Products endpoint returns multiple versions of the same LID in a single query

[jordanpadams]

Checked for duplicates

Yes - I've already checked

🐛 Describe the bug

The /products endpoint returns multiple versions of the same LID in a single query result, even when no version filter is applied. This is unexpected — callers querying for context products generally expect to receive the latest version of each LID, not all historical versions.

📜 To Reproduce

The following query returns two versions of urn:nasa:pds:context:facility:laboratory.relab (::1.0 and ::1.1):

https://pds.mcp.nasa.gov/api/search/1/products?q=(lid%20eq%20%22urn:nasa:pds:context:facility:laboratory.relab%22)

A broader query for all Product_Context products reveals 9 LIDs with multiple versions returned:

LID	Versions returned
urn:nasa:pds:context:facility:laboratory.relab	1.0, 1.1
urn:nasa:pds:context:instrument:relab.bd-vnir	1.0, 1.2
urn:nasa:pds:context:instrument:relab.bc-ftir1	1.0, 1.2
urn:nasa:pds:context:instrument:relab.bcf-ftir2	1.0, 1.2
urn:nasa:pds:context:investigation:other_investigation.relab_speclib	1.0, 2.1
urn:esa:psa:context:instrument_host:spacecraft.vex	1.1, 1.2
urn:esa:psa:context:investigation:mission.venus_express	1.1, 1.1 (same LIDVID from two nodes)
urn:esa:psa:context:instrument:vex.mag	1.2, 1.2 (same LIDVID from two nodes)
urn:esa:psa:context:instrument:vex.virtis	1.2, 1.2 (same LIDVID from two nodes)

The last three (vex.mag, vex.virtis, mission.venus_express) are particularly concerning — the exact same LIDVID appears twice, harvested from two different registry nodes (PDS_ENG and PSA).

This was discovered while investigating a downstream bug in the validate tool (NASA-PDS/validate#1632), where validate -u downloads all context products and the presence of older versions alongside newer ones causes false context_ref_mismatch warnings for users with valid labels.

🕵️ Expected behavior

A query for Product_Context products should return at most one entry per LID — either the latest version only, or deduplicate exact-LIDVID duplicates arising from multi-node harvesting.

🖥 Environment Info

Production API: https://pds.mcp.nasa.gov/api/search/1/

📚 Version of Software Used

PDS Search API (production)

🩺 Additional context

• The ops:Provenance.ops:superseded_by field on all returned products is set to the string "null" — even on genuinely superseded older versions — so there is currently no way for API clients to filter out old versions server-side.
• A latest-only endpoint (/products/latest) returns HTTP 404.
• Adding version/latest filters to the q= parameter returns HTTP 400.

🤖 Generated with Claude Code

[jordanpadams]

This is most likely blocked or potentially fixed by upgraded registry deployment.