#!/usr/bin/env python3
"""Scrape BGA news for new game releases since ID 1103."""
import subprocess
import re
import time
import json

AGENT_BROWSER = "/home/hermes/.hermes/node/lib/node_modules/agent-browser/bin/agent-browser-linux-x64"

def run(cmd, timeout=15):
    result = subprocess.run(cmd, shell=True, capture_output=True, text=True, timeout=timeout)
    return result.stdout + result.stderr

results = []
for news_id in range(1104, 1121):
    # Close any existing session
    run(f'{AGENT_BROWSER} close --all 2>/dev/null', timeout=3)
    
    print(f"Checking ID {news_id}...", flush=True)
    open_out = run(f'{AGENT_BROWSER} open "https://en.boardgamearena.com/news?id={news_id}" --args "--no-sandbox"', timeout=20)
    time.sleep(2)
    
    snap = run(f'{AGENT_BROWSER} snapshot', timeout=15)
    
    # Quick filter: empty pages are very short
    if len(snap) < 2000:
        print(f"  ID {news_id}: SKIP (snapshot length {len(snap)} — no content)", flush=True)
        continue
    
    # Extract heading (level 3 = game title)
    heading_match = re.findall(r'heading\s+"([^"]+)"\s+\[level=3\]', snap)
    # Extract date: "Month DDth YYYY" format
    date_match = re.findall(r'StaticText\s+"(\w+\s+\d+(?:th|st|nd|rd)\s+\d{4})"', snap)
    
    title = heading_match[0] if heading_match else None
    date = date_match[-1] if date_match else None
    
    # Also extract gamepanel links
    gamepanel_links = re.findall(r'link\s+"(https://boardgamearena\.com/gamepanel\?game=[^"]+)"', snap)
    
    print(f"  ID {news_id}: title={title}, date={date}, panels={gamepanel_links}", flush=True)
    
    if title:
        results.append({
            "id": news_id,
            "date": date,
            "title": title,
            "gamepanel_links": gamepanel_links
        })

# Close session
run(f'{AGENT_BROWSER} close --all 2>/dev/null', timeout=3)

print("\n=== RESULTS ===")
print(json.dumps(results, indent=2, ensure_ascii=False))