"""Search Engine — with geo-filtering for US-relevant results."""
import json, re, time, urllib.parse, urllib.request
from pathlib import Path

USER_AGENT = 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36'
PROCESSORS = json.load(open(Path(__file__).parent / 'processors.json'))

SITE_CACHE = {
    "Helcim": [
        {"domain":"redbarnpetproducts.com","industry":"ecommerce","country":"US","notes":"Pet products retailer — Helcim checkout"},
        {"domain":"vitalproteins.com","industry":"nutra","country":"US","notes":"Collagen supplements — Helcim gateway"},
        {"domain":"grizzlycoolers.com","industry":"ecommerce","country":"US","notes":"Premium outdoor coolers — Helcim"},
        {"domain":"woodlanddirect.com","industry":"ecommerce","country":"US","notes":"Fireplaces and grills retailer — Helcim"},
        {"domain":"atomicdefense.com","industry":"firearms","country":"US","notes":"Tactical gear and body armor — Helcim"},
        {"domain":"thegunstorelasvegas.com","industry":"firearms","country":"US","notes":"Las Vegas firearms dealer — Helcim"},
        {"domain":"bishopscustombutchering.com","industry":"ecommerce","country":"US","notes":"Custom meat processing — Helcim"},
        {"domain":"fireplacedoorsonline.com","industry":"ecommerce","country":"US","notes":"Custom fireplace doors — Helcim"},
    ],
    "Stripe": [
        {"domain":"notion.so","industry":"subscription","country":"US","notes":"All-in-one workspace — Stripe billing"},
        {"domain":"linear.app","industry":"subscription","country":"US","notes":"Project management for developers — Stripe"},
        {"domain":"vercel.com","industry":"subscription","country":"US","notes":"Frontend cloud platform — Stripe"},
        {"domain":"figma.com","industry":"subscription","country":"US","notes":"Collaborative design tool — Stripe"},
        {"domain":"substack.com","industry":"subscription","country":"US","notes":"Newsletter platform — Stripe"},
        {"domain":"patreon.com","industry":"subscription","country":"US","notes":"Creator membership platform — Stripe"},
        {"domain":"replit.com","industry":"subscription","country":"US","notes":"Browser-based IDE — Stripe"},
        {"domain":"dub.co","industry":"subscription","country":"US","notes":"Link management — Stripe"},
        {"domain":"cal.com","industry":"subscription","country":"US","notes":"Open-source scheduling — Stripe"},
        {"domain":"raycast.com","industry":"subscription","country":"US","notes":"Mac productivity launcher — Stripe"},
        {"domain":"shopfiy.com","industry":"ecommerce","country":"US","notes":"Ecommerce platform — Stripe payments"},
    ],
    "Adyen": [
        {"domain":"spotify.com","industry":"subscription","country":"US","notes":"Music streaming — Adyen payments"},
        {"domain":"uber.com","industry":"travel","country":"US","notes":"Ride & food delivery — Adyen"},
        {"domain":"ebay.com","industry":"ecommerce","country":"US","notes":"Global marketplace — Adyen"},
        {"domain":"etsy.com","industry":"ecommerce","country":"US","notes":"Handmade & vintage marketplace — Adyen"},
        {"domain":"mcdonalds.com","industry":"retail","country":"US","notes":"Fast food — Adyen US"},
        {"domain":"booking.com","industry":"travel","country":"EU","notes":"Hotel & travel booking — Adyen (EU HQ)"},
        {"domain":"zalando.com","industry":"ecommerce","country":"EU","notes":"European fashion platform — Adyen"},
        {"domain":"deliveroo.com","industry":"travel","country":"EU","notes":"Food delivery UK — Adyen"},
    ],
    "PaymentCloud": [
        {"domain":"kratomspot.com","industry":"cbd","country":"US","notes":"Kratom products — high-risk PaymentCloud"},
        {"domain":"purecbdvapors.com","industry":"cbd","country":"US","notes":"CBD vape products — PaymentCloud"},
        {"domain":"d8gas.com","industry":"cbd","country":"US","notes":"Delta-8 THC products — PaymentCloud"},
        {"domain":"tfcannabis.com","industry":"cbd","country":"US","notes":"Cannabis accessories — PaymentCloud"},
        {"domain":"bwsofficefurniture.com","industry":"ecommerce","country":"US","notes":"Office furniture — PaymentCloud"},
        {"domain":"elementvape.com","industry":"cbd","country":"US","notes":"Vape products — high-risk processor"},
    ],
    "eMerchantBroker": [
        {"domain":"jackpotcitycasino.com","industry":"gambling","country":"INTL","notes":"Online casino — EMB gateway"},
        {"domain":"spinpalacecasino.com","industry":"gambling","country":"INTL","notes":"Online casino — EMB processing"},
        {"domain":"rubyfortune.com","industry":"gambling","country":"INTL","notes":"Casino — EMB payment gateway"},
        {"domain":"gamingclub.com","industry":"gambling","country":"INTL","notes":"Online casino — EMB"},
