DomGod/app/scorer.py

import os
import json
import aiosqlite
from app.db import SQLITE_PATH

KNOWN_CMS = {"wordpress", "joomla", "drupal", "wix", "squarespace", "shopify", "prestashop", "magento", "typo3", "opencart"}
TARGET_COUNTRIES = set(os.getenv("TARGET_COUNTRIES", "ES,GB,DE,FR,RO,PT,AD,IT").split(","))

LOCAL_BIZ_KEYWORDS = {
    "restaurant", "cafe", "shop", "store", "salon", "plumber", "electrician",
    "dentist", "clinic", "garage", "hotel", "bakery", "bar", "gym", "spa",
    "fontanero", "electricista", "dentista", "clínica", "taller",
    "panadería", "peluquería", "tienda", "abogado", "gestor", "inmobili",
}


def local_biz_keywords(title: str | None) -> bool:
    if not title:
        return False
    title_lower = title.lower()
    return any(kw in title_lower for kw in LOCAL_BIZ_KEYWORDS)


def score(domain_row: dict) -> int:
    # Dead sites are unreachable — cap at 5 regardless of other signals
    if not domain_row.get("is_live") and not domain_row.get("reachable"):
        return 5

    s = 20  # Live site base

    ssl_days = domain_row.get("ssl_expiry_days")
    if ssl_days is not None and ssl_days < 30:
        s += 15  # SSL expiring / expired — urgent upsell
    if not domain_row.get("ssl_valid"):
        s += 15  # No valid SSL

    cms = (domain_row.get("cms") or "").lower()
    if cms in KNOWN_CMS:
        s += 15  # Known CMS — maintenance / migration opportunity

    if not domain_row.get("has_mx"):
        s += 10  # No email = needs professional email setup

    if domain_row.get("ip_country") in TARGET_COUNTRIES:
        s += 10

    server = (domain_row.get("server") or "").lower()
    if "shared" in server:
        s += 5

    if local_biz_keywords(domain_row.get("page_title")):
        s += 5

    # Kit Digital: proven buyer of IT services (Gemini-confirmed takes precedence)
    if domain_row.get("kit_digital"):
        s += 20

    # Social media / GMB presence signals
    try:
        ci = json.loads(domain_row.get("contact_info") or "{}")
        has_social = bool(ci.get("social"))
        has_contact = bool(ci.get("emails") or ci.get("phones") or ci.get("whatsapp"))
    except Exception:
        has_social = False
        has_contact = False

    if not has_social:
        s += 5   # No social = opportunity to build presence
    if has_contact:
        s += 3   # Reachable lead — we can actually pitch them

    return min(s, 100)


async def run_scoring():
    async with aiosqlite.connect(SQLITE_PATH) as db:
        db.row_factory = aiosqlite.Row
        async with db.execute("SELECT * FROM enriched_domains") as cur:
            rows = [dict(r) async for r in cur]

        updates = [(score(r), r["domain"]) for r in rows]
        await db.executemany(
            "UPDATE enriched_domains SET score = ? WHERE domain = ?", updates
        )
        await db.executemany(
            """INSERT INTO scores (domain, score) VALUES (?, ?)
               ON CONFLICT(domain) DO UPDATE SET score=excluded.score, scored_at=datetime('now')""",
            updates,
        )
        await db.commit()

    return {"scored": len(updates)}