#!/usr/bin/env python3
"""Runs the cohort's queries through the live Google SERP routes once each and stores one raw response
per call, to count how many organic results come back as a google.com/goto link. Resumable; one worker
per provider; polls the Treg ledger every 15 s (other loops share the balance, so cost is matched by call_id) because `treg balance --limit` caps at the last 200 rows.
Deterministic; no LLM.
Usage: scripts/run-goto-check.py evidence/fixtures/seo-cohort-v1.json <raw-dir>
Score with scripts/score-goto-check.py (no provider call)."""
import json, sys, os, subprocess, time, threading, concurrent.futures as cf
cohort = json.load(open(sys.argv[1])); out = sys.argv[2]; os.makedirs(out, exist_ok=True)
LOC = 2840
ledger_path = os.path.join(out, 'ledger.json')
ledger = json.load(open(ledger_path)) if os.path.exists(ledger_path) else {}
stop = threading.Event()
def poll_once():
    try:
        p = subprocess.run(['treg', 'balance', '--limit', '200', '--json'], capture_output=True, text=True, timeout=45)
        d = json.loads(p.stdout)
    except Exception as e:
        print('ledger poll failed', e, flush=True); return
    for e in d['entries']['items']: ledger[e['id']] = e
    json.dump(ledger, open(ledger_path + '.tmp', 'w')); os.replace(ledger_path + '.tmp', ledger_path)
def poller():
    while not stop.is_set():
        poll_once(); stop.wait(15)
    poll_once()
def call(label, argv, meta):
    f = os.path.join(out, label + '.json')
    if os.path.exists(f): return
    t0 = time.time()
    try:
        p = subprocess.run(['treg', 'call', *argv], capture_output=True, text=True, timeout=180)
        rc, so, se = p.returncode, p.stdout, p.stderr
    except subprocess.TimeoutExpired:
        rc, so, se = -1, '', 'timeout after 180s'
    ms = int((time.time() - t0) * 1000)
    raw = so if so.strip() else se
    try: body = json.loads(raw)
    except Exception: body = {'_nonjson': raw[:2000]}
    rec = {'_run': {**meta, 'label': label, 'latency_ms': ms, 'exit': rc, 'ts': time.strftime('%Y-%m-%dT%H:%M:%SZ', time.gmtime())}, 'response': body}
    json.dump(rec, open(f, 'w'))
    print(label, ms, 'ms', 'exit', rc, flush=True)
jobs = []
for i, q in enumerate(cohort['queries']):
    k = f'q{i+1:02d}'
    def add(prov, argv): jobs.append((f'serp-{prov}-{k}', argv, {'task': 'serp', 'provider': prov, 'q': q}))
    add('cloro', ['cloro.google.serp.organic', '--method', 'POST', '--data', json.dumps({'query': q, 'gl': 'US', 'pages': 1})])
    add('dataforseo', ['dataforseo.x.serp-google-organic-live-advanced', '--method', 'POST', '--data', json.dumps([{'language_code': 'en', 'location_code': LOC, 'keyword': q, 'depth': 10}])])
    add('serpapi', ['serpapi.google.serp.organic', '--query', 'engine=google', '--query', f'q={q}', '--query', 'gl=us', '--query', 'hl=en', '--query', 'num=10'])
    add('treg-routed', ['treg.google.serp.organic', '--method', 'POST', '--data', json.dumps({'q': q})])
    add('anyapi', ['anyapi.google.serp.organic', '--method', 'POST', '--data', json.dumps({'query': q, 'gl': 'us', 'hl': 'en', 'limit': 10})])
    add('limadata', ['limadata.google.serp.organic', '--method', 'POST', '--data', json.dumps({'query': q, 'page': 1})])
    add('serper', ['serper.web.search', '--method', 'POST', '--data', json.dumps({'q': q, 'gl': 'us', 'hl': 'en'})])
    add('scrapecreators', ['scrapecreators.x.v1-google-search', '--query', f'query={q}', '--query', 'region=US'])
by_prov = {}
for j in jobs: by_prov.setdefault(j[2]['provider'], []).append(j)
def run_prov(lst):
    for label, argv, meta in lst: call(label, argv, meta)
t = threading.Thread(target=poller, daemon=True); t.start()
with cf.ThreadPoolExecutor(max_workers=len(by_prov)) as ex:
    list(ex.map(run_prov, by_prov.values()))
stop.set(); t.join(60)
print('done', len(jobs), 'jobs', flush=True)
