fix(nspd-geo): MultiPolygon coercion + wide building schema (on-demand ingest) #1045

Merged
bot-backend merged 1 commit from fix/on-demand-ingest-multipolygon into main 2026-06-04 08:11:54 +00:00
4 changed files with 430 additions and 38 deletions

View file

@ -404,30 +404,42 @@ def upsert_features(
for feature in features:
cat = feature.category_id
if cat == CAT_PARCEL:
upsert_parcel(db, feature, source=source)
counts["parcels"] += 1
elif cat == CAT_BUILDING:
upsert_building(db, feature, source=source)
counts["buildings"] += 1
elif cat == CAT_CONSTRUCTION:
upsert_construction(db, feature, source=source)
counts["constructions"] += 1
elif cat == CAT_ONC:
upsert_onc(db, feature, source=source)
counts["oncs"] += 1
elif cat == CAT_ENK:
upsert_enk(db, feature, source=source)
counts["enks"] += 1
elif cat in CAT_ZOUIT:
upsert_zouit(db, feature, source=source)
counts["zouit"] += 1
elif cat == CAT_QUARTER_STATS:
pass # обрабатывается отдельно в upsert_quarter_stats
else:
if cat is not None:
logger.debug("upsert_features: неизвестный category_id=%s, skip", cat)
counts["skipped"] += 1
# Per-feature SAVEPOINT (backend.md SAVEPOINT rule): один битый feature (edge-case
# GeoJSON / NOT NULL violation / type mismatch) НЕ должен ронять весь snapshot-tx
# квартала. Зеркалит _grid_walk_category (~368) и _save_territorial_zones (~1234):
# begin_nested() rollback'ит только этот feature, остальные сохраняются. Счётчик
# инкрементим ТОЛЬКО при успехе (внутри блока) → counts остаются точными.
try:
with db.begin_nested():
if cat == CAT_PARCEL:
upsert_parcel(db, feature, source=source)
counts["parcels"] += 1
elif cat == CAT_BUILDING:
upsert_building(db, feature, source=source)
counts["buildings"] += 1
elif cat == CAT_CONSTRUCTION:
upsert_construction(db, feature, source=source)
counts["constructions"] += 1
elif cat == CAT_ONC:
upsert_onc(db, feature, source=source)
counts["oncs"] += 1
elif cat == CAT_ENK:
upsert_enk(db, feature, source=source)
counts["enks"] += 1
elif cat in CAT_ZOUIT:
upsert_zouit(db, feature, source=source)
counts["zouit"] += 1
elif cat == CAT_QUARTER_STATS:
pass # обрабатывается отдельно в upsert_quarter_stats
else:
if cat is not None:
logger.debug("upsert_features: неизвестный category_id=%s, skip", cat)
counts["skipped"] += 1
except Exception as e:
# Логируем + skip (не re-raise) — один битый feature не отравляет квартал.
logger.warning(
"upsert_features: feature failed (cat=%s, id=%s): %s", cat, feature.id, e
)
return counts

