meshai/work/tests/test_incident_refactor.py

532 lines
23 KiB
Python
Raw Normal View History

"""Phase-2 incident/roads refactor tests — TIER-A (byte-identical goldens).
chore: excise the dead Central NATS consumer path (-11,328 LOC) (#144) * chore: excise the dead Central NATS consumer path Central was retired and its database dropped 2026-07-15; its NATS broker no longer exists. Verified against the live CT108 deployment: all 12 adapters run feed_source=native, zero on central, and central.enabled is False (default, never overridden). The consumer and its handlers were unreachable. Removed: - central/consumer.py and 6 dead handlers (nws, quake, swpc, nwis, avy, incident) -- their handle_* entrypoints were reachable only from the consumer's dispatch - the Central wiring in main.py (init, guarded start, retry loop, stop path) - the dead config surface: CentralConsumerConfig, EnvironmentalConfig.central, adapter_config ("central","severity_thresholds") and its display block - the nats-py dependency (consumer.py was its only importer) - 19 test files that exercised only the dead path KEPT -- these live under central/ but are imported directly by native adapters, and deleting them would break production: - wfigs_handler.py: firms_handler._handle_pass_boundary() calls its _render() on the live FIRMS growth-fire path (env/firms.py -> ingest_hotspot_pixel) - firms_handler, satpass_handler, tle_handler: split files whose handle_* entrypoints are dead but whose engines are live. Left intact; splitting them is separate work. - pass_predictor, budget, idaho_gauge_sites: fully live. The usgs_quake keys global_mag_floor / regional_mag_floor / regional_centroid / regional_radius_mi / broadcast_pager_alerts are NOT removed despite comments labelling them "CENTRAL-PATH ONLY" -- notifications/gating/quake.py reads them unconditionally in the native path. Those comments are corrected separately. Test-count note: the suite drops ~425 tests. Most were migration PARITY tests whose sole purpose was proving the native rewrite byte-matched the Central handler (golden byte-parity, cross-source identity, gate-sequence replay). With the handler deleted there is nothing left to compare against, so they cannot exist. Native-only tests were kept and reworked where a test reached for a central symbol incidentally. This is a real coverage loss, accepted deliberately: the parity harness proved the refactor faithful, and git history preserves the originals. Suite: 1984 passed, 6 failed -- the same 6 pre-existing failures as main (stale SCHEMA_VERSION x3, expired TLE fixtures x2, one order-dependent), all being fixed on fix/green-test-suite. No new failures. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(nws): restore native-only golden coverage for the wire formatter Commit ca751fb5 deleted the Central nws_handler parity harness along with the handler itself, which took the ONLY tests that pinned formatters.nws .format()'s literal wire output. Gate-sequence and schema-conformance tests already survived natively; the formatter's actual rendered text did not have any native-only regression net. Add TestFormatterGolden to test_nws_refactor.py: 3 real-fixture cases plus 6 hand-built pathological cases mined from the deleted test_nws_handler.py (SVR path-sampling, the "no dangling separator" regression, TOR on-ground vs radar-indicated, FFW flood-cause detection). Every literal was verified by temporarily restoring the pre-excision central.nws_handler._render() from git history (ca751fb5^) in a throwaway, uncommitted script, confirming byte-identical output against the current native format() for all 37 real fixtures (nws/ + nws_last/) and all 9 pathological cases, then pinning the confirmed-matching string as the literal -- not a blind snapshot of current behavior. quake/swpc/avalanche/hydro/incident/fire were checked and already carry equivalent native-only golden coverage (added directly in ca751fb5), so no changes were needed there. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Matt Johnson <mj@k7zvx.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-17 14:03:40 -06:00
The Central `incident_handler` module (`_parse_tomtom_incident()`,
`_parse_itd_511_incident()`, `_render()`) has been deleted along with the
rest of the Central NATS consumer path. The TomTom-incidents and ITD-511
golden-fixture parity groups that called those deleted parsers directly to
build a "golden" wire and compare it byte-for-byte against
formatters.incident.format() have been removed in full (46 tests across
TestTomtomGolden, TestItd511IncidentGolden, and
TestSchemaConformance::test_incident_canonical_keys_from_parser) every
assertion in those tests was generated FROM the deleted parser output, with
no independent hand-written expected content to fall back to, and no
native adapter exists that parses the *TomTom Incident Details* or Central
ITD-511-envelope raw shapes those fixtures capture (env/traffic.py is the
TomTom traffic-*flow* adapter, a different feed; env/roads511.py parses
ITD 511's own REST API shape, not the Central envelope fixtures here).
Original diffs are preserved in git history. This is a real production gap
flagged for Matt: TomTom road-incident ingestion in particular has no
native replacement.
chore: delete dead renderers/work_zone.py format_work_zone_mesh() had exactly one production caller, the central/consumer.py NATS bridge deleted in the prior Central-excision pass. Its live replacement, formatters.incident._render_work_zone() (registered for category "work_zone" in formatters/__init__.py), is an already-shipped byte-identical replica per that module's own docstring. All remaining references to renderers.work_zone were prose/comments describing the replica relationship, not imports. Test fallout: - tests/test_work_zone_renderer.py tested only the dead renderer in isolation (17 cases). Deleted — the live path has its own coverage (test_adapter_wzdx.py's formatter-integration tests, plus TestCrossSourceIdentity::test_work_zone_category_uses_wz_renderer and TestWorkZoneGolden in test_incident_refactor.py). - tests/test_itd_511_work_zone.py::test_itd_511_work_zone_renderer_produces_wire only smoke-tested the dead renderer's wire output for itd_511 data; redundant with TestWorkZoneGolden's byte-identical fixture coverage for the same adapter. Deleted. - tests/test_incident_refactor.py::TestWorkZoneGolden compared the live formatters.incident.format() output against a golden computed by calling the dead renderer live on two real fixtures. Mirroring the precedent already in test_nws_refactor.py for this exact situation (golden generator deleted out from under a parity test), the two golden strings were captured by running format_work_zone_mesh() against these fixtures immediately before deletion and are now pinned as literals — same coverage, no live dependency on the dead module. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-17 21:03:35 +00:00
Work-zone parity: `meshai.central_normalizer` (a top-level, non-Central-NATS
module; note the name is legacy) was never part of the deleted consumer path
and remains live. `meshai.notifications.renderers.work_zone` was ALSO never
part of the deleted consumer path, but was itself dead code (zero production
callers) once formatters.incident absorbed it as `_render_work_zone()`; it
was removed in a later ripout pass. See TestWorkZoneGolden below for how its
golden-parity coverage was preserved as pinned literals.
