Model/tests/repositories/geospatial/test_geospatial_repository.py
Khalim Conn-Kowlessar 95719dd587 feat(geospatial): batch coordinates_for_uprns lookup (#1227)
Adds GeospatialRepository.coordinates_for_uprns(uprns) -> dict — a batch
coordinate lookup returning only covered UPRNs. The S3 adapter overrides it
to read the meta once, group UPRNs by their covering partition, and read each
partition once for all the UPRNs it covers; co-located (closely-numbered)
UPRNs share a partition, so an EPC Prediction cohort is typically one or two
reads instead of one per neighbour. Default port impl is a per-UPRN loop.

Feeds the EPC Prediction geo-proximity work: a cohort's UPRNs resolve to
coordinates in a couple of reads (validated at corpus scale: 170 partition
reads for 2683 UPRNs).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-15 14:35:32 +00:00

190 lines
6.4 KiB
Python

"""GeospatialRepo resolves a Property's coordinates from the OS Open-UPRN data.
A reference-data lookup, not a Fetcher (ADR-0011): no live OS API call. The
adapter reads the partitioned Open-UPRN parquet via an injected reader, so the
test exercises the partition lookup + filter against real fixture parquets with
no network.
"""
from __future__ import annotations
from collections.abc import Callable
from pathlib import Path
import pandas as pd
from typing import Optional
from domain.geospatial.coordinates import Coordinates
from domain.geospatial.planning_restrictions import PlanningRestrictions
from domain.geospatial.spatial_reference import SpatialReference
from repositories.geospatial.geospatial_s3_repository import GeospatialS3Repository
def _reader(base: Path) -> Callable[[str], pd.DataFrame]:
def read(key: str) -> pd.DataFrame:
return pd.read_parquet(base / key)
return read
def _write_open_uprn(base: Path) -> None:
spatial = base / "spatial"
spatial.mkdir(parents=True, exist_ok=True)
pd.DataFrame(
{"lower": [0], "upper": [100000], "filenames": ["0_100000.parquet"]}
).to_parquet(spatial / "filename_meta.parquet")
pd.DataFrame(
{
"UPRN": [12345, 12346],
"LATITUDE": [51.5074, 51.6000],
"LONGITUDE": [-0.1278, -0.2000],
# Planning flags co-located with the coordinates in the partition
# (legacy column names — confirm exact names in the S3 deep-dive).
"conservation_status": [True, False],
"is_listed_building": [False, True],
"is_heritage_building": [False, False],
}
).to_parquet(spatial / "0_100000.parquet")
def test_coordinates_for_returns_lon_lat(tmp_path: Path) -> None:
# Arrange
_write_open_uprn(tmp_path)
repo = GeospatialS3Repository(_reader(tmp_path))
# Act
coords = repo.coordinates_for(12345)
# Assert
assert coords == Coordinates(longitude=-0.1278, latitude=51.5074)
def test_coordinates_for_returns_none_when_uprn_absent(tmp_path: Path) -> None:
# Arrange
_write_open_uprn(tmp_path)
repo = GeospatialS3Repository(_reader(tmp_path))
# Act / Assert — uprn inside the partition range but not present in the data
assert repo.coordinates_for(99999) is None
def test_coordinates_for_returns_none_when_no_partition_covers_uprn(
tmp_path: Path,
) -> None:
# Arrange
_write_open_uprn(tmp_path)
repo = GeospatialS3Repository(_reader(tmp_path))
# Act / Assert — uprn beyond every partition's range
assert repo.coordinates_for(500000) is None
def test_planning_restrictions_for_reads_the_co_located_flags(tmp_path: Path) -> None:
# Arrange — same partition, planning flags alongside the coordinates.
_write_open_uprn(tmp_path)
repo = GeospatialS3Repository(_reader(tmp_path))
# Act
restrictions = repo.planning_restrictions_for(12345)
# Assert — the three flags come back as the Property's PlanningRestrictions.
assert restrictions == PlanningRestrictions(
in_conservation_area=True, is_listed=False, is_heritage=False
)
def test_planning_restrictions_for_returns_none_when_uprn_absent(
tmp_path: Path,
) -> None:
# Arrange
_write_open_uprn(tmp_path)
repo = GeospatialS3Repository(_reader(tmp_path))
# Act / Assert
assert repo.planning_restrictions_for(99999) is None
def test_spatial_for_returns_coordinates_and_restrictions_together(
tmp_path: Path,
) -> None:
# Arrange — one partition row carries the coordinates and the planning flags.
_write_open_uprn(tmp_path)
repo = GeospatialS3Repository(_reader(tmp_path))
# Act — a single reference lookup yields both, so Ingestion reads the row once.
reference: Optional[SpatialReference] = repo.spatial_for(12346)
# Assert
assert reference == SpatialReference(
coordinates=Coordinates(longitude=-0.2000, latitude=51.6000),
restrictions=PlanningRestrictions(
in_conservation_area=False, is_listed=True, is_heritage=False
),
)
def test_spatial_for_returns_none_when_uprn_absent(tmp_path: Path) -> None:
# Arrange
_write_open_uprn(tmp_path)
repo = GeospatialS3Repository(_reader(tmp_path))
# Act / Assert
assert repo.spatial_for(99999) is None
def _write_two_partition_open_uprn(base: Path) -> None:
"""Two UPRN-range partitions, so the batch lookup must span both."""
spatial = base / "spatial"
spatial.mkdir(parents=True, exist_ok=True)
pd.DataFrame(
{
"lower": [0, 100001],
"upper": [100000, 200000],
"filenames": ["0_100000.parquet", "100001_200000.parquet"],
}
).to_parquet(spatial / "filename_meta.parquet")
pd.DataFrame(
{"UPRN": [10, 11], "LATITUDE": [51.0, 51.1], "LONGITUDE": [-1.0, -1.1]}
).to_parquet(spatial / "0_100000.parquet")
pd.DataFrame(
{"UPRN": [150000], "LATITUDE": [52.0], "LONGITUDE": [-2.0]}
).to_parquet(spatial / "100001_200000.parquet")
def test_coordinates_for_uprns_resolves_a_batch_across_partitions(
tmp_path: Path,
) -> None:
# Arrange — UPRNs spanning two partitions, plus one absent and one off-scale.
_write_two_partition_open_uprn(tmp_path)
repo = GeospatialS3Repository(_reader(tmp_path))
# Act
resolved = repo.coordinates_for_uprns([10, 11, 150000, 99999, 500000])
# Assert — present UPRNs resolved; absent (99999) and uncovered (500000) omitted.
assert resolved == {
10: Coordinates(longitude=-1.0, latitude=51.0),
11: Coordinates(longitude=-1.1, latitude=51.1),
150000: Coordinates(longitude=-2.0, latitude=52.0),
}
def test_coordinates_for_uprns_reads_each_partition_once(tmp_path: Path) -> None:
# Arrange — count reads so co-located UPRNs don't re-read their partition.
_write_two_partition_open_uprn(tmp_path)
reads: list[str] = []
def counting_reader(key: str) -> pd.DataFrame:
reads.append(key)
return pd.read_parquet(tmp_path / key)
repo = GeospatialS3Repository(counting_reader)
# Act — two UPRNs share partition 0; one is in partition 1.
repo.coordinates_for_uprns([10, 11, 150000])
# Assert — the meta once + each of the two partitions once (3 reads, not 4).
assert reads.count("spatial/0_100000.parquet") == 1
assert reads.count("spatial/100001_200000.parquet") == 1
assert reads.count("spatial/filename_meta.parquet") == 1