[python/hotfix] Fix IndexError when reading manifest with empty _MIN_VALUES

paimon-python/pypaimon/table/row/binary_row.py

@@ -31,11 +31,19 @@ def __init__(self, data: bytes, fields: List[DataField]):
         Initialize BinaryRow with raw binary data and field definitions.
         """
         self.data = data
-        self.arity = int.from_bytes(data[:4], 'big')
-        # Skip the arity prefix (4 bytes) if present
-        self.actual_data = data[4:] if len(data) >= 4 else data
+        if len(data) < 4:
+            self.arity = 0
+            self.actual_data = b''
+        else:
+            self.arity = int.from_bytes(data[:4], 'big')
+            self.actual_data = data[4:] if len(data) >= 4 else data
+
         self.fields = fields
-        self.row_kind = RowKind(self.actual_data[0])
+
+        if len(self.actual_data) == 0:
+            self.row_kind = RowKind.INSERT
+        else:
+            self.row_kind = RowKind(self.actual_data[0])
 
     def get_field(self, index: int) -> Any:
         from pypaimon.table.row.generic_row import GenericRowDeserializer

paimon-python/pypaimon/tests/binary_row_test.py

@@ -27,7 +27,9 @@
 from pypaimon.manifest.schema.manifest_entry import ManifestEntry
 from pypaimon.manifest.schema.simple_stats import SimpleStats
 from pypaimon.read.scanner.full_starting_scanner import FullStartingScanner
+from pypaimon.table.row.binary_row import BinaryRow
 from pypaimon.table.row.generic_row import GenericRow, GenericRowDeserializer
+from pypaimon.table.row.row_kind import RowKind
 
 
 def _random_format():
@@ -332,3 +334,22 @@ def _overwrite_manifest_entry(self, table):
                 [0],
             )
         starting_scanner.manifest_file_manager.write(manifest_files[0].file_name, manifest_entries)
+
+    def test_binary_row_with_empty_actual_data(self):
+        empty_data = b''
+        binary_row = BinaryRow(empty_data, [])
+        self.assertEqual(binary_row.arity, 0)
+        self.assertEqual(binary_row.actual_data, b'')
+        self.assertEqual(binary_row.get_row_kind(), RowKind.INSERT)
+
+        arity_zero_data = (0).to_bytes(4, 'big')  # b'\x00\x00\x00\x00'
+        binary_row = BinaryRow(arity_zero_data, [])
+        self.assertEqual(binary_row.arity, 0)
+        self.assertEqual(binary_row.actual_data, b'')
+        self.assertEqual(binary_row.get_row_kind(), RowKind.INSERT)
+
+        normal_empty_row = (0).to_bytes(4, 'big') + b'\x00' * 8  # 12 bytes total
+        binary_row = BinaryRow(normal_empty_row, [])
+        self.assertEqual(binary_row.arity, 0)
+        self.assertEqual(len(binary_row.actual_data), 8)
+        self.assertEqual(binary_row.get_row_kind(), RowKind.INSERT)

paimon-python/pypaimon/tests/reader_base_test.py

@@ -249,6 +249,23 @@ def test_full_data_types(self):
                                                                 []).values
             self.assertEqual(min_value_stats, [])
             self.assertEqual(max_value_stats, [])
+
+            self.assertGreater(
+                len(manifest_entries[0].file.value_stats.min_values.data), 0,
+                "MIN_VALUES.data should have bytes written, not empty (0 bytes)")
+            self.assertGreaterEqual(
+                len(manifest_entries[0].file.value_stats.min_values.data), 12,
+                f"MIN_VALUES.data should be at least 12 bytes "
+                f"(4 bytes arity + 8 bytes fixed part for empty GenericRow), "
+                f"but got {len(manifest_entries[0].file.value_stats.min_values.data)} bytes")
+            self.assertGreater(
+                len(manifest_entries[0].file.value_stats.max_values.data), 0,
+                "MAX_VALUES.data should have bytes written, not empty (0 bytes)")
+            self.assertGreaterEqual(
+                len(manifest_entries[0].file.value_stats.max_values.data), 12,
+                f"MAX_VALUES.data should be at least 12 bytes "
+                f"(4 bytes arity + 8 bytes fixed part for empty GenericRow), "
+                f"but got {len(manifest_entries[0].file.value_stats.max_values.data)} bytes")
 
     def test_write_wrong_schema(self):
         self.catalog.create_table('default.test_wrong_schema',
@@ -623,6 +640,26 @@ def test_value_stats_empty_when_stats_disabled(self):
             len(file_meta.value_stats.null_counts), len(empty_stats.null_counts),
             "value_stats.null_counts should be empty (same as SimpleStats.empty_stats()) when stats are disabled"
         )
+
+        # 验证Python确实写入了字节数，而不是空的MIN_VALUES
+        # 即使min_values是empty (GenericRow([], []))，Python也应该序列化为12字节
+        # (4字节arity + 8字节fixed part)，而不是0字节或4字节
+        self.assertGreater(
+            len(file_meta.value_stats.min_values.data), 0,
+            "MIN_VALUES.data should have bytes written, not empty (0 bytes)")
+        self.assertGreaterEqual(
+            len(file_meta.value_stats.min_values.data), 12,
+            f"MIN_VALUES.data should be at least 12 bytes "
+            f"(4 bytes arity + 8 bytes fixed part for empty GenericRow), "
+            f"but got {len(file_meta.value_stats.min_values.data)} bytes")
+        self.assertGreater(
+            len(file_meta.value_stats.max_values.data), 0,
+            "MAX_VALUES.data should have bytes written, not empty (0 bytes)")
+        self.assertGreaterEqual(
+            len(file_meta.value_stats.max_values.data), 12,
+            f"MAX_VALUES.data should be at least 12 bytes "
+            f"(4 bytes arity + 8 bytes fixed part for empty GenericRow), "
+            f"but got {len(file_meta.value_stats.max_values.data)} bytes")
 
     def test_types(self):
         data_fields = [