fix: 安全兼容采购规格繁简匹配 (#251)

2026-08-17 15:20:35 +08:00
parent 1927e3edb6
commit 134514e382
4 changed files with 373 additions and 56 deletions
@@ -1,6 +1,11 @@
+import ctypes
+from functools import lru_cache
 import re
+import sys
 import time
+import unicodedata
 import xml.etree.ElementTree as ET
+from ctypes import wintypes
 from typing import Any, Optional, Union


@@ -10,6 +15,7 @@ Bounds = tuple[int, int, int, int]
 _BOUNDS_PATTERN = re.compile(r"^\[(\d+),(\d+)\]\[(\d+),(\d+)\]$")
 _COLOR_HEADINGS = ("颜色分类", "颜色", "款式", "颜色款式", "花色")
 _PACKAGE_HEADINGS = ("套餐",)
+_LCMAP_SIMPLIFIED_CHINESE = 0x02000000


 def _parse_xml(xml_data: XmlData) -> ET.Element:
@@ -41,21 +47,97 @@ def _node_labels(node: ET.Element) -> tuple[str, str]:
    return node.get("text", "").strip(), node.get("content-desc", "").strip()


+@lru_cache(maxsize=512)
+def normalize_spec_text(value: str) -> str:
+    """生成采购规格比较键；Windows 下把繁体转换为简体。
+
+    原始规格文字仍用于点击、显示和保存。本函数只生成比较键；Windows API
+    不可用时安全退化为 Unicode 规范化，不做模糊匹配。
+    """
+
+    normalized = unicodedata.normalize("NFKC", str(value or "")).strip()
+    if not normalized or sys.platform != "win32":
+        return normalized.casefold()
+
+    try:
+        function = ctypes.windll.kernel32.LCMapStringEx
+        function.argtypes = [
+            wintypes.LPCWSTR,
+            wintypes.DWORD,
+            wintypes.LPCWSTR,
+            ctypes.c_int,
+            wintypes.LPWSTR,
+            ctypes.c_int,
+            ctypes.c_void_p,
+            ctypes.c_void_p,
+            wintypes.LPARAM,
+        ]
+        function.restype = ctypes.c_int
+        required = function(
+            "zh-CN",
+            _LCMAP_SIMPLIFIED_CHINESE,
+            normalized,
+            -1,
+            None,
+            0,
+            None,
+            None,
+            0,
+        )
+        if required <= 0:
+            return normalized.casefold()
+        buffer = ctypes.create_unicode_buffer(required)
+        written = function(
+            "zh-CN",
+            _LCMAP_SIMPLIFIED_CHINESE,
+            normalized,
+            -1,
+            buffer,
+            required,
+            None,
+            None,
+            0,
+        )
+        if written <= 0:
+            return normalized.casefold()
+        return buffer.value.casefold()
+    except (AttributeError, OSError, TypeError, ValueError):
+        return normalized.casefold()
+
+
+def _label_matches(label: str, target: str) -> bool:
+    if label.casefold() == target.casefold():
+        return True
+
+    pattern = rf"(?:^|[\s,，:：;；]){re.escape(target)}(?:$|[\s,，:：;；])"
+    return re.search(pattern, label, flags=re.IGNORECASE) is not None
+
+
+def _node_match_kind(node: ET.Element, target: str) -> int:
+    """返回 2=原文精确匹配、1=繁简规范化匹配、0=不匹配。"""
+
+    checked_target = target.strip()
+    if not checked_target:
+        return 0
+
+    labels = _node_labels(node)
+    if any(_label_matches(label, checked_target) for label in labels):
+        return 2
+
+    normalized_target = normalize_spec_text(checked_target)
+    if not normalized_target:
+        return 0
+    for label in labels:
+        normalized_label = normalize_spec_text(label)
+        if normalized_label and _label_matches(normalized_label, normalized_target):
+            return 1
+    return 0
+
+
 def _matches_target(node: ET.Element, target: str) -> bool:
-    target = target.strip()
-    if not target:
+    if not target.strip():
        return False
-
-    for label in _node_labels(node):
-        if label.casefold() == target.casefold():
-            return True
-
-        # 兼容“颜色 黑色，已选中”或“尺码: XL”等无障碍描述，
-        # 同时避免 XL 错误匹配到 2XL。
-        pattern = rf"(?:^|[\s,，:：;；]){re.escape(target)}(?:$|[\s,，:：;；])"
-        if re.search(pattern, label, flags=re.IGNORECASE):
-            return True
-    return False
+    return _node_match_kind(node, target) > 0


 def _parent_map(root: ET.Element) -> dict[ET.Element, ET.Element]:
@@ -328,17 +410,45 @@ def _safe_horizontal_target(bounds: Bounds, region: Bounds) -> bool:
    )


