Add support for BYTEA/BLOB (#511)

Fixes #464

src/pgduckdb_filter.cpp

@@ -83,6 +83,8 @@ FilterOperationSwitch(const Datum &value, const duckdb::Value &constant, Oid typ
 	case TEXTOID:
 	case VARCHAROID:
 		return StringFilterOperation<OP>(value, constant, type_oid == BPCHAROID);
+	case BYTEAOID:
+		return StringFilterOperation<OP>(value, constant, false);
 	default:
 		throw duckdb::InvalidTypeException(
 		    duckdb::string("(DuckDB/FilterOperationSwitch) Unsupported duckdb type: " + std::to_string(type_oid)));

src/pgduckdb_types.cpp

@@ -2,6 +2,7 @@
 #include "duckdb/common/shared_ptr.hpp"
 #include "duckdb/common/extra_type_info.hpp"
 #include "duckdb/common/types/uuid.hpp"
+#include "duckdb/common/types/blob.hpp"
 
 #include "pgduckdb/pgduckdb_types.hpp"
 #include "pgduckdb/pgduckdb_utils.hpp"
@@ -199,6 +200,17 @@ ConvertVarCharDatum(const duckdb::Value &value) {
 	return PointerGetDatum(result);
 }
 
+static Datum
+ConvertBinaryDatum(const duckdb::Value &value) {
+	auto str = value.GetValueUnsafe<duckdb::string_t>();
+	auto blob_len = str.GetSize();
+	auto blob = str.GetDataUnsafe();
+	bytea* result = (bytea *)palloc0(blob_len + VARHDRSZ);
+	SET_VARSIZE(result, blob_len + VARHDRSZ);
+	memcpy(VARDATA(result), blob, blob_len);
+	return PointerGetDatum(result);
+}
+
 inline Datum
 ConvertDateDatum(const duckdb::Value &value) {
 	duckdb::date_t date = value.GetValue<duckdb::date_t>();
@@ -505,6 +517,19 @@ struct PostgresTypeTraits<VARCHAROID> {
 	}
 };
 
+// BLOB type
+template <>
+struct PostgresTypeTraits<BYTEAOID> {
+	static constexpr int16_t typlen = -1; // variable-length
+	static constexpr bool typbyval = false;
+	static constexpr char typalign = 'i';
+
+	static inline Datum
+	ToDatum(const duckdb::Value &val) {
+		return ConvertBinaryDatum(val);
+	}
+};
+
 template <int32_t OID>
 struct PostgresOIDMapping {
 	static constexpr int32_t postgres_oid = OID;
@@ -545,6 +570,7 @@ using TimestampArray = PODArray<PostgresOIDMapping<TIMESTAMPOID>>;
 using UUIDArray = PODArray<PostgresOIDMapping<UUIDOID>>;
 using VarCharArray = PODArray<PostgresOIDMapping<VARCHAROID>>;
 using NumericArray = PODArray<PostgresOIDMapping<NUMERICOID>>;
+using ByteArray = PODArray<PostgresOIDMapping<BYTEAOID>>;
 
 static idx_t
 GetDuckDBListDimensionality(const duckdb::LogicalType &list_type, idx_t depth = 0) {
@@ -733,6 +759,10 @@ ConvertDuckToPostgresValue(TupleTableSlot *slot, duckdb::Value &value, idx_t col
 		slot->tts_values[col] = ConvertUUIDDatum(value);
 		break;
 	}
+	case BYTEAOID: {
+		slot->tts_values[col] = ConvertBinaryDatum(value);
+		break;
+	}
 	case BOOLARRAYOID: {
 		ConvertDuckToPostgresArray<BoolArray>(slot, value, col);
 		break;
@@ -784,6 +814,10 @@ ConvertDuckToPostgresValue(TupleTableSlot *slot, duckdb::Value &value, idx_t col
 		ConvertDuckToPostgresArray<UUIDArray>(slot, value, col);
 		break;
 	}
+	case BYTEAARRAYOID: {
+		ConvertDuckToPostgresArray<ByteArray>(slot, value, col);
+		break;
+	}
 	default:
 		elog(WARNING, "(PGDuckDB/ConvertDuckToPostgresValue) Unsuported pgduckdb type: %d", oid);
 		return false;
@@ -866,6 +900,9 @@ ConvertPostgresToBaseDuckColumnType(Form_pg_attribute &attribute) {
 	case REGCLASSOID:
 	case REGCLASSARRAYOID:
 		return duckdb::LogicalTypeId::UINTEGER;
+	case BYTEAOID:
+	case BYTEAARRAYOID:
+		return duckdb::LogicalTypeId::BLOB;
 	default:
 		return duckdb::LogicalType::USER("UnsupportedPostgresType (Oid=" + std::to_string(attribute->atttypid) + ")");
 	}
@@ -920,6 +957,8 @@ GetPostgresArrayDuckDBType(const duckdb::LogicalType &type) {
 		return NUMERICARRAYOID;
 	case duckdb::LogicalTypeId::UUID:
 		return UUIDARRAYOID;
+	case duckdb::LogicalTypeId::BLOB:
+		return BYTEAARRAYOID;
 	default: {
 		elog(WARNING, "(PGDuckDB/GetPostgresDuckDBType) Unsupported `LIST` subtype %d to Postgres type",
 		     static_cast<uint8_t>(type.id()));
@@ -974,6 +1013,8 @@ GetPostgresDuckDBType(const duckdb::LogicalType &type) {
 		}
 		return GetPostgresArrayDuckDBType(*duck_type);
 	}
+	case duckdb::LogicalTypeId::BLOB:
+		return BYTEAOID;
 	default: {
 		elog(WARNING, "(PGDuckDB/GetPostgresDuckDBType) Could not convert DuckDB type: %s to Postgres type",
 		     type.ToString().c_str());
@@ -1222,6 +1263,14 @@ ConvertPostgresToDuckValue(Oid attr_type, Datum value, duckdb::Vector &result, i
 		Append(result, duckdb_uuid, offset);
 		break;
 	}
+	case duckdb::LogicalTypeId::BLOB: {
+		const char *bytea_data = VARDATA_ANY(value);
+		size_t bytea_length = VARSIZE_ANY_EXHDR(value);
+		const duckdb::string_t s(bytea_data, bytea_length);
+		auto data = duckdb::FlatVector::GetData<duckdb::string_t>(result);
+		data[offset] = duckdb::StringVector::AddString(result, s);
+		break;
+	}
 	case duckdb::LogicalTypeId::LIST: {
 		// Convert Datum to ArrayType
 		auto array = DatumGetArrayTypeP(value);

