X-Financial/server/tests/test_expense_case_legacy_backfill.py

from __future__ import annotations

from datetime import UTC, datetime, timedelta
from decimal import Decimal

import pytest
from sqlalchemy import create_engine, func, select
from sqlalchemy.orm import Session, sessionmaker
from sqlalchemy.pool import StaticPool

from app.db.base import Base
from app.models.expense_case import BusinessEvent, ExpenseCase, ExpenseCaseLink
from app.models.financial_record import ExpenseClaim
from app.services.expense_case_legacy_backfill import (
    HISTORICAL_CLAIM_IMPORTED_EVENT,
    HISTORICAL_IMPORT_DELIVERY_STATUS,
    ExpenseCaseLegacyBackfillService,
    LegacyBackfillDisposition,
)

CUTOFF = datetime(2026, 7, 14, tzinfo=UTC)


def build_session() -> Session:
    engine = create_engine(
        "sqlite+pysqlite:///:memory:",
        connect_args={"check_same_thread": False},
        poolclass=StaticPool,
    )
    Base.metadata.create_all(bind=engine)
    return sessionmaker(bind=engine, autoflush=False, autocommit=False)()


def build_claim(claim_no: str, *, created_at: datetime | None = None) -> ExpenseClaim:
    return ExpenseClaim(
        claim_no=claim_no,
        employee_name="张三",
        department_name="市场部",
        project_code="PRJ-BACKFILL",
        expense_type="travel",
        reason="历史客户拜访",
        location="上海",
        amount=Decimal("128.50"),
        currency="CNY",
        invoice_count=1,
        occurred_at=datetime(2026, 7, 10, 9, 0, tzinfo=UTC),
        submitted_at=datetime(2026, 7, 11, 10, 0, tzinfo=UTC),
        status="submitted",
        approval_stage="直属领导审批",
        risk_flags_json=[],
        created_at=created_at or datetime(2026, 7, 10, 8, 0, tzinfo=UTC),
        updated_at=datetime(2026, 7, 11, 10, 0, tzinfo=UTC),
    )


def seed_link(db: Session, claim: ExpenseClaim, *, tenant_id: str = "default") -> None:
    expense_case = ExpenseCase(
        tenant_id=tenant_id,
        case_no=f"CASE-{claim.claim_no}",
        scene_code="travel",
        title=claim.reason,
        current_stage="reviewing",
        status="active",
    )
    db.add(expense_case)
    db.flush()
    db.add(
        ExpenseCaseLink(
            tenant_id=tenant_id,
            expense_case_id=expense_case.id,
            resource_type="expense_claim",
            resource_id=claim.id,
            relation_type="claim",
        )
    )
    db.flush()


def seed_event_without_link(db: Session, claim: ExpenseClaim) -> None:
    expense_case = ExpenseCase(
        tenant_id="default",
        case_no=f"CASE-ORPHAN-{claim.claim_no}",
        scene_code="travel",
        title=claim.reason,
        current_stage="reviewing",
        status="active",
    )
    db.add(expense_case)
    db.flush()
    db.add(
        BusinessEvent(
            tenant_id="default",
            expense_case_id=expense_case.id,
            aggregate_type="expense_claim",
            aggregate_id=claim.id,
            event_type="claim_submitted",
            event_version=1,
            idempotency_key=f"orphan:{claim.id}",
            correlation_id="orphan-event",
            actor_id="system",
            actor_type="system",
            payload_json={},
            delivery_status="pending",
            occurred_at=datetime(2026, 7, 11, tzinfo=UTC),
        )
    )
    db.flush()


def table_counts(db: Session) -> tuple[int, int, int]:
    return (
        db.scalar(select(func.count()).select_from(ExpenseCase)) or 0,
        db.scalar(select(func.count()).select_from(ExpenseCaseLink)) or 0,
        db.scalar(select(func.count()).select_from(BusinessEvent)) or 0,
    )


def test_requires_explicit_tenant_and_timezone_aware_cutoff() -> None:
    with build_session() as db:
        with pytest.raises(ValueError, match="tenant_id"):
            ExpenseCaseLegacyBackfillService(db, tenant_id="", cutoff=CUTOFF)
        with pytest.raises(ValueError, match="timezone"):
            ExpenseCaseLegacyBackfillService(
                db,
                tenant_id="default",
                cutoff=datetime(2026, 7, 14),
            )


