Merge pull request #355 from edwinwzhe/avoid_unnecessary_tz_conversion

Avoid unnecessary timezone conversion (#354)

clickhouse_driver/columns/numpy/datetimecolumn.py

@@ -21,12 +21,12 @@ def __init__(self, timezone=None, offset_naive=True, local_timezone=None,
     def apply_timezones_after_read(self, dt):
         timezone = self.timezone if self.timezone else self.local_timezone
 
-        ts = pd.to_datetime(dt, utc=True).tz_convert(timezone)
-
-        if self.offset_naive:
+        if self.offset_naive and timezone.zone != 'UTC':
+            ts = pd.to_datetime(dt, utc=True).tz_convert(timezone)
             ts = ts.tz_localize(None)
+            return ts.to_numpy(self.datetime_dtype)
 
-        return ts.to_numpy(self.datetime_dtype)
+        return dt
 
     def apply_timezones_before_write(self, items):
         if isinstance(items, pd.DatetimeIndex):
@@ -120,24 +120,24 @@ def create_numpy_datetime_column(spec, column_options):
 
     context = column_options['context']
 
-    tz_name = timezone = None
+    tz_name = None
     offset_naive = True
 
     # As Numpy do not use local timezone for converting timestamp to
     # datetime we need always detect local timezone for manual converting.
-    local_timezone = get_localzone_name_compat()
+    local_tz_name = get_localzone_name_compat()
 
     # Use column's timezone if it's specified.
     if spec and spec[-1] == ')':
         tz_name = spec[1:-2]
         offset_naive = False
     else:
         if not context.settings.get('use_client_time_zone', False):
-            if local_timezone != context.server_info.timezone:
+            if local_tz_name != context.server_info.timezone:
                 tz_name = context.server_info.timezone
 
-    if tz_name:
-        timezone = get_timezone(tz_name)
+    timezone = get_timezone(tz_name) if tz_name else None
+    local_timezone = get_timezone(local_tz_name) if local_tz_name else None
 
     return cls(timezone=timezone, offset_naive=offset_naive,
                local_timezone=local_timezone, **column_options)

tests/numpy/columns/test_datetime.py

@@ -570,6 +570,78 @@ def test_datetime_with_timezone_column_use_client_timezone(self):
                     inserted[0], self.make_tz_numpy_array(dt, self.col_tz_name)
                 )
 
+    @require_server_version(1, 1, 54337)
+    def test_read_tz_aware_column(self):
+        # read data from tz aware column Asia/Novosibirsk
+        # offset_naive is False -> tz convert not needed
+
+        with self.create_table(self.table_columns(with_tz=True)):
+            with patch.object(
+                pd, 'to_datetime', wraps=pd.to_datetime
+            ) as to_datetime_spy:
+                self.client.execute(
+                    'INSERT INTO test (a) VALUES', [self.dt_arr], columnar=True
+                )
+
+                self.emit_cli(
+                    "INSERT INTO test (a) VALUES ('2017-07-14 05:40:00')",
+                )
+
+                to_datetime_calls_before_read = to_datetime_spy.call_count
+
+                query = 'SELECT * FROM test'
+                inserted = self.client.execute(query, columnar=True)
+
+                self.assertEqual(
+                    to_datetime_calls_before_read,
+                    to_datetime_spy.call_count
+                )
+
+                self.assertArraysEqual(
+                    inserted[0],
+                    self.make_tz_numpy_array(self.dt, self.col_tz_name)
+                )
+
+    @require_server_version(1, 1, 54337)
+    def test_read_tz_naive_column_with_client_timezone(self):
+        # read data from column without timezone
+        # client timezone = Asia/Novosibirsk
+        # offset_naive is True and timezone is not UTC -> tz convert needed
+
+        settings = {'use_client_time_zone': True}
+
+        with patch_env_tz('Asia/Novosibirsk'):
+            with self.create_table(self.table_columns()):
+                with patch.object(
+                    pd, 'to_datetime', wraps=pd.to_datetime
+                ) as to_datetime_spy:
+                    self.client.execute(
+                        'INSERT INTO test (a) VALUES', [self.dt_arr],
+                        settings=settings, columnar=True
+                    )
+
+                    self.emit_cli(
+                       "INSERT INTO test (a) VALUES ('2017-07-14 05:40:00')",
+                       use_client_time_zone=1
+                    )
+
+                    to_datetime_calls_before_read = to_datetime_spy.call_count
+
+                    query = 'SELECT * FROM test'
+                    inserted = self.client.execute(
+                        query, settings=settings, columnar=True
+                    )
+
+                    self.assertEqual(
+                        to_datetime_calls_before_read + 2,
+                        to_datetime_spy.call_count
+                    )
+
+                    self.assertArraysEqual(
+                        inserted[0],
+                        self.make_numpy_d64ns([self.dt_str] * 2)
+                    )
+
 
 class DateTime64TimezonesTestCase(DateTimeTimezonesTestCase):
     dt_type = 'DateTime64'