fix(scrapers): address PR #98 auto-review minor feedback

Backend (nspd_client.py):
1. (line 100+) bbox_around_point_m signature split на multi-line.
2. (lazy imports) import json, time → module-level.
3. (duplicate SSL ctx) Reuse _SSL_CTX из nspd_lite через explicit import.
4. (shape defense) list_layers чекает {"data": [...]} wrapper + non-dict/list garbage → warning + empty.
5. (cad_num docstring) search_by_cad документирует формат + ссылку на validate_cad_format.

Tests: +5 mock тестов (19/19 passed):
- get_feature_info URL builder + parsing
- get_features_in_bbox bbox propagation
- list_layers_walks_tree_response
- list_layers_handles_data_wrapper
- list_layers_handles_garbage_response

Backend (parcels.py): docstring '~25с' → '~15с'.

Per auto-review on be14388.
This commit is contained in:
lekss361 2026-05-12 09:45:45 +03:00
parent be143880d0
commit 6ce73221a9
3 changed files with 156 additions and 9 deletions

View file

@ -1020,7 +1020,7 @@ def analyze_parcel(
Issue #93 — Graceful fallback при отсутствии geometry:
- Не возвращаем 404 сразу. Вместо: enqueue NSPD on-demand fetch, ждём
inline до _INLINE_FETCH_WAIT_S (~25с). Если за это время геометрия
inline до _INLINE_FETCH_WAIT_S (~15с). Если за это время геометрия
появилась в БД продолжаем analyze (fast path).
- Иначе 202 Accepted + {status, job_id, eta_seconds} для polling
через GET /fetch-status.

View file

@ -31,9 +31,10 @@ WMS endpoints (per #94 issue body, TIER 1-6 каталог слоёв):
from __future__ import annotations
import json
import logging
import math
import ssl
import time
import urllib.error
import urllib.parse
import urllib.request
@ -41,6 +42,7 @@ from dataclasses import dataclass
from typing import Any
from app.services.scrapers.nspd_lite import (
_SSL_CTX,
HEADERS,
NspdLiteError,
NspdLiteWafError,
@ -109,8 +111,8 @@ LAYERS: dict[str, int] = {
# Default rate limit (мс между запросами) — баланс между скоростью и WAF
DEFAULT_RATE_MS = 600
# SSL context — reuse from nspd_lite pattern (WAF-compatible)
_SSL_CTX = ssl._create_unverified_context()
# SSL context — reuse `_SSL_CTX` из nspd_lite (one source of truth для WAF-
# compatible TLS settings; нет дублирования если что-то поменяется).
# ── Typed responses ──────────────────────────────────────────────────────────
@ -178,9 +180,6 @@ def _http_get_json(url: str, *, timeout: int = 15, rate_ms: int = 0) -> Any:
layers-tree). Reuse HEADERS + SSL context из nspd_lite общая защита от
WAF блокировки по TLS fingerprint.
"""
import json
import time
if rate_ms > 0:
time.sleep(rate_ms / 1000.0)
req = urllib.request.Request(url, headers=HEADERS)
@ -212,7 +211,11 @@ def lonlat_to_3857(lon: float, lat: float) -> tuple[float, float]:
return (x, y)
def bbox_around_point_m(lon: float, lat: float, buffer_m: int) -> tuple[float, float, float, float]:
def bbox_around_point_m(
lon: float,
lat: float,
buffer_m: int,
) -> tuple[float, float, float, float]:
"""Микро-bbox вокруг точки в EPSG:3857 — для WMS GetFeatureInfo.
Returns (xmin, ymin, xmax, ymax). buffer_m по m в каждую сторону.
@ -241,7 +244,15 @@ class NSPDClient:
"""REST поиск /v2/search/geoportal — 1 запрос = geometry + ВРИ + кост.
Args:
cad_num: '66:41:0204016' (квартал) или '66:41:0204016:10' (участок).
cad_num: канонический формат `NN:NN:NNNNNN[:NN[:NN]]`, цифры через `:`.
Примеры:
- 3-сегментный квартал: `'66:41:0204016'`
- 4-сегментный участок: `'66:41:0204016:10'`
- 5-сегментное здание: `'66:41:0204016:10:1'`
Спецсимволы / пробелы корректно URL-encode'нутся (нет SQL/URL
injection), но НСПД молча вернёт empty result. Caller обязан
сам валидировать формат если нужна defensive обработка (см.
`app.services.site_finder.cadastre_fetch.validate_cad_format`).
thematic_id: 1 ЗУ / 2 квартал / 5 здание / 7 территориальная зона.
Default 1 соответствует use-case G2 ВРИ.
@ -373,6 +384,18 @@ class NSPDClient:
"""
url = f"{NSPD_LAYERS_TREE}?themeId={theme_id}"
data = _http_get_json(url, timeout=self.timeout, rate_ms=self.rate_ms)
# Shape defense: NSPD иногда оборачивает в {"data": [...]} — паттерн уже
# ловили в Bug_Nspd_Geo_Str_Object_No_Get_Fixed. Без проверки walker
# молча вернул бы [], а caller увидел бы пустой catalog.
if isinstance(data, dict) and "data" in data and not data.get("children"):
data = data.get("data") or []
if not isinstance(data, list | dict):
logger.warning(
"list_layers: unexpected NSPD response type %s for theme_id=%s",
type(data).__name__,
theme_id,
)
return []
layers: list[NSPDLayer] = []
# Tree response: nested {"children": [...], "id": N, "title": "..."}
# Используем walker по subtree чтобы вытащить все leaf layers

