[SPARK-6902] [SQL] [PYSPARK] Row should be read-only

Raise an read-only exception when user try to mutable a Row.

Author: Davies Liu <davies@databricks.com>

Closes #8009 from davies/readonly_row and squashes the following commits:

8722f3f [Davies Liu] add tests
05a3d36 [Davies Liu] Row should be read-only

python/pyspark/sql/tests.py

@@ -179,6 +179,21 @@ class SQLTests(ReusedPySparkTestCase):
         ReusedPySparkTestCase.tearDownClass()
         shutil.rmtree(cls.tempdir.name, ignore_errors=True)
 
+    def test_row_should_be_read_only(self):
+        row = Row(a=1, b=2)
+        self.assertEqual(1, row.a)
+
+        def foo():
+            row.a = 3
+        self.assertRaises(Exception, foo)
+
+        row2 = self.sqlCtx.range(10).first()
+        self.assertEqual(0, row2.id)
+
+        def foo2():
+            row2.id = 2
+        self.assertRaises(Exception, foo2)
+
     def test_range(self):
         self.assertEqual(self.sqlCtx.range(1, 1).count(), 0)
         self.assertEqual(self.sqlCtx.range(1, 0, -1).count(), 1)

python/pyspark/sql/types.py

@@ -1246,6 +1246,11 @@ class Row(tuple):
         except ValueError:
             raise AttributeError(item)
 
+    def __setattr__(self, key, value):
+        if key != '__fields__':
+            raise Exception("Row is read-only")
+        self.__dict__[key] = value
+
     def __reduce__(self):
         """Returns a tuple so Python knows how to pickle Row."""
         if hasattr(self, "__fields__"):