Fix production ARR reads and add audited source field review

integrations/ohip/DATA_SOURCE.md

@@ -58,7 +58,7 @@ result = source.fetch(report_date="2026-09-15", request_id="0123456789abcdef0123
 | 当日预订列表 | `searchHotelReservations` | 分页查询，取完后再次核对 |
 | 预订详情 | `getReservation` | 每笔预订；明确排除 Traces 请求 |
 | 指定日有效价 | `searchRateInfo` | 每笔预订、用户所选日期 |
-| 主客完整姓名 | `searchProfiles` | 已关联主客的精确档案编号，重复档案复用 |
+| 主客完整姓名 | `getProfiles` | GET 精确档案编号，`summaryInfo=true&limit=1&offset=0`，重复档案复用 |
 | 档案详情 | `getProfile` | 关联公司名称或主客姓名组成资料未嵌入预订时补查 |
 | 团队代码 | `getBlock` | 已有关联团队编号，但未返回代码时补查 |
 | 房号及换房资料 | `getRoomCalendar` | 预订资料不能确定房号时，分页查找同一笔预订 |
@@ -72,6 +72,24 @@ result = source.fetch(report_date="2026-09-15", request_id="0123456789abcdef0123
 以及 `getBlock` 返回的 `blocks.blockInfo[].block`，另按 Oracle 固定版本接口说明核对，
 没有从模拟服务反推实际接口结构。
 
+2026-10-08 生产兼容修正：2026-10-07 到店日的只读捕获证明，带可选
+`orderBy=ConfirmationNo/sortOrder=Asc` 的查询被拒绝；省略这两个参数后查询成功。
+结构化数据入口只发送到店起止日期、`limit`、`offset`，保留每页原始顺序及
+`source_sequence`，不自行重排。生产默认排序规则尚未获得证明；完整分页后仍逐项比较
+首尾两次列表（含顺序），并验证预订身份、重复 ID、分页总数与末页完整性。
+这能检出已观察到的批次变化，但不构成原子快照，也不证明未来分页中的默认顺序固定。
+旧的候选捕获/探针入口保留原请求，未静默改变历史证据的含义。
+
+同日生产核对中 `searchProfiles` POST 不可用，`getProfiles` GET 按已关联主客的唯一档案 ID
+读取成功。结构化入口使用 GET 并严格核对返回的 `operation_id=getProfiles`、酒店、
+Oracle 请求编号、唯一档案 ID、结果数量和姓名组成，不按名称搜索或拼接完整姓名。
+这些生产捕获只用于核对请求兼容性，不代表完整日报已验收。
+
+原生可选字段省略仍保留为待确认：`reservationBlock`、`reservationPackages`、
+`reservationProfiles` 未返回时不自动视为空，套餐金额为零也不能证明没有套餐。
+`roomCalendar={}` 不能证明所选日期没有房号，继续保留房号缺口；
+Cancelled/NoShow 等状态不会在取数阶段导致预订被丢弃。
+
 ## 返回的数据
 
 `result` 包含 `status`、`collection_complete`、`input_complete`、记录数、各字段状态数量、

integrations/ohip/arr_data.py

@@ -98,8 +98,9 @@ def _name(detail, reader):
         target = next(g for g in working["reservationGuests"] if g.get("primary") is True)
         target["profileInfo"]["profile"] = profile
         guest = profiles.primary_guest(working, include_prefix=True)
-    summary = reader.query(profiles.OPERATION, identity=identity)
-    inspected = profiles.inspect_summary(summary.raw, guest, reader.hotel_id, include_prefix=True)
+    summary = reader.query(profiles.GET_OPERATION, identity=identity)
+    inspected = profiles.inspect_summary(summary.raw, guest, reader.hotel_id, include_prefix=True,
+                                         operation=profiles.GET_OPERATION)
     require(inspected["name_type"] in (None, "Primary"), "profile_nonprimary_summary_name")
     require(all(v in {"equal", "both_absent"} for v in inspected["component_comparisons"].values()),
             "profile_name_components_mismatch")
@@ -348,7 +349,7 @@ def collect(reader):
                                           "source_kind": reader.source_kind, "max_requests": reader.max_requests})
     complete = False
     try:
-        searches = base.search_day(reader, options)
+        searches = base.search_day(reader, options, server_sort=False)
         search_refs = list(reader.used)
         for index, search in enumerate(searches, 1):
             before = len(reader.used)
@@ -392,7 +393,7 @@ def collect(reader):
                     row["related"]["room_history"] = occurrences
                     row["fields"]["DISP_ROOM_NO"] = observe(lambda: _history_room(occurrences, options.arrival_date))
                 row["sources"] = list(dict.fromkeys(row["sources"] + reader.used[before:]))
-        require(base.search_day(reader, options) == searches, "source_changed_during_collection")
+        require(base.search_day(reader, options, server_sort=False) == searches, "source_changed_during_collection")
         complete = True
     except base.CollectionError as error:
         fatal = str(error)

