"""机器账号的策略标记。

这个文件钉住的核心是**为什么它只能是策略、不能是自动判定**:

1. GitHub 的 `type` 字段在本租户里全部报 `User`(七个 bot 账号实测),
   所以没有权威信号可用;
2. 只能靠命名约定,而命名约定**会误伤真人** —— `robot-liu` 命中 `*robot*`;
3. 因此写入永远经过人确认,没有定时自动版本。

第 2 条有一条专门的用例。它不是理论上的担心:把一个员工标成机器账号,
后果是他离职时不被回收 —— 和挂错人一样,是一次静默的权限泄漏。
"""
from __future__ import annotations

import uuid

import pytest

from app.domains.github_org import bot_policy, report
from app.domains.github_org import constants as ghc
from app.domains.github_org import service as gh
from app.identity.person_models import IdentityPerson

pytestmark = pytest.mark.asyncio


async def _org(db, slug="pingcap"):
    o = await gh.create_org(db, slug=slug, display_name=slug)
    await db.commit()
    return o


async def _m(db, org, login, *, profile=None):
    m = await gh.upsert_membership(
        db, org=org, github_login=login, person_id=None,
        state=ghc.MEMBERSHIP_STATE_ACTIVE, source="reconcile",
        actor_kind="integration", trace_id="t",
    )
    if profile is not None:
        m.github_profile = profile
    await db.flush()
    return m


# ── 命名策略 ─────────────────────────────────────────────────────


def test_the_production_bot_handles_all_match():
    """生产台账里真实存在的那批,策略必须都认出来。"""
    pats = list(bot_policy.DEFAULT_PATTERNS)
    for h in ("sre-bot", "TestInfra-Bot", "ti-chi-bot", "ti-srebot",
              "tidb-ci-bot", "tidbcloud-bot", "pingcap-zendesk-bot"):
        assert bot_policy.match_reason(h, pats, None) is not None, h


def test_real_contributor_handles_do_not_match():
    """生产上真实的外部贡献者 handle,不能被误判。"""
    pats = list(bot_policy.DEFAULT_PATTERNS)
    for h in ("Ehco1996", "zz-jason", "hawkingrei", "suzaku", "cicada-lewis",
              "birdstorm", "CalvinNeo", "anotherrachel", "innerr"):
        assert bot_policy.match_reason(h, pats, None) is None, h


def test_a_human_named_robot_is_a_known_false_positive():
    """🔴 这条用例是整个「只建议、不自动落库」设计的理由本身。

    `robot-liu` 命中 `*robot*`。策略**必须**会误伤 —— 我们没有更好的信号
    (GitHub 的 type 字段在本租户全报 User)。所以这里断言的不是「它不匹配」,
    而是「它确实匹配,因此必须有人看一眼」。

    如果哪天有人给这条加了自动标记的 cron,这个注释就是那次改动要面对的
    反例:一个真名叫 robot 的员工被标成机器账号,他离职时不会被回收。
    """
    assert bot_policy.match_reason(
        "robot-liu", list(bot_policy.DEFAULT_PATTERNS), None,
    ) is not None


def test_the_profile_name_strengthens_the_reason_but_never_creates_it():
    """资料名只能加强理由,不能单独构成理由 —— 一个叫「CI Bot」的人
    如果 handle 完全不像机器账号,不该被建议。"""
    pats = list(bot_policy.DEFAULT_PATTERNS)
    strong = bot_policy.match_reason("sre-bot", pats, {"name": "SRE bot"})
    assert "资料名" in strong
    assert bot_policy.match_reason("zhang.san", pats, {"name": "CI Bot"}) is None


def test_a_real_github_app_bot_type_outranks_the_naming_convention():
    """GitHub App 身份确实会报 type=Bot。本租户从没命中过,但留着,
    并且它的理由应该压过命名约定 —— 那是权威信号。"""
    r = bot_policy.match_reason(
        "dependabot", list(bot_policy.DEFAULT_PATTERNS),
        {"account_type": "Bot"},
    )
    assert "GitHub 账号类型就是 Bot" in r


