Support the PostgreSQL domain

src/pgduckdb_types.cpp

@@ -32,6 +32,7 @@ extern "C" {
 
 #include "pgduckdb/pgduckdb_filter.hpp"
 #include "pgduckdb/pgduckdb_detoast.hpp"
+#include "pgduckdb/pgduckdb_process_lock.hpp"
 
 namespace pgduckdb {
 
@@ -917,10 +918,41 @@ ConvertPostgresToBaseDuckColumnType(Form_pg_attribute &attribute) {
 
 duckdb::LogicalType
 ConvertPostgresToDuckColumnType(Form_pg_attribute &attribute) {
+	int dimensions = -1;
+	Oid save_typoid = InvalidOid;
+	std::lock_guard<std::mutex> lock(GlobalProcessLock::GetLock());
+
+	if (get_typtype(attribute->atttypid) == TYPTYPE_DOMAIN) {
+		/* If the domain is an array type, you need to obtain the corresponding array dimension information */
+		if (type_is_array_domain(attribute->atttypid)) {
+			HeapTuple typeTuple = SearchSysCache1(TYPEOID, ObjectIdGetDatum(attribute->atttypid));
+			dimensions = ((Form_pg_type)GETSTRUCT(typeTuple))->typndims;
+			ReleaseSysCache(typeTuple);
+		}
+
+		save_typoid = attribute->atttypid;
+		/* It is a domain type that needs to be reduced to its base type */
+		attribute->atttypid = getBaseType(attribute->atttypid);
+	} else if (type_is_array(attribute->atttypid)) {
+		Oid elem_type = get_base_element_type(attribute->atttypid);
+		if (OidIsValid(elem_type) && get_typtype(elem_type) == TYPTYPE_DOMAIN) {
+			save_typoid = attribute->atttypid;
+			/* When the member type of an array is domain, you need to build a base array type */
+			attribute->atttypid = get_array_type(getBaseType(elem_type));
+		}
+	}
+
 	auto base_type = ConvertPostgresToBaseDuckColumnType(attribute);
-	auto dimensions = attribute->attndims;
-	if (dimensions == 0) {
-		return base_type;
+
+	if (save_typoid != InvalidOid) {
+		attribute->atttypid = save_typoid;
+	}
+
+	if (dimensions == -1) {
+		dimensions = attribute->attndims;
+		if (dimensions == 0) {
+			return base_type;
+		}
 	}
 
 	for (int i = 0; i < dimensions; i++) {

test/regression/expected/domain.out

@@ -0,0 +1,109 @@
+create domain domainvarchar varchar(5) check (value is not null);
+create domain domainnumeric numeric(8,2)  check (value is not null);
+create domain domainint4 int4 check (value > 0);
+create domain domaintext text check (value is not null);
+-- Test tables using domains
+create table basictest
+           ( testint4 domainint4
+           , testtext domaintext
+           , testvarchar domainvarchar
+           , testnumeric domainnumeric
+           );
+INSERT INTO basictest values ('88', 'haha', 'short', '123.12');      -- Good
+INSERT INTO basictest values ('88', 'haha', 'short text', '123.12'); -- Bad varchar
+ERROR:  value too long for type character varying(5)
+INSERT INTO basictest values ('88', 'haha', 'short', '123.1212');    -- Truncate numeric
+-- domain check
+INSERT INTO basictest values ('-1', 'haha', 'short', '123.1212');   -- Bad int4
+ERROR:  value for domain domainint4 violates check constraint "domainint4_check"
+INSERT INTO basictest values ('88', NULL, 'short', '123.1212');    -- Bad text
+ERROR:  value for domain domaintext violates check constraint "domaintext_check"
+INSERT INTO basictest values ('88', 'haha', NULL, '123.1212');   -- Bad varchar
+ERROR:  value for domain domainvarchar violates check constraint "domainvarchar_check"
+INSERT INTO basictest values ('88', 'haha', 'short', NULL);    -- Bad numeric
+ERROR:  value for domain domainnumeric violates check constraint "domainnumeric_check"
+SELECT 5::domainint4; -- Good
+ domainint4 
+------------
+          5
+(1 row)
+
+SELECT (-5)::domainint4; -- Bad int4
+ERROR:  value for domain domainint4 violates check constraint "domainint4_check"
+-- not support. It will be converted to the following statement
+-- SELECT ('-5'::integer)::domainint4 AS domainint4 FROM pgduckdb.xxx.basictest
+SELECT (-5)::domainint4 FROM basictest;
+WARNING:  (PGDuckDB/CreatePlan) Prepared query returned an error: 'Catalog Error: Type with name domainint4 does not exist!
+Did you mean "tinyint"?
+ERROR:  value for domain domainint4 violates check constraint "domainint4_check"
+select * from basictest;
+ testint4 | testtext | testvarchar | testnumeric 
+----------+----------+-------------+-------------
+       88 | haha     | short       |      123.12
+       88 | haha     | short       |      123.12
+(2 rows)
+
+select testtext || testvarchar as concat, testnumeric + 42 as sum
+from basictest;
+  concat   |   sum   
+-----------+---------
+ hahashort | 165.120
+ hahashort | 165.120
+(2 rows)
+
+select * from basictest where testtext = 'haha';
+ testint4 | testtext | testvarchar | testnumeric 
+----------+----------+-------------+-------------
+       88 | haha     | short       |      123.12
+       88 | haha     | short       |      123.12
+(2 rows)
+
+select * from basictest where testvarchar = 'short';
+ testint4 | testtext | testvarchar | testnumeric 
+----------+----------+-------------+-------------
+       88 | haha     | short       |      123.12
+       88 | haha     | short       |      123.12
+(2 rows)
+
+-- array_domain
+create domain domain_int_array as INT[];
+CREATE TABLE domain_int_array_1d(a domain_int_array);
+INSERT INTO domain_int_array_1d SELECT CAST(a as domain_int_array) FROM (VALUES
+    ('{1, 2, 3}'),
+    (NULL),
+    ('{4, 5, NULL, 7}'),
+    ('{}')
+) t(a);
+SELECT * FROM domain_int_array_1d;
+      a       
+--------------
+ {1,2,3}
+
+ {4,5,NULL,7}
+ {}
+(4 rows)
+
+CREATE TABLE domain_int_array_2d(a domainint4[]);
+INSERT INTO domain_int_array_2d SELECT CAST(a as domain_int_array) FROM (VALUES
+    ('{1, 2, 3}'),
+    (NULL),
+    ('{4, 5, NULL, 7}'),
+    ('{}')
+) t(a);
+SELECT * FROM domain_int_array_2d;
+      a       
+--------------
+ {1,2,3}
+
+ {4,5,NULL,7}
+ {}
+(4 rows)
+
+drop table domain_int_array_2d;
+drop table domain_int_array_1d;
+drop domain domain_int_array;
+drop table basictest;
+drop domain domainvarchar restrict;
+drop domain domainnumeric restrict;
+drop domain domainint4 restrict;
+drop domain domaintext;

