Add INSERT support for OBJECT

src/snowflake/sqlalchemy/base.py

@@ -12,6 +12,8 @@
 from sqlalchemy.sql.elements import quoted_name
 from sqlalchemy.util.compat import string_types
 
+from snowflake.sqlalchemy.custom_types import OBJECT
+
 from .custom_commands import AWSBucket, AzureContainer, ExternalStage
 
 RESERVED_WORDS = frozenset(
@@ -150,6 +152,36 @@ def _split_schema_by_dot(self, schema):
 
 
 class SnowflakeCompiler(compiler.SQLCompiler):
+    def visit_insert(self, stmt, **kw):
+        # https://github.com/sqlalchemy/sqlalchemy/discussions/7894#discussioncomment-2520337
+        insert_sql = super().visit_insert(stmt, **kw)
+
+        columns = self.column_keys
+        if columns is None:
+            columns = stmt.table.columns.keys()
+
+        # look in the columns being inserted, see if there's
+        # JSON being inserted.  also can just look at the INSERT string
+        # and look for the json function
+
+        use_json = any(isinstance(stmt.table.c[key].type, OBJECT) for key in columns)
+
+        if not use_json:
+            return insert_sql
+
+        stmt_reg = re.match(
+            r"^INSERT INTO (.+?) \((.+?)\) VALUES \((.+)\)$", insert_sql
+        )
+        if not stmt_reg:
+            return insert_sql
+
+        # rewrite INSERT as per
+        # https://docs.snowflake.com/en/sql-reference/sql/insert.html#usage-notes
+        return (
+            f"INSERT INTO {stmt_reg.group(1)} "
+            f"({stmt_reg.group(2)}) SELECT {stmt_reg.group(3)}"
+        )
+
     def visit_sequence(self, sequence, **kw):
         return self.dialect.identifier_preparer.format_sequence(sequence) + ".nextval"