        {"domain":"luxurycasino.com","industry":"gambling","country":"INTL","notes":"High-limit casino — EMB processing"},
    ],
    "MuchBetter": [
        {"domain":"stake.com","industry":"gambling","country":"INTL","notes":"Largest crypto casino — MuchBetter"},
        {"domain":"bc.game","industry":"gambling","country":"INTL","notes":"Crypto casino — MuchBetter option"},
        {"domain":"cloudbet.com","industry":"gambling","country":"INTL","notes":"Bitcoin sportsbook — MuchBetter"},
        {"domain":"bitstarz.com","industry":"gambling","country":"INTL","notes":"Award-winning casino — MuchBetter"},
        {"domain":"fortunejack.com","industry":"gambling","country":"INTL","notes":"Crypto casino — MuchBetter payments"},
    ],
    "AstroPay": [
        {"domain":"bet365.com","industry":"gambling","country":"EU","notes":"World's largest sportsbook — AstroPay"},
        {"domain":"1xbet.com","industry":"gambling","country":"INTL","notes":"International sportsbook — AstroPay"},
        {"domain":"betway.com","industry":"gambling","country":"EU","notes":"Sports & casino — AstroPay option"},
        {"domain":"22bet.com","industry":"gambling","country":"INTL","notes":"Sportsbook — AstroPay deposits"},
    ],
    "Trustly": [
        {"domain":"leovegas.com","industry":"gambling","country":"EU","notes":"Mobile casino Sweden — Trustly Pay N Play"},
        {"domain":"comeon.com","industry":"gambling","country":"EU","notes":"Sportsbook — Trustly deposits"},
        {"domain":"unibet.com","industry":"gambling","country":"EU","notes":"Major European bookmaker — Trustly"},
        {"domain":"betsson.com","industry":"gambling","country":"EU","notes":"Nordic casino group — Trustly"},
        {"domain":"mrgreen.com","industry":"gambling","country":"EU","notes":"Award-winning casino — Trustly"},
        {"domain":"ninjacasino.com","industry":"gambling","country":"EU","notes":"Casino — Trustly payments"},
    ],
    "CardX": [
        {"domain":"chicago.gov","industry":"government","country":"US","notes":"City of Chicago payments — CardX"},
        {"domain":"harvard.edu","industry":"education","country":"US","notes":"Harvard University — CardX tuition"},
        {"domain":"floridarevenue.com","industry":"government","country":"US","notes":"Florida tax payments — CardX"},
        {"domain":"laccd.edu","industry":"education","country":"US","notes":"LA Community College District — CardX"},
    ],
    "CDGcommerce": [
        {"domain":"bluedogink.com","industry":"ecommerce","country":"US","notes":"Custom printing — CDGcommerce"},
        {"domain":"shoprattan.com","industry":"ecommerce","country":"US","notes":"Rattan furniture — CDGcommerce"},
        {"domain":"sportsmemorabilia.com","industry":"ecommerce","country":"US","notes":"Sports collectibles — CDGcommerce"},
        {"domain":"thebouqs.com","industry":"ecommerce","country":"US","notes":"Farm-fresh flowers — CDGcommerce"},
    ],
    "PayKings": [
        {"domain":"guns.com","industry":"firearms","country":"US","notes":"Firearms marketplace — PayKings"},
        {"domain":"palmettostatearmory.com","industry":"firearms","country":"US","notes":"AR-15 & firearms retailer — PayKings"},
        {"domain":"cheaperthandirt.com","industry":"firearms","country":"US","notes":"Discount ammunition — PayKings"},
    ],
    "Zimpler": [
        {"domain":"leovegas.com","industry":"gambling","country":"EU","notes":"Mobile casino — Zimpler payments"},
        {"domain":"casumo.com","industry":"gambling","country":"EU","notes":"Adventure casino — Zimpler"},
    ],
    "Interac": [
        {"domain":"bet365.com","industry":"gambling","country":"EU","notes":"Sportsbook — Interac for Canada"},
        {"domain":"betway.com","industry":"gambling","country":"EU","notes":"Casino — Interac e-Transfer"},
    ],
    "Soar Payments": [
        {"domain":"cbdistillery.com","industry":"cbd","country":"US","notes":"Major CBD brand — Soar gateway"},
        {"domain":"medterracbd.com","industry":"cbd","country":"US","notes":"CBD products — Soar processing"},
        {"domain":"katsbotanicals.com","industry":"cbd","country":"US","notes":"Kratom & CBD — Soar Payments"},
    ],
    "Durango Merchant": [
        {"domain":"adultempire.com","industry":"adult","country":"US","notes":"Adult retailer — Durango processing"},
    ],
    "Forte": [
        {"domain":"plannedparenthood.org","industry":"healthcare","country":"US","notes":"Healthcare — Forte gateway"},
        {"domain":"habitat.org","industry":"nonprofit","country":"US","notes":"Habitat for Humanity — Forte"},
        {"domain":"ymca.net","industry":"nonprofit","country":"US","notes":"YMCA — Forte processing"},
    ],
}

