gendesign/backend/app/services/site_finder/pzz_loader.py

"""Импорт ПЗЗ территориальных зон ЕКБ из Росреестр PKK6."""

import asyncio
import json
import logging

import httpx
from sqlalchemy import text

from app.core.db import SessionLocal

logger = logging.getLogger(__name__)

PKK6_URL = "https://pkk.rosreestr.ru/arcgis/rest/services/PKK6/ZONES/MapServer/5/query"
# bbox ЕКБ: (xmin, ymin, xmax, ymax) в WGS84
EKB_BBOX = (60.5, 56.7, 60.75, 56.95)


async def fetch_pkk6_zones() -> list[dict]:
    """Запросить ВСЕ территориальные зоны для bbox ЕКБ с пагинацией.

    PKK6 maxRecordCount = 1000 per request. Пагинируем через resultOffset.
    Rate-limit: asyncio.sleep(0.5) между страницами.
    """
    xmin, ymin, xmax, ymax = EKB_BBOX
    geometry = json.dumps(
        {
            "xmin": xmin,
            "ymin": ymin,
            "xmax": xmax,
            "ymax": ymax,
            "spatialReference": {"wkid": 4326},
        }
    )

    headers = {
        "User-Agent": "GenDesign-SiteFinder/1.0 (+https://gendsgn.ru)",
        "Accept": "application/json",
    }
    all_features: list[dict] = []
    offset = 0
    page_size = 1000

    # verify=False: Росреестр PKK6 self-signed cert. Публичный read-only API,
    # MITM risk минимален. follow_redirects=True: PKK6 возвращает 301 на
    # некоторые запросы (вероятно legacy URL redirect).
    async with httpx.AsyncClient(
        timeout=60, headers=headers, verify=False, follow_redirects=True
    ) as client:
        while True:
            params = {
                "where": "1=1",
                "geometry": geometry,
                "geometryType": "esriGeometryEnvelope",
                "inSR": "4326",
                "outSR": "4326",
                "outFields": "*",
                "f": "geojson",
                "resultOffset": offset,
                "resultRecordCount": page_size,
            }
            r = await client.get(PKK6_URL, params=params)
            r.raise_for_status()
            data = r.json()
            features = data.get("features", [])
            if not features:
                break
            all_features.extend(features)
            logger.info("PKK6 page offset=%d → %d features", offset, len(features))
            if len(features) < page_size:
                break
            offset += page_size
            await asyncio.sleep(0.5)  # вежливый rate-limit

    logger.info("PKK6: total %d zones for EKB bbox", len(all_features))
    return all_features


def sync_pzz_zones_to_db() -> dict[str, int]:
    """Загрузить зоны из PKK6 и UPSERT в pzz_zones_ekb.

    Возвращает словарь {fetched, inserted, updated, skipped}.

    Каждый UPSERT обёрнут в SAVEPOINT (db.begin_nested), чтобы ошибка
    одной записи не откатывала всю транзакцию и счётчики оставались
    согласованными с тем, что реально записано в DB.
    """
    features = asyncio.run(fetch_pkk6_zones())
    inserted = 0
    updated = 0
    skipped = 0
    db = SessionLocal()
    try:
        for f in features:
            props = f.get("properties") or {}
            geom = f.get("geometry")
            if not geom:
                skipped += 1
                continue
            rosreestr_id = props.get("OBJECTID") or props.get("objectid")
            zone_code = (
                props.get("zone_number")
                or props.get("zone_code")
                or props.get("identification_number")
                or None
            )
            zone_name = (
                props.get("description") or props.get("name") or props.get("zone_name") or None
            )
            description = props.get("description")
            try:
                with db.begin_nested():  # SAVEPOINT — откат только этой записи
                    result = db.execute(
                        text("""
                            INSERT INTO pzz_zones_ekb
                                (rosreestr_id, zone_code, zone_name, description, raw_props, geom)
                            VALUES (
                                :rid, :code, :nm, :desc, CAST(:props AS jsonb),
                                ST_Multi(ST_SetSRID(ST_GeomFromGeoJSON(:g), 4326))
                            )
                            ON CONFLICT (rosreestr_id) DO UPDATE
                               SET zone_code   = EXCLUDED.zone_code,
                                   zone_name   = EXCLUDED.zone_name,
                                   description = EXCLUDED.description,
                                   raw_props   = EXCLUDED.raw_props,
                                   geom        = EXCLUDED.geom,
                                   fetched_at  = NOW()
                            RETURNING (xmax = 0) AS is_insert
                        """),
                        {
                            "rid": rosreestr_id,
                            "code": zone_code,
                            "nm": zone_name,
                            "desc": description,
                            "props": json.dumps(props, ensure_ascii=False),
                            "g": json.dumps(geom),
                        },
                    ).scalar()
                if result:
                    inserted += 1
                else:
                    updated += 1
            except Exception as e:
                logger.warning("PZZ insert failed for rosreestr_id=%s: %s", rosreestr_id, e)
                skipped += 1
        db.commit()
    except Exception as e:
        db.rollback()
        logger.exception("pzz_loader: unexpected error, outer tx rolled back: %s", e)
        raise
    finally:
        db.close()
    return {
        "fetched": len(features),
        "inserted": inserted,
        "updated": updated,
        "skipped": skipped,
    }