optimize

Author: xiaohongbo

paimon-python/pypaimon/manifest/schema/data_file_meta.py

@@ -17,7 +17,7 @@
 ################################################################################
 
 from dataclasses import dataclass
-from datetime import datetime
+from datetime import datetime, timezone
 from typing import List, Optional
 import time
 
@@ -61,12 +61,12 @@ def get_creation_time(self) -> Optional[Timestamp]:
     def creation_time_epoch_millis(self) -> Optional[int]:
         if self.creation_time is None:
             return None
-        millis = self.creation_time.get_millisecond()
-        if time.daylight:
-            tz_offset_seconds = -time.altzone
-        else:
-            tz_offset_seconds = -time.timezone
-        return millis - (tz_offset_seconds * 1000)
+        local_dt = self.creation_time.to_local_date_time()
+        local_time_struct = local_dt.timetuple()
+        local_timestamp = time.mktime(local_time_struct)
+        local_time_struct_utc = time.gmtime(local_timestamp)
+        utc_timestamp = time.mktime(local_time_struct_utc)
+        return int(utc_timestamp * 1000)
 
     def creation_time_as_datetime(self) -> Optional[datetime]:
         if self.creation_time is None:

paimon-python/pypaimon/tests/file_store_commit_test.py

@@ -104,12 +104,7 @@ def test_generate_partition_statistics_single_partition_single_file(
         self.assertEqual(stat.record_count, 10000)
         self.assertEqual(stat.file_count, 1)
         self.assertEqual(stat.file_size_in_bytes, 1024 * 1024)
-        expected_time = creation_time.get_millisecond()
-        if time.daylight:
-            tz_offset_seconds = -time.altzone
-        else:
-            tz_offset_seconds = -time.timezone
-        expected_time = expected_time - (tz_offset_seconds * 1000)
+        expected_time = file_meta.creation_time_epoch_millis()
         self.assertEqual(stat.last_file_creation_time, expected_time)
 
     def test_generate_partition_statistics_multiple_files_same_partition(
@@ -173,12 +168,7 @@ def test_generate_partition_statistics_multiple_files_same_partition(
         self.assertEqual(stat.record_count, 25000)  # 10000 + 15000
         self.assertEqual(stat.file_count, 2)
         self.assertEqual(stat.file_size_in_bytes, 3 * 1024 * 1024)  # 1MB + 2MB
-        expected_time = creation_time_2.get_millisecond()
-        if time.daylight:
-            tz_offset_seconds = -time.altzone
-        else:
-            tz_offset_seconds = -time.timezone
-        expected_time = expected_time - (tz_offset_seconds * 1000)
+        expected_time = file_meta_2.creation_time_epoch_millis()
         self.assertEqual(stat.last_file_creation_time, expected_time)
 
     def test_generate_partition_statistics_multiple_partitions(

paimon-python/pypaimon/tests/reader_primary_key_test.py

@@ -289,11 +289,12 @@ def test_manifest_creation_time_timestamp(self):
                     self.assertGreater(epoch_millis, 0)
                     import time
                     expected_epoch_millis = creation_time.get_millisecond()
-                    if time.daylight:
-                        tz_offset_seconds = -time.altzone
-                    else:
-                        tz_offset_seconds = -time.timezone
-                    expected_epoch_millis = expected_epoch_millis - (tz_offset_seconds * 1000)
+                    local_dt = creation_time.to_local_date_time()
+                    local_time_struct = local_dt.timetuple()
+                    local_timestamp = time.mktime(local_time_struct)
+                    local_time_struct_utc = time.gmtime(local_timestamp)
+                    utc_timestamp = time.mktime(local_time_struct_utc)
+                    expected_epoch_millis = int(utc_timestamp * 1000)
                     self.assertEqual(epoch_millis, expected_epoch_millis)
                     creation_times_found.append(epoch_millis)