COUNTRIES = {"US": "🇺🇸 United States", "EU": "🇪🇺 Europe", "INTL": "🌍 International"}

def search_duckduckgo(query, max_results=6):
    url = f"https://html.duckduckgo.com/html/?q={urllib.parse.quote(query)}"
    req = urllib.request.Request(url, headers={'User-Agent': USER_AGENT})
    try:
        resp = urllib.request.urlopen(req, timeout=12)
        html = resp.read().decode('utf-8', errors='ignore')
    except:
        return []
    if 'captcha' in html.lower() or 'sorry' in html.lower():
        return []
    links = re.findall(r'<a[^>]*class="result__a"[^>]*href="([^"]+)"[^>]*>([^<]+)</a>', html)
    snippets = re.findall(r'<a[^>]*class="result__snippet"[^>]*>([^<]+(?:<[^>]+>[^<]*)*)</a>', html)
    results = []
    for i, (url, title) in enumerate(links[:max_results]):
        url = urllib.parse.unquote(url)
        if 'duckduckgo.com' in url or not url.startswith('http'): continue
        domain = urllib.parse.urlparse(url).netloc.replace('www.', '')
        snippet = re.sub(r'<[^>]+>', '', snippets[i]) if i < len(snippets) else ''
        results.append({'url': url, 'domain': domain, 'title': title.strip(), 'snippet': snippet.strip()})
    return results

def get_from_cache(processor_name, industry=None, country=None):
    cache = SITE_CACHE.get(processor_name, [])
    if industry: cache = [s for s in cache if s['industry'] == industry]
    if country: cache = [s for s in cache if s.get('country') == country]
    proc = PROCESSORS['processors'].get(processor_name, {})
    return [{
        'url': f"https://{s['domain']}", 'domain': s['domain'],
        'title': s['domain'], 'snippet': s['notes'],
        'industry': s['industry'], 'country': s.get('country', 'US'),
        'processor': processor_name, 'processor_type': proc.get('type', ''), 'cached': True
    } for s in cache]

def search_processor(processor_name, industry=None, country=None, max_results=12):
    proc = PROCESSORS['processors'].get(processor_name)
    if not proc: return {'error': f'Unknown: {processor_name}', 'results': []}
    
    results, seen = [], set()
    
    # Try live DDG
    for dork in proc['dorks'][:2]:
        query = dork
        if industry: query += f' {industry}'
        if country == 'US': query += ' site:*.com'
        try:
            for r in search_duckduckgo(query, max_results=3):
                if r['domain'] not in seen:
                    seen.add(r['domain'])
                    r['processor'] = processor_name
                    r['processor_type'] = proc['type']
                    r['industry'] = industry or proc['industries'][0]
                    r['country'] = country or 'US'
                    r['cached'] = False
                    results.append(r)
        except: pass
        time.sleep(0.5)
    
    # Cache fallback
    for c in get_from_cache(processor_name, industry, country):
        if c['domain'] not in seen and len(results) < max_results:
            seen.add(c['domain'])
            results.append(c)
    
    return {
        'processor': processor_name, 'type': proc['type'],
        'industries': proc['industries'], 'dorks': proc['dorks'],
        'result_count': len(results),
        'live_results': sum(1 for r in results if not r.get('cached')),
        'cached_results': sum(1 for r in results if r.get('cached')),
        'results': results[:max_results]
    }

def search_all(industry=None, country=None, max_per_processor=6):
    all_results = {}
    for name in PROCESSORS['processors']:
        if industry and industry not in PROCESSORS['processors'][name]['industries']: continue
        result = search_processor(name, industry, country, max_per_processor)
        all_results[name] = result
        time.sleep(0.2)
    return all_results