736 lines
30 KiB
Python
736 lines
30 KiB
Python
"""Geocoder service — address → lat/lon.
|
||
|
||
Стратегия:
|
||
- Cache lookup в `geocode_cache` (Postgres) — TTL 90 дней
|
||
- Cache miss → Yandex Geocoder (если есть key) → fallback Nominatim
|
||
- Результат сохраняется в кэш для последующих вызовов
|
||
|
||
Используется в:
|
||
- /api/v1/trade-in/estimate (вход — адрес от пользователя)
|
||
- /api/v1/geocode/lookup (debug endpoint)
|
||
- scraper jobs (когда нужно по адресу определить координаты)
|
||
"""
|
||
|
||
from __future__ import annotations
|
||
|
||
import asyncio
|
||
import logging
|
||
from dataclasses import dataclass
|
||
from typing import Literal
|
||
|
||
import httpx
|
||
from sqlalchemy import text
|
||
from sqlalchemy.orm import Session
|
||
from tenacity import retry, stop_after_attempt, wait_exponential
|
||
|
||
from app.core.config import settings
|
||
|
||
logger = logging.getLogger(__name__)
|
||
|
||
|
||
# ── Result type ──────────────────────────────────────────────────────────────
|
||
@dataclass(frozen=True, slots=True)
|
||
class GeocodeResult:
|
||
lat: float
|
||
lon: float
|
||
full_address: str
|
||
provider: Literal["nominatim", "yandex", "cache"]
|
||
confidence: Literal["exact", "approximate", "locality"] = "approximate"
|
||
|
||
|
||
# ── Address normalisation ───────────────────────────────────────────────────
|
||
def normalize_address(address: str) -> str:
|
||
"""Нормализация для cache lookup: lowercase + trim + collapse whitespace.
|
||
|
||
« Ул. МАЛЫШЕВА, 30 » → «ул. малышева, 30»
|
||
"""
|
||
return " ".join(address.lower().strip().split())
|
||
|
||
|
||
# Согласные, которые часто пишут с одной буквой вместо двух (RU typos).
|
||
_DOUBLE_CONSONANTS = "лнмссккттпп"
|
||
|
||
|
||
def _typo_variants(query: str, limit: int = 6) -> list[str]:
|
||
"""Генерация вариантов с удвоением согласных — для случая когда пользователь
|
||
написал «Цвилинга» вместо «Цвиллинга», «Толстова» вместо «Толстого» итд.
|
||
|
||
Каждая позиция где есть одинокая согласная — кандидат на удвоение.
|
||
Возвращаем до `limit` вариантов в порядке вероятности (ближе к началу слова → выше).
|
||
"""
|
||
if len(query) < 3:
|
||
return []
|
||
variants: list[str] = []
|
||
chars = list(query)
|
||
for i in range(1, len(chars) - 1):
|
||
c = chars[i].lower()
|
||
if c not in _DOUBLE_CONSONANTS:
|
||
continue
|
||
prev_c = chars[i - 1].lower()
|
||
next_c = chars[i + 1].lower()
|
||
# Удваиваем только если соседи — гласные/'ь'/'ъ' (характерно для русских типо)
|
||
if prev_c in "аеёиоуыэюяьъ" and next_c in "аеёиоуыэюяьъ":
|
||
variant = query[:i + 1] + chars[i] + query[i + 1:]
|
||
if variant != query and variant not in variants:
|
||
variants.append(variant)
|
||
if len(variants) >= limit:
|
||
break
|
||
return variants
|
||
|
||
|
||
# ── Cache ────────────────────────────────────────────────────────────────────
|
||
def _cache_get(db: Session, address_norm: str) -> GeocodeResult | None:
|
||
row = db.execute(
|
||
text(
|
||
"""
|
||
SELECT lat, lon, full_address, provider, confidence
|
||
FROM geocode_cache
|
||
WHERE address_normalized = :addr
|
||
AND expires_at > NOW()
|
||
"""
|
||
),
|
||
{"addr": address_norm},
|
||
).fetchone()
|
||
if row is None:
|
||
return None
|
||
return GeocodeResult(
|
||
lat=row.lat,
|
||
lon=row.lon,
|
||
full_address=row.full_address,
|
||
provider="cache",
|
||
confidence=row.confidence or "approximate",
|
||
)
|
||
|
||
|
||
def _cache_put(db: Session, address_norm: str, result: GeocodeResult) -> None:
