Use dateutil.tz.tzlocal directly.

Author: ueshin

python/pyspark/sql/dataframe.py

@@ -1763,11 +1763,13 @@ def toPandas(self):
 
             if self.sql_ctx.getConf("spark.sql.execution.pandas.timeZoneAware", "false").lower() \
                     == "true":
+                from dateutil import tz
+                tzlocal = tz.tzlocal()
                 timezone = self.sql_ctx.getConf("spark.sql.session.timeZone")
                 for field in self.schema:
                     if type(field.dataType) == TimestampType:
                         pdf[field.name] = pdf[field.name].apply(
-                            lambda ts: ts.tz_localize('tzlocal()').tz_convert(timezone))
+                            lambda ts: ts.tz_localize(tzlocal).tz_convert(timezone))
 
             return pdf
 

python/pyspark/sql/tests.py

@@ -2510,6 +2510,8 @@ def test_to_pandas(self):
     @unittest.skipIf(not _have_pandas, "Pandas not installed")
     def test_to_pandas_timezone_aware(self):
         import pandas as pd
+        from dateutil import tz
+        tzlocal = tz.tzlocal()
         ts = datetime.datetime(1970, 1, 1)
         pdf = pd.DataFrame.from_records([[ts]], columns=['ts'])
 
@@ -2530,7 +2532,7 @@ def test_to_pandas_timezone_aware(self):
 
         pdf_pst_naive = pdf_pst.copy()
         pdf_pst_naive['ts'] = pdf_pst_naive['ts'].apply(
-            lambda ts: ts.tz_convert('tzlocal()').tz_localize(None))
+            lambda ts: ts.tz_convert(tzlocal).tz_localize(None))
         self.assertTrue(pdf_pst_naive.equals(pdf))
 
         self.spark.conf.unset('spark.sql.execution.pandas.timeZoneAware')