Add pg_get_domain_ddl() function to reconstruct CREATE DOMAIN statements

[timwaizenegger]

Add pg_get_domain_ddl() function to reconstruct CREATE DOMAIN statements
This patch introduces a new system function pg_get_domain_ddl() that
reconstructs the CREATE DOMAIN statement for a given domain. The function
takes a regtype parameter and returns the complete DDL statement including
the domain name, base type, default value, and all associated constraints.

The function follows the same pattern as other DDL reconstruction functions
like pg_get_functiondef() and pg_get_constraintdef(), providing a
decompiled reconstruction rather than the original command text.

Key features:

• Supports domains with default values
• Includes all domain constraints (CHECK, NOT NULL)
• Properly quotes identifiers and schema names
• Handles complex constraint expressions

A new documentation section "Get Object DDL Functions" has been created
to group DDL reconstruction functions, starting with pg_get_domain_ddl().
This provides a foundation for future DDL functions for other object types.

Comprehensive regression tests are included covering various domain
configurations.

Reference: PG-151
Author: Florin Irion florin.irion@enterprisedb.com
Author: Tim Waizenegger tim.waizenegger@enterprisedb.com

[irionr]

Not sure about this one...

[alvherre]

If I read the comments in pg_type.h right, the default value is in typdefaultbin, not in typdefault, so it looks like this shouldn't actually work. Maybe your test cases for defaults are all too simplistic, or maybe the comments in pg_type.h are wrong.

[irionr]

• typdefaultbin: stores the nodeToString representation of a default expression
• typdefault: stores the human-readable version of the default expression represented by typdefaultbin

So for our case it's better to use the human-readable version. no?
I think the comments are correct but also our implementation, no?

here is an example:

# CREATE DOMAIN public.regress_seq_domain2 AS integer 
     DEFAULT nextval('regress_test_seq'::regclass) 
     CONSTRAINT "regress_Constraint B" CHECK (VALUE > 10) 
     CONSTRAINT "regress_ConstraintC" CHECK (VALUE <> 55);

# SELECT pg_get_domain_ddl('regress_seq_domain2');
                                                                                                pg_get_domain_ddl
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
 CREATE DOMAIN public.regress_seq_domain2 AS integer DEFAULT nextval('regress_test_seq') CONSTRAINT "regress_Constraint B" CHECK (VALUE > 10) CONSTRAINT "regress_ConstraintC" CHECK (VALUE <> 55);
 
 
 # SELECT typdefaultbin, typdefault 
      FROM  pg_catalog.pg_type 
      WHERE typname ='regress_seq_domain2'  ;
-[ RECORD 1 ]-+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
typdefaultbin | {FUNCEXPR :funcid 480 :funcresulttype 23 :funcretset false :funcvariadic false :funcformat 2 :funccollid 0 :inputcollid 0 :args ({FUNCEXPR :funcid 1574 :funcresulttype 20 :funcretset false :funcvariadic false :funcformat 0 :funccollid 0 :inputcollid 0 :args ({CONST :consttype 2205 :consttypmod -1 :constcollid 0 :constlen 4 :constbyval true :constisnull false :location -1 :constvalue 4 [ 89 64 0 0 0 0 0 0 ]}) :location -1}) :location -1}
typdefault    | nextval('regress_test_seq'::regclass)

@alvherre Please explain a bit more if I didn't understand correctly.

[alvherre]

yeah, you should be able to use get_rule_expr() to obtain a human-readable expression when given the typdefaultbin. My concern here is that it is possible that domains exist where the default is not stored in the human-readable form. I don't know how can that happen TBH. Maybe an example to try is to create a domain that defaults to a sequence, and later rename the sequence. I think the typdefault would continue to use the old name, which would be bogus.

[irionr]

Indeed, it was not using the renamed sequence!
Thanks for catching it.
Adjusted.