Groups
------
chore: excise the dead Central NATS consumer path (-11,328 LOC) (#144) * chore: excise the dead Central NATS consumer path Central was retired and its database dropped 2026-07-15; its NATS broker no longer exists. Verified against the live CT108 deployment: all 12 adapters run feed_source=native, zero on central, and central.enabled is False (default, never overridden). The consumer and its handlers were unreachable. Removed: - central/consumer.py and 6 dead handlers (nws, quake, swpc, nwis, avy, incident) -- their handle_* entrypoints were reachable only from the consumer's dispatch - the Central wiring in main.py (init, guarded start, retry loop, stop path) - the dead config surface: CentralConsumerConfig, EnvironmentalConfig.central, adapter_config ("central","severity_thresholds") and its display block - the nats-py dependency (consumer.py was its only importer) - 19 test files that exercised only the dead path KEPT -- these live under central/ but are imported directly by native adapters, and deleting them would break production: - wfigs_handler.py: firms_handler._handle_pass_boundary() calls its _render() on the live FIRMS growth-fire path (env/firms.py -> ingest_hotspot_pixel) - firms_handler, satpass_handler, tle_handler: split files whose handle_* entrypoints are dead but whose engines are live. Left intact; splitting them is separate work. - pass_predictor, budget, idaho_gauge_sites: fully live. The usgs_quake keys global_mag_floor / regional_mag_floor / regional_centroid / regional_radius_mi / broadcast_pager_alerts are NOT removed despite comments labelling them "CENTRAL-PATH ONLY" -- notifications/gating/quake.py reads them unconditionally in the native path. Those comments are corrected separately. Test-count note: the suite drops ~425 tests. Most were migration PARITY tests whose sole purpose was proving the native rewrite byte-matched the Central handler (golden byte-parity, cross-source identity, gate-sequence replay). With the handler deleted there is nothing left to compare against, so they cannot exist. Native-only tests were kept and reworked where a test reached for a central symbol incidentally. This is a real coverage loss, accepted deliberately: the parity harness proved the refactor faithful, and git history preserves the originals. Suite: 1984 passed, 6 failed -- the same 6 pre-existing failures as main (stale SCHEMA_VERSION x3, expired TLE fixtures x2, one order-dependent), all being fixed on fix/green-test-suite. No new failures. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(nws): restore native-only golden coverage for the wire formatter Commit ca751fb5 deleted the Central nws_handler parity harness along with the handler itself, which took the ONLY tests that pinned formatters.nws .format()'s literal wire output. Gate-sequence and schema-conformance tests already survived natively; the formatter's actual rendered text did not have any native-only regression net. Add TestFormatterGolden to test_nws_refactor.py: 3 real-fixture cases plus 6 hand-built pathological cases mined from the deleted test_nws_handler.py (SVR path-sampling, the "no dangling separator" regression, TOR on-ground vs radar-indicated, FFW flood-cause detection). Every literal was verified by temporarily restoring the pre-excision central.nws_handler._render() from git history (ca751fb5^) in a throwaway, uncommitted script, confirming byte-identical output against the current native format() for all 37 real fixtures (nws/ + nws_last/) and all 9 pathological cases, then pinning the confirmed-matching string as the literal -- not a blind snapshot of current behavior. quake/swpc/avalanche/hydro/incident/fire were checked and already carry equivalent native-only golden coverage (added directly in ca751fb5), so no changes were needed there. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Matt Johnson <mj@k7zvx.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-17 14:03:40 -06:00
1. Work-zone golden byte-parity (traffic_last/0002 itd_511, traffic_last/0003
wzdx) calls normalize() directly, same as the production consumer.
chore: excise the dead Central NATS consumer path (-11,328 LOC) (#144) * chore: excise the dead Central NATS consumer path Central was retired and its database dropped 2026-07-15; its NATS broker no longer exists. Verified against the live CT108 deployment: all 12 adapters run feed_source=native, zero on central, and central.enabled is False (default, never overridden). The consumer and its handlers were unreachable. Removed: - central/consumer.py and 6 dead handlers (nws, quake, swpc, nwis, avy, incident) -- their handle_* entrypoints were reachable only from the consumer's dispatch - the Central wiring in main.py (init, guarded start, retry loop, stop path) - the dead config surface: CentralConsumerConfig, EnvironmentalConfig.central, adapter_config ("central","severity_thresholds") and its display block - the nats-py dependency (consumer.py was its only importer) - 19 test files that exercised only the dead path KEPT -- these live under central/ but are imported directly by native adapters, and deleting them would break production: - wfigs_handler.py: firms_handler._handle_pass_boundary() calls its _render() on the live FIRMS growth-fire path (env/firms.py -> ingest_hotspot_pixel) - firms_handler, satpass_handler, tle_handler: split files whose handle_* entrypoints are dead but whose engines are live. Left intact; splitting them is separate work. - pass_predictor, budget, idaho_gauge_sites: fully live. The usgs_quake keys global_mag_floor / regional_mag_floor / regional_centroid / regional_radius_mi / broadcast_pager_alerts are NOT removed despite comments labelling them "CENTRAL-PATH ONLY" -- notifications/gating/quake.py reads them unconditionally in the native path. Those comments are corrected separately. Test-count note: the suite drops ~425 tests. Most were migration PARITY tests whose sole purpose was proving the native rewrite byte-matched the Central handler (golden byte-parity, cross-source identity, gate-sequence replay). With the handler deleted there is nothing left to compare against, so they cannot exist. Native-only tests were kept and reworked where a test reached for a central symbol incidentally. This is a real coverage loss, accepted deliberately: the parity harness proved the refactor faithful, and git history preserves the originals. Suite: 1984 passed, 6 failed -- the same 6 pre-existing failures as main (stale SCHEMA_VERSION x3, expired TLE fixtures x2, one order-dependent), all being fixed on fix/green-test-suite. No new failures. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(nws): restore native-only golden coverage for the wire formatter Commit ca751fb5 deleted the Central nws_handler parity harness along with the handler itself, which took the ONLY tests that pinned formatters.nws .format()'s literal wire output. Gate-sequence and schema-conformance tests already survived natively; the formatter's actual rendered text did not have any native-only regression net. Add TestFormatterGolden to test_nws_refactor.py: 3 real-fixture cases plus 6 hand-built pathological cases mined from the deleted test_nws_handler.py (SVR path-sampling, the "no dangling separator" regression, TOR on-ground vs radar-indicated, FFW flood-cause detection). Every literal was verified by temporarily restoring the pre-excision central.nws_handler._render() from git history (ca751fb5^) in a throwaway, uncommitted script, confirming byte-identical output against the current native format() for all 37 real fixtures (nws/ + nws_last/) and all 9 pathological cases, then pinning the confirmed-matching string as the literal -- not a blind snapshot of current behavior. quake/swpc/avalanche/hydro/incident/fire were checked and already carry equivalent native-only golden coverage (added directly in ca751fb5), so no changes were needed there. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Matt Johnson <mj@k7zvx.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-17 14:03:40 -06:00
2. Gate sequence: decide() lifecycle transitions (new cold-dup
suppress-on-update-False magnitude-up suppress-no-change).
chore: excise the dead Central NATS consumer path (-11,328 LOC) (#144) * chore: excise the dead Central NATS consumer path Central was retired and its database dropped 2026-07-15; its NATS broker no longer exists. Verified against the live CT108 deployment: all 12 adapters run feed_source=native, zero on central, and central.enabled is False (default, never overridden). The consumer and its handlers were unreachable. Removed: - central/consumer.py and 6 dead handlers (nws, quake, swpc, nwis, avy, incident) -- their handle_* entrypoints were reachable only from the consumer's dispatch - the Central wiring in main.py (init, guarded start, retry loop, stop path) - the dead config surface: CentralConsumerConfig, EnvironmentalConfig.central, adapter_config ("central","severity_thresholds") and its display block - the nats-py dependency (consumer.py was its only importer) - 19 test files that exercised only the dead path KEPT -- these live under central/ but are imported directly by native adapters, and deleting them would break production: - wfigs_handler.py: firms_handler._handle_pass_boundary() calls its _render() on the live FIRMS growth-fire path (env/firms.py -> ingest_hotspot_pixel) - firms_handler, satpass_handler, tle_handler: split files whose handle_* entrypoints are dead but whose engines are live. Left intact; splitting them is separate work. - pass_predictor, budget, idaho_gauge_sites: fully live. The usgs_quake keys global_mag_floor / regional_mag_floor / regional_centroid / regional_radius_mi / broadcast_pager_alerts are NOT removed despite comments labelling them "CENTRAL-PATH ONLY" -- notifications/gating/quake.py reads them unconditionally in the native path. Those comments are corrected separately. Test-count note: the suite drops ~425 tests. Most were migration PARITY tests whose sole purpose was proving the native rewrite byte-matched the Central handler (golden byte-parity, cross-source identity, gate-sequence replay). With the handler deleted there is nothing left to compare against, so they cannot exist. Native-only tests were kept and reworked where a test reached for a central symbol incidentally. This is a real coverage loss, accepted deliberately: the parity harness proved the refactor faithful, and git history preserves the originals. Suite: 1984 passed, 6 failed -- the same 6 pre-existing failures as main (stale SCHEMA_VERSION x3, expired TLE fixtures x2, one order-dependent), all being fixed on fix/green-test-suite. No new failures. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(nws): restore native-only golden coverage for the wire formatter Commit ca751fb5 deleted the Central nws_handler parity harness along with the handler itself, which took the ONLY tests that pinned formatters.nws .format()'s literal wire output. Gate-sequence and schema-conformance tests already survived natively; the formatter's actual rendered text did not have any native-only regression net. Add TestFormatterGolden to test_nws_refactor.py: 3 real-fixture cases plus 6 hand-built pathological cases mined from the deleted test_nws_handler.py (SVR path-sampling, the "no dangling separator" regression, TOR on-ground vs radar-indicated, FFW flood-cause detection). Every literal was verified by temporarily restoring the pre-excision central.nws_handler._render() from git history (ca751fb5^) in a throwaway, uncommitted script, confirming byte-identical output against the current native format() for all 37 real fixtures (nws/ + nws_last/) and all 9 pathological cases, then pinning the confirmed-matching string as the literal -- not a blind snapshot of current behavior. quake/swpc/avalanche/hydro/incident/fire were checked and already carry equivalent native-only golden coverage (added directly in ca751fb5), so no changes were needed there. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Matt Johnson <mj@k7zvx.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-17 14:03:40 -06:00
3. _anchor.resolve_anchor: DB hit and Photon fallback path.
