parquet: Add support for user-provided metadata loaders
- Dominant language
- Rust
- Stars
- 9.3k
- Forks
- 2.4k
- Avg merge
- 3d 7h
- Merged PRs (30d)
- 344
Description
### Is your feature request related to a problem or challenge?
This allows users to implement #12547 (caching metadata, especially the Page Index) themselves without reimplementing all of `ParquetOpener` and `ParquetExec`
### Describe the solution you'd like
Add a method to the object returned by `ParquetFileReaderFactory` that loads metadata, so that users who overwrite the `DefaultParquerFileReaderFactory` with `SessionState::register_file_format(..., true)` can provide their own implementation.
### Describe alternatives you've considered
Adding caching support directly in Datafusion. I tried in #12548 but it went nowhere.
Support for caching directly in Datafusion can be added later, based on this feature.
### Additional context
_No response_
Contributor guide
Research direction
Trace ParquetFileReaderFactory, DefaultParquerFileReaderFactory, ParquetOpener, and ParquetExec, then follow the SessionState::register_file_format(..., true) registration path. Done means a custom factory can provide metadata-loading behavior, including caching, without reimplementing the opener and execution components.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- rust
- Domain
- backend-api-design
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 45/100