integrations/ohip/collect_arr_source.py

@@ -304,14 +304,17 @@ def validate_note_counts(search: dict, detail: dict) -> None:
             require(integer(expected[0], "invalid_indicator_count") == len(values), "note_or_trace_count_mismatch")
 
 
-def search_day(reader: Reader, options: Options) -> list[dict]:
+def search_day(reader: Reader, options: Options, *, server_sort: bool = True) -> list[dict]:
+    require(type(server_sort) is bool, "invalid_search_sort_option")
     records, seen = [], set()
     expected_counts = None
     for page_number in range(options.max_pages):
         offset = page_number * options.page_size
-        data = reader.read(SEARCH, body={"arrivalStartDate": options.arrival_date,
-            "arrivalEndDate": options.arrival_date, "limit": options.page_size, "offset": offset,
-            "orderBy": ["ConfirmationNo"], "sortOrder": ["Asc"]})
+        body = {"arrivalStartDate": options.arrival_date, "arrivalEndDate": options.arrival_date,
+                "limit": options.page_size, "offset": offset}
+        if server_sort:
+            body.update(orderBy=["ConfirmationNo"], sortOrder=["Asc"])
+        data = reader.read(SEARCH, body=body)
         page = data.get("reservations")
         require(isinstance(page, dict), "invalid_search_page")
         rows = page.get("reservationInfo")

integrations/ohip/data_client.py

@@ -15,7 +15,7 @@ from . import profile_summary, rate_info
 
 
 FETCH = tuple(value for value in base.FETCH if value != "Traces")
-OPERATIONS = (base.SEARCH, base.DETAIL, rate_info.POST, profile_summary.OPERATION,
+OPERATIONS = (base.SEARCH, base.DETAIL, rate_info.POST, profile_summary.GET_OPERATION,
               "getProfile", "getBlock", "getRoomCalendar", "getPackage")
 require = base.require
 
@@ -83,11 +83,12 @@ class DataReader:
         day = self.options.arrival_date
         if operation == base.SEARCH:
             require(identity is None and params is None and type(body) is dict, "invalid_search_request")
-            require(set(body) == {"arrivalStartDate", "arrivalEndDate", "limit", "offset", "orderBy", "sortOrder"}
+            # Optional sorting is rejected by production. Retain the returned
+            # order and the full end-of-batch recheck rather than guessing a sort.
+            require(set(body) == {"arrivalStartDate", "arrivalEndDate", "limit", "offset"}
                     and body["arrivalStartDate"] == body["arrivalEndDate"] == day
                     and body["limit"] == self.options.page_size and type(body["limit"]) is int
-                    and type(body["offset"]) is int and 0 <= body["offset"] < self.options.max_pages * self.options.page_size
-                    and body["orderBy"] == ["ConfirmationNo"] and body["sortOrder"] == ["Asc"],
+                    and type(body["offset"]) is int and 0 <= body["offset"] < self.options.max_pages * self.options.page_size,
                     "invalid_search_request")
             return "POST", "/api/v1/reservations/searches", body
         require(body is None, "unexpected_query_body")
@@ -123,9 +124,8 @@ class DataReader:
         if operation == rate_info.POST:
             method, path, _payload, value = rate_info.request(identity, day)
             return method, path, value
-        if operation == profile_summary.OPERATION:
-            method, path, payload = profile_summary.request(identity)
-            return method, path, base.strict_json(payload)
+        if operation == profile_summary.GET_OPERATION:
+            return profile_summary.get_request(identity)
         if operation == "getProfile":
             return "GET", f"/api/v1/profiles/{identity}", None
         return "GET", f"/api/v1/blocks/{identity}?fetchInstructions=Block", None
@@ -176,7 +176,7 @@ class DataReader:
             # Reservation details/day prices are used once, and calendars once
             # per page. Bound reusable related-object caching in both bytes and
             # entries so a large day cannot retain every raw response in RAM.
-            if operation in (profile_summary.OPERATION, "getProfile", "getBlock", "getPackage"):
+            if operation in (profile_summary.GET_OPERATION, "getProfile", "getBlock", "getPackage"):
                 while self.cache and (len(self.cache) >= 128 or self.cache_bytes + len(raw) > 32 * 1024 * 1024):
                     _key, evicted = self.cache.popitem(last=False)
                     self.cache_bytes -= len(evicted.raw)