4. Schema conformance: canonical data dict has all expected keys.
5. Cross-source identity: same render fields same output regardless of
source string.
"""
from __future__ import annotations
import calendar
import json
import pathlib
import time
from datetime import datetime
from typing import Optional
import pytest
from tests.harness.goldens import assert_byte_identical
# ── Constants ────────────────────────────────────────────────────────────────
_FIXTURE_DIR = pathlib.Path(__file__).parent / "fixtures"
# Pinned clock epoch for deterministic golden comparison (all work-zone tests).
# Using captured_epoch of the traffic_last fixtures (1783206522).
_AT_WZ = 1_783_206_522.0
# Expected canonical data keys for work-zone events.
_WZ_CANONICAL_KEYS = frozenset({
"road", "direction", "mile_start", "mile_end", "sub_type", "impact",
"ends_at_epoch", "town", "distance_mi", "bearing", "lat", "lon",
})
# ── Fixture loaders (module-level, before any DB touch) ──────────────────────
def _load_dir(hazard: str):
"""Load all fixtures from tests/fixtures/<hazard>/ sorted by name."""
d = _FIXTURE_DIR / hazard
if not d.is_dir():
return []
out = []
for p in sorted(d.glob("*.json")):
with open(p, encoding="utf-8") as f:
out.append((p.name, json.load(f)))
return out
_TRAFFIC_LAST_FX = _load_dir("traffic_last") # 6 files (mixed adapters)
chore: excise the dead Central NATS consumer path (-11,328 LOC) (#144) * chore: excise the dead Central NATS consumer path Central was retired and its database dropped 2026-07-15; its NATS broker no longer exists. Verified against the live CT108 deployment: all 12 adapters run feed_source=native, zero on central, and central.enabled is False (default, never overridden). The consumer and its handlers were unreachable. Removed: - central/consumer.py and 6 dead handlers (nws, quake, swpc, nwis, avy, incident) -- their handle_* entrypoints were reachable only from the consumer's dispatch - the Central wiring in main.py (init, guarded start, retry loop, stop path) - the dead config surface: CentralConsumerConfig, EnvironmentalConfig.central, adapter_config ("central","severity_thresholds") and its display block - the nats-py dependency (consumer.py was its only importer) - 19 test files that exercised only the dead path KEPT -- these live under central/ but are imported directly by native adapters, and deleting them would break production: - wfigs_handler.py: firms_handler._handle_pass_boundary() calls its _render() on the live FIRMS growth-fire path (env/firms.py -> ingest_hotspot_pixel) - firms_handler, satpass_handler, tle_handler: split files whose handle_* entrypoints are dead but whose engines are live. Left intact; splitting them is separate work. - pass_predictor, budget, idaho_gauge_sites: fully live. The usgs_quake keys global_mag_floor / regional_mag_floor / regional_centroid / regional_radius_mi / broadcast_pager_alerts are NOT removed despite comments labelling them "CENTRAL-PATH ONLY" -- notifications/gating/quake.py reads them unconditionally in the native path. Those comments are corrected separately. Test-count note: the suite drops ~425 tests. Most were migration PARITY tests whose sole purpose was proving the native rewrite byte-matched the Central handler (golden byte-parity, cross-source identity, gate-sequence replay). With the handler deleted there is nothing left to compare against, so they cannot exist. Native-only tests were kept and reworked where a test reached for a central symbol incidentally. This is a real coverage loss, accepted deliberately: the parity harness proved the refactor faithful, and git history preserves the originals. Suite: 1984 passed, 6 failed -- the same 6 pre-existing failures as main (stale SCHEMA_VERSION x3, expired TLE fixtures x2, one order-dependent), all being fixed on fix/green-test-suite. No new failures. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(nws): restore native-only golden coverage for the wire formatter Commit ca751fb5 deleted the Central nws_handler parity harness along with the handler itself, which took the ONLY tests that pinned formatters.nws .format()'s literal wire output. Gate-sequence and schema-conformance tests already survived natively; the formatter's actual rendered text did not have any native-only regression net. Add TestFormatterGolden to test_nws_refactor.py: 3 real-fixture cases plus 6 hand-built pathological cases mined from the deleted test_nws_handler.py (SVR path-sampling, the "no dangling separator" regression, TOR on-ground vs radar-indicated, FFW flood-cause detection). Every literal was verified by temporarily restoring the pre-excision central.nws_handler._render() from git history (ca751fb5^) in a throwaway, uncommitted script, confirming byte-identical output against the current native format() for all 37 real fixtures (nws/ + nws_last/) and all 9 pathological cases, then pinning the confirmed-matching string as the literal -- not a blind snapshot of current behavior. quake/swpc/avalanche/hydro/incident/fire were checked and already carry equivalent native-only golden coverage (added directly in ca751fb5), so no changes were needed there. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Matt Johnson <mj@k7zvx.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-17 14:03:40 -06:00
# ── Helper: build canonical work-zone dict from normalize() output ───────────
def _n_to_canonical_workzone(n: dict) -> dict:
"""Build canonical work-zone data dict from a normalize() result.
ends_at_epoch is stored as calendar.timegm(naive_dt.timetuple()) which
treats the naive datetime as UTC. The formatter reconstructs via
datetime.utcfromtimestamp() TZ-independent round-trip.
"""
ends_at: Optional[datetime] = n.get("ends_at")
ends_at_epoch: Optional[float] = None
if ends_at is not None:
try:
# Strip tzinfo first (mirrors _format_end_short's strip).
naive = ends_at.replace(tzinfo=None) if ends_at.tzinfo is not None else ends_at
ends_at_epoch = float(calendar.timegm(naive.timetuple()))
except Exception:
ends_at_epoch = None
return {
"road": n.get("road"),
"direction": n.get("direction"),
"mile_start": n.get("mile_start"),
"mile_end": n.get("mile_end"),
"sub_type": n.get("sub_type"),
"impact": n.get("impact"),
"ends_at_epoch": ends_at_epoch,
"town": n.get("town"),
"distance_mi": n.get("distance_mi"),
"bearing": n.get("bearing"),
"lat": None, # anchor already resolved in normalizer
"lon": None,
}
# ── Helper: minimal Event for formatter calls ────────────────────────────────
def _make_event(category: str, data: dict):
from meshai.notifications.events import Event
return Event(category=category, data=data)
# ── pytest fixtures: adapter_config overrides ────────────────────────────────
@pytest.fixture()
def broadcast_on_update_on():
"""Enable incident.broadcast_on_update for gate-sequence tests."""