View file

@ -27,6 +27,7 @@ from sqlalchemy import text
from sqlalchemy.orm import Session
from app.core.db import SessionLocal
from app.services.cadastre.bulk_harvest import _derive_quarter_cad, _safe_int, _safe_numeric
from app.services.scrapers.nspd_lite import (
NspdLiteError,
NspdLiteWafError,
@ -136,7 +137,12 @@ def _save_quarter(db: Session, payload: dict, cad_num: str) -> int:
text(
"""
INSERT INTO cad_quarters_geom (cad_number, geom, raw_props, fetched_at)
VALUES (:cad, ST_Transform(ST_SetSRID(ST_GeomFromGeoJSON(:g), 3857), 4326),
VALUES (:cad,
-- cad_quarters_geom.geom = geometry(MultiPolygon,4326) NOT NULL (migration 58).
-- NSPD может вернуть одноконтурный Polygon ST_Multi coerce'ит в MultiPolygon,
-- иначе "Geometry type (Polygon) does not match column type (MultiPolygon)".
-- 38574326 transform как в upsert_quarter_geom_from_feature (Mercator).
ST_Multi(ST_Transform(ST_SetSRID(ST_GeomFromGeoJSON(:g), 3857), 4326)),
CAST(:props AS jsonb), NOW())
ON CONFLICT (cad_number) DO UPDATE
SET geom = EXCLUDED.geom,
@ -154,7 +160,14 @@ def _save_quarter(db: Session, payload: dict, cad_num: str) -> int:
def _save_building(db: Session, payload: dict, cad_num: str) -> int:
"""Парсит NSPD-response для здания и UPSERT'ит в cad_buildings."""
"""Парсит NSPD-response для здания и UPSERT'ит в cad_buildings (wide schema, migration 92).
cad_buildings (migration 92) широкая схема:
- quarter_cad_number (НЕ quarter_cad_num),
- floors INT (NSPD может отдать "5-7" _safe_int NULL, иначе InvalidTextRepresentation),
- source TEXT NOT NULL (нет дефолта) обязателен,
- geom GEOMETRY(Geometry, 4326) полиморфный, ST_Multi НЕ нужен (mirror upsert_building).
"""
feats = (payload.get("data") or {}).get("features") or []
if not feats:
return 0
@ -167,40 +180,42 @@ def _save_building(db: Session, payload: dict, cad_num: str) -> int:
# Только этого здания (NSPD может вернуть здания всего квартала)
if opts.get("cad_num") != cad_num:
continue
import json as _json
quarter_cad = opts.get("quarter_cad_number") or _derive_quarter_cad(cad_num)
db.execute(
text(
"""
INSERT INTO cad_buildings (
cad_num, quarter_cad_num, geom, purpose, building_name,
readable_address, area, floors, raw_props
cad_num, quarter_cad_number, geom, purpose, building_name,
readable_address, area, floors, raw_props, source
) VALUES (
:cad, :qcad,
ST_Transform(ST_SetSRID(ST_GeomFromGeoJSON(:g), 3857), 4326),
:purpose, :name, :addr, :area, :floors, CAST(:props AS jsonb)
:purpose, :name, :addr, :area, :floors, CAST(:props AS jsonb), :source
)
ON CONFLICT (cad_num) DO UPDATE
SET quarter_cad_num = EXCLUDED.quarter_cad_num,
SET quarter_cad_number = EXCLUDED.quarter_cad_number,
geom = EXCLUDED.geom,
purpose = EXCLUDED.purpose,
building_name = EXCLUDED.building_name,
readable_address = EXCLUDED.readable_address,
area = EXCLUDED.area,
floors = EXCLUDED.floors,
raw_props = EXCLUDED.raw_props
raw_props = EXCLUDED.raw_props,
source = EXCLUDED.source
"""
),
{
"cad": cad_num,
"qcad": opts.get("quarter_cad_number"),
"g": _json.dumps(geom_geojson, ensure_ascii=False),
"qcad": quarter_cad,
"g": json.dumps(geom_geojson, ensure_ascii=False),
"purpose": opts.get("purpose"),
"name": opts.get("building_name") or opts.get("name"),
"addr": opts.get("readable_address"),
"area": opts.get("area"),
"floors": opts.get("floors"),
"props": _json.dumps(opts, ensure_ascii=False),
"area": _safe_numeric(opts.get("area")),
"floors": _safe_int(opts.get("floors")),
"props": json.dumps(opts, ensure_ascii=False),
"source": "nspd",
},
)
n += 1
@ -226,7 +241,14 @@ def _save_parcel(db: Session, payload: dict, cad_num: str) -> int:
text(
"""
INSERT INTO cad_parcels_geom (cad_num, geom, raw_props, fetched_at)
VALUES (:cad, ST_SetSRID(ST_GeomFromGeoJSON(:g), 4326),
VALUES (:cad,
-- cad_parcels.geom = geometry(MultiPolygon,4326) (migration 93;
-- cad_parcels_geom теперь back-compat VIEW над cad_parcels).
-- NSPD/rosreestr2coord отдаёт одноконтурный Polygon ST_Multi coerce'ит
-- в MultiPolygon, иначе "Geometry type (Polygon) does not match column
-- type (MultiPolygon)". БЕЗ ST_Transform: parcel GeoJSON уже WGS84
-- (см. docstring) только ST_SetSRID.
ST_Multi(ST_SetSRID(ST_GeomFromGeoJSON(:g), 4326)),
CAST(:props AS jsonb), NOW())
ON CONFLICT (cad_num) DO UPDATE
SET geom = EXCLUDED.geom,

