chore: sync main — CM duplicate-ISO guard (#290) - #294
Merged
Merged
Conversation
…de (#290) The bundled metadata carries dead/live country pairs sharing an ISO code (Sudan 59/245, Serbia 230/233, … — 11 groups, 9 of which join the shapefile). If a forecast frame carries both members, the isoab→ADM0_A3 merge yields two rows per polygon and `pivot_table(aggfunc="first")` plus the hover-props `drop_duplicates` silently pick one: reproduced — with [59, 245] the map drew the DEAD Sudan's value and the hover reported 59. Ascending country_id makes the dead id win in 4 of the 9 joinable pairs; the pick is arbitrary in all nine. Guard at the single CM chokepoint (build_mapping_dataframe, after the geometry check, before the frame is stored): >1 distinct country_id per ADM0_A3 → logger.error + ValueError naming each collided code and its ids (register C-227; ADR-008). Raise, not warn: a forecast frame's entity set is fixed at its origin, so it never legitimately carries both. Per CODE, not per (code, month): real transitions put both ids in the same month anyway (Sudan 59+245 at 379), and the hover props key on code alone. Detection only — resolution belongs to the producer (pipeline-core ADR-064). PGM raster/image paths untouched (unreachable at CM). RED→GREEN in tests/test_mapping_characterization.py::TestDuplicateIsoIsLoud against the REAL bundle (both 59 and 245 resolve to SDN there); control: 245 alone renders one row. The guard's first catch was our own test double: the 191-country fixture e2e tests labelled the map via `mock_isoab_for_index`, which cycles 10 ISO codes positionally — ~19 country_ids per code, a collision real data never has. Those tests now use the real bundled metadata (offline since C-22; zero collisions on the fixture); the double carries a docstring warning. Latent today (every real forecast fixture carries one member per pair); elevate C-227 to Tier 1 on the first observed dead-over-live draw. Register: C-227 registered + resolved, Cluster F; 91 concerns (68 resolved, 23 open). Closes #290 Co-Authored-By: Claude Opus 5 (1M context) <[email protected]> Claude-Session: https://claude.ai/code/session_01N7YBtiS2kSVj27spMKDYfh
…loud fix(mapping): refuse a CM frame with two country_ids under one ISO code (#290)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Brings #293 to main: the C-227 guard in
build_mapping_dataframe(two country_ids under one ISO code → loud refusal), the RED→GREEN test against the real bundle, the fixture e2e tests moved off the 10-code label double, register C-227. Diff is exactly the reviewed commit.🤖 Generated with Claude Code
https://claude.ai/code/session_01N7YBtiS2kSVj27spMKDYfh