integrations/ohip/profile_summary.py

@@ -8,12 +8,15 @@ from __future__ import annotations
 
 from dataclasses import dataclass, field
 import re
+from urllib.parse import urlencode
 
 from . import collect_arr_source as source
 
 
 OPERATION = "searchProfiles"
 PATH = "/api/v1/profiles/searches"
+GET_OPERATION = "getProfiles"
+GET_PATH = "/api/v1/profiles"
 VERSION = "arr-profile-summary-probe/v1"
 COMPONENTS = {"surname": "name", "givenName": "givenName", "middleName": "middleName",
               "nameTitle": "nameTitle"}
@@ -73,12 +76,20 @@ def request(profile_id: str) -> tuple[str, str, bytes]:
     return "POST", PATH, source.json_bytes({"profileIds": [profile_id], "summaryInfo": True, "limit": 1, "offset": 0})
 
 
-def summary_row(raw: bytes, profile_id: str, hotel_id: str) -> dict:
+def get_request(profile_id: str) -> tuple[str, str, None]:
+    """Exact-ID summary read supported by the production GET contract."""
+    _profile_id([{"type": "Profile", "id": profile_id}])
+    query = urlencode({"profileIds": profile_id, "summaryInfo": "true", "limit": 1, "offset": 0})
+    return "GET", GET_PATH + "?" + query, None
+
+
+def summary_row(raw: bytes, profile_id: str, hotel_id: str, *, operation: str = OPERATION) -> dict:
     """Validate response identity/coverage independently of available name fields."""
-    request(profile_id)
+    require(operation in (OPERATION, GET_OPERATION), "profile_invalid_operation")
+    _profile_id([{"type": "Profile", "id": profile_id}])
     require(type(raw) is bytes and len(raw) <= source.MAX_RESPONSE_BYTES, "profile_response_too_large")
     document = source.strict_json(raw)
-    require(document.get("operation_id") == OPERATION and document.get("hotel_id") == hotel_id,
+    require(document.get("operation_id") == operation and document.get("hotel_id") == hotel_id,
             "profile_response_context_mismatch")
     require(type(document.get("oracle_request_id")) is str and bool(document["oracle_request_id"].strip()),
             "profile_missing_oracle_request_id")
@@ -108,9 +119,10 @@ def summary_row(raw: bytes, profile_id: str, hotel_id: str) -> dict:
     return row
 
 
-def inspect_summary(raw: bytes, guest: PrimaryGuest, hotel_id: str, *, include_prefix: bool = False) -> dict:
+def inspect_summary(raw: bytes, guest: PrimaryGuest, hotel_id: str, *, include_prefix: bool = False,
+                    operation: str = OPERATION) -> dict:
     require(type(guest) is PrimaryGuest, "profile_primary_guest_required")
-    row = summary_row(raw, guest.profile_id, hotel_id)
+    row = summary_row(raw, guest.profile_id, hotel_id, operation=operation)
     profile = row.get("profile")
     require(type(profile) is dict, "profile_missing_summary_profile")
     name = profile.get("formerName")