|
||
db.execute(
|
||
text(
|
||
"""
|
||
INSERT INTO geocode_cache
|
||
(address_normalized, lat, lon, full_address, provider, confidence)
|
||
VALUES (:addr, :lat, :lon, :full, :provider, :conf)
|
||
ON CONFLICT (address_normalized) DO UPDATE
|
||
SET lat = EXCLUDED.lat,
|
||
lon = EXCLUDED.lon,
|
||
full_address = EXCLUDED.full_address,
|
||
provider = EXCLUDED.provider,
|
||
confidence = EXCLUDED.confidence,
|
||
created_at = NOW(),
|
||
expires_at = NOW() + interval '90 days'
|
||
"""
|
||
),
|
||
{
|
||
"addr": address_norm,
|
||
"lat": result.lat,
|
||
"lon": result.lon,
|
||
"full": result.full_address,
|
||
"provider": result.provider,
|
||
"conf": result.confidence,
|
||
},
|
||
)
|
||
db.commit()
|
||
|
||
|
||
# ── Provider: Nominatim (OSM, без ключа) ────────────────────────────────────
|
||
async def _nominatim_query(client: httpx.AsyncClient, address: str) -> dict | None:
|
||
"""Single Nominatim search. Возвращает первый item или None.
|
||
|
||
ВАЖНО: фильтруем результаты по ЕКБ bbox прямо тут, чтобы при опечатках
|
||
не возвращать Пермский край / Челябинск.
|
||
"""
|
||
response = await client.get(
|
||
"https://nominatim.openstreetmap.org/search",
|
||
params={
|
||
"q": address,
|
||
"format": "json",
|
||
"limit": "3",
|
||
"countrycodes": "ru",
|
||
"addressdetails": "1",
|
||
"viewbox": EKB_BBOX["viewbox"],
|
||
"bounded": "1", # строго в ЕКБ bbox
|
||
},
|
||
)
|
||
response.raise_for_status()
|
||
data = response.json()
|
||
for item in data:
|
||
try:
|
||
lat_f = float(item["lat"])
|
||
lon_f = float(item["lon"])
|
||
if 60.40 <= lon_f <= 60.85 and 56.65 <= lat_f <= 56.95:
|
||
return item
|
||
except Exception:
|
||
continue
|
||
return None
|
||
|
||
|
||
@retry(stop=stop_after_attempt(3), wait=wait_exponential(multiplier=1, min=1, max=8))
|
||
async def _nominatim_lookup(address: str) -> GeocodeResult | None:
|
||
"""OSM Nominatim — бесплатно, без ключа, 1 req/sec policy.
|
||
|
||
Бан-policy: User-Agent с email обязателен.
|
||
Tier 1: bounded ЕКБ на оригинальный адрес.
|
||
Tier 2: bounded ЕКБ на typo-варианты (Цвилинга → Цвиллинга).
|
||
"""
|
||
headers = {
|
||
"User-Agent": f"TradeInMVP/0.1 (contact: {settings.contact_email})",
|
||
"Accept": "application/json",
|
||
"Accept-Language": "ru,en;q=0.8",
|
||
"Referer": "https://tradein-mvp.local/",
|
||
}
|
||
async with httpx.AsyncClient(timeout=10.0, headers=headers) as client:
|
||
# Tier 1: оригинал
|
||
item = await _nominatim_query(client, address)
|
||
|
||
# Tier 2: typo-variants
|
||
if item is None:
|
||
for variant in _typo_variants(address, limit=4):
|
||
await asyncio.sleep(1.0) # Nominatim 1 req/sec policy
|
||
item = await _nominatim_query(client, variant)
|
||
if item is not None:
|
||
logger.info("nominatim typo-fixed: %s → %s", address, variant)
|
||
break
|
||
|
||
if item is None:
|
||
return None
|
||
|
||
confidence = "exact" if item.get("class") == "building" else "approximate"
|
||
return GeocodeResult(
|
||
lat=float(item["lat"]),
|
||
lon=float(item["lon"]),
|
||
full_address=item.get("display_name", address),
|
||
provider="nominatim",
|
||
confidence=confidence,
|
||
)
|
||
|
||
|
||
# ── Provider: Yandex Geocoder (требует key, лучшее покрытие РФ) ─────────────
|
||
@retry(stop=stop_after_attempt(3), wait=wait_exponential(multiplier=1, min=1, max=8))
|
||
async def _yandex_lookup(address: str, api_key: str) -> GeocodeResult | None:
|
||
"""Yandex Geocoder — 25K req/day free для самопод, лучше РФ.
