[HIVE-26685] Improve Path name escaping / unescaping performance

• Type: Improvement
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: All Versions
• Fix Version/s: 4.0.0-beta-1
• Component/s: Hive
• Labels: pull-request-available
• External issue URL: https://github.com/apache/hive/pull/3721
• Assignee: James Petty
• Reporter: James Petty
• Created: 01/Nov/22 21:49
• Updated: 15/Aug/23 06:33
• Resolved: 05/Dec/22 18:08

Description

When escaping / unescaping partition path part names, the existing logic incurs significant avoidable overhead by copying each character sequentially into a new StringBuilder even when no escaping/unescaping is necessary as well as using String.format to escape characters inside of the inner loop.

 

The included patch to improve the performance of these operations refactors two static method implementations, but requires no external API surface or user-visible behavior changes. This change is applicable and portable to a wide range of Hive versions from branch-0.6 onward when the initial method implementations were added.

Issue Links

• links to: GitHub Pull Request #3721