from meshai.adapter_config._accessor import set_runtime_override, _overrides
set_runtime_override("incident", "broadcast_on_update", True)
yield
_overrides.pop(("incident", "broadcast_on_update"), None)
chore: excise the dead Central NATS consumer path (-11,328 LOC) (#144) * chore: excise the dead Central NATS consumer path Central was retired and its database dropped 2026-07-15; its NATS broker no longer exists. Verified against the live CT108 deployment: all 12 adapters run feed_source=native, zero on central, and central.enabled is False (default, never overridden). The consumer and its handlers were unreachable. Removed: - central/consumer.py and 6 dead handlers (nws, quake, swpc, nwis, avy, incident) -- their handle_* entrypoints were reachable only from the consumer's dispatch - the Central wiring in main.py (init, guarded start, retry loop, stop path) - the dead config surface: CentralConsumerConfig, EnvironmentalConfig.central, adapter_config ("central","severity_thresholds") and its display block - the nats-py dependency (consumer.py was its only importer) - 19 test files that exercised only the dead path KEPT -- these live under central/ but are imported directly by native adapters, and deleting them would break production: - wfigs_handler.py: firms_handler._handle_pass_boundary() calls its _render() on the live FIRMS growth-fire path (env/firms.py -> ingest_hotspot_pixel) - firms_handler, satpass_handler, tle_handler: split files whose handle_* entrypoints are dead but whose engines are live. Left intact; splitting them is separate work. - pass_predictor, budget, idaho_gauge_sites: fully live. The usgs_quake keys global_mag_floor / regional_mag_floor / regional_centroid / regional_radius_mi / broadcast_pager_alerts are NOT removed despite comments labelling them "CENTRAL-PATH ONLY" -- notifications/gating/quake.py reads them unconditionally in the native path. Those comments are corrected separately. Test-count note: the suite drops ~425 tests. Most were migration PARITY tests whose sole purpose was proving the native rewrite byte-matched the Central handler (golden byte-parity, cross-source identity, gate-sequence replay). With the handler deleted there is nothing left to compare against, so they cannot exist. Native-only tests were kept and reworked where a test reached for a central symbol incidentally. This is a real coverage loss, accepted deliberately: the parity harness proved the refactor faithful, and git history preserves the originals. Suite: 1984 passed, 6 failed -- the same 6 pre-existing failures as main (stale SCHEMA_VERSION x3, expired TLE fixtures x2, one order-dependent), all being fixed on fix/green-test-suite. No new failures. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(nws): restore native-only golden coverage for the wire formatter Commit ca751fb5 deleted the Central nws_handler parity harness along with the handler itself, which took the ONLY tests that pinned formatters.nws .format()'s literal wire output. Gate-sequence and schema-conformance tests already survived natively; the formatter's actual rendered text did not have any native-only regression net. Add TestFormatterGolden to test_nws_refactor.py: 3 real-fixture cases plus 6 hand-built pathological cases mined from the deleted test_nws_handler.py (SVR path-sampling, the "no dangling separator" regression, TOR on-ground vs radar-indicated, FFW flood-cause detection). Every literal was verified by temporarily restoring the pre-excision central.nws_handler._render() from git history (ca751fb5^) in a throwaway, uncommitted script, confirming byte-identical output against the current native format() for all 37 real fixtures (nws/ + nws_last/) and all 9 pathological cases, then pinning the confirmed-matching string as the literal -- not a blind snapshot of current behavior. quake/swpc/avalanche/hydro/incident/fire were checked and already carry equivalent native-only golden coverage (added directly in ca751fb5), so no changes were needed there. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Matt Johnson <mj@k7zvx.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-17 14:03:40 -06:00
# ── Helper: determine adapter type ────────────────────────────────────────────
def _adapter_for(fx: dict) -> str:
return (fx["envelope"]["data"] or {}).get("adapter") or ""
chore: excise the dead Central NATS consumer path (-11,328 LOC) (#144) * chore: excise the dead Central NATS consumer path Central was retired and its database dropped 2026-07-15; its NATS broker no longer exists. Verified against the live CT108 deployment: all 12 adapters run feed_source=native, zero on central, and central.enabled is False (default, never overridden). The consumer and its handlers were unreachable. Removed: - central/consumer.py and 6 dead handlers (nws, quake, swpc, nwis, avy, incident) -- their handle_* entrypoints were reachable only from the consumer's dispatch - the Central wiring in main.py (init, guarded start, retry loop, stop path) - the dead config surface: CentralConsumerConfig, EnvironmentalConfig.central, adapter_config ("central","severity_thresholds") and its display block - the nats-py dependency (consumer.py was its only importer) - 19 test files that exercised only the dead path KEPT -- these live under central/ but are imported directly by native adapters, and deleting them would break production: - wfigs_handler.py: firms_handler._handle_pass_boundary() calls its _render() on the live FIRMS growth-fire path (env/firms.py -> ingest_hotspot_pixel) - firms_handler, satpass_handler, tle_handler: split files whose handle_* entrypoints are dead but whose engines are live. Left intact; splitting them is separate work. - pass_predictor, budget, idaho_gauge_sites: fully live. The usgs_quake keys global_mag_floor / regional_mag_floor / regional_centroid / regional_radius_mi / broadcast_pager_alerts are NOT removed despite comments labelling them "CENTRAL-PATH ONLY" -- notifications/gating/quake.py reads them unconditionally in the native path. Those comments are corrected separately. Test-count note: the suite drops ~425 tests. Most were migration PARITY tests whose sole purpose was proving the native rewrite byte-matched the Central handler (golden byte-parity, cross-source identity, gate-sequence replay). With the handler deleted there is nothing left to compare against, so they cannot exist. Native-only tests were kept and reworked where a test reached for a central symbol incidentally. This is a real coverage loss, accepted deliberately: the parity harness proved the refactor faithful, and git history preserves the originals. Suite: 1984 passed, 6 failed -- the same 6 pre-existing failures as main (stale SCHEMA_VERSION x3, expired TLE fixtures x2, one order-dependent), all being fixed on fix/green-test-suite. No new failures. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(nws): restore native-only golden coverage for the wire formatter Commit ca751fb5 deleted the Central nws_handler parity harness along with the handler itself, which took the ONLY tests that pinned formatters.nws .format()'s literal wire output. Gate-sequence and schema-conformance tests already survived natively; the formatter's actual rendered text did not have any native-only regression net. Add TestFormatterGolden to test_nws_refactor.py: 3 real-fixture cases plus 6 hand-built pathological cases mined from the deleted test_nws_handler.py (SVR path-sampling, the "no dangling separator" regression, TOR on-ground vs radar-indicated, FFW flood-cause detection). Every literal was verified by temporarily restoring the pre-excision central.nws_handler._render() from git history (ca751fb5^) in a throwaway, uncommitted script, confirming byte-identical output against the current native format() for all 37 real fixtures (nws/ + nws_last/) and all 9 pathological cases, then pinning the confirmed-matching string as the literal -- not a blind snapshot of current behavior. quake/swpc/avalanche/hydro/incident/fire were checked and already carry equivalent native-only golden coverage (added directly in ca751fb5), so no changes were needed there. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Matt Johnson <mj@k7zvx.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-17 14:03:40 -06:00
# ── 1. Work-zone golden byte-parity ─────────────────────────────────────────
class TestWorkZoneGolden:
"""traffic_last/0002 (itd_511 work_zone) and traffic_last/0003 (wzdx)
must produce byte-identical output from the new formatter.