|
||
|
||
Docs: https://yandex.ru/dev/maps/geocoder/doc/desc/concepts/input_params.html
|
||
|
||
Запрашиваем с ll+spn (центр ЕКБ) для приоритизации местных результатов,
|
||
но БЕЗ rspn — чтобы fuzzy matching работал при опечатках.
|
||
"""
|
||
# Не запихиваем "Екатеринбург" если оно уже есть в адресе (типичный кейс из suggest)
|
||
geocode_query = address if "екатеринбург" in address.lower() else f"Екатеринбург, {address}"
|
||
async with httpx.AsyncClient(timeout=10.0) as client:
|
||
response = await client.get(
|
||
"https://geocode-maps.yandex.ru/1.x/",
|
||
params={
|
||
"apikey": api_key,
|
||
"geocode": geocode_query,
|
||
"format": "json",
|
||
"results": 5, # берем top-5, отфильтруем по ЕКБ bbox ниже
|
||
"lang": "ru_RU",
|
||
"ll": EKB_BBOX["ll"],
|
||
"spn": EKB_BBOX["spn"],
|
||
},
|
||
)
|
||
response.raise_for_status()
|
||
data = response.json()
|
||
|
||
members = data.get("response", {}).get("GeoObjectCollection", {}).get("featureMember", [])
|
||
if not members:
|
||
return None
|
||
|
||
# Фильтруем top-5 по ЕКБ bbox — игнорируем Челябинск/Уфу/Москву при опечатке
|
||
best = None
|
||
for m in members:
|
||
obj = m.get("GeoObject", {})
|
||
try:
|
||
lon_str, lat_str = obj["Point"]["pos"].split()
|
||
lat_f, lon_f = float(lat_str), float(lon_str)
|
||
if 60.40 <= lon_f <= 60.85 and 56.65 <= lat_f <= 56.95:
|
||
best = obj
|
||
break
|
||
except Exception:
|
||
continue
|
||
|
||
if best is None:
|
||
# Никто из top-5 не попал в ЕКБ → берем первый «как есть» (вне ЕКБ — но хоть что-то)
|
||
best = members[0]["GeoObject"]
|
||
|
||
lon_str, lat_str = best["Point"]["pos"].split()
|
||
precision_raw = (
|
||
best.get("metaDataProperty", {})
|
||
.get("GeocoderMetaData", {})
|
||
.get("precision", "other")
|
||
)
|
||
confidence_map = {
|
||
"exact": "exact",
|
||
"number": "exact",
|
||
"near": "approximate",
|
||
"range": "approximate",
|
||
"street": "approximate",
|
||
}
|
||
return GeocodeResult(
|
||
lat=float(lat_str),
|
||
lon=float(lon_str),
|
||
full_address=best.get("metaDataProperty", {})
|
||
.get("GeocoderMetaData", {})
|
||
.get("text", address),
|
||
provider="yandex",
|
||
confidence=confidence_map.get(precision_raw, "approximate"),
|
||
)
|
||
|
||
|
||
# ── Suggest (автокомплит) ───────────────────────────────────────────────────
|
||
# ЕКБ bounding box (приблизительно): юг 56.65, запад 60.40, север 56.95, восток 60.85
|
||
EKB_BBOX = {
|
||
"viewbox": "60.40,56.95,60.85,56.65", # Nominatim format: lon1,lat1,lon2,lat2 (NW,SE)
|
||
"ll": "60.605,56.838", # Yandex center (lon,lat)
|
||
"spn": "0.45,0.30", # Yandex span (lon,lat)
|
||
}
|
||
|
||
|
||
@dataclass(frozen=True, slots=True)
|
||
class GeocodeSuggestion:
|
||
label: str # формат "Малышева 30, Октябрьский район"
|
||
full_address: str # полный из геокодера
|
||
lat: float
|
||
lon: float
|
||
kind: str # 'house' / 'street' / 'locality'
|
||
|
||
|
||
def _parse_yandex_members(members: list[dict]) -> list[GeocodeSuggestion]:
|
||
"""Yandex geocode_members → list[GeocodeSuggestion]. Чистим описание от мусора."""
