# ----------------------------------------------------------------------------- # Lou-Ka — Agrégateur de logements à louer (province de Québec) # Auteur : Simon-Pierre Boucher — contact@spboucher.ai # db.py : persistance SQLite — upsert avec détection de changements, # cycle de vie avec délai de grâce (2 syncs), détection de dérive, # cache des pages détail, cache de géocodage. # ----------------------------------------------------------------------------- from __future__ import annotations import json import sqlite3 import statistics import time from pathlib import Path from .schema import Listing DB_PATH = Path(__file__).resolve().parent.parent / "data" / "louka.db" # Nombre d'exécutions consécutives où une annonce doit être absente de la # source avant d'être désactivée (délai de grâce contre les ratés ponctuels). MISS_GRACE = 2 # Dérive : si une source retourne <= DRIFT_RATIO × sa médiane historique # (médiane >= DRIFT_MIN_BASE annonces), on alerte et on suspend les retraits. DRIFT_RATIO = 0.25 DRIFT_MIN_BASE = 8 DRIFT_HISTORY = 5 _SCHEMA = """ CREATE TABLE IF NOT EXISTS listings ( uid TEXT PRIMARY KEY, source TEXT NOT NULL, external_id TEXT NOT NULL, url TEXT, title TEXT, address TEXT, sector TEXT, city TEXT, unit_type TEXT, price REAL, price_label TEXT, availability TEXT, availability_date TEXT, area_sqft REAL, pets TEXT, furnished INTEGER, description TEXT, amenities TEXT, -- JSON (liste de textes source) details TEXT, -- JSON (champs structurés : inclusions, parking…) images TEXT, -- JSON lat REAL, lng REAL, geocode_failed INTEGER DEFAULT 0, content_hash TEXT, first_seen REAL, last_seen REAL, updated_at REAL, miss_count INTEGER DEFAULT 0, active INTEGER DEFAULT 1 ); CREATE INDEX IF NOT EXISTS idx_listings_source ON listings(source); CREATE INDEX IF NOT EXISTS idx_listings_city ON listings(city); CREATE INDEX IF NOT EXISTS idx_listings_active ON listings(active); CREATE TABLE IF NOT EXISTS sync_log ( id INTEGER PRIMARY KEY AUTOINCREMENT, source TEXT, ts REAL, found INTEGER, added INTEGER, updated INTEGER, removed INTEGER, ok INTEGER, message TEXT, stats TEXT -- JSON : taux de champs null, missed, alerte… ); CREATE TABLE IF NOT EXISTS detail_cache ( source TEXT NOT NULL, external_id TEXT NOT NULL, key TEXT, -- hash du contenu « liste » de l'annonce payload TEXT, -- JSON opaque propre au connecteur fetched_at REAL, PRIMARY KEY (source, external_id) ); CREATE TABLE IF NOT EXISTS price_log ( uid TEXT NOT NULL, ts REAL NOT NULL, price REAL -- prix observé (NULL = retiré de l'affichage) ); CREATE INDEX IF NOT EXISTS idx_price_log_uid ON price_log(uid); CREATE TABLE IF NOT EXISTS poi_cache ( coord_key TEXT PRIMARY KEY, -- "lat,lng" arrondi à 4 décimales (~11 m) lat REAL, lng REAL, pois TEXT, -- JSON : [{cat, name, dist_m}] (plus proche/catégorie) ts REAL ); CREATE TABLE IF NOT EXISTS geocode_cache ( address TEXT PRIMARY KEY, -- adresse normalisée (clé de cache) lat REAL, lng REAL, provider TEXT, failed INTEGER DEFAULT 0, ts REAL ); """ # Colonnes ajoutées après la v1 — migration automatique des bases existantes. _MIGRATIONS = { "listings": { "availability_date": "TEXT", "area_sqft": "REAL", "pets": "TEXT", "furnished": "INTEGER", "details": "TEXT", "geocode_failed": "INTEGER DEFAULT 0", "miss_count": "INTEGER DEFAULT 0", "dauid": "TEXT", # aire de diffusion 2021 (stats de quartier) "digest": "TEXT", # JSON louka/textmine.py (description structurée) }, "sync_log": { "stats": "TEXT", }, } def connect() -> sqlite3.Connection: DB_PATH.parent.mkdir(parents=True, exist_ok=True) con = sqlite3.connect(DB_PATH) con.row_factory = sqlite3.Row con.executescript(_SCHEMA) for table, cols in _MIGRATIONS.items(): existing = {r["name"] for r in con.execute(f"PRAGMA table_info({table})")} for col, decl in cols.items(): if col not in existing: con.execute(f"ALTER TABLE {table} ADD COLUMN {col} {decl}") con.commit() return con # --------------------------------------------------------------------------- # Synchronisation d'une source # --------------------------------------------------------------------------- def _drift_alert(con: sqlite3.Connection, source: str, found: int, null_price_rate: float) -> str | None: """Détecte une dérive du connecteur (chute du volume ou des prix extraits). Retourne un message d'alerte, ou None si tout est normal. """ hist = con.execute( "SELECT found, stats FROM sync_log WHERE source=? AND ok=1" " ORDER BY ts DESC LIMIT ?", (source, DRIFT_HISTORY)).fetchall() if len(hist) < 3: return None med_found = statistics.median(r["found"] for r in hist) if med_found >= DRIFT_MIN_BASE and found <= DRIFT_RATIO * med_found: return (f"dérive: {found} annonce(s) trouvée(s) contre une médiane de " f"{med_found:.0f} — retraits suspendus, vérifier le connecteur") if found >= DRIFT_MIN_BASE and null_price_rate >= 0.8: rates = [] for r in hist: try: rates.append(json.loads(r["stats"] or "{}")["null_price_rate"]) except (KeyError, ValueError, TypeError): continue if rates and statistics.median(rates) <= 0.3: return (f"dérive: {null_price_rate:.0%} des annonces sans prix " f"(habituellement {statistics.median(rates):.0%}) — " "le format de la source a probablement changé") return None def sync_source(con: sqlite3.Connection, source: str, listings: list[Listing]) -> dict: """Synchronise les annonces d'une source. - nouvelle annonce -> insertion - annonce modifiée -> mise à jour (comparaison de content_hash) - annonce disparue -> miss_count += 1, puis active=0 après MISS_GRACE exécutions consécutives (délai de grâce) - dérive détectée -> alerte consignée, retraits suspendus """ now = time.time() added = updated = 0 seen_uids = set() n = len(listings) null_price = sum(1 for l in listings if l.price is None) null_addr = sum(1 for l in listings if not l.address) null_price_rate = round(null_price / n, 3) if n else 0.0 alert = _drift_alert(con, source, n, null_price_rate) for lst in listings: seen_uids.add(lst.uid) h = lst.content_hash() row = con.execute("SELECT content_hash, price FROM listings WHERE uid=?", (lst.uid,)).fetchone() params = dict( uid=lst.uid, source=lst.source, external_id=lst.external_id, url=lst.url, title=lst.title, address=lst.address, sector=lst.sector, city=lst.city, unit_type=lst.unit_type, price=lst.price, price_label=lst.price_label, availability=lst.availability, availability_date=lst.availability_date, area_sqft=lst.area_sqft, pets=lst.pets, furnished=(None if lst.furnished is None else int(lst.furnished)), description=lst.description, digest=(json.dumps(lst.digest, ensure_ascii=False) if getattr(lst, "digest", None) else None), amenities=json.dumps(lst.amenities, ensure_ascii=False), details=json.dumps(lst.details, ensure_ascii=False), images=json.dumps(lst.images, ensure_ascii=False), lat=lst.lat, lng=lst.lng, content_hash=h, now=now, ) if row is None: con.execute( """INSERT INTO listings (uid, source, external_id, url, title, address, sector, city, unit_type, price, price_label, availability, availability_date, area_sqft, pets, furnished, description, digest, amenities, details, images, lat, lng, content_hash, first_seen, last_seen, updated_at, miss_count, active) VALUES (:uid,:source,:external_id,:url,:title,:address, :sector,:city,:unit_type,:price,:price_label,:availability, :availability_date,:area_sqft,:pets,:furnished, :description,:digest,:amenities,:details,:images,:lat,:lng, :content_hash,:now,:now,:now,0,1)""", params) if