chore: delete dead renderers/work_zone.py format_work_zone_mesh() had exactly one production caller, the central/consumer.py NATS bridge deleted in the prior Central-excision pass. Its live replacement, formatters.incident._render_work_zone() (registered for category "work_zone" in formatters/__init__.py), is an already-shipped byte-identical replica per that module's own docstring. All remaining references to renderers.work_zone were prose/comments describing the replica relationship, not imports. Test fallout: - tests/test_work_zone_renderer.py tested only the dead renderer in isolation (17 cases). Deleted — the live path has its own coverage (test_adapter_wzdx.py's formatter-integration tests, plus TestCrossSourceIdentity::test_work_zone_category_uses_wz_renderer and TestWorkZoneGolden in test_incident_refactor.py). - tests/test_itd_511_work_zone.py::test_itd_511_work_zone_renderer_produces_wire only smoke-tested the dead renderer's wire output for itd_511 data; redundant with TestWorkZoneGolden's byte-identical fixture coverage for the same adapter. Deleted. - tests/test_incident_refactor.py::TestWorkZoneGolden compared the live formatters.incident.format() output against a golden computed by calling the dead renderer live on two real fixtures. Mirroring the precedent already in test_nws_refactor.py for this exact situation (golden generator deleted out from under a parity test), the two golden strings were captured by running format_work_zone_mesh() against these fixtures immediately before deletion and are now pinned as literals — same coverage, no live dependency on the dead module. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-17 21:03:35 +00:00
Originally the golden was computed live via normalize()
format_work_zone_mesh() (old renderers.work_zone path) and compared
against normalize() canonical data formatters.incident.format()
(new path). renderers.work_zone.py has since been deleted (dead code,
zero production callers post-Central-excision; formatters.incident's
_render_work_zone() is its byte-identical live replacement see that
module's docstring). Mirroring the approach in test_nws_refactor.py for
the same situation: the golden strings below were captured by running
format_work_zone_mesh() against these exact fixtures immediately before
its deletion (confirmed byte-identical to the new formatter's output at
that time) and are now pinned as literals, so this test exercises the
LIVE formatters.incident.format() path only.
now is pinned to captured_epoch (1783206522) for both paths so the
ends-at segment is deterministic.
"""
chore: delete dead renderers/work_zone.py format_work_zone_mesh() had exactly one production caller, the central/consumer.py NATS bridge deleted in the prior Central-excision pass. Its live replacement, formatters.incident._render_work_zone() (registered for category "work_zone" in formatters/__init__.py), is an already-shipped byte-identical replica per that module's own docstring. All remaining references to renderers.work_zone were prose/comments describing the replica relationship, not imports. Test fallout: - tests/test_work_zone_renderer.py tested only the dead renderer in isolation (17 cases). Deleted — the live path has its own coverage (test_adapter_wzdx.py's formatter-integration tests, plus TestCrossSourceIdentity::test_work_zone_category_uses_wz_renderer and TestWorkZoneGolden in test_incident_refactor.py). - tests/test_itd_511_work_zone.py::test_itd_511_work_zone_renderer_produces_wire only smoke-tested the dead renderer's wire output for itd_511 data; redundant with TestWorkZoneGolden's byte-identical fixture coverage for the same adapter. Deleted. - tests/test_incident_refactor.py::TestWorkZoneGolden compared the live formatters.incident.format() output against a golden computed by calling the dead renderer live on two real fixtures. Mirroring the precedent already in test_nws_refactor.py for this exact situation (golden generator deleted out from under a parity test), the two golden strings were captured by running format_work_zone_mesh() against these fixtures immediately before deletion and are now pinned as literals — same coverage, no live dependency on the dead module. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-17 21:03:35 +00:00
_GOLDEN = {
"0002.json": "🚧 US-91, near Chubbuck: southbound, road construction, ends Aug 17",
"0003.json": "🚧 US-95, near Wilder: southbound, ends Jul 19",
}
def _run_wz(self, fixture_name: str, adapter_expected: str):
from meshai.central_normalizer import normalize
from meshai.notifications.formatters.incident import format as fmt
# Find the fixture by name
fx_map = dict(_TRAFFIC_LAST_FX)
assert fixture_name in fx_map, f"Fixture {fixture_name!r} not found"
fx = fx_map[fixture_name]
adapter = _adapter_for(fx)
assert adapter == adapter_expected, (
f"Expected adapter={adapter_expected!r}, got {adapter!r}"
)
envelope = fx["envelope"]
now_epoch = float(fx.get("captured_epoch", time.time()))
n = normalize(envelope)
assert n is not None, f"normalize() returned None for {fixture_name!r}"
chore: delete dead renderers/work_zone.py format_work_zone_mesh() had exactly one production caller, the central/consumer.py NATS bridge deleted in the prior Central-excision pass. Its live replacement, formatters.incident._render_work_zone() (registered for category "work_zone" in formatters/__init__.py), is an already-shipped byte-identical replica per that module's own docstring. All remaining references to renderers.work_zone were prose/comments describing the replica relationship, not imports. Test fallout: - tests/test_work_zone_renderer.py tested only the dead renderer in isolation (17 cases). Deleted — the live path has its own coverage (test_adapter_wzdx.py's formatter-integration tests, plus TestCrossSourceIdentity::test_work_zone_category_uses_wz_renderer and TestWorkZoneGolden in test_incident_refactor.py). - tests/test_itd_511_work_zone.py::test_itd_511_work_zone_renderer_produces_wire only smoke-tested the dead renderer's wire output for itd_511 data; redundant with TestWorkZoneGolden's byte-identical fixture coverage for the same adapter. Deleted. - tests/test_incident_refactor.py::TestWorkZoneGolden compared the live formatters.incident.format() output against a golden computed by calling the dead renderer live on two real fixtures. Mirroring the precedent already in test_nws_refactor.py for this exact situation (golden generator deleted out from under a parity test), the two golden strings were captured by running format_work_zone_mesh() against these fixtures immediately before deletion and are now pinned as literals — same coverage, no live dependency on the dead module. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-17 21:03:35 +00:00
golden = self._GOLDEN[fixture_name]
# New formatter
canonical = _n_to_canonical_workzone(n)
event = _make_event("work_zone", canonical)
new_out = fmt(event, now=now_epoch, budget=140)
assert_byte_identical(new_out, golden)
def test_itd511_workzone_0002(self):
self._run_wz("0002.json", "itd_511")
def test_wzdx_workzone_0003(self):
self._run_wz("0003.json", "wzdx")
chore: excise the dead Central NATS consumer path (-11,328 LOC) (#144) * chore: excise the dead Central NATS consumer path Central was retired and its database dropped 2026-07-15; its NATS broker no longer exists. Verified against the live CT108 deployment: all 12 adapters run feed_source=native, zero on central, and central.enabled is False (default, never overridden). The consumer and its handlers were unreachable. Removed: - central/consumer.py and 6 dead handlers (nws, quake, swpc, nwis, avy, incident) -- their handle_* entrypoints were reachable only from the consumer's dispatch - the Central wiring in main.py (init, guarded start, retry loop, stop path) - the dead config surface: CentralConsumerConfig, EnvironmentalConfig.central, adapter_config ("central","severity_thresholds") and its display block - the nats-py dependency (consumer.py was its only importer) - 19 test files that exercised only the dead path KEPT -- these live under central/ but are imported directly by native adapters, and deleting them would break production: - wfigs_handler.py: firms_handler._handle_pass_boundary() calls its _render() on the live FIRMS growth-fire path (env/firms.py -> ingest_hotspot_pixel) - firms_handler, satpass_handler, tle_handler: split files whose handle_* entrypoints are dead but whose engines are live. Left intact; splitting them is separate work. - pass_predictor, budget, idaho_gauge_sites: fully live. The usgs_quake keys global_mag_floor / regional_mag_floor / regional_centroid / regional_radius_mi / broadcast_pager_alerts are NOT removed despite comments labelling them "CENTRAL-PATH ONLY" -- notifications/gating/quake.py reads them unconditionally in the native path. Those comments are corrected separately. Test-count note: the suite drops ~425 tests. Most were migration PARITY tests whose sole purpose was proving the native rewrite byte-matched the Central handler (golden byte-parity, cross-source identity, gate-sequence replay). With the handler deleted there is nothing left to compare against, so they cannot exist. Native-only tests were kept and reworked where a test reached for a central symbol incidentally. This is a real coverage loss, accepted deliberately: the parity harness proved the refactor faithful, and git history preserves the originals. Suite: 1984 passed, 6 failed -- the same 6 pre-existing failures as main (stale SCHEMA_VERSION x3, expired TLE fixtures x2, one order-dependent), all being fixed on fix/green-test-suite. No new failures. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(nws): restore native-only golden coverage for the wire formatter Commit ca751fb5 deleted the Central nws_handler parity harness along with the handler itself, which took the ONLY tests that pinned formatters.nws .format()'s literal wire output. Gate-sequence and schema-conformance tests already survived natively; the formatter's actual rendered text did not have any native-only regression net. Add TestFormatterGolden to test_nws_refactor.py: 3 real-fixture cases plus 6 hand-built pathological cases mined from the deleted test_nws_handler.py (SVR path-sampling, the "no dangling separator" regression, TOR on-ground vs radar-indicated, FFW flood-cause detection). Every literal was verified by temporarily restoring the pre-excision central.nws_handler._render() from git history (ca751fb5^) in a throwaway, uncommitted script, confirming byte-identical output against the current native format() for all 37 real fixtures (nws/ + nws_last/) and all 9 pathological cases, then pinning the confirmed-matching string as the literal -- not a blind snapshot of current behavior. quake/swpc/avalanche/hydro/incident/fire were checked and already carry equivalent native-only golden coverage (added directly in ca751fb5), so no changes were needed there. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Matt Johnson <mj@k7zvx.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-17 14:03:40 -06:00
# ── 2. Gate sequence ──────────────────────────────────────────────────────────
class TestGateSequence:
"""decide() lifecycle transitions:
step 1: new external_id lifecycle="new", broadcast=True
step 2: same id, cold-dup (commit not called) lifecycle="new", broadcast=True
step 3: same id, last_broadcast_at set; broadcast_on_update=False suppress
step 4: enable broadcast_on_update; magnitude up lifecycle="update"
step 5: same magnitude, no other change suppress
"""
_BASE_DATA = {
"external_id": "test-tti-abc123",
"source": "tomtom_incidents",
"sub_type": "jam",
"road": "I-84",
"direction": "W",
"from_loc": None,
"to_loc": None,
"mile_start": None,
"mile_end": None,
"county": "Ada",
"state": "ID",
"lat": 43.5,
"lon": -116.2,
"impact": None,
"start_at": 1783200000,
"end_at": None,
"magnitude": 3,
"delay_seconds": 180,
"icon_category": "jam",
}
def _decide(self, data: dict, now: float = 1_783_200_000.0):
from meshai.notifications.gating.incident import decide
return decide(data, source="tomtom_incidents", now=now)
def test_step1_new(self):
"""First sight → new, broadcast=True, commit is callable."""
result = self._decide(dict(self._BASE_DATA))
assert result.broadcast is True
assert result.lifecycle == "new"
assert callable(result.commit)
assert result.data_patch.get("is_update") is False
def test_step2_cold_dup_no_commit(self):
"""Row exists (from step1 INSERT) but last_broadcast_at = NULL
cold-start still lifecycle="new", broadcast=True."""
data = dict(self._BASE_DATA)
# INSERT without committing (simulate dispatcher drop)
self._decide(data, now=1_783_200_000.0)
# Second call: row exists, last_broadcast_at still NULL
result = self._decide(data, now=1_783_200_001.0)
assert result.broadcast is True
assert result.lifecycle == "new"
def test_step3_suppress_when_update_false(self):
"""After commit (last_broadcast_at set), broadcast_on_update=False → suppress."""
data = dict(self._BASE_DATA)
# Step 1: new + commit
r1 = self._decide(data, now=1_783_200_000.0)
assert r1.broadcast is True
r1.commit(1_783_200_000.0) # sets last_broadcast_at
# Step 3: same data, no magnitude/delay/icon change, update=False (default)
r2 = self._decide(data, now=1_783_200_010.0)
assert r2.broadcast is False
assert r2.lifecycle == "suppress"
def test_step4_magnitude_up_triggers_update(self, broadcast_on_update_on):
"""With broadcast_on_update=True + magnitude stepped up → lifecycle="update"."""
data = dict(self._BASE_DATA)
# new + commit
r1 = self._decide(data, now=1_783_200_000.0)
r1.commit(1_783_200_000.0)
# Higher magnitude
data_updated = dict(self._BASE_DATA, magnitude=4)
result = self._decide(data_updated, now=1_783_200_020.0)
assert result.broadcast is True
assert result.lifecycle == "update"
assert result.data_patch.get("is_update") is True
def test_step5_no_change_suppressed(self, broadcast_on_update_on):
"""After update commit, same magnitude → suppress."""
data = dict(self._BASE_DATA)
# new + commit
r1 = self._decide(data, now=1_783_200_000.0)
r1.commit(1_783_200_000.0)
# mag-up + commit
data_up = dict(self._BASE_DATA, magnitude=4)
r2 = self._decide(data_up, now=1_783_200_020.0)
assert r2.broadcast is True
r2.commit(1_783_200_020.0)
# Same magnitude again → no update condition
r3 = self._decide(data_up, now=1_783_200_030.0)
assert r3.broadcast is False
assert r3.lifecycle == "suppress"
def test_native_adapter_always_broadcasts(self):
"""external_id=None → native path, always broadcast lifecycle='native'."""
data = dict(self._BASE_DATA, external_id=None)
result = self._decide(data)
assert result.broadcast is True
assert result.lifecycle == "native"
assert result.commit is None
chore: excise the dead Central NATS consumer path (-11,328 LOC) (#144) * chore: excise the dead Central NATS consumer path Central was retired and its database dropped 2026-07-15; its NATS broker no longer exists. Verified against the live CT108 deployment: all 12 adapters run feed_source=native, zero on central, and central.enabled is False (default, never overridden). The consumer and its handlers were unreachable. Removed: - central/consumer.py and 6 dead handlers (nws, quake, swpc, nwis, avy, incident) -- their handle_* entrypoints were reachable only from the consumer's dispatch - the Central wiring in main.py (init, guarded start, retry loop, stop path) - the dead config surface: CentralConsumerConfig, EnvironmentalConfig.central, adapter_config ("central","severity_thresholds") and its display block - the nats-py dependency (consumer.py was its only importer) - 19 test files that exercised only the dead path KEPT -- these live under central/ but are imported directly by native adapters, and deleting them would break production: - wfigs_handler.py: firms_handler._handle_pass_boundary() calls its _render() on the live FIRMS growth-fire path (env/firms.py -> ingest_hotspot_pixel) - firms_handler, satpass_handler, tle_handler: split files whose handle_* entrypoints are dead but whose engines are live. Left intact; splitting them is separate work. - pass_predictor, budget, idaho_gauge_sites: fully live. The usgs_quake keys global_mag_floor / regional_mag_floor / regional_centroid / regional_radius_mi / broadcast_pager_alerts are NOT removed despite comments labelling them "CENTRAL-PATH ONLY" -- notifications/gating/quake.py reads them unconditionally in the native path. Those comments are corrected separately. Test-count note: the suite drops ~425 tests. Most were migration PARITY tests whose sole purpose was proving the native rewrite byte-matched the Central handler (golden byte-parity, cross-source identity, gate-sequence replay). With the handler deleted there is nothing left to compare against, so they cannot exist. Native-only tests were kept and reworked where a test reached for a central symbol incidentally. This is a real coverage loss, accepted deliberately: the parity harness proved the refactor faithful, and git history preserves the originals. Suite: 1984 passed, 6 failed -- the same 6 pre-existing failures as main (stale SCHEMA_VERSION x3, expired TLE fixtures x2, one order-dependent), all being fixed on fix/green-test-suite. No new failures. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(nws): restore native-only golden coverage for the wire formatter Commit ca751fb5 deleted the Central nws_handler parity harness along with the handler itself, which took the ONLY tests that pinned formatters.nws .format()'s literal wire output. Gate-sequence and schema-conformance tests already survived natively; the formatter's actual rendered text did not have any native-only regression net. Add TestFormatterGolden to test_nws_refactor.py: 3 real-fixture cases plus 6 hand-built pathological cases mined from the deleted test_nws_handler.py (SVR path-sampling, the "no dangling separator" regression, TOR on-ground vs radar-indicated, FFW flood-cause detection). Every literal was verified by temporarily restoring the pre-excision central.nws_handler._render() from git history (ca751fb5^) in a throwaway, uncommitted script, confirming byte-identical output against the current native format() for all 37 real fixtures (nws/ + nws_last/) and all 9 pathological cases, then pinning the confirmed-matching string as the literal -- not a blind snapshot of current behavior. quake/swpc/avalanche/hydro/incident/fire were checked and already carry equivalent native-only golden coverage (added directly in ca751fb5), so no changes were needed there. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Matt Johnson <mj@k7zvx.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-17 14:03:40 -06:00
# ── 3. _anchor.resolve_anchor ────────────────────────────────────────────────
class TestAnchorResolve:
"""resolve_anchor() returns a result from the town_anchors DB when a row
is within max_mi, and falls through to nearest_town() otherwise."""
