#!/usr/bin/env python3
"""Pool builder for BENCHMARK_COUNTERPARTY_V1 (goal 2026-09-30 ChatGPT Space). Real Dutch companies from the official
register; read-only calls. Zoeken is free; each basisprofiel costs 0.02 EUR on AstraNL's existing KVK subscription and is
capped here at MAX_PAID calls. VIES is free. Output pool.json with the observed source state and the exact spend."""
import json, random, sys, time, urllib.parse, urllib.request
sys.path.insert(0, '/opt/astranl/core'); sys.path.insert(0, '/opt/astranl/organism')
import kvk_lookup
import counterparty_verification as cv

MAX_PAID = 30
R = random.Random(20260930)
WORDS = ['installatie', 'transport', 'techniek', 'logistiek', 'bouw', 'schilders', 'metaal', 'elektro', 'loodgieter', 'dakdekker',
         'machines', 'verhuur', 'groothandel', 'onderhoud', 'staalbouw', 'koeltechniek', 'grondwerk', 'isolatie']
KEY = kvk_lookup._key()


def zoeken(params):
    q = urllib.parse.urlencode(dict(params, resultatenPerPagina=100))
    req = urllib.request.Request('https://api.kvk.nl/api/v2/zoeken?' + q, headers={'apikey': KEY, 'Accept': 'application/json'})
    try:
        return json.loads(urllib.request.urlopen(req, timeout=20).read()).get('resultaten') or []
    except Exception as e:
        return [] if getattr(e, 'code', 0) == 404 else None


def all_rows(kvk):
    return zoeken({'kvkNummer': kvk, 'inclusiefInactieveRegistraties': 'true'})


def main():
    paid = 0
    active, inactive = {}, {}
    for w in WORDS:
        rows = zoeken({'naam': w, 'type': 'rechtspersoon', 'inclusiefInactieveRegistraties': 'true'}) or []
        for x in rows:
            k, n = x.get('kvkNummer'), x.get('naam') or ''
            if not k or 'b.v' not in n.lower():
                continue
            (inactive if str(x.get('actief', 'Ja')).lower() == 'nee' else active)[k] = n
        time.sleep(0.3)
    act = sorted(active.items()); R.shuffle(act)
    ina = sorted(inactive.items()); R.shuffle(ina)
    companies = []
    for k, n in act:
        if paid >= MAX_PAID or len(companies) >= 24:
            break
        rows = all_rows(k)
        if not rows or any(str(r.get('actief', 'Ja')).lower() == 'nee' for r in rows if r.get('type') == 'hoofdvestiging'):
            continue
        bp = kvk_lookup.lookup(k); paid += 1
        rsin = bp.get('rsin') if bp.get('ok') else None
        if not rsin:
            continue
        vat = 'NL%09dB01' % int(rsin)
        v = cv.check_vat(vat)
        if v.get('status') != 'ok' or not v.get('valid'):
            continue
        companies.append({'kvk': k, 'register_names': sorted({r.get('naam') for r in rows if r.get('naam')}), 'name': n, 'vat': vat,
                          'vies_name': v.get('name'), 'observed_at': v.get('checked_at')})
        time.sleep(0.5)
    dead = []
    for k, n in ina:
        if len(dead) >= 10:
            break
        rows = all_rows(k)
        if rows and all(str(r.get('actief', 'Ja')).lower() == 'nee' for r in rows):
            dead.append({'kvk': k, 'name': n, 'register_names': sorted({r.get('naam') for r in rows if r.get('naam')})})
        time.sleep(0.3)
    missing = []
    while len(missing) < 10:
        k = '%08d' % R.randint(10000000, 99999999)
        rows = all_rows(k)
        if rows == []:
            missing.append(k)
        time.sleep(0.3)
    out = {'built_at': time.strftime('%Y-%m-%dT%H:%M:%SZ', time.gmtime()), 'paid_basisprofiel_calls': paid, 'spend_eur': round(paid * 0.02, 2),
           'companies_active_with_valid_vat': companies, 'companies_all_inactive': dead, 'kvk_numbers_not_in_register': missing}
    json.dump(out, open('/opt/astranl/var/benchmark_counterparty_v1/pool.json', 'w'), ensure_ascii=False, indent=1)
    print(json.dumps({'active_valid_vat': len(companies), 'inactive': len(dead), 'missing': len(missing), 'paid': paid}))


if __name__ == '__main__':
    main()