|
||
out: list[GeocodeSuggestion] = []
|
||
for m in members:
|
||
obj = m.get("GeoObject", {})
|
||
try:
|
||
lon_str, lat_str = obj["Point"]["pos"].split()
|
||
meta = obj.get("metaDataProperty", {}).get("GeocoderMetaData", {})
|
||
kind = meta.get("kind", "other")
|
||
full = meta.get("text", obj.get("name", ""))
|
||
name = obj.get("name", full)
|
||
desc = obj.get("description", "")
|
||
desc_parts = [
|
||
p.strip() for p in desc.split(",")
|
||
if p.strip() and p.strip() not in {"Россия", "Свердловская область"}
|
||
]
|
||
label = name if not desc_parts else f"{name} · {', '.join(desc_parts)}"
|
||
out.append(GeocodeSuggestion(
|
||
label=label, full_address=full,
|
||
lat=float(lat_str), lon=float(lon_str), kind=kind,
|
||
))
|
||
except Exception:
|
||
continue
|
||
return out
|
||
|
||
|
||
async def _yandex_geocode_request(
|
||
client: httpx.AsyncClient, api_key: str, query: str, limit: int, bounded: bool
|
||
) -> list[dict]:
|
||
"""Single Yandex Geocoder request — bounded=True → строго в ЕКБ через rspn=1."""
|
||
params: dict[str, str] = {
|
||
"apikey": api_key,
|
||
"geocode": query,
|
||
"format": "json",
|
||
"results": str(limit),
|
||
"lang": "ru_RU",
|
||
"ll": EKB_BBOX["ll"],
|
||
"spn": EKB_BBOX["spn"],
|
||
}
|
||
if bounded:
|
||
params["rspn"] = "1"
|
||
response = await client.get("https://geocode-maps.yandex.ru/1.x/", params=params)
|
||
response.raise_for_status()
|
||
data = response.json()
|
||
return data.get("response", {}).get("GeoObjectCollection", {}).get("featureMember", [])
|
||
|
||
|
||
@retry(stop=stop_after_attempt(2), wait=wait_exponential(multiplier=1, min=1, max=4))
|
||
async def _yandex_suggest(query: str, api_key: str, limit: int = 8) -> list[GeocodeSuggestion]:
|
||
"""Yandex Geocoder с авто-fallback на typo-tolerant режим.
|
||
|
||
Tier 1: bounded ЕКБ (rspn=1) на оригинальный query.
|
||
Tier 2: bounded ЕКБ на typo-variants (удвоение согласных).
|
||
Tier 3: без rspn — fuzzy по всей стране, фильтр результатов по ЕКБ bbox.
