Python tool for aircraft Doppler audio geolocation using historical ADS-B flight data (Source-available)
Python acoustic-geolocationads-baircraft-trackingairplaneaudio-forensicsdoppler-effectdoppler-shiftgeolocationosintsignal-processing
"""Named refinement reads against actual indexed aircraft observations."""
import json
import os
from pathlib import Path
import shutil
import sqlite3
import numpy as np
import pytest
from dopplergeist.automatic import _refinement_tracks
from dopplergeist.data_access import DataUnavailable, FlightIndex
@pytest.fixture(scope="module")
def recorded_index():
source = os.environ.get("DOPPLERGEIST_REAL_CPA_FIXTURE")
if not source:
pytest.skip("set DOPPLERGEIST_REAL_CPA_FIXTURE to a recorded-data diagnostic")
fixture = json.loads(Path(source).read_text())
index = FlightIndex(fixture["real_index"])
assert index.report["input_hash"] == fixture["index_input_hash"]
bounds, times = fixture["query_bounds"], fixture["query_time_s"]
observed = list(index.query(bounds, times))
assert observed, "The recorded workload contains no trajectories"
assert [track.id for track in observed] == sorted(fixture["track_ids"])
return index, bounds, times, observed
def same_tracks(actual, expected):
assert [track.id for track in actual] == [track.id for track in expected]
for left, right in zip(actual, expected):
for name in ("time", "lat", "lon", "altitude", "xyz", "relative_time", "velocity"):
a, b = getattr(left, name), getattr(right, name)
if a is None or b is None:
assert a is b
else:
np.testing.assert_array_equal(a, b, strict=True)
np.testing.assert_array_equal(left.spline.c, right.spline.c, strict=True)
np.testing.assert_array_equal(left.derivative.c, right.derivative.c, strict=True)
assert left.origin == right.origin
for name in ("typecode", "altitude_basis", "aircraft_class", "velocity_inferred_components"):
assert getattr(left, name) == getattr(right, name)
def test_named_envelope_matches_broad_query_across_sql_batches(recorded_index):
index, bounds, times, observed = recorded_index
with index.connect() as database:
names = [row[0] for row in database.execute("SELECT name FROM segments ORDER BY name LIMIT 650")]
# The real index contains enough names to cross the 500-parameter batch.
assert len(names) > 500
names = sorted(set(names) | {track.id for track in observed[:6]})
requested = names[::-1] + names[:3] + ["missing-recorded-segment"]
selected = set(requested)
expected = [track for track in observed if track.id in selected]
actual = list(index.lookup(requested, bounds=bounds, time_range=times))
same_tracks(actual, expected)
def test_named_envelope_preserves_spatial_time_filters_and_old_index_fallback(recorded_index):
original, bounds, times, observed = recorded_index
index = FlightIndex(original.path)
names = [track.id for track in observed[:6]]
# A disjoint geographic envelope and out-of-day time query expose accidental
# removal of either predicate without manufacturing trajectory observations.
disjoint = (bounds[0] + 20, bounds[1], bounds[2] + 20, bounds[3])
outside_time = (index.report["coverage"]["time_range_s"][1] + 86400,) * 2
for box, window in ((disjoint, times), (bounds, outside_time),
(bounds, (observed[0].time[0], observed[0].time[0]))):
expected = [track for track in index.query(box, window) if track.id in names]
if window == outside_time:
assert not expected
if window == (observed[0].time[0], observed[0].time[0]):
assert observed[0].id in {track.id for track in expected}
same_tracks(list(index.lookup(names, bounds=box, time_range=window)), expected)
# Older index metadata must disable spatial pruning in both entry points.
index.report["continuous_geometry_bounds"] = False
expected = [track for track in index.query(disjoint, times) if track.id in names]
assert [track.id for track in expected] == names
same_tracks(list(index.lookup(names, bounds=disjoint, time_range=times)), expected)
index.report["continuous_geometry_bounds"] = True
index.report["normalizer_sha256"] = "different-normalizer-version"
same_tracks(list(index.lookup(names, bounds=disjoint, time_range=times)), expected)
def test_refinement_adapter_duplicates_and_sample_budget(recorded_index):
index, bounds, times, observed = recorded_index
needed = {track.id for track in observed[:6]}
expected = [track for track in observed if track.id in needed]
sample_limit = sum(len(track.time) for track in expected)
class QueryAdapter:
def query(self, box, window):
assert tuple(box) == tuple(bounds) and tuple(window) == tuple(times)
yield from observed
for indexes in ([index], [index, index], [QueryAdapter()], [index, QueryAdapter()]):
actual = _refinement_tracks(indexes, needed, bounds, times, sample_limit)
same_tracks(list(actual.values()), expected)
with pytest.raises(ValueError, match="^Refinement trajectory sample budget exceeded$"):
_refinement_tracks(indexes, needed, bounds, times, sample_limit - 1)
assert _refinement_tracks([index], {"missing-recorded-segment"}, bounds, times, 1) == {}
def test_named_envelope_requires_complete_scope_and_checks_payload(recorded_index, tmp_path):
index, bounds, times, observed = recorded_index
name = observed[0].id
for settings in ({"bounds": bounds}, {"time_range": times}):
with pytest.raises(ValueError, match="must be supplied together"):
list(index.lookup([name], **settings))
path = tmp_path / "corrupt-index.sqlite"
shutil.copyfile(index.path, path)
with sqlite3.connect(path) as database:
database.execute("UPDATE segments SET payload_hash=? WHERE name=?", ("corrupt", name))
with pytest.raises(DataUnavailable, match="Corrupt flight payload"):
list(FlightIndex(path).lookup([name], bounds=bounds, time_range=times))