def test_fingerprint_is_stable_and_idempotency_key_is_tenant_scoped() -> None:
    with build_session() as db:
        claim = build_claim("EXP-HISTORY-FINGERPRINT")
        db.add(claim)
        db.flush()
        default_service = ExpenseCaseLegacyBackfillService(
            db,
            tenant_id="default",
            cutoff=CUTOFF,
        )
        tenant_service = ExpenseCaseLegacyBackfillService(
            db,
            tenant_id="tenant-b",
            cutoff=CUTOFF,
        )

        first_fingerprint = default_service.source_fingerprint(claim)
        assert default_service.source_fingerprint(claim) == first_fingerprint
        assert default_service.idempotency_key(claim.id) == (
            f"historical-import:v1:default:{claim.id}"
        )
        assert tenant_service.idempotency_key(claim.id) == (
            f"historical-import:v1:tenant-b:{claim.id}"
        )

        claim.status = "approved"
        assert default_service.source_fingerprint(claim) != first_fingerprint


def test_long_tenant_idempotency_key_is_stably_hashed_to_database_limit() -> None:
    with build_session() as db:
        service = ExpenseCaseLegacyBackfillService(
            db,
            tenant_id="t" * 64,
            cutoff=CUTOFF,
        )

        first = service.idempotency_key("c" * 36)
        second = service.idempotency_key("c" * 36)

        assert first == second
        assert first.startswith("historical-import:v1:sha256:")
        assert len(first) <= 120


def test_preview_is_read_only_and_classifies_linked_and_orphan_event_conflicts() -> None:
    with build_session() as db:
        eligible = build_claim("EXP-HISTORY-ELIGIBLE")
        linked = build_claim("EXP-HISTORY-LINKED", created_at=CUTOFF - timedelta(days=3))
        conflict = build_claim("EXP-HISTORY-CONFLICT", created_at=CUTOFF - timedelta(days=2))
        future = build_claim("EXP-HISTORY-FUTURE", created_at=CUTOFF + timedelta(seconds=1))
        db.add_all([eligible, linked, conflict, future])
        db.flush()
        seed_link(db, linked)
        seed_event_without_link(db, conflict)
        db.commit()

        counts_before = table_counts(db)
        preview = ExpenseCaseLegacyBackfillService(
            db,
            tenant_id="default",
            cutoff=CUTOFF,
        ).preview(batch_size=10)

        assert preview.inspected == 3
        assert preview.eligible == 1
        assert preview.linked == 1
        assert preview.conflicts == 1
        assert not preview.has_more
        assert table_counts(db) == counts_before
        dispositions = {item.claim_no: item.disposition for item in preview.items}
        assert dispositions == {
            "EXP-HISTORY-ELIGIBLE": LegacyBackfillDisposition.ELIGIBLE,
            "EXP-HISTORY-LINKED": LegacyBackfillDisposition.LINKED,
            "EXP-HISTORY-CONFLICT": LegacyBackfillDisposition.CONFLICT,
        }


def test_apply_batch_writes_honest_snapshot_without_committing_and_is_idempotent() -> None:
    with build_session() as db:
        claim = build_claim("EXP-HISTORY-APPLY")
        db.add(claim)
        db.commit()
        backfilled_at = datetime(2026, 7, 14, 12, 30, tzinfo=UTC)
        service = ExpenseCaseLegacyBackfillService(
            db,
            tenant_id="default",
            cutoff=CUTOFF,
        )

        result = service.apply_batch(
            run_id="legacy-backfill-run-1",
            batch_size=10,
            backfilled_at=backfilled_at,
        )

        assert result.created == 1
        assert result.skipped_linked == 0
        assert result.conflicts == 0
        event = db.scalar(select(BusinessEvent))
        assert event is not None
        assert event.event_type == HISTORICAL_CLAIM_IMPORTED_EVENT
        assert event.idempotency_key == (f"historical-import:v1:default:{claim.id}")
        assert event.idempotency_key == service.idempotency_key(claim.id)
        assert event.actor_id == "system"
        assert event.actor_type == "system"
        assert event.delivery_status == HISTORICAL_IMPORT_DELIVERY_STATUS
        assert event.occurred_at == backfilled_at
        assert event.payload_json["history_reconstructed"] is False
        assert event.payload_json["backfill_version"] == 1
        assert event.payload_json["backfilled_at"] == "2026-07-14T12:30:00Z"
        assert event.payload_json["performed_by"] == "expense_case_legacy_backfill_cli"
        assert event.payload_json["source_times"] == {
            "occurred_at": "2026-07-10T09:00:00Z",
            "submitted_at": "2026-07-11T10:00:00Z",
            "created_at": "2026-07-10T08:00:00Z",
            "updated_at": "2026-07-11T10:00:00Z",
        }
        assert event.payload_json["source_fingerprint"].startswith("sha256:")
        assert "未重建" in event.payload_json["reason"]

