"""总览报表 —— 「做成了多少」,以及它必须和治理待办说同一件事。

这个文件最重要的不是某个数字对不对,是**两处口径不能分叉**:
报表里的「还没认出是谁」和治理面的 `count_unlinked` 必须永远相等。
它们分叉的表现是界面上两个 tab 显示两个数,而谁都说不清该信哪个 ——
而这正是本域已经踩过一次的形状(「认不出是谁会自己收敛」那句话在四处
文档里各写各的,直到实测把它证伪)。
"""
from __future__ import annotations

import uuid

import pytest

from app.domains.github_org import constants as ghc
from app.domains.github_org import report
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", **kw):
    o = await gh.create_org(db, slug=slug, display_name=slug, **kw)
    await db.commit()
    return o


async def _person(db, *, status="active"):
    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="Some One", employment_status=status,
    )
    db.add(p)
    await db.flush()
    return p


async def _m(db, org, login, *, person=None, source="import",
             state=ghc.MEMBERSHIP_STATE_ACTIVE):
    row = await gh.upsert_membership(
        db, org=org, github_login=login,
        person_id=str(person.id) if person else None,
        state=state, source=source, actor_kind="integration", trace_id="t",
    )
    await db.flush()
    return row


# ── 口径一致性:报表不能和治理面分叉 ────────────────────────────


async def test_not_yet_identified_matches_the_governance_count(db_session):
    """🔴 同一个数,两个页面,必须相等。

    报表的分母口径若和 ``count_unlinked`` 不同,用户会在「待处理」看到一个
    数、在「总览」看到另一个数 —— 然后不再信任两者中的任何一个。
    """
    org = await _org(db_session)
    p = await _person(db_session)
    await _m(db_session, org, "known", person=p)
    a = await _m(db_session, org, "stranger-a")
    await _m(db_session, org, "stranger-b")
    await db_session.commit()

    # 把其中一条确认成外部人员 —— 它必须同时从两处的缺口里消失
    await gh.confirm_external(db_session, a, actor_person_id=None)
    await db_session.commit()

    rep = await report.build_report(db_session)
    assert rep["totals"]["not_yet_identified"] == await gh.count_unlinked(db_session)
    assert rep["totals"]["confirmed_external"] == await gh.count_confirmed_external(db_session)
    assert rep["totals"]["not_yet_identified"] == 1


async def test_leavers_still_in_orgs_matches_the_governance_number(db_session):
    """这个数是这套东西的成败本身,两处必须一致。"""
    org = await _org(db_session, "pingcap-inc")
    leaver = await _person(db_session, status="left")
    await _m(db_session, org, "goneperson", person=leaver)
    await db_session.commit()

    rep = await report.build_report(db_session)
    assert rep["offboarding"]["leavers_still_in_orgs"] == 1
    assert rep["offboarding"]["leavers_still_in_orgs"] == (
        await gh.count_leavers_still_in_orgs(db_session)
    )


async def test_exempt_orgs_are_excluded_from_the_leaver_number(db_session):
    """豁免组织不算 —— 与治理面同一条规则。把它们算进来只会训练人忽略这个数。"""
    org = await _org(
        db_session, "tikv", offboarding_policy=ghc.OFFBOARDING_POLICY_EXEMPT,
    )
    leaver = await _person(db_session, status="left")
    await _m(db_session, org, "maintainer", person=leaver)
    await db_session.commit()
    rep = await report.build_report(db_session)
    assert rep["offboarding"]["leavers_still_in_orgs"] == 0


# ── 来源分布:这一页真正要说的事 ──────────────────────────────


async def test_source_split_separates_automation_from_discovery(db_session):
    """🔴 报表的核心断言。

    `import` / `reconcile` 证明的恰恰是「在此之前没有任何流程管过」;
    只有 `approval` / `auto_join` 才是这套自动化真的交付了的部分。
    把它们混在一个「成员总数」里,这个区别就消失了。
    """
    org = await _org(db_session)
    await _m(db_session, org, "a", source=ghc.SOURCE_IMPORT)
    await _m(db_session, org, "b", source=ghc.SOURCE_IMPORT)
    await _m(db_session, org, "c", source=ghc.SOURCE_RECONCILE)
    await _m(db_session, org, "d", source=ghc.SOURCE_APPROVAL)
    await _m(db_session, org, "e", source=ghc.SOURCE_MANUAL)
    await db_session.commit()

    src = (await report.build_report(db_session))["by_source"]
    assert src["total"] == 5
    assert src["discovered"] == 3      # import ×2 + reconcile ×1
    assert src["automated"] == 1       # approval ×1
    assert src["manual"] == 1
    assert src["automated_pct"] == 20.0


async def test_an_empty_ledger_reports_none_not_a_fake_zero_percent(db_session):
    """没有任何成员时,「自动化覆盖率」是**未知**,不是 0%。