View file

@ -195,3 +195,127 @@ def test_nspd_layer_dataclass() -> None:
# Frozen
with pytest.raises(AttributeError):
layer.title = "x" # type: ignore[misc]
# ── Mock tests for WMS/layers methods (no network) ──────────────────────────
def test_get_feature_info_parses_response(monkeypatch: pytest.MonkeyPatch) -> None:
"""get_feature_info вызывает _http_get_json с правильным URL + парсит features."""
captured_urls: list[str] = []
def fake_http(url: str, **kwargs: Any) -> dict[str, Any]:
captured_urls.append(url)
return {
"type": "FeatureCollection",
"features": [
{
"id": "tz-1",
"geometry": {"type": "Polygon", "coordinates": []},
"properties": {"zone_code": "Ж-1"},
}
],
}
monkeypatch.setattr("app.services.scrapers.nspd_client._http_get_json", fake_http)
client = NSPDClient()
feats = client.get_feature_info(875838, lon=60.605, lat=56.838, buffer_m=100)
assert len(feats) == 1
assert feats[0].properties["zone_code"] == "Ж-1"
# Verify URL structure
assert len(captured_urls) == 1
url = captured_urls[0]
assert "/api/aeggis/v4/875838/wms?" in url
assert "REQUEST=GetFeatureInfo" in url
assert "INFO_FORMAT=application%2Fjson" in url
assert "STYLES=" in url # required per NSPD WAF
assert "CRS=EPSG%3A3857" in url
# bbox should reflect 100m buffer (200m side)
assert "BBOX=" in url
def test_get_features_in_bbox_url_builds_correctly(monkeypatch: pytest.MonkeyPatch) -> None:
"""get_features_in_bbox использует переданный bbox без преобразования."""
captured_urls: list[str] = []
def fake_http(url: str, **kwargs: Any) -> dict[str, Any]:
captured_urls.append(url)
return {"features": []}
monkeypatch.setattr("app.services.scrapers.nspd_client._http_get_json", fake_http)
client = NSPDClient()
bbox = (6700000.0, 7700000.0, 6800000.0, 7800000.0) # ЕКБ region in 3857
feats = client.get_features_in_bbox(875838, bbox)
assert feats == []
assert len(captured_urls) == 1
url = captured_urls[0]
# bbox passed directly into BBOX param (URL-encoded comma = %2C)
assert "BBOX=6700000.0%2C7700000.0%2C6800000.0%2C7800000.0" in url
# Large width/height для bulk fetch
assert "WIDTH=4096" in url
assert "HEIGHT=4096" in url
# I/J pointer в центр виртуального тайла
assert "I=2048" in url
assert "J=2048" in url
def test_list_layers_walks_tree_response(monkeypatch: pytest.MonkeyPatch) -> None:
"""list_layers вызывает /layers-theme-tree и распаковывает в NSPDLayer."""
fake_tree = [
{
"id": 1,
"title": "PKK тема",
"children": [
{
"id": 875838,
"title": "Территориальные зоны",
"layerType": "wms",
"children": [],
},
{
"id": 37577,
"title": "ЗОУИТ ОКН",
"layerType": "wms",
"children": [],
},
],
}
]
monkeypatch.setattr(
"app.services.scrapers.nspd_client._http_get_json",
lambda url, **kw: fake_tree,
)
client = NSPDClient()
layers = client.list_layers(theme_id=1)
layer_ids = [layer.layer_id for layer in layers]
# Root + 2 children = 3 leaves
assert 875838 in layer_ids
assert 37577 in layer_ids
# Sample layer metadata preserved
tz_layer = next(layer for layer in layers if layer.layer_id == 875838)
assert tz_layer.title == "Территориальные зоны"
assert tz_layer.layer_type == "wms"
def test_list_layers_handles_data_wrapper(monkeypatch: pytest.MonkeyPatch) -> None:
"""Shape defense — NSPD иногда оборачивает в {"data": [...]}."""
fake_wrapped = {
"data": [{"id": 36048, "title": "ЗУ", "children": []}],
"meta": {"count": 1},
}
monkeypatch.setattr(
"app.services.scrapers.nspd_client._http_get_json",
lambda url, **kw: fake_wrapped,
)
layers = NSPDClient().list_layers(theme_id=1)
assert len(layers) >= 1
assert any(layer.layer_id == 36048 for layer in layers)
def test_list_layers_handles_garbage_response(monkeypatch: pytest.MonkeyPatch) -> None:
"""Не dict/list response → empty list + warning log (no crash)."""
monkeypatch.setattr(
"app.services.scrapers.nspd_client._http_get_json",
lambda url, **kw: "garbage string",
)
assert NSPDClient().list_layers(theme_id=1) == []