Repository navigation
S1 — Coverage contract: complete-coverage invariant + cell-count logging (C-34) #53
Description
Activity
Build shape (per the epic's post-
/falsifydesign contract, 2026-06-26)Do not add coverage logic to the pandas
_validategod-method. Build it as:views_postprocessing/delivery/coverage.py(representation-free): a pure function, e.g.assert_complete_coverage(received_gids: set[int], expected_count: int) -> None(raise on mismatch), and the region→expected-count constant lives alongside (delivery/constants). No pandas import here.views_postprocessing/unfao/extraction.py:cells_of(df) -> set[int]— the only pandas-aware line for this guard.- Manager:
assert_complete_coverage(cells_of(self._df), EXPECTED[region])— extract → call → raise. The guard is called, not a method.
Two checks (from the earlier deep-dive): (a) internal consistency — received gids survive to delivery (expected = the gids that arrived); (b) region intent — count matches the intended region (expected = declared constant, sourced from datafactory's region metadata when it publishes it — same companion as #52's
last_valid_month_id).Tests live against
delivery/coverage.pyon plain sets/ints (no framework). The design-contract stubs intests/test_input_integrity_design_contract.pyalready enforce the placement + pandas-free rules.- changed the title
[-]S2 — Coverage contract: assert expected cell count + log cell counts (C-34)[/-][+]S1 — Coverage contract: complete-coverage invariant + cell-count logging (C-34)[/+]on Jun 26, 2026 Delivered and merged to
developmentvia #64 (input-integrity sprint S0–S6). ruff clean, 126 passed / 44 xfailed. Closing — the auto-close keyword did not fire because #64 merged intodevelopment, not the default branch. Live-run validation of the residual contracts (C-25 identity name/loa, C-43 enrichment equivalence) is tracked in the risk register and exercised by the go-global run (views-models#127).- added a commit that references this issue
on Aug 11, 2026
Epic: #51 · S1 · depends on #61 (S0) ·
storyimplementationProblem (C-34)
Nothing asserts how many cells the FAO pipeline should process; a wrong/stale region delivers partial coverage to FAO with no signal.
_validatechecks only metadata nulls.Build shape (per the epic design contract)
delivery/coverage.py(representation-free):assert_complete_coverage(received_gids: set[int], expected_count: int) -> None(raise on mismatch); region→expected-count constant alongside (delivery/constants). No pandas.unfao/extraction.py:cells_of(df) -> set[int]— the only pandas line.assert_complete_coverage(cells_of(df), EXPECTED[region])via the S0 hook; logs fetched/enriched/unmapped counts.Two checks: (a) internal consistency — received gids survive to delivery (expected = the gids that arrived); (b) region intent — count matches the intended region (expected = declared constant; to be sourced from datafactory region metadata when it publishes it — same companion shape as S2's
last_valid_month_id).Acceptance criteria
delivery/coverage.pyis a pure, pandas-free function with the region constant; unit-tested on plain sets/ints.Files
views_postprocessing/delivery/coverage.py(new),views_postprocessing/unfao/extraction.py(extend),unfao/managers/unfao.py(call),tests/.