def test_db_hit_within_range(self):
"""Insert a town_anchors row closer than any seeded row → resolve_anchor returns it."""
import time as _time
from meshai.persistence import get_db
from meshai.notifications.formatters._anchor import resolve_anchor
conn = get_db()
# Use extreme southern coordinates so no seeded anchor is closer.
# Insert our test anchor right next to the event coords.
conn.execute(
"INSERT OR IGNORE INTO town_anchors(name, lat, lon, state, enabled, updated_at) "
"VALUES (?, ?, ?, ?, ?, ?)",
("testville", -33.8688, 151.2093, "NSW", 1, _time.time()), # Sydney
)
# Event very close to Sydney
result = resolve_anchor(-33.870, 151.210, max_mi=50.0)
assert result is not None
assert result["town"] == "Testville" # title-cased
assert isinstance(result["distance_mi"], int)
assert result["bearing"] in {"N", "NE", "E", "SE", "S", "SW", "W", "NW"}
def test_db_hit_out_of_range_returns_none(self):
"""Only a far-away town_anchors row exists → max_mi filter → None from DB step."""
import time as _time
from meshai.persistence import get_db
from meshai.notifications.formatters._anchor import resolve_anchor
conn = get_db()
# Clear seeded anchors so only our controlled row exists.
conn.execute("DELETE FROM town_anchors")
conn.execute(
"INSERT INTO town_anchors(name, lat, lon, state, enabled, updated_at) "
"VALUES (?, ?, ?, ?, ?, ?)",
("fartown", 47.0, -116.2, "ID", 1, _time.time()), # ~380 km N
)
# max_mi=10 — fartown is way out of range; nearest_town fallback will
# also fail in the test env (no Photon at these coords) → None
result = resolve_anchor(43.615, -116.205, max_mi=10.0)
assert result is None
def test_photon_fallback(self, monkeypatch):
"""With all town_anchors cleared, nearest_town() fallback is exercised."""
import time as _time
from meshai.persistence import get_db
from meshai.notifications.formatters._anchor import resolve_anchor
import meshai.central_normalizer as cn_mod
# Clear all seeded town_anchors so the DB step finds nothing.
conn = get_db()
conn.execute("DELETE FROM town_anchors")
called = []
def fake_nearest_town(lat, lon, max_distance_mi=50.0):
called.append((lat, lon))
return {"name": "Photon City", "distance_mi": 5, "bearing": "NW"}
monkeypatch.setattr(cn_mod, "nearest_town", fake_nearest_town)
result = resolve_anchor(43.615, -116.205, max_mi=50.0)
assert result is not None
assert result["town"] == "Photon City"
assert result["distance_mi"] == 5
assert result["bearing"] == "NW"
assert called # ensure fallback was called
def test_none_coords_returns_none(self):
from meshai.notifications.formatters._anchor import resolve_anchor
assert resolve_anchor(None, -116.2, max_mi=50.0) is None
assert resolve_anchor(43.6, None, max_mi=50.0) is None
chore: excise the dead Central NATS consumer path (-11,328 LOC) (#144) * chore: excise the dead Central NATS consumer path Central was retired and its database dropped 2026-07-15; its NATS broker no longer exists. Verified against the live CT108 deployment: all 12 adapters run feed_source=native, zero on central, and central.enabled is False (default, never overridden). The consumer and its handlers were unreachable. Removed: - central/consumer.py and 6 dead handlers (nws, quake, swpc, nwis, avy, incident) -- their handle_* entrypoints were reachable only from the consumer's dispatch - the Central wiring in main.py (init, guarded start, retry loop, stop path) - the dead config surface: CentralConsumerConfig, EnvironmentalConfig.central, adapter_config ("central","severity_thresholds") and its display block - the nats-py dependency (consumer.py was its only importer) - 19 test files that exercised only the dead path KEPT -- these live under central/ but are imported directly by native adapters, and deleting them would break production: - wfigs_handler.py: firms_handler._handle_pass_boundary() calls its _render() on the live FIRMS growth-fire path (env/firms.py -> ingest_hotspot_pixel) - firms_handler, satpass_handler, tle_handler: split files whose handle_* entrypoints are dead but whose engines are live. Left intact; splitting them is separate work. - pass_predictor, budget, idaho_gauge_sites: fully live. The usgs_quake keys global_mag_floor / regional_mag_floor / regional_centroid / regional_radius_mi / broadcast_pager_alerts are NOT removed despite comments labelling them "CENTRAL-PATH ONLY" -- notifications/gating/quake.py reads them unconditionally in the native path. Those comments are corrected separately. Test-count note: the suite drops ~425 tests. Most were migration PARITY tests whose sole purpose was proving the native rewrite byte-matched the Central handler (golden byte-parity, cross-source identity, gate-sequence replay). With the handler deleted there is nothing left to compare against, so they cannot exist. Native-only tests were kept and reworked where a test reached for a central symbol incidentally. This is a real coverage loss, accepted deliberately: the parity harness proved the refactor faithful, and git history preserves the originals. Suite: 1984 passed, 6 failed -- the same 6 pre-existing failures as main (stale SCHEMA_VERSION x3, expired TLE fixtures x2, one order-dependent), all being fixed on fix/green-test-suite. No new failures. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(nws): restore native-only golden coverage for the wire formatter Commit ca751fb5 deleted the Central nws_handler parity harness along with the handler itself, which took the ONLY tests that pinned formatters.nws .format()'s literal wire output. Gate-sequence and schema-conformance tests already survived natively; the formatter's actual rendered text did not have any native-only regression net. Add TestFormatterGolden to test_nws_refactor.py: 3 real-fixture cases plus 6 hand-built pathological cases mined from the deleted test_nws_handler.py (SVR path-sampling, the "no dangling separator" regression, TOR on-ground vs radar-indicated, FFW flood-cause detection). Every literal was verified by temporarily restoring the pre-excision central.nws_handler._render() from git history (ca751fb5^) in a throwaway, uncommitted script, confirming byte-identical output against the current native format() for all 37 real fixtures (nws/ + nws_last/) and all 9 pathological cases, then pinning the confirmed-matching string as the literal -- not a blind snapshot of current behavior. quake/swpc/avalanche/hydro/incident/fire were checked and already carry equivalent native-only golden coverage (added directly in ca751fb5), so no changes were needed there. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Matt Johnson <mj@k7zvx.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-17 14:03:40 -06:00
# ── 4. Schema conformance ────────────────────────────────────────────────────
class TestSchemaConformance:
"""Canonical data dicts produced by to_event() and the bridge must
contain all expected keys."""
def test_workzone_canonical_keys_from_normalize(self):
"""All work-zone canonical keys present in extraction from normalize()."""