        db.rollback()
        assert table_counts(db) == (0, 0, 0)

        first_apply = service.apply_batch(
            run_id="legacy-backfill-run-2",
            batch_size=10,
            backfilled_at=backfilled_at,
        )
        db.commit()
        repeated_apply = service.apply_batch(
            run_id="legacy-backfill-run-3",
            batch_size=10,
            backfilled_at=backfilled_at + timedelta(minutes=1),
        )

        assert first_apply.created == 1
        assert repeated_apply.created == 0
        assert repeated_apply.skipped_linked == 1
        assert table_counts(db) == (1, 1, 1)


def test_apply_batch_reports_event_without_link_as_conflict() -> None:
    with build_session() as db:
        claim = build_claim("EXP-HISTORY-ORPHAN")
        db.add(claim)
        db.flush()
        seed_event_without_link(db, claim)
        db.commit()

        result = ExpenseCaseLegacyBackfillService(
            db,
            tenant_id="default",
            cutoff=CUTOFF,
        ).apply_batch(
            run_id="legacy-backfill-conflict",
            batch_size=10,
        )

        assert result.created == 0
        assert result.conflicts == 1
        assert result.items[0].disposition is LegacyBackfillDisposition.CONFLICT
        assert "without an expense case link" in result.items[0].reason
        assert db.scalar(select(func.count()).select_from(ExpenseCaseLink)) == 0


def test_preview_rejects_damaged_link_tenant_and_unlinked_case() -> None:
    with build_session() as db:
        damaged_link_claim = build_claim("EXP-HISTORY-DAMAGED-LINK")
        unlinked_case_claim = build_claim(
            "EXP-HISTORY-UNLINKED-CASE",
            created_at=CUTOFF - timedelta(days=2),
        )
        db.add_all([damaged_link_claim, unlinked_case_claim])
        db.flush()

        foreign_case = ExpenseCase(
            tenant_id="other-tenant",
            case_no=f"CASE-{damaged_link_claim.claim_no}",
            scene_code="travel",
            title=damaged_link_claim.reason,
            current_stage="reviewing",
            status="active",
        )
        orphan_case = ExpenseCase(
            tenant_id="default",
            case_no=f"CASE-{unlinked_case_claim.claim_no}",
            scene_code="travel",
            title=unlinked_case_claim.reason,
            current_stage="reviewing",
            status="active",
        )
        db.add_all([foreign_case, orphan_case])
        db.flush()
        db.add(
            ExpenseCaseLink(
                tenant_id="default",
                expense_case_id=foreign_case.id,
                resource_type="expense_claim",
                resource_id=damaged_link_claim.id,
                relation_type="claim",
            )
        )
        db.commit()

        preview = ExpenseCaseLegacyBackfillService(
            db,
            tenant_id="default",
            cutoff=CUTOFF,
        ).preview(batch_size=10)

        assert preview.conflicts == 2
        reasons = {item.claim_no: item.reason for item in preview.items}
        assert "invalid tenant" in reasons["EXP-HISTORY-DAMAGED-LINK"]
        assert "without an expense claim link" in reasons["EXP-HISTORY-UNLINKED-CASE"]


def test_preview_cursor_provides_stable_keyset_batches() -> None:
    with build_session() as db:
        first = build_claim("EXP-HISTORY-001", created_at=CUTOFF - timedelta(days=3))
        second = build_claim("EXP-HISTORY-002", created_at=CUTOFF - timedelta(days=2))
        db.add_all([first, second])
        db.commit()
        service = ExpenseCaseLegacyBackfillService(
            db,
            tenant_id="default",
            cutoff=CUTOFF,
        )

        page_one = service.preview(batch_size=1)
        page_two = service.preview(batch_size=1, after=page_one.next_cursor)

        assert page_one.has_more
        assert page_one.items[0].claim_no == "EXP-HISTORY-001"
        assert not page_two.has_more
        assert page_two.items[0].claim_no == "EXP-HISTORY-002"


def test_standard_event_delivery_status_remains_pending_by_default() -> None:
    with build_session() as db:
        claim = build_claim("EXP-NORMAL-EVENT")
        db.add(claim)
        db.commit()

        _expense_case, event = ExpenseCaseLegacyBackfillService(
            db,
            tenant_id="default",
            cutoff=CUTOFF,
        ).expense_cases.record_claim_event(
            claim,
            event_type="claim_submitted",
            actor_id="owner@example.com",
            tenant_id="default",
        )

        assert event.delivery_status == "pending"