feat(client): project sanitized SKU price candidates

client/src/cmbuyer_client/device/sku_evidence_sanitizer.py

@@ -1,6 +1,7 @@
 """T-103 原始规格面板证据的本机确定性隐私脱敏。
 
-此模块只处理人工采集的本地文件：不连接设备、不理解拼多多页面，也不识别规格或价格。
+此模块只处理人工采集的本地文件：不连接设备、不识别规格；仅可按已取证的固定
+几何和严格格式，将跨隐私边界的价格叶节点投影到派生 XML。
 """
 
 from __future__ import annotations
@@ -21,7 +22,7 @@ from ..pdd.product_url import ProductUrl, ProductUrlError, parse_product_url
 from ..pdd.sku_panel_state import HUMAN_DECLARED_STATES
 
 
-SANITIZER_VERSION = "t103-privacy-v3"
+SANITIZER_VERSION = "t103-privacy-v4"
 EXPECTED_GOODS_ID = "937122477375"
 EXPECTED_PDD_VERSION = "8.17.0"
 EXPECTED_DEVICE_MODEL = "PKG110"
@@ -37,6 +38,21 @@ _FULL_PHONE_RE = re.compile(r"(?:\+?86)?1[3-9]\d{9}")
 _MASKED_PHONE_RE = re.compile(r"1[3-9]\d\*{4}\d{4}")
 _MASK_TRANSLATION = str.maketrans({"＊": "*", "•": "*", "·": "*", "×": "*", "x": "*", "X": "*"})
 _SEPARATOR_RE = re.compile(r"[\s\-‐‑‒–—―()（）]+")
+# 这两个槽位来自 T-103 当前第一态、1080×2376 XML 坐标的人工审查。它们不是通用
+# 页面判据；坐标、文本或结构任何变化都停止发布，交由人重新取证。
+_CROSSING_PRICE_BOUNDS = frozenset({(396, 503, 712, 570), (730, 503, 895, 570)})
+_PRICE_PROJECTION_ATTRIBUTES = (
+    "bounds",
+    "text",
+    "package",
+    "class",
+    "clickable",
+    "enabled",
+    "visible-to-user",
+)
+# 仅接受普通 ASCII 空格，且每个可分隔位置最多一个；禁止换行、折扣、支付/提交文案和
+# 任何其它字符。前缀捕获组用于区分当前价与至多一个划线/原价候选。
+_CROSSING_PRICE_TEXT_RE = re.compile(r" {0,1}(?:(快卖光) {0,1})?[¥￥] {0,1}[1-9]\d*\.\d{2} {0,1}\Z")
 
 
 class SkuEvidenceSanitizationError(RuntimeError):
@@ -49,9 +65,12 @@ class _CleanupStats:
 
     removed_nodes: int = 0
     cleared_crossing_nodes: int = 0
+    preserved_crossing_price_nodes: int = 0
     retained_below_nodes: int = 0
     max_right: int = 0
     max_bottom: int = 0
+    current_price_candidates: int = 0
+    original_price_candidates: int = 0
 
 
 @dataclass(frozen=True)
@@ -271,6 +290,7 @@ def _sanitize_hierarchy(source: Path, target: Path) -> _CleanupStats:
     _require_expected_xml_coordinate_space(stats)
     if stats.removed_nodes < 1 or stats.retained_below_nodes < 1:
         raise SkuEvidenceSanitizationError("原始节点树未满足隐私几何结构。")
+    _require_safe_crossing_price_projection(stats)
     if _contains_phone(root):
         raise SkuEvidenceSanitizationError("派生节点树仍包含手机号，拒绝发布。")
     ElementTree.ElementTree(root).write(target, encoding="utf-8", xml_declaration=True)
@@ -294,9 +314,12 @@ def _sanitize_node(parent: ElementTree.Element, node: ElementTree.Element, stats
         parent.remove(node)
         return
     if position == "crossing":
-        # 全屏/跨界容器可保留其下方子节点，但自身所有属性和文本都可能含地址或手机号。
-        _clear_node_text(node)
-        stats.cleared_crossing_nodes += 1
+        if bounds in _CROSSING_PRICE_BOUNDS and node.get("text"):
+            _project_crossing_price_node(node, stats)
+        else:
+            # 全屏/跨界容器可保留其下方子节点，但自身所有属性和文本都可能含地址或手机号。
+            _clear_node_text(node)
+            stats.cleared_crossing_nodes += 1
     else:
         stats.retained_below_nodes += 1
     for child in list(node):
@@ -340,6 +363,47 @@ def _vertical_position(bounds: tuple[int, int, int, int]) -> str:
     return "crossing"
 
 
+def _project_crossing_price_node(node: ElementTree.Element, stats: _CleanupStats) -> None:
+    """投影唯一允许的跨界价格叶节点；任何结构漂移一律拒绝发布。"""
+
+    if (
+        len(node) != 0
+        or node.get("package") != "com.xunmeng.pinduoduo"
+        or node.get("class") != "android.widget.TextView"
+        or node.get("clickable") != "false"
+        or node.get("enabled") != "true"
+        or node.get("visible-to-user") != "true"
+    ):
+        raise SkuEvidenceSanitizationError("跨界价格节点结构不匹配，拒绝发布。")
+    text = node.get("text")
+    if text is None:
+        raise SkuEvidenceSanitizationError("跨界价格节点文本不匹配，拒绝发布。")
+    match = _CROSSING_PRICE_TEXT_RE.fullmatch(text)
+    if match is None:
+        raise SkuEvidenceSanitizationError("跨界价格节点文本不匹配，拒绝发布。")
+
+    # 只有这七项经上述检查后可进入派生 XML；尤其不复制 content-desc、resource-id 等原始属性。
+    node.attrib = {attribute: node.attrib[attribute] for attribute in _PRICE_PROJECTION_ATTRIBUTES}
+    node.text = None
+    node.tail = None
+    stats.preserved_crossing_price_nodes += 1
+    if match.group(1) is not None:
+        stats.current_price_candidates += 1
+    else:
+        stats.original_price_candidates += 1
+
+
+def _require_safe_crossing_price_projection(stats: _CleanupStats) -> None:
+    """当前价必须唯一；原价仅可选且唯一，避免把任意金额释放为价格证据。"""
+
+    if (
+        stats.current_price_candidates != 1
+        or stats.original_price_candidates > 1
+        or stats.preserved_crossing_price_nodes != stats.current_price_candidates + stats.original_price_candidates
+    ):
+        raise SkuEvidenceSanitizationError("跨界价格候选不唯一或缺失，拒绝发布。")
+
+
 def _clear_node_text(node: ElementTree.Element) -> None:
     node.attrib = {"bounds": node.attrib["bounds"]} if "bounds" in node.attrib else {}
     node.text = None
@@ -415,6 +479,7 @@ def _derived_manifest(
         "privacy_cleanup": {
             "removed_nodes": cleanup_stats.removed_nodes,
             "cleared_crossing_nodes": cleanup_stats.cleared_crossing_nodes,
+            "preserved_crossing_price_nodes": cleanup_stats.preserved_crossing_price_nodes,
             "retained_below_nodes": cleanup_stats.retained_below_nodes,
             "max_right": cleanup_stats.max_right,
             "max_bottom": cleanup_stats.max_bottom,