from meshai.central_normalizer import normalize
with open(_FIXTURE_DIR / "traffic_last" / "0002.json", encoding="utf-8") as f:
fx = json.load(f)
n = normalize(fx["envelope"])
assert n is not None
canonical = _n_to_canonical_workzone(n)
assert _WZ_CANONICAL_KEYS == set(canonical.keys())
def test_roads511_to_event_canonical_keys(self):
"""Roads511Adapter.to_event() emits all incident canonical keys."""
from meshai.env.roads511 import Roads511Adapter
class _Cfg:
api_key = ""; base_url = ""; endpoints = []; bbox = []; tick_seconds = 300
adapter = Roads511Adapter(_Cfg())
# Minimal internal event dict
evt = {
"source": "511",
"event_id": "511_test001",
"event_type": "Incident",
"headline": "Test: Road Event",
"description": "Debris on roadway",
"severity": "routine",
"lat": 43.6, "lon": -116.2,
"expires": 1_783_300_000.0,
"fetched_at": 1_783_200_000.0,
"properties": {
"roadway": "I-84",
"is_closure": False,
"last_updated": None,
},
}
event = adapter.to_event(evt)
assert event is not None
d = event.data
# All required keys present (sub-set — not all fields are set by native adapter)
for key in ("external_id", "source", "sub_type", "road", "lat", "lon"):
assert key in d, f"Missing key {key!r} in Roads511Adapter canonical data"
chore: excise the dead Central NATS consumer path (-11,328 LOC) (#144) * chore: excise the dead Central NATS consumer path Central was retired and its database dropped 2026-07-15; its NATS broker no longer exists. Verified against the live CT108 deployment: all 12 adapters run feed_source=native, zero on central, and central.enabled is False (default, never overridden). The consumer and its handlers were unreachable. Removed: - central/consumer.py and 6 dead handlers (nws, quake, swpc, nwis, avy, incident) -- their handle_* entrypoints were reachable only from the consumer's dispatch - the Central wiring in main.py (init, guarded start, retry loop, stop path) - the dead config surface: CentralConsumerConfig, EnvironmentalConfig.central, adapter_config ("central","severity_thresholds") and its display block - the nats-py dependency (consumer.py was its only importer) - 19 test files that exercised only the dead path KEPT -- these live under central/ but are imported directly by native adapters, and deleting them would break production: - wfigs_handler.py: firms_handler._handle_pass_boundary() calls its _render() on the live FIRMS growth-fire path (env/firms.py -> ingest_hotspot_pixel) - firms_handler, satpass_handler, tle_handler: split files whose handle_* entrypoints are dead but whose engines are live. Left intact; splitting them is separate work. - pass_predictor, budget, idaho_gauge_sites: fully live. The usgs_quake keys global_mag_floor / regional_mag_floor / regional_centroid / regional_radius_mi / broadcast_pager_alerts are NOT removed despite comments labelling them "CENTRAL-PATH ONLY" -- notifications/gating/quake.py reads them unconditionally in the native path. Those comments are corrected separately. Test-count note: the suite drops ~425 tests. Most were migration PARITY tests whose sole purpose was proving the native rewrite byte-matched the Central handler (golden byte-parity, cross-source identity, gate-sequence replay). With the handler deleted there is nothing left to compare against, so they cannot exist. Native-only tests were kept and reworked where a test reached for a central symbol incidentally. This is a real coverage loss, accepted deliberately: the parity harness proved the refactor faithful, and git history preserves the originals. Suite: 1984 passed, 6 failed -- the same 6 pre-existing failures as main (stale SCHEMA_VERSION x3, expired TLE fixtures x2, one order-dependent), all being fixed on fix/green-test-suite. No new failures. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(nws): restore native-only golden coverage for the wire formatter Commit ca751fb5 deleted the Central nws_handler parity harness along with the handler itself, which took the ONLY tests that pinned formatters.nws .format()'s literal wire output. Gate-sequence and schema-conformance tests already survived natively; the formatter's actual rendered text did not have any native-only regression net. Add TestFormatterGolden to test_nws_refactor.py: 3 real-fixture cases plus 6 hand-built pathological cases mined from the deleted test_nws_handler.py (SVR path-sampling, the "no dangling separator" regression, TOR on-ground vs radar-indicated, FFW flood-cause detection). Every literal was verified by temporarily restoring the pre-excision central.nws_handler._render() from git history (ca751fb5^) in a throwaway, uncommitted script, confirming byte-identical output against the current native format() for all 37 real fixtures (nws/ + nws_last/) and all 9 pathological cases, then pinning the confirmed-matching string as the literal -- not a blind snapshot of current behavior. quake/swpc/avalanche/hydro/incident/fire were checked and already carry equivalent native-only golden coverage (added directly in ca751fb5), so no changes were needed there. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Matt Johnson <mj@k7zvx.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-17 14:03:40 -06:00
# ── 5. Cross-source identity ─────────────────────────────────────────────────
class TestCrossSourceIdentity:
"""Same render-relevant canonical fields → same formatter output regardless
of source string or other metadata fields."""
def test_same_fields_different_source(self):
"""Two events with identical display fields produce identical output."""
from meshai.notifications.formatters.incident import format as fmt
shared_fields = {
"sub_type": "accident",
"road": "I-84",
"direction": "W",
"geocoder_city": "Boise",
"state": "ID",
"mile_marker": None,
"from_loc": None,
"to_loc": None,
"lanes_affected": "Two left lanes closed",
"comment": "Multi-vehicle crash",
"impact": None,
"county": None,
}
data_a = dict(shared_fields, external_id="TTI-001", source="tomtom_incidents",
lat=43.6, lon=-116.2, magnitude=4, delay_seconds=300,
icon_category="accident", start_at=None, end_at=None,
mile_start=None, mile_end=None, cause=None, landclass=None)
data_b = dict(shared_fields, external_id="itd-511:9999", source="itd_511",
lat=43.61, lon=-116.21, magnitude=None, delay_seconds=None,
icon_category="accident", start_at=None, end_at=None,
mile_start=None, mile_end=None, cause=None, landclass=None)
event_a = _make_event("road_incident", data_a)
event_b = _make_event("road_incident", data_b)
out_a = fmt(event_a, now=1_783_200_000.0, budget=140)
out_b = fmt(event_b, now=1_783_200_000.0, budget=140)
assert_byte_identical(out_a, out_b)
def test_work_zone_category_uses_wz_renderer(self):
"""event.category='work_zone' selects the work-zone path, not incident."""
from meshai.notifications.formatters.incident import format as fmt
data = {
"road": "US-20", "direction": "eastbound",
"mile_start": None, "mile_end": None,
"sub_type": "paving", "impact": "partial",
"ends_at_epoch": None, "town": "Arco",
"distance_mi": 3, "bearing": "NW",
"lat": None, "lon": None,
}
event = _make_event("work_zone", data)
out = fmt(event, now=1_783_200_000.0, budget=140)
# Should start with work-zone emoji
assert out.startswith("🚧")
# When road is present AND town/distance are set, both appear in the output
# (suppress_distance_seg=False when raw_road is set).
assert "US-20" in out
assert "paving" in out
# Town distance segment is included when road + town are both present
assert "Arco" in out or "mi" in out or "paving" in out # flexible
def test_incident_category_uses_incident_renderer(self):
"""event.category='road_incident' selects the incident path."""
from meshai.notifications.formatters.incident import format as fmt
data = {
"sub_type": "accident", "road": "I-84", "direction": "W",
"geocoder_city": "Boise", "state": "ID",
"from_loc": None, "to_loc": None, "mile_marker": None,
"lanes_affected": None, "comment": None, "impact": None,
"county": None, "external_id": "x", "source": "tomtom_incidents",
"lat": 43.6, "lon": -116.2, "magnitude": 4, "delay_seconds": None,
"icon_category": "accident", "start_at": None, "end_at": None,
"mile_start": None, "mile_end": None, "cause": None, "landclass": None,
}
event = _make_event("road_incident", data)
out = fmt(event, now=1_783_200_000.0, budget=140)
assert out.startswith("🚨") # accident emoji
assert "Crash" in out
assert "Boise" in out