View file

@ -353,6 +353,70 @@ def test_upsert_features_quarter_stats_not_counted() -> None:
assert counts["skipped"] == 0
def _db_with_real_savepoints() -> tuple[MagicMock, list[Any]]:
"""Mock Session где begin_nested() — настоящий context manager (для SAVEPOINT-теста).
Возвращает (db, savepoint_calls) savepoint_calls фиксирует каждый __enter__,
т.е. сколько раз был открыт per-feature SAVEPOINT.
"""
savepoint_calls: list[Any] = []
db = MagicMock()
def _begin_nested() -> Any:
cm = MagicMock()
cm.__enter__ = MagicMock(side_effect=lambda: savepoint_calls.append(1))
cm.__exit__ = MagicMock(return_value=False)
return cm
db.begin_nested = MagicMock(side_effect=_begin_nested)
return db, savepoint_calls
def test_upsert_features_wraps_each_row_in_savepoint() -> None:
"""backend.md SAVEPOINT rule: upsert_features открывает begin_nested() per feature."""
parcel = _make_parcel_feature("66:41:0303161:1")
building = _make_building_feature("66:41:0303161:1:1")
db, savepoint_calls = _db_with_real_savepoints()
with (
patch("app.services.cadastre.bulk_harvest.upsert_parcel"),
patch("app.services.cadastre.bulk_harvest.upsert_building"),
):
upsert_features(db, [parcel, building], source="search")
# По одному SAVEPOINT на каждый из 2 features
assert db.begin_nested.call_count == 2
assert len(savepoint_calls) == 2
def test_upsert_features_one_bad_feature_does_not_abort_others() -> None:
"""Один битый feature (raise в upsert_*) НЕ роняет весь snapshot — остальные сохраняются,
counts остаются точными (инкремент только при успехе).
Раньше: без per-row begin_nested один битый feature отравлял всю tx квартала
все features rollback'ились (а счётчики уже инкрементированы = inconsistent).
"""
parcel_ok = _make_parcel_feature("66:41:0303161:1")
building_bad = _make_building_feature("66:41:0303161:1:1")
parcel_ok2 = _make_parcel_feature("66:41:0303161:2")
db, _savepoint_calls = _db_with_real_savepoints()
with (
patch("app.services.cadastre.bulk_harvest.upsert_parcel") as mock_parcel,
patch("app.services.cadastre.bulk_harvest.upsert_building") as mock_building,
):
# Здание падает (например type mismatch / NOT NULL), парцелы — ок
mock_building.side_effect = Exception("geom type mismatch / NOT NULL violation")
counts = upsert_features(db, [parcel_ok, building_bad, parcel_ok2], source="search")
# Оба парцела обработаны несмотря на падение здания между ними
assert mock_parcel.call_count == 2
mock_building.assert_called_once()
# Счётчики точные: 2 успешных парцела, 0 зданий (битое не считается)
assert counts["parcels"] == 2
assert counts["buildings"] == 0
# ── upsert_parcel SQL params ─────────────────────────────────────────────────

