fix: 识别带价格的纵向套餐续页 (#168)

client/src/pdd_collect_service.py

@@ -1740,9 +1740,12 @@ class PddCollectService:
                     size_name = dimension.name
                     size_found = True
                     for value in dimension.values:
-                        if self._is_second_dimension_action(value.text):
+                        option_text = self._second_dimension_option_text(value.text)
+                        if option_text is None:
                             continue
-                        sizes[value.text] = sizes.get(value.text, False) or value.available
+                        sizes[option_text] = (
+                            sizes.get(option_text, False) or value.available
+                        )
 
             if size_found:
                 break
@@ -1827,9 +1830,10 @@ class PddCollectService:
                 )
             size_name = size_dimension.name
             for value in size_dimension.values:
-                if self._is_second_dimension_action(value.text):
+                option_text = self._second_dimension_option_text(value.text)
+                if option_text is None:
                     continue
-                sizes[value.text] = sizes.get(value.text, False) or value.available
+                sizes[option_text] = sizes.get(option_text, False) or value.available
 
             signature = self._second_dimension_signature(xml_data, size_dimension)
             stable_edge_reads = (
@@ -1928,14 +1932,24 @@ class PddCollectService:
         """返回已确认的第二规格签名，不要求当前屏仍显示规格标题。"""
 
         root = _parse_xml(xml_data)
+        targets = {
+            normalized
+            for value in dimension.values
+            if (normalized := self._second_dimension_option_text(value.text))
+            is not None
+        }
         result = []
-        for value in dimension.values:
-            node = self._find_option_node(root, value.text)
-            if node is None:
+        for node in root.iter("node"):
+            if node.get("clickable") != "true":
+                continue
+            text = self._second_dimension_option_text(
+                _preferred_or_descendant_label(node).strip()
+            )
+            if text not in targets:
                 continue
             bounds = _parse_bounds(node.get("bounds", ""))
             if bounds is not None:
-                result.append((value.text, bounds))
+                result.append((text, bounds))
         return tuple(result)
 
     def _build_second_dimension_context(
@@ -2009,13 +2023,11 @@ class PddCollectService:
                 continue
             text = _preferred_or_descendant_label(node).strip()
             bounds = _parse_bounds(node.get("bounds", ""))
+            option_text = self._second_dimension_option_text(text)
             if (
-                not text
-                or text in seen
+                option_text is None
+                or option_text in seen
                 or bounds is None
-                or self._is_second_dimension_action(text)
-                or bounds[2] - bounds[0]
-                >= (container_bounds[2] - container_bounds[0]) * 0.85
             ):
                 continue
             if text.endswith(("…", "...")):
@@ -2023,8 +2035,8 @@ class PddCollectService:
                     "PDD_DATA_SKU_NAME_TRUNCATED",
                     f"规格名称被截断，无法安全采集：{text}",
                 )
-            seen.add(text)
-            values.append(DimensionValue(text, _is_available(node)))
+            seen.add(option_text)
+            values.append(DimensionValue(option_text, _is_available(node)))
         if not values:
             return None
         return SpecDimension("size", context.name, tuple(values))
@@ -2078,13 +2090,11 @@ class PddCollectService:
         )
 
     @staticmethod
-    def _is_second_dimension_action(text: str) -> bool:
-        """排除规格面板中容易和选项结构相同的操作按钮。"""
+    def _second_dimension_option_text(text: str) -> Optional[str]:
+        """返回去掉尾部价格的规格文字；操作项或纯价格返回 ``None``。"""
 
         compact = text.replace(" ", "")
-        if "¥" in compact or "￥" in compact:
-            return True
-        return compact.startswith(("已选", "请选择")) or any(
+        if not compact or compact.startswith(("已选", "请选择")) or any(
             marker in compact
             for marker in (
                 "增加数量",
@@ -2097,7 +2107,19 @@ class PddCollectService:
                 "确认购买",
                 "关闭",
             )
-        )
+        ):
+            return None
+        if re.fullmatch(
+            r"(?:券后|到手价|拼单价|单买价|价格)?[¥￥]\d+(?:\.\d{1,2})?",
+            compact,
+        ):
+            return None
+        normalized = re.sub(
+            r"\s*[¥￥]\s*\d+(?:\.\d{1,2})?\s*$",
+            "",
+            text,
+        ).strip()
+        return normalized or None
 
     def _second_dimension_region(
         self,