lst.price is not None: # prix initial = point de départ de l'historique con.execute("INSERT INTO price_log (uid, ts, price) VALUES (?,?,?)", (lst.uid, now, lst.price)) added += 1 elif row["content_hash"] != h: # COALESCE : ne jamais écraser des coordonnées géocodées par null con.execute( """UPDATE listings SET url=:url, title=:title, address=:address, sector=:sector, city=:city, unit_type=:unit_type, price=:price, price_label=:price_label, availability=:availability, availability_date=:availability_date, area_sqft=:area_sqft, pets=:pets, furnished=:furnished, description=:description, digest=:digest, amenities=:amenities, details=:details, images=:images, lat=COALESCE(:lat, lat), lng=COALESCE(:lng, lng), content_hash=:content_hash, last_seen=:now, updated_at=:now, miss_count=0, active=1 WHERE uid=:uid""", params) if lst.price != row["price"]: # changement de prix -> historique con.execute("INSERT INTO price_log (uid, ts, price) VALUES (?,?,?)", (lst.uid, now, lst.price)) updated += 1 else: con.execute( "UPDATE listings SET last_seen=?, miss_count=0, active=1 WHERE uid=?", (now, lst.uid)) # Annonces de cette source qui n'apparaissent plus : délai de grâce, # puis désactivation. Suspendu si une dérive est détectée. removed = missed = 0 if not alert: for r in con.execute( "SELECT uid, miss_count FROM listings WHERE source=? AND active=1", (source,)).fetchall(): if r["uid"] in seen_uids: continue missed += 1 if r["miss_count"] + 1 >= MISS_GRACE: con.execute( "UPDATE listings SET active=0, miss_count=?, updated_at=?" " WHERE uid=?", (r["miss_count"] + 1, now, r["uid"])) removed += 1 else: con.execute("UPDATE listings SET miss_count=miss_count+1 WHERE uid=?", (r["uid"],)) stats = { "null_price_rate": null_price_rate, "null_address_rate": round(null_addr / n, 3) if n else 0.0, "missed": missed, } if alert: stats["alert"] = alert con.execute( "INSERT INTO sync_log (source, ts, found, added, updated, removed, ok," " message, stats) VALUES (?,?,?,?,?,?,1,?,?)", (source, now, n, added, updated, removed, alert or "ok", json.dumps(stats, ensure_ascii=False))) con.commit() out = {"source": source, "found": n, "added": added, "updated": updated, "removed": removed} if alert: out["alert"] = alert return out def log_failure(con: sqlite3.Connection, source: str, message: str) -> None: con.execute( "INSERT INTO sync_log (source, ts, found, added, updated, removed, ok, message)" " VALUES (?,?,0,0,0,0,0,?)", (source, time.time(), message)) con.commit() # --------------------------------------------------------------------------- # Cache des pages détail (« détail si nouveau/modifié ») # --------------------------------------------------------------------------- def get_cached_detail(con: sqlite3.Connection, source: str, external_id: str, key: str) -> dict | None: """Payload détail mis en cache si la clé (hash liste) n'a pas changé.""" row = con.execute( "SELECT key, payload FROM detail_cache WHERE source=? AND external_id=?", (source, external_id)).fetchone() if row and row["key"] == key and row["payload"]: try: return json.loads(row["payload"]) except ValueError: return None return None def put_cached_detail(con: sqlite3.Connection, source: str, external_id: str, key: str, payload: dict) -> None: con.execute( "INSERT INTO detail_cache (source, external_id, key, payload, fetched_at)" " VALUES (?,?,?,?,?)" " ON CONFLICT(source, external_id) DO UPDATE SET" " key=excluded.key, payload=excluded.payload, fetched_at=excluded.fetched_at", (source, external_id, key, json.dumps(payload, ensure_ascii=False), time.time())) con.commit()