fixed hex stuff

Author: fivetran-kwoodbeck

sqlglot/dialects/duckdb.py

@@ -563,7 +563,7 @@ def _cast_to_boolean(arg: t.Optional[exp.Expression]) -> t.Optional[exp.Expressi
 
 
 def _is_binary(arg: exp.Expression) -> bool:
-    return arg.is_type(
+    return isinstance(arg, exp.HexString) or arg.is_type(
         exp.DataType.Type.BINARY,
         exp.DataType.Type.VARBINARY,
         exp.DataType.Type.BLOB,
@@ -771,7 +771,7 @@ def _bitshift_sql(
         this = annotate_types(this, dialect=self.dialect)
 
     # Deal with binary separately, remember the original type, cast back later
-    if _is_binary(this) or isinstance(this, exp.HexString):
+    if _is_binary(this):
         original_type = this.to if isinstance(this, exp.Cast) else exp.DataType.build("BLOB")
         expression.set("this", exp.cast(this, exp.DataType.Type.BIT))
     elif expression.args.get("requires_int128"):
@@ -2273,16 +2273,8 @@ def format_sql(self, expression: exp.Format) -> str:
         def hexstring_sql(
             self, expression: exp.HexString, binary_function_repr: t.Optional[str] = None
         ) -> str:
-            from_hex = super().hexstring_sql(expression, binary_function_repr="FROM_HEX")
-
-            if expression.args.get("is_integer"):
-                return from_hex
-
-            # `from_hex` has transpiled x'ABCD' (BINARY) to DuckDB's '\xAB\xCD' (BINARY)
-            # `to_hex` & CASTing transforms it to "ABCD" (BINARY) to match representation
-            to_hex = exp.cast(self.func("TO_HEX", from_hex), exp.DataType.Type.BLOB)
-
-            return self.sql(to_hex)
+            # UNHEX('FF') correctly produces blob \xFF in DuckDB
+            return super().hexstring_sql(expression, binary_function_repr="UNHEX")
 
         def timestamptrunc_sql(self, expression: exp.TimestampTrunc) -> str:
             unit = unit_to_str(expression)

sqlglot/typing/__init__.py

@@ -42,6 +42,8 @@
         for expr_type in {
             exp.FromBase32,
             exp.FromBase64,
+            exp.HexString,
+            exp.Unhex,
         }
     },
     **{

tests/dialects/test_mysql.py

@@ -482,7 +482,7 @@ def test_hexadecimal_literal(self):
             "clickhouse": UnsupportedError,
             "databricks": "SELECT X'CC'",
             "drill": "SELECT 204",
-            "duckdb": "SELECT CAST(HEX(FROM_HEX('CC')) AS VARBINARY)",
+            "duckdb": "SELECT UNHEX('CC')",
             "hive": "SELECT 204",
             "mysql": "SELECT x'CC'",
             "oracle": "SELECT 204",
@@ -503,7 +503,7 @@ def test_hexadecimal_literal(self):
             "clickhouse": UnsupportedError,
             "databricks": "SELECT X'0000CC'",
             "drill": "SELECT 204",
-            "duckdb": "SELECT CAST(HEX(FROM_HEX('0000CC')) AS VARBINARY)",
+            "duckdb": "SELECT UNHEX('0000CC')",
             "hive": "SELECT 204",
             "mysql": "SELECT x'0000CC'",
             "oracle": "SELECT 204",

tests/dialects/test_snowflake.py

@@ -1743,14 +1743,14 @@ def test_snowflake(self):
             "SELECT BITSHIFTLEFT(X'002A'::BINARY, 1)",
             write={
                 "snowflake": "SELECT BITSHIFTLEFT(CAST(x'002A' AS BINARY), 1)",
-                "duckdb": "SELECT CAST(CAST(UNHEX('002A') AS BIT) << 1 AS BLOB)",
+                "duckdb": "SELECT CAST(CAST(CAST(UNHEX('002A') AS BLOB) AS BIT) << 1 AS BLOB)",
             },
         )
         self.validate_all(
             "SELECT BITSHIFTRIGHT(X'002A'::BINARY, 1)",
             write={
                 "snowflake": "SELECT BITSHIFTRIGHT(CAST(x'002A' AS BINARY), 1)",
-                "duckdb": "SELECT CAST(CAST(UNHEX('002A') AS BIT) >> 1 AS BLOB)",
+                "duckdb": "SELECT CAST(CAST(CAST(UNHEX('002A') AS BLOB) AS BIT) >> 1 AS BLOB)",
             },
         )
 
@@ -2171,7 +2171,7 @@ def test_snowflake(self):
             "SELECT x'ABCD'",
             write={
                 "snowflake": "SELECT x'ABCD'",
-                "duckdb": "SELECT CAST(HEX(FROM_HEX('ABCD')) AS VARBINARY)",
+                "duckdb": "SELECT UNHEX('ABCD')",
             },
         )