[ARROW-17320] [Python] Refine pyarrow.parquet API exposure

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 10.0.0
• Component/s: Parquet, Python
• Labels: pull-request-available
• External issue URL: https://github.com/apache/arrow/issues/32597

Language

Python

• Assignee: Miles Granger
• Reporter: Miles Granger
• Created: 05/Aug/22 12:28
• Updated: 11/Jan/23 11:50
• Resolved: 27/Sep/22 09:12

Description

Spawning from ARROW-17106, moving code from `pyarrow/parquet/init` to `pyarrow/parquet/core` and re-exporting in `_init{_}` to maintain the same functionality.

pyarrow.initis very careful about what is exposed through the public API by prefixing private symbols with underscores, even imports. 

What's exposed at the top level of `pyarrow.parquet`, however, is not so careful. API calls such as `pq.FileSystem`, `pq.pa.Array`, `pq.json` are all valid and should probably be designated as private attributes in pyarrow.parquet.

Issue Links

• relates to: ARROW-17106 [Python] Move parquet code from __init__.py and expose only API
• relates to: ARROW-9672 [Python][Parquet] Expose _filters_to_expression
• links to: GitHub Pull Request #14096