test/regression/expected/array_type_support.out

@@ -318,6 +318,19 @@ SELECT * FROM varchar_array_2d;
  {{some,strings},{NULL,last}}
 (5 rows)
 
+-- BYTEA (single dimension)
+CREATE TABLE bytea_array_1d (a bytea[]);
+INSERT INTO bytea_array_1d (a) 
+VALUES 
+    (ARRAY[decode('01020304', 'hex'), decode('aabbccdd', 'hex')]),
+    (ARRAY[decode('11223344', 'hex'), decode('55667788', 'hex')]);
+SELECT * FROM bytea_array_1d;
+               a               
+-------------------------------
+ {"\\x01020304","\\xaabbccdd"}
+ {"\\x11223344","\\x55667788"}
+(2 rows)
+
 -- TIMESTAMP (two dimensions)
 CREATE TABLE timestamp_array_2d(a TIMESTAMP[][]);
 INSERT INTO timestamp_array_2d VALUES

test/regression/expected/type_support.out

@@ -308,6 +308,33 @@ SELECT * FROM json_tbl;
  {}
 (4 rows)
 
+-- BLOB
+CREATE TABLE blob_tbl(a bytea);
+INSERT INTO blob_tbl SELECT CAST(a as bytea) FROM (VALUES
+    ('\x'),
+    ('\x110102030405060708090a0b0c0d0e0f'),
+    (''),
+    ('\x00'),
+    ('\x07'),
+    (NULL)
+) t(a);
+SELECT * from blob_tbl;
+                 a                  
+------------------------------------
+ \x
+ \x110102030405060708090a0b0c0d0e0f
+ \x
+ \x00
+ \x07
+
+(6 rows)
+
+SELECT * from blob_tbl where a = '\x07';
+  a   
+------
+ \x07
+(1 row)
+
 -- REGCLASSOID
 CREATE TABLE regclass_tbl (a REGCLASS);
 INSERT INTO regclass_tbl VALUES (42), (3000000000);
@@ -337,4 +364,5 @@ DROP TABLE bigint_numeric;
 DROP TABLE hugeint_numeric;
 DROP TABLE uuid_tbl;
 DROP TABLE json_tbl;
+DROP TABLE blob_tbl;
 DROP TABLE regclass_tbl;

test/regression/sql/array_type_support.sql

@@ -195,6 +195,15 @@ INSERT INTO varchar_array_2d VALUES
     ('{{"some","strings"},{NULL,"last"}}');
 SELECT * FROM varchar_array_2d;
 
+-- BYTEA (single dimension)
+CREATE TABLE bytea_array_1d (a bytea[]);
+
+INSERT INTO bytea_array_1d (a) 
+VALUES 
+    (ARRAY[decode('01020304', 'hex'), decode('aabbccdd', 'hex')]),
+    (ARRAY[decode('11223344', 'hex'), decode('55667788', 'hex')]);
+SELECT * FROM bytea_array_1d;
+
 -- TIMESTAMP (two dimensions)
 CREATE TABLE timestamp_array_2d(a TIMESTAMP[][]);
 INSERT INTO timestamp_array_2d VALUES

test/regression/sql/type_support.sql

@@ -154,6 +154,19 @@ INSERT INTO json_tbl SELECT CAST(a as JSON) FROM (VALUES
 ) t(a);
 SELECT * FROM json_tbl;
 
+-- BLOB
+CREATE TABLE blob_tbl(a bytea);
+INSERT INTO blob_tbl SELECT CAST(a as bytea) FROM (VALUES
+    ('\x'),
+    ('\x110102030405060708090a0b0c0d0e0f'),
+    (''),
+    ('\x00'),
+    ('\x07'),
+    (NULL)
+) t(a);
+SELECT * from blob_tbl;
+SELECT * from blob_tbl where a = '\x07';
+
 -- REGCLASSOID
 CREATE TABLE regclass_tbl (a REGCLASS);
 INSERT INTO regclass_tbl VALUES (42), (3000000000);
@@ -178,4 +191,5 @@ DROP TABLE bigint_numeric;
 DROP TABLE hugeint_numeric;
 DROP TABLE uuid_tbl;
 DROP TABLE json_tbl;
+DROP TABLE blob_tbl;
 DROP TABLE regclass_tbl;