"""Otomasyon modülü: kademe/rastgele seçim, dış test ekleme, uçtan uca kural tabanlı çalıştırma ve raporlar."""
import io
import json
import zipfile
from collections import Counter

import pytest

from otomasyon import calistir, isler, katalog, modeller
from otomasyon.metin import M


def test_tiers():
    assert modeller.tier("claude|claude-opus-5") == 1
    assert modeller.tier("claude|claude-fable-5-1") == 1
    assert modeller.tier("claude|claude-sonnet-5-5") == 2
    assert modeller.tier("claude|claude-haiku-4-5-20251001") == 3
    assert modeller.tier("claude|claude-3-opus-20240229") == 3
    assert modeller.tier("openai|gpt-6-astra") == 1
    assert modeller.tier("openai|gpt-5-mini") == 2
    assert modeller.tier("openai|gpt-4.1-nano") == 3
    assert modeller.tier("ollama|qwen2.5-coder:32b") == 1
    assert modeller.tier("ollama|qwen2.5-coder:14b") == 2
    assert modeller.tier("ollama|qwen2.5-coder:7b") == 3
    assert modeller.tier("ollama|x:latest", {"note": "32.8B · Q4_K_M"}) == 1
    assert modeller.tier("kural") == 3


def test_picker_prefers_top_tier():
    pool = [{"secim": "a", "kademe": 1}, {"secim": "b", "kademe": 2}, {"secim": "c", "kademe": 3}]
    pk = modeller.Picker(pool, seed=1, bias=6)
    c = Counter(pk.pick()["secim"] for _ in range(3000))
    assert c["a"] > c["b"] > c["c"]
    assert c["a"] / 3000 > 0.5
    pk.drop("a")
    assert all(pk.pick()["secim"] != "a" for _ in range(50))
    assert pk.pick(exclude={"b", "c"}) is None


def test_metin_has_both_languages():
    for k, v in M.items():
        assert v.get("tr") and v.get("en"), k


def test_validate_questions():
    ok = [{"soru": "x?", "tip": "sayi", "cevap": 5, "belge": "a.pdf"}, {"soru": "y?", "tip": "cevapsiz"}]
    assert katalog.validate_questions(ok, {"a.pdf"}, False) == []
    bad = [{"soru": "x?", "tip": "sayi", "cevap": "beş"}, {"soru": "", "tip": "metin"},
           {"soru": "z?", "tip": "liste", "cevap": "a"}, {"soru": "w?", "tip": "metin", "cevap": "k", "belge": "yok.pdf"}]
    errs = katalog.validate_questions(bad, {"a.pdf"}, False)
    assert len(errs) == 4


@pytest.fixture()
def tmp_dirs(tmp_path, monkeypatch):
    monkeypatch.setattr(katalog, "EXT_DIR", tmp_path / "testler")
    monkeypatch.setattr(isler, "JOBS_DIR", tmp_path / "isler")
    return tmp_path


def _sample_zip() -> bytes:
    csv = "İl;2023;2024\nAnkara;1.234,5;1.300,0\nİzmir;980,0;1.010,5\nToplam;2.214,5;2.310,5\n"
    qs = {"sorular": [
        {"soru": "Ankara'nın 2024 değeri nedir?", "tip": "sayi", "cevap": 1300.0, "belge": "iller.csv"},
        {"soru": "2024'te en yüksek değere sahip il hangisidir?", "tip": "metin", "cevap": "Ankara", "belge": "iller.csv"},
        {"soru": "Bursa'nın 2024 değeri nedir?", "tip": "cevapsiz", "belge": "iller.csv"}]}
    buf = io.BytesIO()
    with zipfile.ZipFile(buf, "w") as z:
        z.writestr("paket/iller.csv", csv.encode("utf-8"))
        z.writestr("paket/sorular.json", json.dumps(qs, ensure_ascii=False))
        z.writestr("paket/test.json", json.dumps({"ad": "İl değerleri", "aciklama": "küçük CSV testi"}))
    return buf.getvalue()