|
||
"""
|
||
async with httpx.AsyncClient(timeout=8.0) as client:
|
||
# Tier 1: strict bounded на оригинал
|
||
members = await _yandex_geocode_request(
|
||
client, api_key, f"Екатеринбург, {query}", limit, bounded=True,
|
||
)
|
||
results = _parse_yandex_members(members)
|
||
if results:
|
||
return results
|
||
|
||
# Tier 2: bounded на typo-варианты
|
||
for variant in _typo_variants(query, limit=4):
|
||
members = await _yandex_geocode_request(
|
||
client, api_key, f"Екатеринбург, {variant}", limit, bounded=True,
|
||
)
|
||
results = _parse_yandex_members(members)
|
||
if results:
|
||
return results
|
||
|
||
# Tier 3: без rspn — даём fuzzy сделать своё дело, фильтр по bbox
|
||
members = await _yandex_geocode_request(
|
||
client, api_key, f"Екатеринбург, {query}", limit, bounded=False,
|
||
)
|
||
results = _parse_yandex_members(members)
|
||
in_ekb = [
|
||
r for r in results
|
||
if 60.40 <= r.lon <= 60.85 and 56.65 <= r.lat <= 56.95
|
||
]
|
||
return in_ekb
|
||
|
||
|
||
async def _nominatim_query_multi(
|
||
client: httpx.AsyncClient, query: str, limit: int
|
||
) -> list[dict]:
|
||
"""Один Nominatim search с фильтром по ЕКБ bbox. Возвращает up to N items."""
|
||
response = await client.get(
|
||
"https://nominatim.openstreetmap.org/search",
|
||
params={
|
||
"q": query,
|
||
"format": "json",
|
||
"limit": str(limit),
|
||
"countrycodes": "ru",
|
||
"viewbox": EKB_BBOX["viewbox"],
|
||
"bounded": "1",
|
||
"addressdetails": "1",
|
||
},
|
||
)
|
||
response.raise_for_status()
|
||
data = response.json()
|
||
return data if isinstance(data, list) else []
|
||
|
||
|
||
@retry(stop=stop_after_attempt(2), wait=wait_exponential(multiplier=1, min=1, max=4))
|
||
async def _nominatim_suggest(query: str, limit: int = 8) -> list[GeocodeSuggestion]:
|
||
"""Nominatim в режиме suggest. С typo-fallback (для случаев когда Yandex недоступен)."""
|
||
headers = {
|
||
"User-Agent": f"TradeInMVP/0.1 (contact: {settings.contact_email})",
|
||
"Accept": "application/json",
|
||
"Accept-Language": "ru,en;q=0.8",
|
||
}
|
||
async with httpx.AsyncClient(timeout=8.0, headers=headers) as client:
|
||
# Tier 1: оригинальный query
|
||
data = await _nominatim_query_multi(client, f"{query}, Екатеринбург", limit)
|
||
|
||
# Tier 2: typo-варианты если оригинал пустой
|
||
if not data:
|
||
for variant in _typo_variants(query, limit=3):
|
||
await asyncio.sleep(1.0) # Nominatim 1 req/sec
|
||
data = await _nominatim_query_multi(client, f"{variant}, Екатеринбург", limit)
|
||
if data:
|
||
logger.info("nominatim suggest typo-fixed: %s → %s", query, variant)
|
||
break
|
||
|
||
out: list[GeocodeSuggestion] = []
|
||
for item in data:
|
||
display = item.get("display_name", "")
|
||
addr = item.get("address", {}) or {}
|
||
# Компактный лейбл: street + house_number / locality / district
|
||
street = addr.get("road") or addr.get("street") or ""
|
||
house = addr.get("house_number", "")
|
||
district = addr.get("suburb") or addr.get("city_district") or addr.get("borough") or ""
|
||
parts = []
|
||
if street:
|
||
parts.append(f"{street}{f', {house}' if house else ''}")
|
||
elif item.get("name"):
|
||
parts.append(item["name"])
|
||
if district:
|
||
parts.append(district)
|
||
label = " · ".join(parts) if parts else display[:80]
|
||
kind = "house" if house else ("street" if street else "locality")
|
||
out.append(GeocodeSuggestion(
|
||
label=label, full_address=display,
|
||
lat=float(item["lat"]), lon=float(item["lon"]), kind=kind,
|
||
))
|
||
return out
|
||
|
||
|
||
def _cadastral_forward_sync(
|
||
db: Session, query: str, limit: int = 5
|
||
) -> list[GeocodeSuggestion]:
|
||
"""Forward search via gendesign_cad_buildings FDW.
|
||
|
||
Uses ILIKE prefilter (FDW-pushable) + pg_trgm similarity ranking locally.
