foodlinkk-command-center/tools-api/app/connectors/food_trends.py

"""Halal vlees trends, top gerechten en food concept seeds."""
from __future__ import annotations

import re
from datetime import datetime, timezone
from typing import Any

from app.db import fetch_all

HALAL_MEAT_KEYWORDS = (
    "halal vlees", "halal meat", "halal kip", "halal lam", "halal rund",
    "halal gehakt", "halal chicken", "halal beef", "halal slacht",
    "halal certific", "vleesvervanger halal",
)

TOP_DISH_KEYWORDS = (
    "kant-en-klaar", "kant en klaar", "ready meal", "maaltijd", "gerecht",
    "curry", "stamppot", "biryani", "tagine", "lasagne", "schotel",
    "meal prep", "microwave meal", "diepvries maaltijd",
)

CONCEPT_SEEDS = [
    "Halal {dish} single-serve voor {chain} schappen in regio's met score >{score}",
    "Premium halal {meat} maaltijdlijn — inspelen op trend: {trend}",
    "Seizoens {dish} tray (4-portions) voor Plus/Jumbo non-listed partnership pitch",
    "AH {dish} variant — benchmark tegen {competitor} koers momentum ({pct}%)",
    "Halal-gap fill: {dish} + {meat} combo voor filialen zonder halal schap",
]


def _match_keywords(text: str, keywords: tuple[str, ...]) -> bool:
    blob = (text or "").lower()
    return any(k in blob for k in keywords)


def fetch_halal_meat_trends(limit: int = 15) -> list[dict[str, Any]]:
    rows = fetch_all(
        """SELECT i.title, i.link, i.summary, i.published_at, f.name AS feed_name, f.url AS feed_url
           FROM rss_items i JOIN rss_feeds f ON f.id = i.feed_id AND f.is_active = TRUE
           ORDER BY i.published_at DESC NULLS LAST LIMIT 200"""
    )
    out = []
    for r in rows:
        title = r.get("title") or ""
        summary = r.get("summary") or ""
        if not _match_keywords(f"{title} {summary}", HALAL_MEAT_KEYWORDS):
            continue
        out.append({
            "title": title,
            "link": r.get("link"),
            "summary": (summary or "")[:280],
            "feed_name": r.get("feed_name"),
            "feed_url": r.get("feed_url"),
            "published_at": r.get("published_at").isoformat() if r.get("published_at") else None,
            "category": "halal_vlees",
            "source_url": r.get("link"),
        })
        if len(out) >= limit:
            break
    return out


def fetch_top_dishes(limit: int = 12) -> list[dict[str, Any]]:
    rows = fetch_all(
        """SELECT i.title, i.link, i.summary, i.published_at, f.name AS feed_name, f.url AS feed_url
           FROM rss_items i JOIN rss_feeds f ON f.id = i.feed_id AND f.is_active = TRUE
           ORDER BY i.published_at DESC NULLS LAST LIMIT 250"""
    )
    scored: dict[str, dict[str, Any]] = {}
    for r in rows:
        title = (r.get("title") or "").lower()
        summary = (r.get("summary") or "").lower()
        blob = f"{title} {summary}"
        if not _match_keywords(blob, TOP_DISH_KEYWORDS):
            continue
        for kw in TOP_DISH_KEYWORDS:
            if kw in blob:
                key = kw.strip()
                if key not in scored:
                    scored[key] = {
                        "dish_keyword": key,
                        "mentions": 0,
                        "latest_title": r.get("title"),
                        "latest_link": r.get("link"),
                        "feed_name": r.get("feed_name"),
                        "source_url": r.get("link"),
                    }
                scored[key]["mentions"] += 1
                break
    items = sorted(scored.values(), key=lambda x: x["mentions"], reverse=True)[:limit]
    return items


def fetch_market_trend_rows(limit: int = 8) -> list[dict[str, Any]]:
    rows = fetch_all(
        """SELECT trend_name, description, opportunity_score, source, category, updated_at
           FROM market_trends ORDER BY updated_at DESC NULLS LAST LIMIT %s""",
        (limit,),
    )
    out = []
    for r in rows:
        out.append({
            "trend_name": r.get("trend_name"),
            "description": r.get("description"),
            "opportunity_score": float(r.get("opportunity_score") or 0),
            "source": r.get("source") or "Foodlinkk trends",
            "source_url": "/retail",
            "category": r.get("category") or "markt",
        })
    return out


def generate_concepts(
    dishes: list[dict[str, Any]] | None = None,
    halal_items: list[dict[str, Any]] | None = None,
    market_best: dict[str, Any] | None = None,
    limit: int = 6,
) -> list[dict[str, Any]]:
    dishes = dishes or fetch_top_dishes(5)
    halal_items = halal_items or fetch_halal_meat_trends(5)
    best_chain = (market_best or {}).get("chains", ["Albert Heijn"])[0]
    pct = (market_best or {}).get("change_pct", 0)
    competitor = (market_best or {}).get("name", "Ahold Delhaize")

    concepts = []
    for i, tmpl in enumerate(CONCEPT_SEEDS[:limit]):
        dish = dishes[i % len(dishes)]["dish_keyword"] if dishes else "kant-en-klaar maaltijd"
        meat = "halal kip" if halal_items else "halal vlees"
        trend = halal_items[i % len(halal_items)]["title"][:60] if halal_items else "groei halal convenience"
        text = tmpl.format(
            dish=dish,
            meat=meat,
            chain=best_chain,
            score=75,
            trend=trend,
            competitor=competitor,
            pct=pct,
        )
        concepts.append({
            "id": i + 1,
            "concept": text,
            "based_on": {
                "dish": dish,
                "halal_trend": trend,
                "market_signal": f"{competitor} {pct:+.1f}%" if market_best else "retail DB",
            },
            "source_urls": [
                u for u in [
                    dishes[i % len(dishes)].get("source_url") if dishes else None,
                    halal_items[i % len(halal_items)].get("source_url") if halal_items else None,
                    (market_best or {}).get("source_url"),
                ] if u
            ],
        })
    return concepts


def food_trends_dashboard() -> dict[str, Any]:
    halal = fetch_halal_meat_trends()
    dishes = fetch_top_dishes()
    trends = fetch_market_trend_rows()
    return {
        "halal_meat_trends": halal,
        "top_dishes": dishes,
        "market_trends": trends,
        "updated_at": datetime.now(timezone.utc).isoformat(),
    }