View file

@ -0,0 +1,294 @@
"""Тесты для backend/app/workers/tasks/nspd_geo.py — on-demand НСПД-ингест.
Покрывает 3 prod-confirmed бага «пользователь вводит новый/произвольный реальный участок»:
- Finding 1: _save_parcel / _save_quarter должны ST_Multi() geom (колонки MultiPolygon,
migration 93 / 58) иначе "Geometry type (Polygon) does not match column type (MultiPolygon)".
- Finding 2: _save_building пишет в широкую схему cad_buildings (migration 92):
quarter_cad_number (не quarter_cad_num), source NOT NULL, floors через _safe_int.
Mock DB через unittest.mock без реального PostgreSQL / NSPD сети.
"""
from __future__ import annotations
from typing import Any
from unittest.mock import MagicMock
from app.workers.tasks.nspd_geo import _save_building, _save_parcel, _save_quarter
# ── Фикстуры payload (форма ответа _persist_target: {"data": {"features": [...]}}) ──
def _parcel_payload(cad_num: str = "66:41:0610029:83") -> dict[str, Any]:
"""NSPD-ответ для ЗУ (одноконтурный Polygon — триггер бага type-mismatch)."""
return {
"data": {
"features": [
{
"type": "Feature",
"geometry": {
"type": "Polygon",
"coordinates": [
[
[60.6, 56.8],
[60.61, 56.8],
[60.61, 56.81],
[60.6, 56.81],
[60.6, 56.8],
]
],
},
"properties": {"options": {"cad_num": cad_num}},
}
]
}
}
def _quarter_payload(cad_num: str = "66:41:0610029") -> dict[str, Any]:
"""NSPD-ответ для квартала (одноконтурный Polygon в EPSG:3857)."""
return {
"data": {
"features": [
{
"type": "Feature",
"geometry": {
"type": "Polygon",
"coordinates": [
[
[6090000.0, 7590000.0],
[6090200.0, 7590000.0],
[6090200.0, 7590200.0],
[6090000.0, 7590200.0],
[6090000.0, 7590000.0],
]
],
},
"properties": {"options": {}},
}
]
}
}
def _building_payload(
cad_num: str = "66:41:0610029:83",
floors: Any = "5-7",
quarter: str | None = "66:41:0610029",
) -> dict[str, Any]:
"""NSPD-ответ для здания. floors по умолчанию range-строка "5-7" (триггер бага floors INT)."""
opts: dict[str, Any] = {
"cad_num": cad_num,
"purpose": "Жилое",
"building_name": "Тестовое здание",
"readable_address": "г Екатеринбург, ул Тестовая, д 1",
"area": "1234.56",
"floors": floors,
}
if quarter is not None:
opts["quarter_cad_number"] = quarter
return {
"data": {
"features": [
{
"type": "Feature",
"geometry": {
"type": "Polygon",
"coordinates": [
[
[6090000.0, 7590000.0],
[6090050.0, 7590000.0],
[6090050.0, 7590050.0],
[6090000.0, 7590050.0],
[6090000.0, 7590000.0],
]
],
},
"properties": {"options": opts},
}
]
}
}
def _capturing_db() -> tuple[MagicMock, list[tuple[str, Any]]]:
"""Mock Session, фиксирующий (sql_text, params) каждого db.execute."""
captured: list[tuple[str, Any]] = []
db = MagicMock()
db.execute = lambda stmt, params=None: captured.append((str(stmt), params)) or MagicMock()
return db, captured
def _strip_sql_comments(sql: str) -> str:
"""Убрать `-- ...` комментарии — чтобы assert'ить ИСПОЛНЯЕМЫЙ SQL, не текст комментов.
Комментарии легитимно упоминают ST_Transform/ST_Multi, поэтому проверяем выражения
только в реальном SQL.
"""
lines = []
for line in sql.splitlines():
idx = line.find("--")
lines.append(line if idx == -1 else line[:idx])
return "\n".join(lines)
# ── Finding 1: ST_Multi для MultiPolygon-колонок ─────────────────────────────
def test_save_parcel_sql_wraps_geom_in_st_multi() -> None:
"""cad_parcels.geom = MultiPolygon (migration 93) → SQL должен ST_Multi() обернуть geom.
Иначе одноконтурный Polygon из rosreestr2coord "Geometry type (Polygon) does not
match column type (MultiPolygon)" → on-demand ингест участка крашится.
"""
db, captured = _capturing_db()
n = _save_parcel(db, _parcel_payload(), "66:41:0610029:83")
assert n == 1
assert len(captured) == 1
sql, _params = captured[0]
assert "ST_Multi(" in sql, "SQL _save_parcel должен содержать ST_Multi для MultiPolygon schema"
def test_save_parcel_sql_no_transform_parcel_is_4326() -> None:
"""parcel GeoJSON уже WGS84 (rosreestr2coord) — только ST_SetSRID(...,4326), без ST_Transform.
ST_Transform 38574326 на 4326-геометрии схлопнул бы участок в точку у (0,0).
"""
db, captured = _capturing_db()
_save_parcel(db, _parcel_payload(), "66:41:0610029:83")
sql, _params = captured[0]
executable = _strip_sql_comments(sql)
assert "ST_Transform" not in executable, "parcel уже 4326 — ST_Transform схлопнет в точку"
assert "ST_SetSRID(ST_GeomFromGeoJSON(:g), 4326)" in executable
def test_save_quarter_sql_wraps_geom_in_st_multi() -> None:
"""cad_quarters_geom.geom = MultiPolygon NOT NULL (migration 58) → ST_Multi() обязателен."""
