Files
madminandClaude Sonnet 5 d8f59b7725 Fix external_shipment_id placement and test-mode store_id ticket persistence in the return-label flow
external_shipment_id was at the top level of the POST /v2/labels payload in both
_create_dummy_outbound_label() and create_return_label_from_dummy() - ShipStation's
schema only supports it nested under shipment, so it was silently ignored and
auto-generated ("SEAuto-...") on every label this flow ever created. Moved it inside
the shipment dict in both functions. Confirmed end-to-end on a real production
ticket (AR-166098): Order # now correctly shows the real ticket number, and
ship_from/ship_to are correct for a return.

Also: store_id is now only required in production - ShipStation's sandbox
environment cannot have stores/Order Sources at all (confirmed in the real sandbox
dashboard), so test mode omits it from the request instead of requiring an
impossible value. And three per-ticket write-backs (mark_shipstation_sent,
save_dummy_outbound_label_id, save_pack_data) were hardcoded to source == "jira",
silently no-opping for source == "test" tickets created by Create Test Shipment -
now match on ticket_number alone, since source is only ever "jira" or "test" and
ticket_number is already unique across both.

CLAUDE.md records the full investigation and closes out the long-standing
"return-label flow real-world verification" known-pending item.

Co-Authored-By: Claude Sonnet 5 <[email protected]>
2026-10-01 15:53:54 -05:00
..
2026-08-26 09:15:35 -05:00
2026-08-26 09:15:35 -05:00