|
||
Returns up to `limit` suggestions, or [] on error / no results.
|
||
"""
|
||
try:
|
||
rows = db.execute(
|
||
text("""
|
||
SELECT cad_num, readable_address, lat, lon,
|
||
similarity(readable_address, CAST(:q AS text)) AS sim
|
||
FROM gendesign_cad_buildings
|
||
WHERE readable_address ILIKE CAST('%' || :q || '%' AS text)
|
||
ORDER BY sim DESC, length(readable_address) ASC
|
||
LIMIT CAST(:lim AS integer)
|
||
"""),
|
||
{"q": query, "lim": limit * 4},
|
||
).mappings().all()
|
||
except Exception:
|
||
logger.warning("cadastral forward sync failed for query=%r", query, exc_info=True)
|
||
return []
|
||
|
||
out: list[GeocodeSuggestion] = []
|
||
for r in rows[:limit]:
|
||
out.append(GeocodeSuggestion(
|
||
label=str(r["readable_address"]),
|
||
full_address=str(r["readable_address"]),
|
||
lat=float(r["lat"]),
|
||
lon=float(r["lon"]),
|
||
kind="house",
|
||
))
|
||
return out
|
||
|
||
|
||
def _cadastral_reverse_sync(
|
||
db: Session, lat: float, lon: float, radius_m: int = 200
|
||
) -> str | None:
|
||
"""Reverse lookup via gendesign_cad_buildings FDW.
|
||
|
||
bbox prefilter (±0.0025° lat, ±0.005° lon ≈ 280m) is FDW-pushable.
|
||
Distance calculated locally on small result set. Filters гаражи/СНТ.
|
||
Returns None on no result or error.
|
||
"""
|
||
try:
|
||
row = db.execute(
|
||
text("""
|
||
WITH candidates AS (
|
||
SELECT cad_num, readable_address, lat, lon,
|
||
111320.0 * sqrt(
|
||
pow(CAST(:lat AS double precision) - lat, 2) +
|
||
pow(
|
||
cos(radians(CAST(:lat AS double precision)))
|
||
* (CAST(:lon AS double precision) - lon), 2
|
||
)
|
||
) AS dist_m
|
||
FROM gendesign_cad_buildings
|
||
WHERE lat BETWEEN CAST(:lat AS double precision) - 0.0025
|
||
AND CAST(:lat AS double precision) + 0.0025
|
||
AND lon BETWEEN CAST(:lon AS double precision) - 0.005
|
||
AND CAST(:lon AS double precision) + 0.005
|
||
AND readable_address !~* '(гараж|снт|садовод|товарищ|уч\\.)'
|
||
)
|
||
SELECT readable_address, dist_m
|
||
FROM candidates
|
||
WHERE dist_m < CAST(:radius AS double precision)
|
||
ORDER BY dist_m ASC
|
||
LIMIT 1
|
||
"""),
|
||
{"lat": lat, "lon": lon, "radius": float(radius_m)},
|
||
).first()
|
||
except Exception:
|
||
logger.warning(
|
||
"cadastral reverse sync failed for (%.5f, %.5f)", lat, lon, exc_info=True
|
||
)
|
||
return None
|
||
|
||
if row is None:
|
||
return None
|
||
return str(row.readable_address)
|
||
|
||
|
||
async def suggest(
|
||
query: str, db: Session | None = None, limit: int = 8
|
||
) -> list[GeocodeSuggestion]:
|
||
"""Автокомплит адресов в ЕКБ. Cadastral FDW → Yandex → Nominatim → [].
|
||
|
||
db: если передан — cadastral lookup через gendesign_cad_buildings (первый tier).
|
||
Без кэша (дешёво, провайдеры толерируют автокомплит-запросы).
