"""Green-Acres source — wraps the existing legacy implementation. This is a thin adapter rather than a rewrite: the legacy search_greenacres() in search_properties.py works fine; rewriting it would risk regressions for no behavioral gain. The adapter: - Translates SearchCriteria → the legacy fixed-regions iteration - Yields normalized PropertyHit records - Adds a health() preflight (HEAD against green-acres.fr to detect blocks) When upstream filtering is added (price/size in URL), it should land here and bypass the legacy code path. """ from __future__ import annotations import sys from pathlib import Path from typing import Iterator import requests # Allow `from store import ...` and importing the legacy module SCRIPT_DIR = Path(__file__).resolve().parent.parent if str(SCRIPT_DIR) not in sys.path: sys.path.insert(0, str(SCRIPT_DIR)) from store import load # noqa: E402 from sources._base import PropertyHit, SearchCriteria, Source # noqa: E402 # Region map: SearchCriteria.region slug → (country_tld, region_slug, display_name) # Built from the legacy REGIONS_GREENACRES in search_properties.py REGION_MAP = { # France — top vibe (★★★★★) 'drome': ('fr', 'drome', 'Drôme'), 'ardeche': ('fr', 'ardeche', 'Ardèche'), 'herault': ('fr', 'herault', 'Hérault'), 'cotes-d-armor': ('fr', 'cotes-d-armor', "Côtes-d'Armor"), 'morbihan': ('fr', 'morbihan', 'Morbihan'), 'finistere': ('fr', 'finistere', 'Finistère'), # France — strong vibe (★★★★) 'charente': ('fr', 'charente', 'Charente'), 'charente-maritime': ('fr', 'charente-maritime', 'Charente-Maritime'), # France — good vibe (★★★) 'dordogne': ('fr', 'dordogne', 'Dordogne'), 'gard': ('fr', 'gard', 'Gard'), 'lot': ('fr', 'lot', 'Lot'), 'correze': ('fr', 'correze', 'Corrèze'), 'creuse': ('fr', 'creuse', 'Creuse'), # France — favourite (★★) 'gers': ('fr', 'gers', 'Gers'), 'lot-et-garonne': ('fr', 'lot-et-garonne', 'Lot-et-Garonne'), 'mayenne': ('fr', 'mayenne', 'Mayenne'), # France — adjacent 'orne': ('fr', 'orne', 'Orne'), 'indre': ('fr', 'indre', 'Indre'), 'deux-sevres': ('fr', 'deux-sevres', 'Deux-Sèvres'), 'sarthe': ('fr', 'sarthe', 'Sarthe'), # France — Massif Central + Pyrenean + south-Languedoc (added 2026-05-27) 'cantal': ('fr', 'cantal', 'Cantal'), 'aveyron': ('fr', 'aveyron', 'Aveyron'), 'lozere': ('fr', 'lozere', 'Lozère'), 'tarn': ('fr', 'tarn', 'Tarn'), 'tarn-et-garonne': ('fr', 'tarn-et-garonne', 'Tarn-et-Garonne'), 'pyrenees-atlantiques': ('fr', 'pyrenees-atlantiques', 'Pyrénées-Atlantiques'), 'hautes-pyrenees': ('fr', 'hautes-pyrenees', 'Hautes-Pyrénées'), 'aude': ('fr', 'aude', 'Aude'), 'pyrenees-orientales': ('fr', 'pyrenees-orientales', 'Pyrénées-Orientales'), # Italy 'marche': ('it', 'marche', 'Le Marche'), 'umbria': ('it', 'umbria', 'Umbria'), 'toscana': ('it', 'toscana', 'Tuscany'), 'liguria': ('it', 'liguria', 'Liguria'), } class GreenAcresSource(Source): name = 'greenacres' countries = ['FR', 'IT'] requires_auth = False def health(self) -> tuple[bool, str]: """GET probe against green-acres.fr root. Note: HEAD often returns 405 on these sites, so we use a tiny GET.""" try: r = requests.get('https://www.green-acres.fr/', headers={'User-Agent': 'paradisomatch/1.0'}, timeout=8, allow_redirects=True, stream=True) if r.status_code >= 400: return False, f'HTTP {r.status_code}' r.close() return True, 'reachable' except requests.RequestException as e: return False, f'{type(e).__name__}' def search(self, criteria: SearchCriteria, known_urls: set[str] | None = None) -> Iterator[PropertyHit]: """Run the legacy search for a single region, yield normalized hits. The legacy implementation handles the full regions list internally, so we filter its output to the requested region. For multi-region campaigns, the orchestrator calls search() once per region. IMPORTANT: known_urls is converted to a {url: True} dict and passed as the legacy `store` argument. Without this, the legacy function re-parses every cached card on every run and `limit` counts cached re-discoveries toward the cap (eval bug #2, #3). """ # Import lazily — avoids loading the heavy legacy module unless needed from search_properties import search_greenacres if criteria.region and criteria.region not in REGION_MAP: return # unknown region for this source # Temporarily restrict the legacy regions list to the one we want. # The legacy code reads REGIONS_GREENACRES at call time, so we mutate # the module-level list (restoring after). Safe because the orchestrator # is single-threaded and we restore in a finally block. import search_properties as legacy original_regions = legacy.REGIONS_GREENACRES original_criteria = legacy.CRITERIA.copy() try: if criteria.region: target = REGION_MAP[criteria.region] legacy.REGIONS_GREENACRES = [target] # Push upstream filters by overriding the legacy CRITERIA dict if criteria.min_price is not None: legacy.CRITERIA['price_min'] = criteria.min_price if criteria.max_price is not None: legacy.CRITERIA['price_max'] = criteria.max_price if criteria.min_bedrooms is not None: legacy.CRITERIA['bedrooms_min'] = criteria.min_bedrooms # Pass known URLs as a fake "store" — legacy code only does # `set(store.keys())` then `prop_url in existing`. Limit defaults # to 50 (sane cap) if None, NOT 0 which legacy treats as unlimited. seed_store = {u: True for u in (known_urls or ())} limit = criteria.limit if criteria.limit is not None else 50 results = search_greenacres(seed_store, limit, criteria.max_pages) finally: legacy.REGIONS_GREENACRES = original_regions legacy.CRITERIA = original_criteria for r in results: # Apply min_land_m2 / min_building_m2 downstream filtering # (Green-Acres URL doesn't support these as query params, # so we filter the returned cards instead). if criteria.min_land_m2 and (r.get('land_size') or 0) < criteria.min_land_m2: continue if criteria.min_building_m2 and (r.get('building_size') or 0) < criteria.min_building_m2: continue yield PropertyHit( url=r['url'], source=r['source'], title=r.get('title'), price=r.get('price'), city=r.get('city'), country=r.get('country'), building_size=r.get('building_size'), land_size=r.get('land_size'), bedrooms=r.get('bedrooms'), rooms=r.get('rooms'), thumbnail=r.get('thumbnail'), search_region=r.get('search_region'), )