-def _target_click_bounds(
+def _contains_bounds(outer: Bounds, inner: Bounds) -> bool:
+    return (
+        outer[0] <= inner[0]
+        and outer[1] <= inner[1]
+        and outer[2] >= inner[2]
+        and outer[3] >= inner[3]
+    )
+
+
+def _collapse_nested_bounds(bounds: list[Bounds]) -> list[Bounds]:
+    """把同一规格卡片的外层和文字点击区域合并为外层卡片。"""
+
+    unique = sorted(
+        set(bounds),
+        key=lambda item: (item[2] - item[0]) * (item[3] - item[1]),
+        reverse=True,
+    )
+    result: list[Bounds] = []
+    for candidate in unique:
+        if any(_contains_bounds(existing, candidate) for existing in result):
+            continue
+        result.append(candidate)
+    return result
+
+
+def _target_click_candidates(
    root: ET.Element,
    target: str,
    region: Bounds,
    require_horizontal_safe: bool = False,
-) -> Optional[Bounds]:
+) -> list[tuple[int, int, Bounds]]:
    parents = _parent_map(root)
-    candidates: list[tuple[int, Bounds]] = []
+    candidates: list[tuple[int, int, Bounds]] = []

    for node in root.iter("node"):
-        if not _is_available(node) or not _matches_target(node, target):
+        if not _is_available(node):
+            continue
+        match_kind = _node_match_kind(node, target)
+        if match_kind == 0:
            continue

        bounds = _nearest_click_bounds(node, parents, region)
@@ -348,11 +458,47 @@ def _target_click_bounds(
            continue

        area = (bounds[2] - bounds[0]) * (bounds[3] - bounds[1])
-        candidates.append((area, bounds))
+        candidates.append((match_kind, area, bounds))
+    return candidates

+
+def _normalized_target_is_ambiguous(
+    root: ET.Element,
+    target: str,
+    region: Bounds,
+    require_horizontal_safe: bool = False,
+) -> bool:
+    candidates = _target_click_candidates(
+        root,
+        target,
+        region,
+        require_horizontal_safe=require_horizontal_safe,
+    )
+    if not candidates or any(kind == 2 for kind, _area, _bounds in candidates):
+        return False
+    return len(_collapse_nested_bounds([item[2] for item in candidates])) > 1
+
+
+def _target_click_bounds(
+    root: ET.Element,
+    target: str,
+    region: Bounds,
+    require_horizontal_safe: bool = False,
+) -> Optional[Bounds]:
+    candidates = _target_click_candidates(
+        root,
+        target,
+        region,
+        require_horizontal_safe=require_horizontal_safe,
+    )
    if not candidates:
        return None
-    return max(candidates, key=lambda item: item[0])[1]
+    best_kind = max(item[0] for item in candidates)
+    best = [item for item in candidates if item[0] == best_kind]
+    if best_kind == 1:
+        bounds = _collapse_nested_bounds([item[2] for item in best])
+        return bounds[0] if len(bounds) == 1 else None
+    return max(best, key=lambda item: item[1])[2]


 def _target_is_selected(root: ET.Element, target: str) -> bool:
@@ -391,6 +537,13 @@ def color_selection_failure_reason(xml_data: XmlData, target: str) -> str:
    if not matching_nodes:
        return "target_not_visible"
    region = _horizontal_color_region(root, target)
+    if region is not None and _normalized_target_is_ambiguous(
+        root,
+        target,
+        region,
+        require_horizontal_safe=True,
+    ):
+        return "target_ambiguous"
    if region is None or _target_click_bounds(
        root,
        target,
@@ -401,6 +554,25 @@ def color_selection_failure_reason(xml_data: XmlData, target: str) -> str:
    return "selection_unconfirmed"


+def size_selection_failure_reason(xml_data: XmlData, target: str) -> str:
+    """说明尺码选择失败阶段，繁简等价候选不唯一时明确拒绝。"""
+
+    root = _parse_xml(xml_data)
+    matching_nodes = [
+        node
+        for node in root.iter("node")
+        if _is_available(node) and _matches_target(node, target)
+    ]
+    if not matching_nodes:
+        return "target_not_visible"
+    region = _vertical_panel_region(root)
+    if region is not None and _normalized_target_is_ambiguous(root, target, region):
+        return "target_ambiguous"
+    if region is None or _target_click_bounds(root, target, region) is None:
+        return "safe_target_missing"
+    return "selection_unconfirmed"
+
+
 def _visible_signature(
    root: ET.Element,
    region: Bounds,
@@ -426,6 +598,13 @@ def _click_target_and_verify(
    action_delay: float,
    require_horizontal_safe: bool = False,
 ) -> tuple[bool, ET.Element]:
+    if _normalized_target_is_ambiguous(
+        root,
+        target,
+        region,
+        require_horizontal_safe=require_horizontal_safe,
+    ):
+        return False, root
    if _target_is_selected(root, target):
        return True, root

@@ -444,6 +623,8 @@ def _click_target_and_verify(
    time.sleep(action_delay)

    refreshed_root = _parse_xml(device.dump_hierarchy())
+    # 点击前已经完成唯一候选校验。点击后页面通常会额外出现“已选：目标”
+    # 摘要，不能把摘要和原卡片误判成两个可选候选。
    return _target_is_selected(refreshed_root, target), refreshed_root