|
||
"""
|
||
if not query or len(query.strip()) < 2:
|
||
return []
|
||
|
||
# Tier 0: cadastral FDW (если db доступна) — самый быстрый, без внешних запросов
|
||
if db is not None:
|
||
cad_results = await asyncio.to_thread(_cadastral_forward_sync, db, query.strip(), limit)
|
||
if cad_results:
|
||
return cad_results
|
||
|
||
if settings.yandex_geocoder_api_key:
|
||
try:
|
||
results = await _yandex_suggest(query, settings.yandex_geocoder_api_key, limit)
|
||
if results:
|
||
return results
|
||
except Exception:
|
||
logger.exception("yandex suggest failed, falling back to nominatim")
|
||
|
||
try:
|
||
return await _nominatim_suggest(query, limit)
|
||
except Exception:
|
||
logger.exception("nominatim suggest failed")
|
||
return []
|
||
|
||
|
||
# ── Public API ───────────────────────────────────────────────────────────────
|
||
async def geocode(address: str, db: Session) -> GeocodeResult | None:
|
||
"""Геокодинг с кэшем. Cadastral FDW → Yandex → Nominatim → None.
|
||
|
||
Args:
|
||
address: пользовательский ввод (может быть грязным — нормализуем).
|
||
db: сессия Postgres для cache lookup/write и cadastral FDW lookup.
|
||
|
||
Returns:
|
||
GeocodeResult или None если ни один провайдер не отвечает.
|
||
"""
|
||
if not address or len(address.strip()) < 3:
|
||
return None
|
||
|
||
addr_norm = normalize_address(address)
|
||
|
||
# 1. Cache
|
||
cached = _cache_get(db, addr_norm)
|
||
if cached is not None:
|
||
logger.info("geocode cache hit: %s", addr_norm)
|
||
return cached
|
||
|
||
# 2. Cadastral FDW (прямой запрос к gendesign_cad_buildings — без внешнего API)
|
||
cad_suggestions = await asyncio.to_thread(_cadastral_forward_sync, db, address.strip(), limit=1)
|
||
if cad_suggestions:
|
||
s = cad_suggestions[0]
|
||
result = GeocodeResult(
|
||
lat=s.lat,
|
||
lon=s.lon,
|
||
full_address=s.full_address,
|
||
provider="nominatim", # treat as "local" — same confidence as nominatim
|
||
confidence="exact",
|
||
)
|
||
_cache_put(db, addr_norm, result)
|
||
logger.info(
|
||
"geocode cadastral fdw: %s → (%.5f, %.5f)", addr_norm, result.lat, result.lon
|
||
)
|
||
return result
|
||
|
||
# 3. Yandex (если есть key) с typo-fallback
|
||
if settings.yandex_geocoder_api_key:
|
||
try:
|
||
result = await _yandex_lookup(address, settings.yandex_geocoder_api_key)
|
||
# Если результат вне ЕКБ — пробуем typo-варианты
|
||
in_ekb = (
|
||
result is not None
|
||
and 60.40 <= result.lon <= 60.85
|
||
and 56.65 <= result.lat <= 56.95
|
||
)
|
||
if result is not None and in_ekb:
|
||
_cache_put(db, addr_norm, result)
|
||
logger.info("geocode yandex: %s → (%.5f, %.5f)", addr_norm, result.lat, result.lon)
|
||
return result
|
||
# Tier 2: typo-variants
|
||
for variant in _typo_variants(address, limit=4):
|
||
try:
|
||
result = await _yandex_lookup(variant, settings.yandex_geocoder_api_key)
|
||
except Exception:
|
||
continue
|
||
if result is None:
|
||
continue
|
||
if 60.40 <= result.lon <= 60.85 and 56.65 <= result.lat <= 56.95:
|
||
_cache_put(db, addr_norm, result)
|
||
logger.info(
|
||
"geocode yandex typo-fixed: %s → %s → (%.5f, %.5f)",
|
||
addr_norm, variant, result.lat, result.lon,
|
||
)
|
||
return result
|
||
except Exception:
|
||
logger.exception("yandex geocoder failed, falling back to nominatim")
|
||
|
||
# 4. Nominatim fallback
|
||
try:
|
||
result = await _nominatim_lookup(address)
|
||
if result is not None:
|
||
_cache_put(db, addr_norm, result)
|
||
logger.info(
|
||
"geocode nominatim: %s → (%.5f, %.5f)", addr_norm, result.lat, result.lon
|
||
)
|
||
# Nominatim rate-limit policy: 1 req/sec — спим после успешного запроса
|
||
await asyncio.sleep(1.0)
|
||
return result
|
||
except Exception:
|
||
logger.exception("nominatim geocoder failed")
|
||
|
||
return None
|
||
|
||
|
||
# ── Reverse: координаты → адрес (для map-picker'а) ──────────────────────────
|
||
def _format_reverse_address(addr: dict) -> str | None:
|
||
"""Собирает чистый уличный адрес из Nominatim address-объекта.
