		typdelim BKI_DEFAULT(',');
 
 	/* associated pg_class OID if a composite type, else 0 */
-	Oid			typrelid BKI_DEFAULT(0);
+	Oid			typrelid BKI_DEFAULT(0) BKI_ARRAY_DEFAULT(0);
 
 	/*
 	 * If typelem is not 0 then it identifies another row in pg_type. The
@@ -116,19 +116,19 @@ CATALOG(pg_type,1247,TypeRelationId) BKI_BOOTSTRAP BKI_ROWTYPE_OID(71,TypeRelati
 	 * If there is a "true" array type having this type as element type,
 	 * typarray links to it.  Zero if no associated "true" array type.
 	 */
-	Oid			typarray BKI_DEFAULT(0) BKI_LOOKUP(pg_type);
+	Oid			typarray BKI_DEFAULT(0) BKI_ARRAY_DEFAULT(0) BKI_LOOKUP(pg_type);
 
 	/*
 	 * I/O conversion procedures for the datatype.
 	 */
 
 	/* text format (required) */
-	regproc		typinput BKI_LOOKUP(pg_proc);
-	regproc		typoutput BKI_LOOKUP(pg_proc);
+	regproc		typinput BKI_ARRAY_DEFAULT(array_in) BKI_LOOKUP(pg_proc);
+	regproc		typoutput BKI_ARRAY_DEFAULT(array_out) BKI_LOOKUP(pg_proc);
 
 	/* binary format (optional) */
-	regproc		typreceive BKI_LOOKUP(pg_proc);
-	regproc		typsend BKI_LOOKUP(pg_proc);
+	regproc		typreceive BKI_ARRAY_DEFAULT(array_recv) BKI_LOOKUP(pg_proc);
+	regproc		typsend BKI_ARRAY_DEFAULT(array_send) BKI_LOOKUP(pg_proc);
 
 	/*
 	 * I/O functions for optional type modifiers.
@@ -139,7 +139,7 @@ CATALOG(pg_type,1247,TypeRelationId) BKI_BOOTSTRAP BKI_ROWTYPE_OID(71,TypeRelati
 	/*
 	 * Custom ANALYZE procedure for the datatype (0 selects the default).
 	 */
-	regproc		typanalyze BKI_DEFAULT(-) BKI_LOOKUP(pg_proc);
+	regproc		typanalyze BKI_DEFAULT(-) BKI_ARRAY_DEFAULT(array_typanalyze) BKI_LOOKUP(pg_proc);
 
 	/* ----------------
 	 * typalign is the alignment required when storing a value of this
@@ -177,7 +177,7 @@ CATALOG(pg_type,1247,TypeRelationId) BKI_BOOTSTRAP BKI_ROWTYPE_OID(71,TypeRelati
 	 * 'm' MAIN		  like 'x' but try to keep in main tuple
 	 * ----------------
 	 */
-	char		typstorage BKI_DEFAULT(p);
+	char		typstorage BKI_DEFAULT(p) BKI_ARRAY_DEFAULT(x);
 
 	/*
 	 * This flag represents a "NOT NULL" constraint against this datatype.
@@ -221,7 +221,7 @@ CATALOG(pg_type,1247,TypeRelationId) BKI_BOOTSTRAP BKI_ROWTYPE_OID(71,TypeRelati
 	 * a default expression for the type.  Currently this is only used for
 	 * domains.
 	 */
-	pg_node_tree typdefaultbin BKI_DEFAULT(_null_);
+	pg_node_tree typdefaultbin BKI_DEFAULT(_null_) BKI_ARRAY_DEFAULT(_null_);
 
 	/*
 	 * typdefault is NULL if the type has no associated default value. If
@@ -231,7 +231,7 @@ CATALOG(pg_type,1247,TypeRelationId) BKI_BOOTSTRAP BKI_ROWTYPE_OID(71,TypeRelati
 	 * external representation of the type's default value, which may be fed
 	 * to the type's input converter to produce a constant.
 	 */
-	text		typdefault BKI_DEFAULT(_null_);
+	text		typdefault BKI_DEFAULT(_null_) BKI_ARRAY_DEFAULT(_null_);
 
 	/*
 	 * Access permissions

src/include/catalog/reformat_dat_file.pl

@@ -5,9 +5,9 @@
 #    Perl script that reads in catalog data file(s) and writes out
 #    functionally equivalent file(s) in a standard format.
 #
-#    In each entry of a reformatted file, metadata fields (if any) come
-#    first, with normal attributes starting on the following line, in
-#    the same order as the columns of the corresponding catalog.
+#    In each entry of a reformatted file, metadata fields (if present)
+#    come first, with normal attributes starting on the following line,
+#    in the same order as the columns of the corresponding catalog.
 #    Comments and blank lines are preserved.
 #
 # Portions Copyright (c) 1996-2018, PostgreSQL Global Development Group
@@ -26,6 +26,11 @@ use FindBin;
 use lib "$FindBin::RealBin/../../backend/catalog/";
 use Catalog;
 
+# Names of the metadata fields of a catalog entry.  (line_number is also
+# a metadata field, but we never write it out, so it's not listed here.)
+my @METADATA =
+  ('oid', 'oid_symbol', 'array_type_oid', 'descr', 'autogenerated');
+
 my @input_files;
 my $output_path = '';
 my $full_tuples = 0;
@@ -62,9 +67,6 @@ if ($output_path ne '' && substr($output_path, -1) ne '/')
 	$output_path .= '/';
 }
 
-# Metadata of a catalog entry
-my @METADATA = ('oid', 'oid_symbol', 'descr');
-
 # Read all the input files into internal data structures.
 # We pass data file names as arguments and then look for matching
 # headers to parse the schema from.
@@ -142,6 +144,9 @@ foreach my $catname (@catnames)
 
 			if (!$full_tuples)
 			{
+				# If it's an autogenerated entry, drop it completely.
+				next if $values{autogenerated};
+				# Else, just drop any default/computed fields.
 				strip_default_values(\%values, $schema, $catname);
 			}
 
@@ -162,10 +167,6 @@ foreach my $catname (@catnames)
 			print $dat " },\n";
 		}
 
-		# Strings -- handle accordingly or ignore. It was necessary to
-		# ignore bare commas during the initial data conversion. This
-		# should be a no-op now, but we may as well keep that behavior.
-
 		# Preserve blank lines.
 		elsif ($data =~ /^\s*$/)
 		{
@@ -207,6 +208,14 @@ sub strip_default_values
 	{
 		delete $row->{pronargs} if defined $row->{proargtypes};
 	}
+
+	# If a pg_type entry has an auto-generated array type, then its
+	# typarray field is a computed value too (see GenerateArrayTypes).
+	if ($catname eq 'pg_type')
+	{
+		delete $row->{typarray} if defined $row->{array_type_oid};
+	}
+
 	return;
 }