You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jira@arrow.apache.org by "Max Burke (Jira)" <ji...@apache.org> on 2021/02/01 05:48:00 UTC
[jira] [Created] (ARROW-11452) [Rust] Parquet reader cannot read
file where a struct column has the same name as struct member columns
Max Burke created ARROW-11452:
---------------------------------
Summary: [Rust] Parquet reader cannot read file where a struct column has the same name as struct member columns
Key: ARROW-11452
URL: https://issues.apache.org/jira/browse/ARROW-11452
Project: Apache Arrow
Issue Type: Bug
Reporter: Max Burke
Attachments: structs.parquet
For example, given the schema:
count: struct<min: int64 not null, max: int64 not null, mean: int64 not null, count: uint64 not null, sum: int64 not null, variance: int64 not null> not null
child 0, min: int64 not null
child 1, max: int64 not null
child 2, mean: int64 not null
child 3, count: uint64 not null
child 4, sum: int64 not null
child 5, variance: int64 not null
ul_observation_date: struct<min: timestamp[us] not null, max: timestamp[us] not null, mean: timestamp[us] not null, count: uint64 not null, sum: timestamp[us] not null, variance: timestamp[us] not null> not null
child 0, min: timestamp[us] not null
child 1, max: timestamp[us] not null
child 2, mean: timestamp[us] not null
child 3, count: uint64 not null
child 4, sum: timestamp[us] not null
child 5, variance: timestamp[us] not null
The array reader performs dictionary lookups for the type of columns of types such as ul_observation_date, but when it looks up the field `count` it gets the definition not of the ul_observation_date.count field but of the `count` struct.
Attached is a sample file that exhibits this behavior.
[^structs.parquet]
--
This message was sent by Atlassian Jira
(v8.3.4#803005)