|
||
|
||
Nominatim `display_name` ведёт с названия ближайшего POI/организации
|
||
(«NataliOlympic, 2к1, Трамвайный переулок, …»). Для map-picker'а это
|
||
ломает кейс: такой адрес нельзя forward-геокодировать обратно и оценка
|
||
не находит аналогов. Берём именно улицу + дом + город.
|
||
|
||
Возвращает None если в объекте нет улицы — тогда вызывающий код
|
||
откатывается на display_name.
|
||
"""
|
||
road = (
|
||
addr.get("road")
|
||
or addr.get("pedestrian")
|
||
or addr.get("footway")
|
||
)
|
||
if not road:
|
||
return None
|
||
house = addr.get("house_number")
|
||
city = (
|
||
addr.get("city")
|
||
or addr.get("town")
|
||
or addr.get("village")
|
||
or addr.get("municipality")
|
||
)
|
||
parts: list[str] = [str(road)]
|
||
if house:
|
||
parts.append(str(house))
|
||
if city:
|
||
parts.append(str(city))
|
||
return ", ".join(parts)
|
||
|
||
|
||
@retry(stop=stop_after_attempt(3), wait=wait_exponential(multiplier=1, min=1, max=8))
|
||
async def _nominatim_reverse(lat: float, lon: float) -> str | None:
|
||
"""Nominatim /reverse — extracted for fallback usage. @retry handles transient errors."""
|
||
headers = {
|
||
"User-Agent": f"TradeInMVP/0.1 (contact: {settings.contact_email})",
|
||
"Accept": "application/json",
|
||
"Accept-Language": "ru,en;q=0.8",
|
||
}
|
||
async with httpx.AsyncClient(timeout=10.0, headers=headers) as client:
|
||
response = await client.get(
|
||
"https://nominatim.openstreetmap.org/reverse",
|
||
params={
|
||
"lat": str(lat),
|
||
"lon": str(lon),
|
||
"format": "json",
|
||
"addressdetails": "1",
|
||
"zoom": "18",
|
||
},
|
||
)
|
||
response.raise_for_status()
|
||
data = response.json()
|
||
if not isinstance(data, dict) or "error" in data:
|
||
return None
|
||
addr = data.get("address")
|
||
if isinstance(addr, dict):
|
||
street = _format_reverse_address(addr)
|
||
if street:
|
||
return street
|
||
return data.get("display_name")
|
||
|
||
|
||
async def reverse_geocode(lat: float, lon: float, db: Session | None = None) -> str | None:
|
||
"""Cadastral FDW → Nominatim fallback.
|
||
|
||
Возвращает None если ни один источник не дал адрес. Endpoint
|
||
api/v1/geocode/reverse сам выкинет 404. НЕ даёт выйти HTTPStatusError
|
||
наверх — раньше Nominatim 403 → RetryError → FastAPI 500.
|
||
|
||
db: если передан — cadastral lookup через gendesign_cad_buildings FDW (первый tier).
|
||
"""
|
||
# 1. Cadastral FDW primary (без внешнего API, возвращает жилой дом not POI)
|
||
if db is not None:
|
||
cadastral = await asyncio.to_thread(_cadastral_reverse_sync, db, lat, lon)
|
||
if cadastral:
|
||
return cadastral
|
||
|
||
# 2. Nominatim fallback (wrap to prevent 500 on ban/rate-limit)
|
||
try:
|
||
return await _nominatim_reverse(lat, lon)
|
||
except Exception:
|
||
logger.exception("nominatim reverse failed for (%.5f, %.5f)", lat, lon)
|
||
return None
|