fix: 兼容非滚动PDD规格面板 (#154)

client/src/pdd_collect_service.py

@@ -435,17 +435,80 @@ def _is_dimension_heading(label: str) -> bool:
     # 新版页面会把可选数量写进标题，例如“颜色 (6)”或“颜色（6）”。
     # 数量不是规格名称的一部分，只在判断标题类型时去掉，最终展示仍保留原文。
     compact = re.sub(r"[（(]\d+[)）]$", "", compact)
+    # “确认款式”是新版规格弹层的标题，不是一个可选择的“款式”维度。
+    if compact in ("确认款式", "確認款式"):
+        return False
     return compact in _DIMENSION_NAMES or compact.endswith(
         ("分类", "规格", "尺寸", "尺码", "颜色", "型号", "款式", "容量", "类型", "版本", "口味")
     )
 
 
+def _find_non_scrollable_spec_panel(
+    root: ET.Element,
+    parents: Mapping[ET.Element, ET.Element],
+) -> Optional[ET.Element]:
+    """用多项强证据定位不暴露 scrollable 属性的自绘规格面板。"""
+
+    headings: list[ET.Element] = []
+    summaries: list[ET.Element] = []
+    panel_cues: list[ET.Element] = []
+    confirms: list[ET.Element] = []
+    for node in root.iter("node"):
+        label = _preferred_node_label(node).strip()
+        compact = label.replace(" ", "")
+        if not label or _parse_bounds(node.get("bounds", "")) is None:
+            continue
+        if _is_dimension_heading(label):
+            headings.append(node)
+        if compact.startswith(("已选", "已選", "请选择", "請選擇")):
+            summaries.append(node)
+        if compact in ("确认款式", "確認款式", "关闭", "關閉"):
+            panel_cues.append(node)
+        if compact in ("确定", "確定") and node.get("clickable") == "true":
+            confirms.append(node)
+
+    if not headings or not summaries or not panel_cues or not confirms:
+        return None
+
+    required = [*headings, summaries[0], panel_cues[0], confirms[0]]
+    common = set([required[0], *_ancestors(required[0], parents)])
+    for node in required[1:]:
+        common.intersection_update([node, *_ancestors(node, parents)])
+
+    all_bounds = [
+        bounds
+        for node in root.iter("node")
+        if (bounds := _parse_bounds(node.get("bounds", ""))) is not None
+    ]
+    screen_area = 0
+    if all_bounds:
+        screen_area = max(item[2] for item in all_bounds) * max(
+            item[3] for item in all_bounds
+        )
+
+    candidates: list[tuple[int, ET.Element]] = []
+    for node in common:
+        bounds = _parse_bounds(node.get("bounds", ""))
+        if bounds is None or node.get("visible-to-user", "true") != "true":
+            continue
+        area = (bounds[2] - bounds[0]) * (bounds[3] - bounds[1])
+        if screen_area and area >= screen_area * 0.95:
+            continue
+        candidates.append((area, node))
+    if not candidates:
+        return None
+    return min(candidates, key=lambda item: item[0])[1]
+
+
 def _is_spec_panel_open(xml_data: str | bytes) -> bool:
     """判断规格面板是否已经出现，不要求当前视口已露出规格选项。"""
 
     root = _parse_xml(xml_data)
     labels = _all_labels(root)
     _raise_special_page(labels)
+    parents = {child: parent for parent in root.iter() for child in parent}
+    if _find_non_scrollable_spec_panel(root, parents) is not None:
+        return True
     has_scrollable_region = any(
         node.get("scrollable") == "true"
         and _parse_bounds(node.get("bounds", "")) is not None
@@ -536,11 +599,23 @@ def parse_spec_panel(xml_data: str | bytes) -> SpecSnapshot:
         for node in root.iter("node")
         if node.get("scrollable") == "true" and _parse_bounds(node.get("bounds", ""))
     ]
-    if not scrollables:
+    non_scrollable_outer = _find_non_scrollable_spec_panel(root, parents)
+    panel_scrollables = []
+    if non_scrollable_outer is not None:
+        panel_scrollables = [
+            node
+            for node in scrollables
+            if node is non_scrollable_outer
+            or non_scrollable_outer in _ancestors(node, parents)
+        ]
+    outer_candidates = panel_scrollables or (
+        [non_scrollable_outer] if non_scrollable_outer is not None else scrollables
+    )
+    if not outer_candidates:
         raise PddCollectError("PDD_DATA_SPEC_INCOMPLETE", "规格面板没有可识别的规格区域")
 
     outer = max(
-        scrollables,
+        outer_candidates,
         key=lambda node: (
             (_parse_bounds(node.get("bounds", "")) or (0, 0, 0, 0))[2]
             - (_parse_bounds(node.get("bounds", "")) or (0, 0, 0, 0))[0]
@@ -616,8 +691,9 @@ def parse_spec_panel(xml_data: str | bytes) -> SpecSnapshot:
         dimensions.append(SpecDimension(key, name, tuple(values)))
 
     selected_text = next((label for label in labels if label.startswith("已选")), None)
+    price_boundary = heading_nodes[0][0] if heading_nodes else outer_bounds[1]
     price_cent, raw_price, list_price_cent = _price_from_nodes(
-        root.iter("node"), outer_bounds[1]
+        root.iter("node"), price_boundary
     )
     return SpecSnapshot(
         tuple(dimensions), selected_text, price_cent, raw_price, list_price_cent