[SEDONA-673] Cannot load GeoParquet without bbox metadata when spatial filter is applied

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.7.0
• Labels: pull-request-available
• Assignee: Unassigned
• Reporter: Kristin Cowalcijk
• Created: 14/Nov/24 14:49
• Updated: 24/Nov/24 07:58
• Resolved: 24/Nov/24 07:58

Description

When reading geoparquet which does not have bbox in metadata query like below fails with Index OutofBoundException

sparkSession.read
        .format("geoparquet")
        .load(overtureBBOX)
        .where("ST_Intersects(geometry, ST_PolygonFromEnvelope(0, 0, 1, 1))")
        .count()
The issue was when evaluating predicate pushdowns

How to reproduce

Download data with the command

uv run overturemaps download --bbox 0,0,1,1 -f geoparquet --type=place -o bbox.geoparquet

then try to load it with some predicate like contains or within.

Issue Links

• links to: GitHub Pull Request #1681