def test_add_external_and_run_end_to_end(tmp_dirs):
    test, errs = katalog.add_external_test("", "", False, zip_bytes=_sample_zip())
    assert errs == [] and test is not None
    assert test.name() == "İl değerleri" and test.n_docs() == 1 and len(test.questions()) == 3
    assert any(x.key == test.key for x in katalog.all_tests())
    spec = {"ad": "birim testi", "testler": [test.key, "pilot"], "kip": "single", "model": "kural", "modeller": [],
            "havuz": {}, "ust_agirlik": 6, "sinir": 4, "tohum": 1, "kendini_duzeltme": True, "max_deneme": 3,
            "yedek": True, "diller": ["tr", "en"], "bicimler": ["html", "pdf"]}
    d = isler.create_job(spec)
    calistir.main(d)
    st = isler.read_status(d)
    assert st["faz"] == "bitti", st.get("hata")
    assert [u["durum"] for u in st["birimler"]] == ["bitti", "bitti"]
    u = st["birimler"][0]
    assert u["yapilan"] == 3
    for k in ("tr_html", "tr_pdf", "en_html", "en_pdf"):
        p = d / u["klasor"] / u["raporlar"][k]
        assert p.exists() and p.stat().st_size > 1000
    assert (d / "genel_rapor_tr.pdf").read_bytes()[:4] == b"%PDF"
    assert "İl değerleri" in (d / "genel_rapor_tr.html").read_text(encoding="utf-8")
    z = zipfile.ZipFile(d / "tum_raporlar.zip")
    assert sum(n.endswith(".pdf") for n in z.namelist()) == 6
    # tek başına genel rapor: bağlantı yok; paketteki genel rapor: test raporlarına göreli bağlantı var
    alone = (d / "genel_rapor_tr.html").read_text(encoding="utf-8")
    assert "<a href" not in alone and "t2sTema" in alone
    packed = z.read(f"{d.name}/genel_rapor_tr.html").decode("utf-8")
    hrefs = [h for h in packed.split("href='")[1:]]
    assert hrefs and all(f"{d.name}/" + h.split("'")[0] in z.namelist() for h in hrefs)
    assert "t2sTema" in z.read(f"{d.name}/{u['klasor']}/rapor_en.html").decode("utf-8")
    assert isler.unacked_finished() == d
    isler.ack(d)
    assert isler.unacked_finished() is None
    assert katalog.delete_external_test(test.key)


def test_sampling_is_reproducible():
    qs = [{"i": i} for i in range(100)]
    a = calistir.sample_questions(qs, 10, 5, "x")
    assert a == calistir.sample_questions(qs, 10, 5, "x")
    assert a != calistir.sample_questions(qs, 10, 6, "x")
    assert calistir.sample_questions(qs, 0, 5, "x") == qs


def test_model_broken():
    recs = [{"durum": "cevap", "aciklama": ""}] * 5
    assert not calistir.model_broken({"kayitlar": recs}, 10, 2)
    assert calistir.model_broken({"kayitlar": recs}, 10, 6)
    assert calistir.model_broken(None, 0, 0)
    assert calistir.model_broken({"kayitlar": [{"durum": "istisna", "aciklama": "istisna: x"}] * 5}, 0, 0)


def test_health_check_filters_broken_models(tmp_path):
    import os
    import socket
    import subprocess
    import sys
    import time
    from pathlib import Path

    from t2s.llm import list_models
    from t2s.settings import Settings
    root = Path(__file__).resolve().parent.parent
    s = socket.socket()
    s.bind(("127.0.0.1", 0))
    port = s.getsockname()[1]
    s.close()
    p = subprocess.Popen([sys.executable, str(root / "tests" / "mock_llm_server.py"), str(port)],
                         env={**os.environ, "MOCK_EXTRA": "1"}, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
    try:
        url = f"http://127.0.0.1:{port}"
        for _ in range(50):
            try:
                list_models("ollama", None, url)
                break
            except Exception:  # noqa: BLE001
                time.sleep(0.1)
        cfg = Settings(tmp_path / "ayar.json")
        for prov, base in (("claude", url), ("openai", url + "/v1"), ("ollama", url)):
            cfg.conn(prov)["url"] = base
            if prov != "ollama":
                cfg.set_key(prov, "test")
            cfg.set_models(prov, [{"id": m.id, "label": m.label, "note": m.note}
                                  for m in list_models(prov, "test" if prov != "ollama" else None, base)])
        pool = [c["secim"] for c in modeller.candidates(cfg)]
        res = {r["secim"]: r for r in modeller.health_check(cfg, pool)}
        assert not res["openai|gpt-mock-bozuk"]["ok"]
        assert res["claude|claude-mock-large"]["ok"] and res["ollama|qwen2.5-coder:7b"]["ok"]
        assert res["claude|claude-mock-dengesiz"]["ok"]       # sınamayı geçer, işte çöker
    finally:
        p.terminate()


def test_costly_models_excluded_by_default():
    ch = ["openai|gpt-5", "openai|o3-pro", "openai|gpt-5-pro-2025", "openai|gpt-4o-search-preview",
          "claude|claude-opus-5", "ollama|qwen2.5-coder:7b"]
    assert modeller.default_excluded(ch) == ["openai|o3-pro", "openai|gpt-5-pro-2025", "openai|gpt-4o-search-preview"]