test/regression/schedule

@@ -14,6 +14,7 @@ test: hugeint_conversion
 test: read_functions
 test: duckdb_only_functions
 test: duckdb_recycle
+test: domain
 test: cte
 test: create_schema
 test: create_table_as

test/regression/sql/domain.sql

@@ -0,0 +1,65 @@
+create domain domainvarchar varchar(5) check (value is not null);
+create domain domainnumeric numeric(8,2)  check (value is not null);
+create domain domainint4 int4 check (value > 0);
+create domain domaintext text check (value is not null);
+
+-- Test tables using domains
+create table basictest
+           ( testint4 domainint4
+           , testtext domaintext
+           , testvarchar domainvarchar
+           , testnumeric domainnumeric
+           );
+
+INSERT INTO basictest values ('88', 'haha', 'short', '123.12');      -- Good
+INSERT INTO basictest values ('88', 'haha', 'short text', '123.12'); -- Bad varchar
+INSERT INTO basictest values ('88', 'haha', 'short', '123.1212');    -- Truncate numeric
+
+-- domain check
+INSERT INTO basictest values ('-1', 'haha', 'short', '123.1212');   -- Bad int4
+INSERT INTO basictest values ('88', NULL, 'short', '123.1212');    -- Bad text
+INSERT INTO basictest values ('88', 'haha', NULL, '123.1212');   -- Bad varchar
+INSERT INTO basictest values ('88', 'haha', 'short', NULL);    -- Bad numeric
+SELECT 5::domainint4; -- Good
+SELECT (-5)::domainint4; -- Bad int4
+
+-- not support. It will be converted to the following statement
+-- SELECT ('-5'::integer)::domainint4 AS domainint4 FROM pgduckdb.xxx.basictest
+SELECT (-5)::domainint4 FROM basictest;
+
+select * from basictest;
+
+select testtext || testvarchar as concat, testnumeric + 42 as sum
+from basictest;
+
+select * from basictest where testtext = 'haha';
+select * from basictest where testvarchar = 'short';
+
+-- array_domain
+create domain domain_int_array as INT[];
+CREATE TABLE domain_int_array_1d(a domain_int_array);
+INSERT INTO domain_int_array_1d SELECT CAST(a as domain_int_array) FROM (VALUES
+    ('{1, 2, 3}'),
+    (NULL),
+    ('{4, 5, NULL, 7}'),
+    ('{}')
+) t(a);
+SELECT * FROM domain_int_array_1d;
+
+CREATE TABLE domain_int_array_2d(a domainint4[]);
+INSERT INTO domain_int_array_2d SELECT CAST(a as domain_int_array) FROM (VALUES
+    ('{1, 2, 3}'),
+    (NULL),
+    ('{4, 5, NULL, 7}'),
+    ('{}')
+) t(a);
+SELECT * FROM domain_int_array_2d;
+
+drop table domain_int_array_2d;
+drop table domain_int_array_1d;
+drop domain domain_int_array;
+drop table basictest;
+drop domain domainvarchar restrict;
+drop domain domainnumeric restrict;
+drop domain domainint4 restrict;
+drop domain domaintext;