# ----------------------------------------------------------------------------- # Auto-Ka — Agrégateur de voitures usagées à vendre (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, # historique de prix (baisses de prix = signal clé en auto usagée), # cache des pages détail. # ----------------------------------------------------------------------------- from __future__ import annotations import json import sqlite3 import statistics import time from pathlib import Path from .schema import Vehicle DB_PATH = Path(__file__).resolve().parent.parent / "data" / "autoka.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 vehicles ( uid TEXT PRIMARY KEY, source TEXT NOT NULL, external_id TEXT NOT NULL, url TEXT, title TEXT, make TEXT, model TEXT, trim TEXT, year INTEGER, price REAL, price_label TEXT, mileage_km REAL, mileage_label TEXT, transmission TEXT, fuel TEXT, drivetrain TEXT, body_type TEXT, exterior_color TEXT, interior_color TEXT, engine TEXT, doors INTEGER, seats INTEGER, vin TEXT, stock_number TEXT, dealer_name TEXT, city TEXT, region TEXT, description TEXT, features TEXT, -- JSON (liste de textes source) details TEXT, -- JSON (champs structurés) images TEXT, -- JSON carfax_url TEXT, 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_vehicles_source ON vehicles(source); CREATE INDEX IF NOT EXISTS idx_vehicles_make ON vehicles(make); CREATE INDEX IF NOT EXISTS idx_vehicles_region ON vehicles(region); CREATE INDEX IF NOT EXISTS idx_vehicles_active ON vehicles(active); CREATE INDEX IF NOT EXISTS idx_vehicles_price ON vehicles(price); 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); """ def connect() -> sqlite3.Connection: DB_PATH.parent.mkdir(parents=True, exist_ok=True) con = sqlite3.connect(DB_PATH, timeout=30) con.row_factory = sqlite3.Row # WAL : plusieurs processus de sync peuvent écrire sans se bloquer con.execute("PRAGMA journal_mode=WAL") con.execute("PRAGMA busy_timeout=15000") con.executescript(_SCHEMA) 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).""" 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, vehicles: list[Vehicle]) -> 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) — véhicule vendu ou retiré - dérive détectée -> alerte consignée, retraits suspendus """ now = time.time() added = updated = 0 seen_uids = set() n = len(vehicles) null_price = sum(1 for v in vehicles if v.price is None) null_km = sum(1 for v in vehicles if v.mileage_km is None) null_price_rate = round(null_price / n, 3) if n else 0.0 alert = _drift_alert(con, source, n, null_price_rate) for veh in vehicles: seen_uids.add(veh.uid) h = veh.content_hash() row = con.execute("SELECT content_hash, price FROM vehicles WHERE uid=?", (veh.uid,)).fetchone() params = dict( uid=veh.uid, source=veh.source, external_id=veh.external_id, url=veh.url, title=veh.title, make=veh.make, model=veh.model, trim=veh.trim, year=veh.year, price=veh.price, price_label=veh.price_label, mileage_km=veh.mileage_km, mileage_label=veh.mileage_label, transmission=veh.transmission, fuel=veh.fuel, drivetrain=veh.drivetrain, body_type=veh.body_type, exterior_color=veh.exterior_color, interior_color=veh.interior_color, engine=veh.engine, doors=veh.doors, seats=veh.seats, vin=veh.vin, stock_number=veh.stock_number, dealer_name=veh.dealer_name, city=veh.city, region=veh.region, description=veh.description, features=json.dumps(veh.features, ensure_ascii=False), details=json.dumps(veh.details, ensure_ascii=False), images=json.dumps(veh.images, ensure_ascii=False), carfax_url=veh.carfax_url, content_hash=h, now=now, ) if row is None: con.execute( """INSERT INTO vehicles (uid, source, external_id, url, title, make, model, trim, year, price, price_label, mileage_km, mileage_label, transmission, fuel, drivetrain, body_type, exterior_color, interior_color, engine, doors, seats, vin, stock_number, dealer_name, city, region, description, features, details, images, carfax_url, content_hash, first_seen, last_seen, updated_at, miss_count, active) VALUES (:uid,:source,:external_id,:url,:title,:make,:model, :trim,:year,:price,:price_label,:mileage_km,:mileage_label, :transmission,:fuel,:drivetrain,:body_type,:exterior_color, :interior_color,:engine,:doors,:seats,:vin,:stock_number, :dealer_name,:city,:region,:description,:features,:details, :images,:carfax_url,:content_hash,:now,:now,:now,0,1)""", params) if veh.price is not None: # prix initial = départ de l'historique con.execute("INSERT INTO price_log (uid, ts, price) VALUES (?,?,?)", (veh.uid, now, veh.price)) added += 1 elif row["content_hash"] != h: con.execute( """UPDATE vehicles SET url=:url, title=:title, make=:make, model=:model, trim=:trim, year=:year, price=:price, price_label=:price_label, mileage_km=:mileage_km, mileage_label=:mileage_label, transmission=:transmission, fuel=:fuel, drivetrain=:drivetrain, body_type=:body_type, exterior_color=:exterior_color, interior_color=:interior_color, engine=:engine, doors=:doors, seats=:seats, vin=:vin, stock_number=:stock_number, dealer_name=:dealer_name, city=:city, region=:region, description=:description, features=:features, details=:details, images=:images, carfax_url=:carfax_url, content_hash=:content_hash, last_seen=:now, updated_at=:now, miss_count=0, active=1 WHERE uid=:uid""", params) if veh.price != row["price"]: # changement de prix -> historique con.execute("INSERT INTO price_log (uid, ts, price) VALUES (?,?,?)", (veh.uid, now, veh.price)) updated += 1 else: con.execute( "UPDATE vehicles SET last_seen=?, miss_count=0, active=1 WHERE uid=?", (now, veh.uid)) # Annonces de cette source qui n'apparaissent plus : délai de grâce, # puis désactivation (véhicule vendu). Suspendu si dérive détectée. removed = missed = 0 if not alert: for r in con.execute( "SELECT uid, miss_count FROM vehicles 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 vehicles SET active=0, miss_count=?, updated_at=?" " WHERE uid=?", (r["miss_count"] + 1, now, r["uid"])) removed += 1 else: con.execute("UPDATE vehicles SET miss_count=miss_count+1 WHERE uid=?", (r["uid"],)) stats = { "null_price_rate": null_price_rate, "null_km_rate": round(null_km / 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()