Bob Briody created ARROW-5562:
---------------------------------
Summary: pyarrow parquet writer does not handle negative zero
correctly
Key: ARROW-5562
URL: https://issues.apache.org/jira/browse/ARROW-5562
Project: Apache Arrow
Issue Type: Bug
Components: Python
Affects Versions: 0.13.0
Reporter: Bob Briody
I have the following csv file (note that col_a contains a negative zero value):
{code:java}
col_a,col_b
0.0,0.0
-0.0,0.0{code}
...and process it via:
{code:java}
from pyarrow import csv, parquet
in_csv = 'in.csv'
table = csv.read_csv(in_csv)
parquet.write_to_dataset(table, root_path='./'){code}
The output parquet file is then loaded into S3 and queried via AWS Athena (i.e.
PrestoDB / Hive).
Any query that touches {{col_a}} fails with the following error:
{code:java}
HIVE_CANNOT_OPEN_SPLIT: Error opening Hive split {{REDACTED}} (offset=0,
length=593): low must be less than or equal to high{code}
As a sanity check, I transformed the csv file to parquet using an AWS Glue
Spark Job and I was able to query the output parquet file successfully.
As such, it appears as though the pyarrow writer is producing an invalid
parquet file when a column contains both 0.0 and -0.0, and only 0.0 and -0.0.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
