实现租约恢复核对：verify_on_site 接真实订单列表反查

新增 SiteInteractor.list_recent_orders（分页拉 order.my.rakuten.co.jp 订单列表，按商品 URL 反查）+ GatewayClient.get_task，替换掉恒返回 UNKNOWN 的桩。 NOT_ORDERED 分支目前只有逻辑验证、没有真实多单数据支撑，刻意仍路由到 needs_human，不自动重新下单。全程只读查询，未触发任何真实付款操作。
2026-08-13 23:58:54 +08:00
parent 51e2538438
commit 93c1a83406
8 changed files with 868 additions and 22 deletions
@@ -15,18 +15,26 @@ from typing import Any

 import pytest

+from datetime import datetime, timezone
+
 from app.shared.errors import CartOperationError, InvalidRequestError, OrderOperationError
 from app.trading.worker.models import LeaseTask
 from app.shared.task_state import OrderState
 from app.trading.worker.site_interact import (
    CheckoutSummary,
+    OrderListEntry,
+    OrderListPage,
    OrderStatusSnapshot,
    SiteInteractor,
+    _accumulate_order_list_page,
    _extract_error_message,
    _extract_purchase_fields,
+    _OrderListAccumulator,
    _parse_checkout_summary,
    _parse_initial_state,
+    _parse_order_list,
    _parse_order_status,
+    parse_order_datetime,
 )

 FIXTURES = Path(__file__).parent / "fixtures"
@@ -329,6 +337,186 @@ async def test_check_order_status_without_start_raises():
        await site.check_order_status(_REAL_ORDER_ID)


