Files
cmbuyer/client/src/cmbuyer_client/pdd/sku_selection.py
T

588 lines
26 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
"""T-103:仅限已取证 PDD 8.17.0 的规格面板恢复。"""
from __future__ import annotations
from dataclasses import dataclass
import re
from time import monotonic, sleep
from typing import Any, Callable, Protocol
from xml.etree import ElementTree
from ..device.baseline import PDD_PACKAGE
from .product_open import EXPECTED_PDD_VERSION
from .product_url import parse_product_url
EXPECTED_GOODS_ID = "937122477375"
EXPECTED_UNIT_PRICE = "12.88"
# 任务值不是页面判据;右侧是 v5 取证的唯一 accessibility 文案(空格/全角括号均有意义)。
TASK_TO_UI_SELECTION = {("黑色CHA(纯棉)", "M(建议100-115)"): ("黑色 CHA (纯棉)", "M(建议100-115)")}
_TARGET_COLOR_UI, _TARGET_SIZE_UI = next(iter(TASK_TO_UI_SELECTION.values()))
_ENTRY = "快要抢光 ¥ 12.88"
_ENTRY_PROMOTION_LABEL = "快要抢光"
_ENTRY_TEXT_BOUNDS = "[688,2184][1042,2253]"
_ENTRY_ACTION_DESC = "快要抢光¥12.88"
_ENTRY_ACTION_BOUNDS = "[446,2166][1080,2328]"
_ENTRY_SIBLING = "免拼购买"
_ENTRY_SIBLING_BOUNDS = "[688,2256][856,2305]"
_FORBIDDEN_ENTRY_ACTION_DESC = (
"购买", "下单", "付款", "订单", "单独购买", "直接拼成", "提交订单", "支付",
"先用后付", "0元下单", "0 元下单",
)
_SIZE = "尺码"
_W, _H = 1080, 2376
_PRICE_PARENT = "[396,498][895,570]"
_CURRENT = "[396,503][712,570]"
_ORIGINAL = "[730,503][895,570]"
_SUMMARY = "[396,654][1053,716]"
_COLOR_REGION = "[36,1000][1080,1631]"
_SIZE_LABEL = "[36,1654][114,1700]"
_SIZE_HEADER = "[36,1637][1044,1718]"
_SIZE_OPTIONS = "[36,1730][1044,2045]"
_BOUNDS = re.compile(r"^\[(\d+),(\d+)\]\[(\d+),(\d+)\]$")
_PRICE = re.compile(r"^[^0-9¥¥]*[¥¥]([1-9][0-9]*\.[0-9]{2})$")
_ORIGINAL_PRICE = re.compile(r"^[¥¥][1-9][0-9]*\.[0-9]{2}$")
_BAD_PRICE_ROLE = ("提交订单", "支付", "优惠", "券", "会员", "补贴", "区间", "实付", "到手", "原价", "划线价", "最低", "低至", "起价", "下单", "先用后付", "预估")
class SkuSelectionError(RuntimeError):
"""已取证判据不成立时的脱敏停止。"""
_SKU_ENTRY_FAILURE_STAGES = frozenset(
(
"sku_entry_pre_intent",
"sku_entry_discovery",
"sku_entry_click",
"sku_entry_panel_verify",
)
)
_SKU_ENTRY_FAILURE_MARKER = object()
def _annotate_sku_entry_failure(error: BaseException, stage: str) -> None:
"""把 Flow 实际经过的固定入口子阶段附到原异常,不改变异常类型。"""
if type(stage) is not str or stage not in _SKU_ENTRY_FAILURE_STAGES:
return
try:
# marker 最后写入:若第三方异常拒绝任一属性写入,就不能形成可信诊断。
setattr(error, "_cmbuyer_failure_stage", stage)
setattr(error, "_cmbuyer_sku_entry_failure_marker", _SKU_ENTRY_FAILURE_MARKER)
except BaseException:
pass
def _safe_sku_entry_failure_stage(error: BaseException) -> str | None:
"""只读取由本模块写入的入口子阶段;任意异常自报的值不可信。"""
try:
marker = getattr(error, "_cmbuyer_sku_entry_failure_marker", None)
stage = getattr(error, "_cmbuyer_failure_stage", None)
if marker is not _SKU_ENTRY_FAILURE_MARKER or type(stage) is not str:
return None
return stage if stage in _SKU_ENTRY_FAILURE_STAGES else None
except BaseException:
return None
class SkuPanelDevice(Protocol):
def app_info(self, package_name: str) -> dict[str, Any]: ...
def app_current(self) -> dict[str, Any]: ...
def dump_window_hierarchy(self) -> str: ...
def tap_sku_entry(self, bounds: str) -> None: ...
def tap_sku_option(self, bounds: str) -> None: ...
def leave_sku_panel(self) -> None: ...
@dataclass(frozen=True)
class SkuSelection:
color: str
size: str
def resolve_task_selection(color: str, size: str) -> SkuSelection:
mapped = TASK_TO_UI_SELECTION.get((color, size))
if mapped is None:
raise SkuSelectionError("规格任务值不是已取证的唯一目标,已停止操作。")
return SkuSelection(*mapped)
@dataclass(frozen=True)
class _Node:
element: ElementTree.Element
parent: "_Node | None"
@property
def text(self) -> str: return self.element.get("text", "")
@property
def desc(self) -> str: return self.element.get("content-desc", "")
@property
def bounds(self) -> str: return self.element.get("bounds", "")
class SkuSelectionFlow:
def __init__(self, device: SkuPanelDevice, entry_wait_timeout_seconds: float = 0.2,
entry_poll_interval_seconds: float = 0.2, monotonic_clock: Callable[[], float] = monotonic,
sleep_function: Callable[[float], None] = sleep) -> None:
if entry_wait_timeout_seconds < 0 or entry_poll_interval_seconds <= 0:
raise ValueError("入口等待参数无效。")
self._device, self._entry_timeout, self._poll = device, entry_wait_timeout_seconds, entry_poll_interval_seconds
self._clock, self._sleep = monotonic_clock, sleep_function
self._pending: tuple[str, Callable[[list[_Node]], Any]] | None = None
def open_sku_panel(self, product_url: str, pre_intent_hierarchy: str | None = None) -> None:
try:
if parse_product_url(product_url).goods_id != EXPECTED_GOODS_ID:
raise SkuSelectionError("商品不是已取证目标,已停止操作。")
if pre_intent_hierarchy is not None:
previous_nodes = _parse_nodes(pre_intent_hierarchy)
# intent 前只判断旧页是否已经存在完整入口链;浮层或额外动作节点不能把旧商品
# 伪装成“不安全所以不存在”,否则 intent 后可能误把旧页当成新目标页。
if _physical_entries(previous_nodes):
raise SkuSelectionError("intent 前页面已出现规格入口,已拒绝旧商品误点。")
except BaseException as error:
_annotate_sku_entry_failure(error, "sku_entry_pre_intent")
raise
try:
entry, before = self._wait_for_entry(pre_intent_hierarchy)
_action_bounds(entry.bounds)
except BaseException as error:
_annotate_sku_entry_failure(error, "sku_entry_discovery")
raise
try:
self._pending = (before, _panel)
self._device.tap_sku_entry(entry.bounds)
except BaseException as error:
_annotate_sku_entry_failure(error, "sku_entry_click")
raise
try:
self._wait_after_action(before, _panel)
except BaseException as error:
_annotate_sku_entry_failure(error, "sku_entry_panel_verify")
raise
def select_sku_options(self, selection: SkuSelection) -> None:
if selection not in {SkuSelection(*item) for item in TASK_TO_UI_SELECTION.values()}:
raise SkuSelectionError("规格 UI 文案不是获准目标,已停止操作。")
initial = self._verified_nodes()
_option(initial, "color", selection.color); _option(initial, "size", selection.size)
_selected_label(initial, "color"); _selected_label(initial, "size")
self._restore("color", selection.color)
self._restore("size", selection.size)
def read_sku_unit_price(self) -> str:
return _unit_price(self._verified_nodes())
def verify_target_selection_and_read_price(self, selection: SkuSelection) -> str:
nodes = self._verified_nodes()
_selected(nodes, "color", selection.color)
_selected(nodes, "size", selection.size)
return _unit_price(nodes)
def exit_sku_panel_safely(self) -> None:
self._require_foreground()
before = self._read_hierarchy()
_panel(_parse_nodes(before))
self._device.leave_sku_panel()
deadline = self._clock() + self._entry_timeout
while True:
self._require_foreground()
raw = self._read_hierarchy()
if raw != before:
try:
_panel(_parse_nodes(raw))
except SkuSelectionError:
return
remaining = deadline - self._clock()
if remaining <= 0:
raise SkuSelectionError("安全退出后未确认离开规格面板,未重试返回。")
self._sleep(min(self._poll, remaining))
def reconcile_pending_action(self) -> None:
"""仅只读调和一次已发出但尚未得到后置条件确认的动作。"""
if self._pending is None:
return
before, condition = self._pending
self._wait_after_action(before, condition)
def _restore(self, dimension: str, expected: str) -> None:
self._require_foreground()
before = self._read_hierarchy()
nodes = _panel(_parse_nodes(before))
target = _option(nodes, dimension, expected)
if _selected_label(nodes, dimension) == expected:
return
_action_bounds(target.bounds)
condition: Callable[[list[_Node]], Any]
if dimension == "color":
condition = lambda refreshed: _post_color(refreshed, expected)
else:
condition = lambda refreshed: _post_all_targets(refreshed, expected)
self._pending = (before, condition)
self._device.tap_sku_option(target.bounds)
if dimension == "color":
self._wait_after_action(before, condition)
else:
self._wait_after_action(before, condition)
def _wait_for_entry(self, previous: str | None) -> tuple[_Node, str]:
deadline, stable = self._clock() + self._entry_timeout, None
while True:
self._require_version()
current = self._device.app_current()
if isinstance(current, dict) and current.get("package") == PDD_PACKAGE:
raw = self._read_hierarchy()
nodes = _parse_nodes(raw)
entries = _eligible_entries(nodes)
if len(entries) > 1:
raise SkuSelectionError("商品页规格入口不唯一,已停止操作。")
if len(entries) == 1 and raw != previous:
# 商品详情正文包含倒计时等动态节点,全文 XML 稳定不是已取证入口的安全属性。
# 连续两帧只比较证据绑定的底部入口、直接父容器和不可点击兄弟节点;商品正文及
# 父容器内其他非危险动态节点不是入口身份,不能迫使实现退回坐标兜底。
projection = _entry_projection(entries[0], nodes)
if projection is None:
raise SkuSelectionError("商品页规格入口结构失效,已停止操作。")
if stable == projection:
return entries[0], raw
stable = projection
else:
stable = None
else:
stable = None
remaining = deadline - self._clock()
if remaining <= 0:
raise SkuSelectionError("等待已取证规格入口超时,未执行点击。")
self._sleep(min(self._poll, remaining))
def _wait_after_action(self, previous: str, condition: Callable[[list[_Node]], Any]) -> list[_Node]:
deadline = self._clock() + self._entry_timeout
while True:
self._require_foreground()
raw = self._read_hierarchy()
if raw != previous:
nodes = _parse_nodes(raw)
try:
condition(nodes)
self._pending = None
return nodes
except SkuSelectionError:
pass
remaining = deadline - self._clock()
if remaining <= 0:
raise SkuSelectionError("动作后页面未在限定时间内满足已取证后置条件,未重试动作。")
self._sleep(min(self._poll, remaining))
def _verified_nodes(self) -> list[_Node]:
self._require_foreground()
return _panel(self._read_nodes())
def _require_version(self) -> None:
info = self._device.app_info(PDD_PACKAGE)
version = (info.get("versionName") or info.get("version_name")) if isinstance(info, dict) else None
if version != EXPECTED_PDD_VERSION:
raise SkuSelectionError("拼多多版本与已取证版本不一致,已停止操作。")
def _require_foreground(self) -> None:
self._require_version()
current = self._device.app_current()
if not isinstance(current, dict) or current.get("package") != PDD_PACKAGE:
raise SkuSelectionError("拼多多不在前台,已停止操作。")
def _read_hierarchy(self) -> str:
try: raw = self._device.dump_window_hierarchy()
except Exception as error: raise SkuSelectionError("节点树读取失败,已停止操作。") from error
if not isinstance(raw, str) or not raw: raise SkuSelectionError("节点树不可用,已停止操作。")
return raw
def _read_nodes(self) -> list[_Node]: return _parse_nodes(self._read_hierarchy())
def _parse_nodes(raw: str) -> list[_Node]:
try: root = ElementTree.fromstring(raw)
except ElementTree.ParseError as error: raise SkuSelectionError("节点树格式无效,已停止操作。") from error
if root.tag != "hierarchy": raise SkuSelectionError("节点树根节点无效,已停止操作。")
result: list[_Node] = []
def visit(element: ElementTree.Element, parent: _Node | None) -> None:
node = _Node(element, parent); result.append(node)
for child in element: visit(child, node)
visit(root, None)
return result
def _panel(nodes: list[_Node]) -> list[_Node]:
parent = _one([n for n in nodes if n.bounds == _PRICE_PARENT], "规格面板价格区域不唯一,已停止操作。")
_one([n for n in nodes if n.parent is parent and n.bounds == _ORIGINAL and _readonly(n) and _ORIGINAL_PRICE.fullmatch(n.text)], "规格面板原价槽位不唯一,已停止操作。")
_one([n for n in nodes if n.bounds == _SUMMARY and _readonly(n) and n.text.startswith("已选:")], "规格面板已选摘要不唯一,已停止操作。")
_color_container(nodes); _size_container(nodes)
return nodes
def _unit_price(nodes: list[_Node]) -> str:
parent = _one([n for n in nodes if n.bounds == _PRICE_PARENT], "规格面板价格区域不唯一,已停止读取。")
money = [n for n in nodes if n.parent is parent and _readonly(n) and any(mark in n.text for mark in "¥¥")]
if len(money) != 2: raise SkuSelectionError("规格面板金额槽位不唯一,已停止读取。")
current = _one([n for n in money if n.bounds == _CURRENT and not _clickable_ancestor(n) and not any(word in n.text for word in _BAD_PRICE_ROLE) and _PRICE.fullmatch(n.text)], "规格面板现价不唯一或不符合已取证槽位,已停止读取。")
if not any(n.bounds == _ORIGINAL and _ORIGINAL_PRICE.fullmatch(n.text) for n in money):
raise SkuSelectionError("规格面板原价槽位无效,已停止读取。")
match = _PRICE.fullmatch(current.text)
if match is None: raise SkuSelectionError("规格面板现价格式失效,已停止读取。")
return match.group(1)
def _option(nodes: list[_Node], dimension: str, expected: str) -> _Node:
_panel(nodes)
return _one([n for n in _options(nodes, dimension) if _label(n) == expected], "规格选项不唯一或不是精确匹配,已停止操作。")
def _selected(nodes: list[_Node], dimension: str, expected: str) -> None:
if _selected_label(nodes, dimension) != expected:
raise SkuSelectionError("规格选择后读回的 selected 文案不一致,已停止操作。")
def _selected_label(nodes: list[_Node], dimension: str) -> str:
selected = [n for n in _options(nodes, dimension) if n.element.get("selected") == "true"]
label = _label(_one(selected, "规格维度没有唯一 selected 状态,已停止操作。"))
if label is None: raise SkuSelectionError("规格维度 selected 文案无效,已停止操作。")
return label
def _post_color(nodes: list[_Node], expected: str) -> None:
_panel(nodes)
_selected(nodes, "color", expected)
_selected_label(nodes, "size")
def _post_all_targets(nodes: list[_Node], expected_size: str) -> None:
_panel(nodes)
_selected(nodes, "color", _TARGET_COLOR_UI)
_selected(nodes, "size", expected_size)
def _options(nodes: list[_Node], dimension: str) -> list[_Node]:
container = _color_container(nodes) if dimension == "color" else _size_container(nodes) if dimension == "size" else None
if container is None: raise SkuSelectionError("未知规格维度,已停止操作。")
candidates = [n for n in nodes if _descendant(n, container) and _contained(n, container) and _choice(n) and _label(n) is not None]
return [n for n in candidates if not _labeled_ancestor(n, candidates)]
def _color_container(nodes: list[_Node]) -> _Node:
return _one([n for n in nodes if n.element.get("package") == PDD_PACKAGE and n.element.get("class") == "androidx.recyclerview.widget.RecyclerView" and n.bounds == _COLOR_REGION], "规格面板颜色容器不唯一,已停止操作。")
def _size_container(nodes: list[_Node]) -> _Node:
label = _one([n for n in nodes if n.text == _SIZE and n.bounds == _SIZE_LABEL and _readonly(n)], "规格面板尺码标签不唯一,已停止操作。")
header = label.parent
if header is None or header.element.get("package") != PDD_PACKAGE or header.element.get("class") != "android.widget.LinearLayout" or header.bounds != _SIZE_HEADER or header.parent is None:
raise SkuSelectionError("规格面板尺码标题容器不符合已取证结构,已停止操作。")
return _one([n for n in nodes if n.parent is header.parent and n.element.get("package") == PDD_PACKAGE and n.element.get("class") == "android.widget.LinearLayout" and n.bounds == _SIZE_OPTIONS], "规格面板尺码选项容器不唯一,已停止操作。")
def _label(node: _Node) -> str | None:
values = {value for value in (node.text, node.desc) if value}
return values.pop() if len(values) == 1 else None
def _labeled_ancestor(node: _Node, candidates: list[_Node]) -> bool:
ids, parent = {id(n.element) for n in candidates}, node.parent
while parent is not None:
if id(parent.element) in ids and _label(parent) is not None: return True
parent = parent.parent
return False
def _descendant(node: _Node, ancestor: _Node) -> bool:
parent = node.parent
while parent is not None:
if parent.element is ancestor.element: return True
parent = parent.parent
return False
def _contained(node: _Node, container: _Node) -> bool:
left, top, right, bottom = _action_bounds(node.bounds)
outer_left, outer_top, outer_right, outer_bottom = _action_bounds(container.bounds)
return outer_left <= left < right <= outer_right and outer_top <= top < bottom <= outer_bottom
def _clickable_ancestor(node: _Node) -> bool:
parent = node.parent
while parent is not None:
if parent.element.get("clickable") == "true": return True
parent = parent.parent
return False
def _readonly(node: _Node) -> bool:
return node.element.get("package") == PDD_PACKAGE and node.element.get("class") == "android.widget.TextView" and node.element.get("clickable") == "false" and node.element.get("enabled") == "true" and node.element.get("visible-to-user") == "true"
def _live(node: _Node) -> bool:
return node.element.get("package") == PDD_PACKAGE and node.element.get("clickable") == "true" and node.element.get("enabled") == "true" and node.element.get("visible-to-user") == "true" and bool(node.bounds)
def _choice(node: _Node) -> bool:
return _live(node) and node.element.get("class") == "android.view.ViewGroup" and node.element.get("selected") in {"true", "false"}
def _eligible_entries(nodes: list[_Node]) -> list[_Node]:
# 入口文本本身不可点击:必须证明它仍是已取证底部父容器的直接子节点,但动作坐标继续
# 使用第一行文本的窄 bounds,避免把父容器中心或第二行“免拼购买”变成坐标兜底。
if any(node.bounds == _PRICE_PARENT for node in nodes):
return []
entries = _physical_entries(nodes)
if len(entries) != 1:
return entries
entry = entries[0]
chain = _physical_entry_chain(entry, nodes)
if chain is None or _entry_projection(entry, nodes) is None:
return []
ancestor = chain[1]
left, top, right, bottom = _action_bounds(entry.bounds)
center = (left + (right - left) // 2, top + (bottom - top) // 2)
occupants = _live_clickables_covering(nodes, center)
# RPC 点的是文本中心而不是祖先对象。只有完整入口链自己的动作祖先占用该坐标时才可点击;
# SystemUI 浮层、额外按钮或任意部分覆盖矩形都可能截获触摸,必须零点击失败关闭。
return entries if len(occupants) == 1 and occupants[0].element is ancestor.element else []
def _physical_entries(nodes: list[_Node]) -> list[_Node]:
entry_labels = [
node for node in nodes
if node.text == _ENTRY
and node.element.get("package") == PDD_PACKAGE
and node.element.get("class") == "android.widget.TextView"
]
# 这里故意只证明物理结构。pre-intent 必须识别旧商品,不能让父容器描述或浮层等
# post-intent 安全条件把已经存在的旧入口伪装成“不存在”。
return [node for node in entry_labels if _physical_entry_chain(node, nodes) is not None]
def _physical_entry_chain(node: _Node, nodes: list[_Node]) -> tuple[_Node, ...] | None:
if node.text != _ENTRY or not _exact_entry_node(
node, "android.widget.TextView", _ENTRY_TEXT_BOUNDS, "false"
) or len(node.element) != 0:
return None
ancestor = node.parent
if ancestor is None or not _exact_entry_node(
ancestor, "android.view.ViewGroup", _ENTRY_ACTION_BOUNDS, "true"
):
return None
siblings = [
candidate for candidate in nodes
if candidate.parent is ancestor
and candidate.text == _ENTRY_SIBLING
and _exact_entry_node(
candidate, "android.widget.TextView", _ENTRY_SIBLING_BOUNDS, "false"
)
and not candidate.desc
and len(candidate.element) == 0
]
if len(siblings) != 1:
return None
return node, ancestor, siblings[0]
def _entry_projection(node: _Node, nodes: list[_Node]) -> tuple[object, ...] | None:
chain = _physical_entry_chain(node, nodes)
if chain is None:
return None
_, ancestor, sibling = chain
if node.desc:
return None
# 金额只作为这个已取证入口的不可变身份。这里既不解析也不返回它,价格闸门仍只能读取规格面板。
if ancestor.text or ancestor.desc != _ENTRY_ACTION_DESC:
return None
sibling_mentions = [
candidate for candidate in nodes
if _ENTRY_SIBLING in candidate.text or _ENTRY_SIBLING in candidate.desc
]
if len(sibling_mentions) != 1 or sibling_mentions[0].element is not sibling.element:
return None
subtree = [
candidate for candidate in nodes
if candidate.element is ancestor.element or _descendant(candidate, ancestor)
]
if any(
forbidden in value
for candidate in subtree
for value in (candidate.text, candidate.desc)
if candidate.element is not sibling.element
for forbidden in _FORBIDDEN_ENTRY_ACTION_DESC
):
return None
if any(
value == _ENTRY_PROMOTION_LABEL
for candidate in subtree
for value in (candidate.text, candidate.desc)
):
return None
entry_text_nodes = [candidate for candidate in subtree if candidate.text == _ENTRY]
if len(entry_text_nodes) != 1 or entry_text_nodes[0].element is not node.element:
return None
return tuple(
_entry_node_projection(candidate)
for candidate in (node, ancestor, sibling)
)
def _entry_node_projection(node: _Node) -> tuple[str, ...]:
return (
node.element.tag,
node.element.get("package", ""),
node.element.get("class", ""),
node.bounds,
node.element.get("clickable", ""),
node.element.get("enabled", ""),
node.element.get("visible-to-user", ""),
node.text,
node.desc,
)
def _is_live_clickable(node: _Node) -> bool:
return (
node.element.get("clickable") == "true"
and node.element.get("enabled") == "true"
and node.element.get("visible-to-user") == "true"
)
def _live_clickables_covering(nodes: list[_Node], point: tuple[int, int]) -> list[_Node]:
occupants: list[_Node] = []
x, y = point
for node in nodes:
if not _is_live_clickable(node):
continue
# 活跃可点击节点的 bounds 无法验证时,无法证明它不会截获入口坐标,因此整体失败关闭。
left, top, right, bottom = _action_bounds(node.bounds)
if left <= x < right and top <= y < bottom:
occupants.append(node)
return occupants
def _exact_entry_node(node: _Node, class_name: str, bounds: str, clickable: str) -> bool:
return (
node.element.get("package") == PDD_PACKAGE
and node.element.get("class") == class_name
and node.bounds == bounds
and node.element.get("clickable") == clickable
and node.element.get("enabled") == "true"
and node.element.get("visible-to-user") == "true"
)
def _action_bounds(bounds: str) -> tuple[int, int, int, int]:
match = _BOUNDS.fullmatch(bounds)
if match is None: raise SkuSelectionError("规格节点坐标格式无效,已停止操作。")
left, top, right, bottom = (int(item) for item in match.groups())
if not (0 <= left < right <= _W and 0 <= top < bottom <= _H):
raise SkuSelectionError("规格节点坐标不在已取证屏幕范围内,已停止操作。")
return left, top, right, bottom
def _one(nodes: list[_Node], message: str) -> _Node:
if len(nodes) != 1: raise SkuSelectionError(message)
return nodes[0]