BryanCutler commented on a change in pull request #24070: [SPARK-23961][PYTHON]
Fix error when toLocalIterator goes out of scope
URL: https://github.com/apache/spark/pull/24070#discussion_r264833969
##########
File path: python/pyspark/sql/tests/test_dataframe.py
##########
@@ -676,6 +676,34 @@ def test_repr_behaviors(self):
self.assertEquals(None, df._repr_html_())
self.assertEquals(expected, df.__repr__())
+ def test_to_local_iterator(self):
+ df = self.spark.range(8, numPartitions=4)
+ expected = df.collect()
+ it = df.toLocalIterator()
+ self.assertEqual(expected, list(it))
+
+ # Test DataFrame with empty partition
+ df = self.spark.range(3, numPartitions=4)
+ it = df.toLocalIterator()
+ expected = df.collect()
+ self.assertEqual(expected, list(it))
+
+ def test_to_local_iterator_not_fully_consumed(self):
+ # SPARK-23961: toLocalIterator throws exception when not fully consumed
+ # Create a DataFrame large enough so that write to socket will
eventually block
+ df = self.spark.range(1 << 20, numPartitions=2)
+ it = df.toLocalIterator()
+ self.assertEqual(df.take(1)[0], next(it))
+ with QuietTest(self.sc):
+ it = None # remove iterator from scope, socket is closed when
cleaned up
+ # Make sure normal df operations still work
+ result = []
+ for i, row in enumerate(df.toLocalIterator()):
+ result.append(row)
+ if i == 7:
+ break
+ self.assertEqual(df.take(8), result)
Review comment:
Note: this would not have crashed before, only generated the error and I
don't think it's possible to check if this error happened
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