# ── 策略读写 ─────────────────────────────────────────────────────


async def _seed_bot_label(db):
    """586 起 bot 策略从标签规则派生,所以要先有 ``account:bot`` 这个标签。"""
    from app.domains.github_org import labels as lb

    await lb.upsert_label(
        db, key="account:bot", category="account",
        display_zh="机器账号", display_en="Machine account",
    )
    await db.commit()


async def test_no_bot_rules_falls_back_instead_of_matching_nothing(db_session):
    """🔴 一条 bot 规则都没有时退回内置默认。

    空策略静默地什么都不匹配,而那看起来和「没有机器账号」一模一样。
    586 起这个「空」的含义是「标签规则里没有 account:bot 的 handle_glob」。
    """
    await _seed_bot_label(db_session)
    from app.domains.github_org import labels as lb

    await lb.set_rules(db_session, [])
    await db_session.commit()
    assert await bot_policy.get_patterns(db_session) == list(
        bot_policy.DEFAULT_PATTERNS
    )


async def test_setting_an_empty_policy_is_refused(db_session):
    await _seed_bot_label(db_session)
    with pytest.raises(ValueError):
        await bot_policy.set_patterns(db_session, [], actor_person_id=None)
    with pytest.raises(ValueError):
        await bot_policy.set_patterns(db_session, ["  "], actor_person_id=None)


async def test_patterns_round_trip_through_the_shared_rule_store(db_session):
    """🔴 写进去的是**标签规则**,不是第二个存储。

    两个存储并存的必然结果是它们不一致 —— 同一个问题两个界面给两个答案。
    """
    await _seed_bot_label(db_session)
    await bot_policy.set_patterns(
        db_session, ["*-runner", " *-bot "], actor_person_id=None,
    )
    await db_session.commit()
    assert await bot_policy.get_patterns(db_session) == ["*-runner", "*-bot"]

    # 确认它真的落在标签规则里,而不是自己那个设置项
    from app.domains.github_org import labels as lb

    rules = await lb.get_rules(db_session)
    bot_globs = {
        r["value"] for r in rules
        if r["label"] == "account:bot" and r["match"] == lb.MATCH_HANDLE_GLOB
    }
    assert bot_globs == {"*-runner", "*-bot"}


async def test_changing_bot_patterns_does_not_drop_other_label_rules(db_session):
    """🔴 改 bot 策略不能顺手删掉 employer / community 那些规则。"""
    await _seed_bot_label(db_session)
    from app.domains.github_org import labels as lb

    await lb.upsert_label(
        db_session, key="employer:bytedance", category="employer",
        display_zh="ByteDance", display_en="ByteDance",
    )
    await lb.set_rules(db_session, [
        {"match": "profile_company_contains", "value": "bytedance",
         "label": "employer:bytedance"},
        {"match": "handle_glob", "value": "*-bot", "label": "account:bot"},
    ])
    await db_session.commit()

    await bot_policy.set_patterns(db_session, ["*-runner"], actor_person_id=None)
    await db_session.commit()

    rules = await lb.get_rules(db_session)
    assert any(r["label"] == "employer:bytedance" for r in rules), rules
    assert {r["value"] for r in rules if r["label"] == "account:bot"} == {"*-runner"}


# ── 预演 ─────────────────────────────────────────────────────────


async def test_preview_aggregates_one_handle_across_orgs(db_session):
    """一个 bot 通常在多个组织里(ti-chi-bot 在 5 个)。一次确认应该覆盖
    它的所有行,而不是让人对同一个账号点五次。"""
    a = await _org(db_session, "pingcap")
    b = await _org(db_session, "tikv")
    await _m(db_session, a, "ti-chi-bot")
    await _m(db_session, b, "ti-chi-bot")
    await db_session.commit()

    pv = await bot_policy.preview(db_session)
    assert pv["handles"] == 1
    assert pv["memberships"] == 2
    assert pv["items"][0]["org_slugs"] == ["pingcap", "tikv"]


async def test_preview_skips_rows_already_confirmed(db_session):
    """人做过的判断不能被重新捞出来问一遍。"""
    org = await _org(db_session)
    m = await _m(db_session, org, "sre-bot")
    await db_session.commit()
    assert (await bot_policy.preview(db_session))["handles"] == 1