    0% 读起来像「跑了但一条都没覆盖」,而真相是「还没有东西可覆盖」。
    """
    await _org(db_session)
    rep = await report.build_report(db_session)
    assert rep["by_source"]["total"] == 0
    assert rep["by_source"]["automated_pct"] is None
    assert rep["totals"]["identified_pct"] is None


# ── 趋势:读事件流,不拿当前状态倒推 ──────────────────────────


async def test_trend_counts_events_not_current_state(db_session):
    """🔴 一条「加入后又被移除」的成员资格,在趋势里应该同时出现在两条线上。

    拿当前状态倒推历史会把它算成从没发生过 —— 而那恰恰是最该被看见的一种
    事件(有人进来过又走了)。
    """
    org = await _org(db_session)
    m = await _m(db_session, org, "inandout", state=ghc.MEMBERSHIP_STATE_ACTIVE)
    await db_session.commit()
    await gh.mark_removed(
        db_session, m, reason="offboarding",
        removal_source=ghc.REMOVAL_SOURCE_OFFBOARDING,
        actor_person_id=None, actor_kind="integration", trace_id="t",
    )
    await db_session.commit()

    trend = (await report.build_report(db_session))["trend"]
    total_removed = sum(p["removed"] for p in trend["points"])
    assert total_removed == 1, trend
    # 这条记录当前是 removed,但它「来过」这件事仍然记在事件流里
    assert any(p["joined"] > 0 for p in trend["points"]), trend


async def test_trend_window_is_bounded(db_session):
    """窗口要真的收窄扫描范围,不是查全表再丢掉。"""
    await _org(db_session)
    rep = await report.build_report(db_session, days=7)
    assert rep["trend"]["days"] == 7


# ── 坐席:报表里也不能编数字 ─────────────────────────────────


async def test_uncapped_plans_report_no_remaining_in_the_report_too(db_session):
    """🔴 生产实测:PingCAP-QE 是 free 计划,报 filled=203 / total=92。

    按字面算余量是 **-111**。报表和治理面必须用同一个 `seat_remaining`,
    否则会出现「待办说没事、报表说欠 111 个坐席」这种谁也解释不了的分叉。
    """
    org = await _org(db_session, "pingcap-qe")
    org.seat_plan = "free"
    org.seat_filled = 203
    org.seat_total = 92
    await db_session.commit()

    row = next(
        o for o in (await report.build_report(db_session))["by_org"]
        if o["org_slug"] == "pingcap-qe"
    )
    assert row["seat_remaining"] is None
    assert row["seat_filled"] == 203  # 原始数照实给出,只是不算余量


async def test_capped_plan_reports_a_real_remaining(db_session):
    org = await _org(db_session, "pingcap-inc")
    org.seat_plan = "business"
    org.seat_filled = 260
    org.seat_total = 263
    await db_session.commit()
    row = next(
        o for o in (await report.build_report(db_session))["by_org"]
        if o["org_slug"] == "pingcap-inc"
    )
    assert row["seat_remaining"] == 3


# ── 每组织构成 ───────────────────────────────────────────────


async def test_per_org_identified_pct_excludes_confirmed_external_from_the_gap(
    db_session,
):
    """已确认是外部人员的行是**已处理**,不该继续压低覆盖率。"""
    org = await _org(db_session)
    p = await _person(db_session)
    await _m(db_session, org, "known", person=p)
    ext = await _m(db_session, org, "outsider")
    await db_session.commit()
    await gh.confirm_external(db_session, ext, actor_person_id=None)
    await db_session.commit()

    row = (await report.build_report(db_session))["by_org"][0]
    assert row["live_total"] == 2
    assert row["identified"] == 1
    assert row["confirmed_external"] == 1
    assert row["not_yet_identified"] == 0


async def test_an_org_with_no_members_still_appears(db_session):
    """空组织必须出现在报表里 —— 「一个人都没有」是结论,不是缺数据。"""
    await _org(db_session, "emptyorg")
    rep = await report.build_report(db_session)
    row = next(o for o in rep["by_org"] if o["org_slug"] == "emptyorg")
    assert row["live_total"] == 0
    assert row["identified_pct"] is None