services/google_news.py

"""
services/google_news.py
Ambil berita Google News berdasarkan keyword menggunakan SerpApi.

Data yang diambil:
  - title    → judul berita
  - snippet  → ringkasan/deskripsi berita
  - source   → nama media
  - date     → tanggal terbit
"""

import os

# ── CONFIG ──
SERPAPI_KEY = os.getenv("SERPAPI_KEY")   # set di HF Spaces Secrets
MAX_RESULTS = 30                          # jumlah berita per query


# ────────────────────────────────────────────
#  INTERNAL HELPERS
# ────────────────────────────────────────────

def _extract_texts(news_results: list) -> list[str]:
    """
    Ekstrak teks dari hasil Google News SerpApi.

    Struktur item:
      item["title"]           → judul berita
      item["snippet"]         → ringkasan berita
      item["source"]["name"]  → nama media (opsional)
      item["stories"]         → sub-artikel terkait (opsional)
        .title / .snippet
    """
    texts = []

    for item in news_results:
        # Judul
        title = (item.get("title") or "").strip()
        if title and len(title) > 5:
            texts.append(title)

        # Snippet / ringkasan
        snippet = (item.get("snippet") or item.get("description") or "").strip()
        if snippet and len(snippet) > 10:
            texts.append(snippet)

        # Sub-stories (Google News kadang mengelompokkan berita terkait)
        for story in item.get("stories", []):
            if isinstance(story, dict):
                st = (story.get("title") or "").strip()
                ss = (story.get("snippet") or "").strip()
                if st and len(st) > 5:
                    texts.append(st)
                if ss and len(ss) > 10:
                    texts.append(ss)

    return texts


# ────────────────────────────────────────────
#  PUBLIC INTERFACE
# ────────────────────────────────────────────

def get_google_news(keyword: str) -> list[str]:
    """
    Scrape Google News berdasarkan keyword via SerpApi.
    Return list of string (judul + snippet berita).

    Dipanggil dari services/aggregator.py.
    """
    if not SERPAPI_KEY:
        print("⚠️  SERPAPI_KEY tidak diset — skip Google News")
        return []

    try:
        from serpapi import GoogleSearch
    except ImportError:
        print("❌ serpapi package tidak terinstall — tambahkan ke requirements.txt")
        return []

    try:
        params = {
            "engine":   "google_news",
            "q":        keyword,
            "hl":       "id",          # bahasa Indonesia
            "gl":       "id",          # region Indonesia
            "num":      MAX_RESULTS,
            "api_key":  SERPAPI_KEY,
        }

        search  = GoogleSearch(params)
        results = search.get_dict()

        news_results = results.get("news_results", [])

        if not news_results:
            print(f"⚠️  Google News: tidak ada hasil untuk '{keyword}'")
            return []

        texts = _extract_texts(news_results)
        print(f"✅ Google News: {len(texts)} teks dari {len(news_results)} berita")
        return texts

    except Exception as e:
        print(f"❌ Google News error: {e}")
        return []