    await gh.confirm_external(
        db_session, m, actor_person_id=None, kind=ghc.EXTERNAL_KIND_BOT,
    )
    await db_session.commit()
    assert (await bot_policy.preview(db_session))["handles"] == 0


async def test_preview_skips_rows_already_linked_to_a_person(db_session):
    """已经认出是谁的行,不该再被当成机器账号候选。"""
    org = await _org(db_session)
    p = IdentityPerson(
        id=uuid.uuid4(), source_system="test",
        hris_person_id=f"P{uuid.uuid4().hex[:8]}",
        work_email=f"u{uuid.uuid4().hex[:8]}@pingcap.com",
        display_name="Robot Liu", employment_status="active",
    )
    db_session.add(p)
    await db_session.flush()
    await gh.upsert_membership(
        db_session, org=org, github_login="robot-liu", person_id=str(p.id),
        state=ghc.MEMBERSHIP_STATE_ACTIVE, source="reconcile",
        actor_kind="integration", trace_id="t",
    )
    await db_session.commit()
    assert (await bot_policy.preview(db_session))["handles"] == 0


# ── kind 落库与报表分组 ─────────────────────────────────────────


async def test_kind_is_recorded_and_split_in_the_report(db_session):
    """机器账号和「别家公司的人」在报表里必须分开 —— 混成一格,
    「已确认外部 N 条」就再也回答不了任何问题。"""
    org = await _org(db_session)
    bot = await _m(db_session, org, "sre-bot")
    human = await _m(db_session, org, "someone-else")
    await db_session.commit()
    await gh.confirm_external(
        db_session, bot, actor_person_id=None, kind=ghc.EXTERNAL_KIND_BOT,
    )
    await gh.confirm_external(
        db_session, human, actor_person_id=None,
        kind=ghc.EXTERNAL_KIND_OUTSIDE_HUMAN,
    )
    await db_session.commit()

    t = (await report.build_report(db_session))["totals"]
    assert t["confirmed_external"] == 2
    assert t["bots"] == 1
    assert t["outside_humans"] == 1


async def test_an_unrecognised_kind_lands_as_none_not_stored_verbatim(db_session):
    """🔴 拼错的 'Bot' 不能原样存下来 —— 那会让按 kind 的分组凭空多出
    一类,而那一类在界面上没有任何解释。"""
    org = await _org(db_session)
    m = await _m(db_session, org, "sre-bot")
    await db_session.commit()
    await gh.confirm_external(
        db_session, m, actor_person_id=None, kind="Bot",  # 大写,不在枚举里
    )
    await db_session.commit()
    assert m.external_confirmed_at is not None   # 确认本身有效
    assert m.external_kind is None               # 但 kind 落 None


async def test_clearing_external_also_clears_the_kind(db_session):
    org = await _org(db_session)
    m = await _m(db_session, org, "sre-bot")
    await db_session.commit()
    await gh.confirm_external(
        db_session, m, actor_person_id=None, kind=ghc.EXTERNAL_KIND_BOT,
    )
    await db_session.commit()
    await gh.clear_external(db_session, m, actor_person_id=None)
    await db_session.commit()
    assert m.external_kind is None


async def test_linking_a_person_also_clears_the_kind(db_session):
    """先标成机器账号、后来发现是个人 —— 那个 kind 必须一起清掉。"""
    org = await _org(db_session)
    m = await _m(db_session, org, "robot-liu")
    await db_session.commit()
    await gh.confirm_external(
        db_session, m, actor_person_id=None, kind=ghc.EXTERNAL_KIND_BOT,
    )
    await db_session.commit()

    p = IdentityPerson(
        id=uuid.uuid4(), source_system="test",
        hris_person_id=f"P{uuid.uuid4().hex[:8]}",
        work_email=f"u{uuid.uuid4().hex[:8]}@pingcap.com",
        display_name="Robot Liu", employment_status="active",
    )
    db_session.add(p)
    await db_session.flush()
    await gh.relink_membership(
        db_session, m, person_id=str(p.id), actor_person_id=None,
    )
    await db_session.commit()
    assert m.external_kind is None
    assert m.external_confirmed_at is None