db, captured = _capturing_db()
n = _save_quarter(db, _quarter_payload(), "66:41:0610029")
assert n == 1
assert len(captured) == 1
sql, _params = captured[0]
assert "ST_Multi(" in sql, "SQL _save_quarter должен содержать ST_Multi для MultiPolygon schema"
def test_save_quarter_sql_keeps_3857_to_4326_transform() -> None:
"""quarter GeoJSON в Mercator (3857) → ST_Transform к 4326 сохраняется (mirror bulk path)."""
db, captured = _capturing_db()
_save_quarter(db, _quarter_payload(), "66:41:0610029")
sql, _params = captured[0]
assert "ST_Transform(ST_SetSRID(ST_GeomFromGeoJSON(:g), 3857), 4326)" in sql
def test_save_parcel_empty_features_returns_zero() -> None:
"""Пустой features → 0, без db.execute."""
db, captured = _capturing_db()
assert _save_parcel(db, {"data": {"features": []}}, "66:41:0610029:83") == 0
assert captured == []
# ── Finding 2: _save_building → широкая схема cad_buildings (migration 92) ─────
def test_save_building_uses_quarter_cad_number_column() -> None:
"""migration 92: колонка quarter_cad_number (НЕ quarter_cad_num) → иначе UndefinedColumn."""
db, captured = _capturing_db()
n = _save_building(db, _building_payload(), "66:41:0610029:83")
assert n == 1
sql, params = captured[0]
assert "quarter_cad_number" in sql
# старое имя колонки не должно встречаться (как отдельное слово quarter_cad_num)
assert "quarter_cad_num " not in sql
assert "quarter_cad_num," not in sql
assert "quarter_cad_num=" not in sql
assert params["qcad"] == "66:41:0610029"
def test_save_building_sets_source_not_null() -> None:
"""cad_buildings.source NOT NULL без дефолта (migration 92) → INSERT должен биндить source."""
db, captured = _capturing_db()
_save_building(db, _building_payload(), "66:41:0610029:83")
sql, params = captured[0]
assert ":source" in sql
assert params["source"] == "nspd"
def test_save_building_safe_int_floors_range_string() -> None:
"""floors INT: NSPD range "5-7" → _safe_int → None (иначе InvalidTextRepresentation).
Сырой "5-7" в floors INT крашил бы INSERT.
"""
db, captured = _capturing_db()
_save_building(db, _building_payload(floors="5-7"), "66:41:0610029:83")
_sql, params = captured[0]
assert params["floors"] is None # "5-7" не парсится → None, не сырая строка
def test_save_building_safe_int_floors_numeric() -> None:
"""floors "9" → int 9 (валидное число парсится)."""
db, captured = _capturing_db()
_save_building(db, _building_payload(floors="9"), "66:41:0610029:83")
_sql, params = captured[0]
assert params["floors"] == 9
def test_save_building_derives_quarter_when_missing() -> None:
"""Если NSPD не дал quarter_cad_number → выводим из cad_num (первые 3 сегмента)."""
db, captured = _capturing_db()
_save_building(db, _building_payload(quarter=None), "66:41:0610029:83")
_sql, params = captured[0]
assert params["qcad"] == "66:41:0610029"
def test_save_building_geom_transform_no_st_multi() -> None:
"""cad_buildings.geom = GEOMETRY(Geometry, 4326) полиморфный → ST_Multi НЕ нужен,
transform 38574326 сохраняется (mirror upsert_building)."""
db, captured = _capturing_db()
_save_building(db, _building_payload(), "66:41:0610029:83")
sql, _params = captured[0]
assert "ST_Transform(ST_SetSRID(ST_GeomFromGeoJSON(:g), 3857), 4326)" in sql
assert "ST_Multi(" not in sql # полиморфная колонка — coerce не нужен
def test_save_building_skips_feature_with_other_cad_num() -> None:
"""NSPD может вернуть здания всего квартала — берём только запрошенный cad_num."""
db, captured = _capturing_db()
payload = _building_payload(cad_num="66:41:0610029:99") # чужой cad_num
n = _save_building(db, payload, "66:41:0610029:83")
assert n == 0
assert captured == []
def test_save_building_area_safe_numeric() -> None:
"""area NUMERIC: строковое "1234.56" → float через _safe_numeric."""
db, captured = _capturing_db()
_save_building(db, _building_payload(), "66:41:0610029:83")
_sql, params = captured[0]
assert params["area"] == 1234.56
# ── psycopg v3 CAST-trap guard (backend.md) ─────────────────────────────────
def test_save_sql_uses_cast_not_double_colon() -> None:
"""SQL _save_* использует CAST(:x AS jsonb), не :x::jsonb (SQLAlchemy psycopg v3 trap)."""
db, captured = _capturing_db()
_save_parcel(db, _parcel_payload(), "66:41:0610029:83")
_save_quarter(db, _quarter_payload(), "66:41:0610029")
_save_building(db, _building_payload(), "66:41:0610029:83")
for sql, _params in captured:
assert "::jsonb" not in sql, f"SQL содержит ::jsonb — нужен CAST(:x AS jsonb): {sql[:80]}"
assert "CAST(:props AS jsonb)" in sql