Skip to content

Conversation

@suremarc
Copy link
Collaborator

@suremarc suremarc commented Dec 24, 2024

This PR is mostly a port of internal code from Polygon.io, with some new additions to make it generic so that users with custom TableProvider implementations can use it effectively.

FileMetadata is a custom TableProvider that exposes object store metadata through SQL. It implements RowMetadataSource, a trait that provides metadata which will be used in incremental view maintenance (which is not included in this PR).

In order for these components to accommodate custom table types, we have added a ListingTableLike trait that abstracts away the essential features of a Hive-partitioned table. Users must register their custom table types using the register_listing_table API if they expect this crate to detect and work properly with these tables.

This PR ended up quite a bit bigger than I expected, unfortunately. Most of the length comes from the FileMetadata table provider, which implements filter pushdown.

@suremarc suremarc marked this pull request as ready for review December 24, 2024 22:36
@suremarc suremarc merged commit b7974e8 into datafusion-contrib:main Dec 26, 2024
8 checks passed
This was referenced Dec 31, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant