"""Нагрузочный тест: массовый прогон run_pipeline (сотни контрактов × допников).

Создаёт N контрактов × M допников, прогоняет полный конвейер сверки с Fake LLM
(без сети), реально нагружает SQLite: apply_ops + clear_current + reset.

Проверка: full_replace НЕ дублирует строки в масштабе (у каждого контракта ровно
SERVICES строк, а не SERVICES × M_SUPPS).
"""
import os
import time

import pytest

from db import contracts, documents, supplements, spec_current
from db.connection import query
from services import process

pytestmark = pytest.mark.load

N_CONTRACTS = int(os.getenv("LOAD_CONTRACTS", "100"))
M_SUPPS = int(os.getenv("LOAD_SUPPS", "20"))
SERVICES = int(os.getenv("LOAD_SERVICES", "10"))


def _seed():
    cids = []
    for ci in range(N_CONTRACTS):
        c = contracts.insert(f"03700_{ci}")
        for mi in range(M_SUPPS):
            d = documents.insert(f"d{ci}_{mi}.docx", "application/octet-stream", "raw", batch_id=f"b{ci}")
            documents.set_parsed(d["id"], [{"type": "paragraph", "text": f"услуга {mi}"}])
            supplements.insert(c["id"], d["id"], "initial" if mi == 0 else "additional")
        cids.append(c["id"])
    return cids


def _fake_llm():
    def fake_call(current_spec, doc_text, build_prompt_fn, llm_client=None):
        ops = [
            {"action": "ADD", "new_row": {"name": f"услуга {i}", "price": 100 + i, "qty": 1, "sum": 100 + i}}
            for i in range(SERVICES)
        ]
        return ({"mode": "full_replace", "ops": ops}, "pid")
    return fake_call


def test_mass_pipeline(db, monkeypatch):
    cids = _seed()
    monkeypatch.setattr("services.llm.call_llm", _fake_llm())

    t0 = time.time()
    for cid in cids:
        list(process.run_pipeline(cid, "", lambda cur, txt: ("p", "pid")))
    elapsed = time.time() - t0

    # Целостность: у каждого контракта ровно SERVICES строк (full_replace без дублей)
    bad = 0
    for cid in cids:
        if len(spec_current.list_by_contract(cid)) != SERVICES:
            bad += 1
    assert bad == 0, f"{bad} контрактов с неверным числом строк"

    total_events = query("SELECT count(*) AS c FROM spec_events", ())
    print(
        f"\n[load] контрактов={N_CONTRACTS} допников={M_SUPPS} услуг={SERVICES} "
        f"событий={total_events[0]['c']} время={elapsed:.1f}с"
    )