Files
next-tjwater-server/tests/unit/test_scada_replay_processing.py
T

128 lines
6.0 KiB
Python

from datetime import datetime, timedelta, timezone
from uuid import uuid4
import pytest
from app.domain.schemas.scada_replay import ScadaReplayCreate
from app.services.scada_replay_processing import prepare_cleaning, detect_frame
from app.services.scada_normal_profile import (
CALIBRATION_VERSION, DIAGNOSTIC_POLICY, diagnostic_signal,
)
START = datetime(2026, 10, 8, tzinfo=timezone.utc)
def config(**changes):
return ScadaReplayCreate.model_validate({
"request_id": str(uuid4()), "start_time": START, "end_time": START + timedelta(hours=2),
"device_ids": ["P1"], "rules": [{"rule_id": "pressure", "device_ids": ["P1"],
"kind": "relative_change", "direction": "decrease", "threshold": 30, "unit": "m"}],
**changes,
})
def rows(value, minute=0):
return [{"time": START + timedelta(minutes=minute), "device_id": "P1", "monitored_value": value, "source_hash": "h"}]
def test_quality_does_not_repair_a_real_pressure_drop_or_reverse_flow():
for value in (0, -5, 1):
result = prepare_cleaning(rows(value), config())
assert result[0]["quality"] == "valid"
assert result[0]["cleaned_value"] == value
invalid = prepare_cleaning(rows(-5), config(quality_ranges={"P1": {"lower": 0}}))
assert invalid[0]["quality"] == "invalid"
assert invalid[0]["monitored_value"] == -5
assert invalid[0]["cleaned_value"] is None
def test_persistent_drop_is_one_episode_and_recovery_then_drop_is_new_episode():
cfg = config()
doc = {"replay_id": "run", "histories": {"P1": [10, 10, 10, 10]}}
all_events = []
for minute, value in enumerate((1, 1, None, 10, 10, 1)):
doc, events = detect_frame(doc, prepare_cleaning(rows(value, minute), cfg), cfg, START + timedelta(minutes=minute))
all_events.extend(events)
business = [event for event in all_events if event.get("category") != "data_quality"]
quality = [event for event in all_events if event.get("category") == "data_quality"]
assert [event["change"] for event in business] == ["raised", "resolved", "raised"]
assert [event["change"] for event in quality] == ["raised", "resolved"]
assert business[0]["event_id"] != business[2]["event_id"]
assert doc["anomaly_count"] == 3
assert doc["processed_count"] == 6
def test_water_quality_zero_baseline_uses_explicit_bounds():
cfg = config(rules=[{"rule_id": "tracer", "device_ids": ["P1"], "kind": "bounds", "upper": 0.1, "unit": "m"}])
doc = {"replay_id": "run"}
doc, events = detect_frame(doc, prepare_cleaning(rows(0), cfg), cfg, START)
assert events == []
_, events = detect_frame(doc, prepare_cleaning(rows(0.4, 15), cfg), cfg, START + timedelta(minutes=15))
assert events[0]["change"] == "raised"
def test_empty_clock_ticks_do_not_break_consecutive_observations():
cfg = config(rules=[{"rule_id": "p", "device_ids": ["P1"], "kind": "bounds", "upper": 10,
"unit": "m", "consecutive_samples": 2}])
doc, events = detect_frame({"replay_id": "run"}, prepare_cleaning(rows(20), cfg), cfg, START)
assert events == []
doc, _ = detect_frame(doc, [], cfg, START + timedelta(minutes=1))
_, events = detect_frame(doc, prepare_cleaning(rows(20, 15), cfg), cfg, START + timedelta(minutes=15))
assert events[0]["change"] == "raised"
def weighted_document(low=97., high=103.):
return {"replay_id": "run", "histories": {"P1": [100.] * 4},
"monitoring_config": {"decision_policy": "weighted"},
"config": config().model_dump(mode="json"),
"monitoring_calibration": {
"version": CALIBRATION_VERSION, "diagnostic_policy": DIAGNOSTIC_POLICY,
"knowledge_cutoff": START.isoformat(),
"diagnostic": {"P1": {"phase_origin": START.isoformat(), "interval_seconds": 900,
"lower": [low], "upper": [high]}},
}}
def test_weighted_diagnosis_ignores_normal_diurnal_large_relative_changes():
cfg = config(rules=[{"rule_id": "flow", "device_ids": ["P1"], "kind": "relative_change",
"direction": "both", "threshold": 30, "unit": "m3/h"}])
document = weighted_document(25., 35.)
document["histories"]["P1"] = [10.] * 4
_, events = detect_frame(document, prepare_cleaning(rows(30), cfg), cfg, START)
assert events == [] # +200% is normal at this daily sampling phase.
_, events = detect_frame(document, prepare_cleaning(rows(36), cfg), cfg, START)
assert events[0]["change"] == "raised"
@pytest.mark.parametrize("baseline,value,low,high", [
(100., 95., 97., 103.), (0., -5., -1., 1.),
])
def test_weighted_single_indicator_uses_history_envelope_without_thirty_percent_gate(baseline, value, low, high):
cfg = config()
document = weighted_document(low, high)
document["histories"]["P1"] = [baseline] * 4
_, events = detect_frame(document, prepare_cleaning(rows(value), cfg), cfg, START)
assert len(events) == 1
assert events[0]["diagnostic_policy"] == DIAGNOSTIC_POLICY
assert diagnostic_signal(document, events[0])
assert "阈值 30" not in events[0]["detail"]
@pytest.mark.parametrize("missing", ["version", "diagnostic", "knowledge_cutoff"])
def test_weighted_missing_calibration_does_not_fall_back_to_thirty_percent(missing):
cfg = config()
document = weighted_document()
document["monitoring_calibration"].pop(missing)
_, events = detect_frame(document, prepare_cleaning(rows(1), cfg), cfg, START)
assert events == []
def test_old_active_relative_state_cannot_block_or_supply_new_diagnosis():
cfg = config()
document = weighted_document()
old = {"event_id": "old", "rule_id": "pressure", "device_ids": ["P1"], "status": "active"}
document["rule_states"] = {"pressure": {"event": old, "hits": 1, "recoveries": 0}}
result, events = detect_frame(document, prepare_cleaning(rows(95), cfg), cfg, START)
assert not diagnostic_signal(document, old)
assert result["rule_states"]["pressure"]["event"] == old
assert events[0]["diagnostic_policy"] == DIAGNOSTIC_POLICY