services/trend_monitor.py

"""
Trend monitor: current occasions (AI) + niche news (GNews) for ad trending context.
"""

from gnews import GNews
from typing import List, Dict, Any
from datetime import datetime, timedelta
import asyncio

from services.current_occasions import get_current_occasions

NICHE_KEYWORDS = {
    "home_insurance": ["home insurance", "homeowners insurance", "property insurance", "natural disaster", "insurance rates"],
    "glp1": ["GLP-1", "Ozempic", "Wegovy", "weight loss", "Mounjaro", "Zepbound"],
    "auto_insurance": ["auto insurance", "car insurance", "vehicle insurance", "insurance premiums", "driver discounts"],
}

TREND_CACHE: Dict[str, tuple] = {}
CACHE_DURATION = timedelta(hours=1)

TRIGGER_WORDS = [
    (["crisis", "warning", "danger", "risk", "threat", "disaster"], "Fear"),
    (["shortage", "limited", "running out", "exclusive", "sold out"], "FOMO"),
    (["breakthrough", "solution", "cure", "relief", "success"], "Hope"),
    (["save", "discount", "cheaper", "affordable", "deal"], "Greed"),
    (["new", "innovation", "discover", "reveal", "secret"], "Curiosity"),
    (["urgent", "now", "immediate", "breaking"], "Urgency"),
]


class TrendMonitor:
    def __init__(self, language: str = "en", country: str = "US"):
        self.google_news = GNews(language=language, country=country)
        self.google_news.period = "7d"
        self.google_news.max_results = 10

    def _normalize_niche(self, niche: str) -> str:
        key = niche.lower().strip().replace(" ", "_")
        return key if key in NICHE_KEYWORDS else niche

    async def fetch_trends(self, niche: str) -> List[Dict]:
        cache_key = f"trends_{niche}"
        if cache_key in TREND_CACHE:
            data, cached_at = TREND_CACHE[cache_key]
            if datetime.now() - cached_at < CACHE_DURATION:
                return data
        trends = await self._fetch_news(niche)
        TREND_CACHE[cache_key] = (trends, datetime.now())
        return trends

    async def _fetch_news(self, niche: str) -> List[Dict]:
        if niche not in NICHE_KEYWORDS:
            raise ValueError(f"Unsupported niche: {niche}. Use: {list(NICHE_KEYWORDS.keys())}")
        keywords = NICHE_KEYWORDS[niche][:3]
        all_articles = []
        loop = asyncio.get_event_loop()
        for kw in keywords:
            try:
                articles = await loop.run_in_executor(
                    None, lambda k=kw: self.google_news.get_news(k)
                )
                for a in articles:
                    a["keyword"] = kw
                    a["niche"] = niche
                    all_articles.append(a)
            except Exception as e:
                print(f"⚠️ News fetch failed for '{kw}': {e}")
        if not all_articles:
            return []
        scored = self._score_articles(all_articles, niche)
        return scored[:5]

    def _score_articles(self, articles: List[Dict], niche: str) -> List[Dict]:
        keywords = NICHE_KEYWORDS[niche]
        for a in articles:
            text = f"{a.get('title', '')} {a.get('description', '')}".lower()
            score = sum(2 for k in keywords if k.lower() in text)
            pub = a.get("published date")
            if pub:
                try:
                    from email.utils import parsedate_to_datetime
                    dt = parsedate_to_datetime(str(pub)) if not isinstance(pub, datetime) else pub
                    days = (datetime.now() - dt).days
                    score += 5 if days <= 1 else 3 if days <= 3 else 1
                except Exception:
                    pass
            for word in ["crisis", "warning", "breakthrough", "new", "urgent", "breaking", "major"]:
                if word in text:
                    score += 1
            a["relevance_score"] = score
        return sorted(articles, key=lambda x: x.get("relevance_score", 0), reverse=True)

    def _trigger(self, title: str, description: str) -> str:
        text = f"{title} {description}".lower()
        for words, trigger in TRIGGER_WORDS:
            if any(w in text for w in words):
                return trigger
        return "Emotion"

    async def get_relevant_trends_for_niche(self, niche: str) -> Dict[str, Any]:
        relevant = []
        for occ in await get_current_occasions():
            relevant.append({"title": occ["title"], "summary": occ["summary"], "category": occ.get("category", "Occasion")})
        niche_key = self._normalize_niche(niche)
        if niche_key in NICHE_KEYWORDS:
            try:
                for t in (await self.fetch_trends(niche_key))[:3]:
                    desc = t.get("description", "") or t.get("title", "")
                    relevant.append({
                        "title": t.get("title", ""),
                        "summary": desc,
                        "category": t.get("niche", "News").replace("_", " ").title(),
                    })
            except Exception as e:
                print(f"⚠️ News trends skipped: {e}")
        return {"relevant_trends": relevant}

    async def get_trending_angles(self, niche: str) -> List[Dict]:
        trends = await self.fetch_trends(niche)
        angles = []
        for t in trends[:3]:
            title = t.get("title", "")
            angles.append({
                "key": f"trend_{abs(hash(title)) % 10000}",
                "name": f"Trending: {title[:40]}...",
                "trigger": self._trigger(title, t.get("description", "")),
                "example": (t.get("description") or title)[:100],
                "category": "Trending",
                "source": "google_news",
                "url": t.get("url"),
                "expires": (datetime.now() + timedelta(days=7)).isoformat(),
                "relevance_score": t.get("relevance_score", 0),
            })
        return angles

    def clear_cache(self) -> None:
        global TREND_CACHE
        TREND_CACHE = {}


trend_monitor = TrendMonitor()