fix: 识别失效商品链接返回首页 (#111)

client/src/pdd_u2_purchase_adapter.py

@@ -21,6 +21,22 @@ from .pdd_device_service import (
     current_thread_device_service,
 )
 from .performance_timing import current_performance_trace
+from .pdd_page_classifier import (
+    ACTION_NETWORK_ERROR,
+    ACTION_READY,
+    ACTION_REOPEN,
+    ACTION_UNAVAILABLE,
+    PAGE_CAPTCHA,
+    PAGE_GOODS,
+    PAGE_HOME,
+    PAGE_LOGIN_REQUIRED,
+    PAGE_NETWORK_ERROR,
+    PAGE_PAYMENT,
+    PAGE_RISK_CONTROL,
+    GoodsOpenTracker,
+    PddPageObservation,
+    classify_pdd_page,
+)
 from .pdd_purchase_adapter import (
     PddLivePurchaseAdapter,
     PddPurchaseAdapter,
@@ -103,22 +119,7 @@ def _goods_id_from_url(goods_url: str) -> str:
 
 
 def _page_kind(root: ET.Element, current_package: str) -> str:
-    combined = " ".join(_labels(root))
-    if any(marker in combined for marker in _CAPTCHA_MARKERS):
-        return "captcha"
-    if any(marker in combined for marker in _LOGIN_MARKERS):
-        return "login_required"
-    if any(marker in combined for marker in _RISK_MARKERS):
-        return "risk_control"
-    if any(marker in combined for marker in _PAYMENT_MARKERS):
-        return "payment"
-    if any(marker in combined for marker in _FINAL_SUBMIT_MARKERS):
-        return "order_confirmation"
-    if current_package == PDD_PACKAGE_NAME and any(
-        marker in combined for marker in _READY_MARKERS
-    ):
-        return "goods"
-    return "unknown"
+    return classify_pdd_page(root, current_package).kind
 
 
 def _selected(root: ET.Element, target: str) -> bool:
@@ -270,6 +271,7 @@ class U2PddPurchaseAdapter(PddPurchaseAdapter):
             self._session = self._device_service.connect(self._device_address)
             self._device = self._session.__enter__()
             current = self._session.initial_app_state
+            before_open = self._read_observation(current)
             trace = current_performance_trace()
             if current.get("package") != PDD_PACKAGE_NAME:
                 stage = trace.stage("pdd_start_or_wait") if trace else nullcontext()
@@ -287,7 +289,7 @@ class U2PddPurchaseAdapter(PddPurchaseAdapter):
             stage = trace.stage("open_url") if trace else nullcontext()
             with stage:
                 self._device.open_url(goods_url)
-            self._wait_for_page("goods", self._page_timeout)
+            self._wait_for_goods_page(goods_url, before_open)
         except PddPurchaseError:
             raise
         except PddDeviceError as exc:
@@ -445,12 +447,21 @@ class U2PddPurchaseAdapter(PddPurchaseAdapter):
         if session is not None:
             session.__exit__(None, None, None)
 