+# ---- _parse_order_list：verify_on_site 恢复核对用，2026-08-13 用真实账号跑
+# order.my.rakuten.co.jp/purchase-history/order-list 验证过这套 __INITIAL_STATE__
+# 结构（pageType="ph-list" 时 orderListData 直接是结构化 JSON，不需要正则抠 DOM）。
+# 下面 fixture 里的字段名/嵌套结构与真实响应一致，只是精简掉了大量与匹配逻辑
+# 无关的字段（推荐组件、通知、会员信息等）。
+
+
+def _real_order_list_state(
+    *,
+    orders_found: int = 1,
+    page_size: int = 25,
+    orders: list[dict] | None = None,
+    page_type: str = "ph-list",
+) -> dict:
+    if orders is None:
+        orders = [
+            {
+                "orderNumber": _REAL_ORDER_ID,
+                "orderDate": "2026-08-13T09:47:28.000Z",
+                "shopId": 306087,
+                "shopName": "BACKYARD FAMILY インテリアタウン",
+                "items": [
+                    {
+                        "itemId": 10422789,
+                        "itemName": "ウォールフック 粘着フック",
+                        "itemUrl": "https://item.rakuten.co.jp/moccasin/ds001iwrgesaaa2/?variantId=ds001iwrgesaaa2-1a-2a",
+                        "itemPrice": 297,
+                    }
+                ],
+            }
+        ]
+    return {
+        "pageType": page_type,
+        "orderListData": {
+            "ordersFound": orders_found,
+            "pageSize": page_size,
+            "orderList": orders,
+        },
+    }
+
+
+def test_parse_order_list_extracts_entries_from_real_shape():
+    html = _wrap_state(_real_order_list_state())
+    page = _parse_order_list(html)
+    assert page.orders_found == 1
+    assert page.page_size == 25
+    assert len(page.entries) == 1
+    entry = page.entries[0]
+    assert entry.order_number == _REAL_ORDER_ID
+    assert entry.order_date == "2026-08-13T09:47:28.000Z"
+    assert entry.shop_id == 306087
+    assert len(entry.items) == 1
+    assert entry.items[0].item_url == (
+        "https://item.rakuten.co.jp/moccasin/ds001iwrgesaaa2/?variantId=ds001iwrgesaaa2-1a-2a"
+    )
+
+
+def test_parse_order_list_multiple_items_in_one_order():
+    orders = [
+        {
+            "orderNumber": "111-20260101-0000000001",
+            "orderDate": "2026-01-01T00:00:00.000Z",
+            "items": [
+                {"itemUrl": "https://item.rakuten.co.jp/shop/a/", "itemName": "A"},
+                {"itemUrl": "https://item.rakuten.co.jp/shop/b/", "itemName": "B"},
+            ],
+        }
+    ]
+    html = _wrap_state(_real_order_list_state(orders=orders))
+    page = _parse_order_list(html)
+    assert len(page.entries[0].items) == 2
+
+
+def test_parse_order_list_wrong_page_type_returns_empty_and_no_orders_found():
+    """pageType 不是 ph-list（比如命中了 detail_page_view 的错误分支）：拿不到结构化数据"""
+    html = _wrap_state(_real_order_list_state(page_type="ph-error"))
+    page = _parse_order_list(html)
+    assert page.entries == []
+    assert page.orders_found is None
+
+
+def test_parse_order_list_no_initial_state_returns_empty():
+    page = _parse_order_list("<html>没有 state 的页面</html>")
+    assert page.entries == []
+    assert page.orders_found is None
+
+
+def test_parse_order_list_zero_orders_still_has_orders_found():
+    """账号没有任何订单：orderList 空，但 orders_found=0（不是 None）——区分「真的没有」与「解析不出」"""
+    html = _wrap_state(_real_order_list_state(orders_found=0, orders=[]))
+    page = _parse_order_list(html)
+    assert page.entries == []
+    assert page.orders_found == 0
+
+
+# ---- parse_order_datetime ----
+
+
+def test_parse_order_datetime_parses_real_format():
+    dt = parse_order_datetime("2026-08-13T09:47:28.000Z")
+    assert dt == datetime(2026, 8, 13, 9, 47, 28, tzinfo=timezone.utc)
+
+
+def test_parse_order_datetime_none_and_invalid():
+    assert parse_order_datetime(None) is None
+    assert parse_order_datetime("") is None
+    assert parse_order_datetime("不是日期") is None
+
+
+# ---- _accumulate_order_list_page：list_recent_orders 翻页决策的纯函数核心，
+# 2026-08-13 只有「账号 1 笔订单、1 页」的真实数据支撑第一个分支，其余分支
+# （多页、命中窗口边界、拿不到结构化数据）目前只验证逻辑正确，未经真实多页数据跑过 ----
+
+
+def _page(entries: list[OrderListEntry], *, orders_found: int | None) -> OrderListPage:
+    return OrderListPage(entries=entries, orders_found=orders_found, page_size=25)
+
+
+def _entry(order_number: str, order_date: str) -> OrderListEntry:
+    return OrderListEntry(order_number=order_number, order_date=order_date, items=[])
+
+
+_SINCE = datetime(2026, 8, 1, tzinfo=timezone.utc)
+
+
+def test_accumulate_single_page_all_within_window_and_matches_orders_found():
+    acc = _OrderListAccumulator()
+    page = _page([_entry("o1", "2026-08-10T00:00:00Z")], orders_found=1)
+    new_acc, stop = _accumulate_order_list_page(acc, page, since=_SINCE)
+    assert stop is True
+    assert new_acc.gave_up is False
+    assert [e.order_number for e in new_acc.entries] == ["o1"]
+
+
+def test_accumulate_stops_when_oldest_entry_older_than_since():
+    """翻到比 since 更早的订单：窗口内数据已经看全，停止翻页且视为覆盖完整"""
+    acc = _OrderListAccumulator(entries=[_entry("o1", "2026-08-10T00:00:00Z")], total_found=5, is_first_page=False)
+    page = _page([_entry("o2", "2026-07-01T00:00:00Z")], orders_found=5)
+    new_acc, stop = _accumulate_order_list_page(acc, page, since=_SINCE)
+    assert stop is True
+    assert new_acc.gave_up is False
+    assert [e.order_number for e in new_acc.entries] == ["o1", "o2"]
+
+
+def test_accumulate_continues_when_more_pages_remain():
+    """orders_found 还没凑够、最老的一条也没早于 since：应该继续翻下一页"""
+    acc = _OrderListAccumulator()
+    page = _page([_entry("o1", "2026-08-10T00:00:00Z")], orders_found=3)
+    new_acc, stop = _accumulate_order_list_page(acc, page, since=_SINCE)
+    assert stop is False
+    assert new_acc.total_found == 3
+
+
+def test_accumulate_empty_page_means_fully_covered():
+    """翻到空页（没有更多订单了）：不是错误，视为窗口已覆盖完"""
+    acc = _OrderListAccumulator(total_found=1, is_first_page=False)
+    page = _page([], orders_found=1)
+    new_acc, stop = _accumulate_order_list_page(acc, page, since=_SINCE)
+    assert stop is True
+    assert new_acc.gave_up is False
+
+
+def test_accumulate_first_page_missing_orders_found_gives_up():
+    """第一页就拿不到结构化数据（orders_found=None）：停止翻页，但 gave_up=True，不算覆盖完"""
+    acc = _OrderListAccumulator()
+    page = _page([], orders_found=None)
+    new_acc, stop = _accumulate_order_list_page(acc, page, since=_SINCE)
+    assert stop is True
+    assert new_acc.gave_up is True
+
+
+# ---- list_recent_orders 在没启动 Playwright 时应失败 ----
+
+
+async def test_list_recent_orders_without_start_raises():
+    site = SiteInteractor(auth_session=None, settings=None)  # type: ignore[arg-type]
+    with pytest.raises((AttributeError, TypeError)):
+        await site.list_recent_orders(since=_SINCE)
+
+
 # ---- submit_order / pay：没有 enter_checkout 留存的确认页会话时应报错，不静默成功 ----