#!/usr/bin/env python3
"""
Sentinel Check — Brand Monitoring V1
Hourly check for crisis keywords on Google News RSS.
Collects, runs pipeline, and dispatches alerts immediately.
"""

import feedparser
import sqlite3
import datetime
import urllib.parse
import subprocess
import os
import yaml

DB_PATH = "/opt/ai-os/products/ceo/brand-monitoring/data/brand-monitoring.db"
BRAND_DIR = "/opt/ai-os/products/ceo/brand-monitoring/brands/fulbright"

def load_keywords():
    with open(f"{BRAND_DIR}/keywords.yaml", "r") as f:
        return yaml.safe_load(f)

def run_sentinel():
    keywords = load_keywords()
    triggers = keywords.get("crisis_triggers", [])
    
    # Construct search query
    # E.g. Fulbright (scandal OR "tranh cãi" OR "chỉ trích" OR "phàn nàn" OR "học phí" OR "kiểm định" OR "bê bối")
    query_parts = []
    for t in triggers:
        if " " in t:
            query_parts.append(f'"{t}"')
        else:
            query_parts.append(t)
    
    or_query = " OR ".join(query_parts)
    search_query = f"Fulbright ({or_query})"
    encoded_query = urllib.parse.quote(search_query)
    
    url = f"https://news.google.com/rss/search?q={encoded_query}&hl=vi&gl=VN&ceid=VN:vi"
    print(f"Running Sentinel Check with query: {search_query}")
    print(f"Feed URL: {url}")
    
    feed = feedparser.parse(url)
    print(f"Found {len(feed.entries)} entries.")
    
    conn = sqlite3.connect(DB_PATH)
    cursor = conn.cursor()
    
    new_items_count = 0
    for entry in feed.entries:
        try:
            url_link = entry.link
            title = entry.title
            published = entry.get("published", datetime.datetime.now().isoformat())
            
            cursor.execute("SELECT id FROM raw_items WHERE url = ?", (url_link,))
            if cursor.fetchone():
                continue
                
            cursor.execute("""
            INSERT INTO raw_items (source_id, url, title, published, fetched_at, brand_id, lang)
            VALUES (?, ?, ?, ?, ?, ?, ?)
            """, ("GoogleNews_Sentinel", url_link, title, published, datetime.datetime.now().isoformat(), "fulbright", "vi"))
            new_items_count += 1
        except Exception as e:
            print(f"Error processing entry: {e}")
            
    conn.commit()
    conn.close()
    print(f"Sentinel collection done. {new_items_count} new raw items inserted.")
    
    # If new items, run pipeline and dispatcher
    if new_items_count > 0:
        print("Running pipeline...")
        script_dir = os.path.dirname(os.path.abspath(__file__))
        subprocess.run(["python3", f"{script_dir}/pipeline.py"])
        print("Running alert dispatcher...")
        subprocess.run(["python3", f"{script_dir}/alert_dispatcher.py"])
    else:
        # Check if there are any lingering unprocessed items or pending alerts anyway
        print("No new items, but running pipeline + alert_dispatcher to ensure all processed.")
        script_dir = os.path.dirname(os.path.abspath(__file__))
        subprocess.run(["python3", f"{script_dir}/pipeline.py"])
        subprocess.run(["python3", f"{script_dir}/alert_dispatcher.py"])

if __name__ == "__main__":
    run_sentinel()