-    def _wait_for_page(self, expected: str, timeout: float) -> str:
+    def _wait_for_goods_page(
+        self,
+        goods_url: str,
+        before_open: Optional[PddPageObservation],
+    ) -> str:
+        """确认本次深链进入新商品页；稳定首页时只重开一次。"""
+
         trace = current_performance_trace()
         ready_stage = trace.stage("goods_page_ready") if trace else nullcontext()
         with ready_stage:
-            deadline = self._monotonic() + timeout
+            opened_at = self._monotonic()
+            deadline = opened_at + self._page_timeout
+            tracker = GoodsOpenTracker(before_open, opened_at)
             last_kind = "unknown"
+            last_recorded = ""
             first_dump = True
             while self._monotonic() < deadline:
                 self._check_cancelled("purchase_open_goods")
@@ -463,26 +474,94 @@ class U2PddPurchaseAdapter(PddPurchaseAdapter):
                 else:
                     xml_data = self._dump_hierarchy()
                 root = _parse_xml(xml_data)
-                last_kind = _page_kind(root, str(current.get("package") or ""))
-                if last_kind == expected:
+                observation = classify_pdd_page(
+                    root, str(current.get("package") or "")
+                )
+                last_kind = observation.kind
+                if trace is not None and last_kind != last_recorded:
+                    trace.record(
+                        "pdd_page_transition",
+                        0,
+                        f"attempt_{tracker.attempt}_{last_kind}",
+                    )
+                    last_recorded = last_kind
+                if last_kind in {
+                    PAGE_CAPTCHA,
+                    PAGE_LOGIN_REQUIRED,
+                    PAGE_RISK_CONTROL,
+                    PAGE_PAYMENT,
+                }:
+                    self._raise_special_page(last_kind)
+                decision = tracker.observe(observation, self._monotonic())
+                if decision.action == ACTION_READY:
                     if trace is not None:
                         trace.checkpoint("end_to_end_total")
                     return xml_data
-                if last_kind in {
-                    "captcha",
-                    "login_required",
-                    "risk_control",
-                    "payment",
-                }:
-                    self._raise_special_page(last_kind)
+                if decision.action == ACTION_REOPEN:
+                    with (
+                        trace.stage("open_url_retry")
+                        if trace is not None
+                        else nullcontext()
+                    ):
+                        device.open_url(goods_url)
+                    tracker.reopened(self._monotonic())
+                    last_recorded = ""
+                    continue
+                if decision.action == ACTION_UNAVAILABLE:
+                    raise PddPurchaseError(
+                        "PDD_GOODS_UNAVAILABLE",
+                        "商品链接已失效，PDD 无法打开商品详情页并返回了首页",
+                        step="purchase_open_goods",
+                        retryable=False,
+                        diagnostics={
+                            "page_kind": PAGE_HOME,
+                            "open_attempts": 2,
+                        },
+                    )
+                if decision.action == ACTION_NETWORK_ERROR:
+                    raise PddPurchaseError(
+                        "PDD_PAGE_NETWORK_ERROR",
+                        "PDD 商品页网络或服务异常，请稍后重试",
+                        step="purchase_open_goods",
+                        retryable=True,
+                        diagnostics={"page_kind": PAGE_NETWORK_ERROR},
+                    )
                 self._sleep(0.25)
+            if tracker.stale_goods_seen:
+                raise PddPurchaseError(
+                    "PDD_GOODS_IDENTITY_UNCONFIRMED",
+                    "打开商品链接后仍停留在原商品页，无法确认本次目标商品",
+                    step="purchase_open_goods",
+                    retryable=True,
+                    diagnostics={
+                        "page_kind": PAGE_GOODS,
+                        "open_attempts": tracker.attempt,
+                    },
+                )
             raise PddPurchaseError(
                 "PDD_PAGE_TIMEOUT",
-                f"等待 PDD {expected} 页面超时，最后页面为 {last_kind}",
+                f"等待 PDD 商品详情页超时，最后页面为 {last_kind}",
                 step="purchase_open_goods",
                 retryable=True,
+                diagnostics={
+                    "page_kind": last_kind,
+                    "open_attempts": tracker.attempt,
+                },
             )
 
+    def _read_observation(
+        self, current: Mapping[str, Any]
+    ) -> Optional[PddPageObservation]:
+        """读取打开前页面；失败时不阻止后续按新页面重新判断。"""
+
+        try:
+            root = _parse_xml(self._dump_hierarchy())
+            return classify_pdd_page(
+                root, str(current.get("package") or "")
+            )
+        except (PddPurchaseError, RuntimeError, OSError):
+            return None
+
     def _wait_for_confirmation_panel(self) -> str:
         deadline = self._monotonic() + self._panel_timeout
         while self._monotonic() < deadline: