fix

Author: JingsongLi

paimon-python/pypaimon/tests/py36/ao_predicate_test.py

@@ -40,7 +40,7 @@ def setUpClass(cls):
             ('f1', pa.string()),
         ])
         cls.catalog.create_table('default.test_append', Schema.from_pyarrow_schema(
-            pa_schema, options={'file.format': _random_format()}), False)
+            pa_schema, options={'file.format': _random_format(), 'metadata.stats-mode': 'full'}), False)
 
         df = pd.DataFrame({
             'f0': [1, 2, 3, 4, 5],

paimon-python/pypaimon/tests/py36/rest_ao_read_write_test.py

@@ -17,6 +17,7 @@
 """
 import logging
 import time
+import random
 from datetime import date
 from decimal import Decimal
 from unittest.mock import Mock
@@ -143,7 +144,9 @@ def test_full_data_types(self):
             ('f10', pa.decimal128(10, 2)),
             ('f11', pa.date32()),
         ])
-        schema = Schema.from_pyarrow_schema(simple_pa_schema)
+        stats_enabled = random.random() < 0.5
+        options = {'metadata.stats-mode': 'full'} if stats_enabled else {}
+        schema = Schema.from_pyarrow_schema(simple_pa_schema, options=options)
         self.rest_catalog.create_table('default.test_full_data_types', schema, False)
         table = self.rest_catalog.get_table('default.test_full_data_types')
 
@@ -183,14 +186,25 @@ def test_full_data_types(self):
             manifest_files[0].file_name,
             lambda row: table_scan.starting_scanner._filter_manifest_entry(row),
             drop_stats=False)
-        min_value_stats = GenericRowDeserializer.from_bytes(manifest_entries[0].file.value_stats.min_values.data,
-                                                            table.fields).values
-        max_value_stats = GenericRowDeserializer.from_bytes(manifest_entries[0].file.value_stats.max_values.data,
-                                                            table.fields).values
-        expected_min_values = [col[0].as_py() for col in expect_data]
-        expected_max_values = [col[1].as_py() for col in expect_data]
-        self.assertEqual(min_value_stats, expected_min_values)
-        self.assertEqual(max_value_stats, expected_max_values)
+        # Python write does not produce value stats
+        if stats_enabled:
+            self.assertEqual(manifest_entries[0].file.value_stats_cols, None)
+            min_value_stats = GenericRowDeserializer.from_bytes(manifest_entries[0].file.value_stats.min_values.data,
+                                                                table.fields).values
+            max_value_stats = GenericRowDeserializer.from_bytes(manifest_entries[0].file.value_stats.max_values.data,
+                                                                table.fields).values
+            expected_min_values = [col[0].as_py() for col in expect_data]
+            expected_max_values = [col[1].as_py() for col in expect_data]
+            self.assertEqual(min_value_stats, expected_min_values)
+            self.assertEqual(max_value_stats, expected_max_values)
+        else:
+            self.assertEqual(manifest_entries[0].file.value_stats_cols, [])
+            min_value_stats = GenericRowDeserializer.from_bytes(manifest_entries[0].file.value_stats.min_values.data,
+                                                                []).values
+            max_value_stats = GenericRowDeserializer.from_bytes(manifest_entries[0].file.value_stats.max_values.data,
+                                                                []).values
+            self.assertEqual(min_value_stats, [])
+            self.assertEqual(max_value_stats, [])
 
     def test_mixed_add_and_delete_entries_same_partition(self):
         """Test record_count calculation with mixed ADD/DELETE entries in same partition."""