Files
cmautobuy/client/src/pdd_u2_purchase_adapter.py
T

708 lines
27 KiB
Python
Raw Normal View History

"""uiautomator2 采购 Adapter。
演练 factory 返回没有提交方法的窄接口;live factory 单独返回只允许一次提交的
接口。两条路径都不提供付款、取消订单或绕过安全校验的方法。
"""
from __future__ import annotations
import re
import time
import xml.etree.ElementTree as ET
from contextlib import nullcontext
from decimal import Decimal, InvalidOperation, ROUND_HALF_UP
from typing import Any, Callable, Mapping, Optional
from urllib.parse import parse_qs, urlparse
from .pdd_device_service import (
PDD_PACKAGE_NAME,
PddDeviceError,
PddDeviceService,
current_thread_device_service,
)
from .performance_timing import current_performance_trace
from .pdd_page_classifier import (
ACTION_NETWORK_ERROR,
ACTION_READY,
ACTION_REOPEN,
ACTION_UNAVAILABLE,
PAGE_CAPTCHA,
PAGE_GOODS,
PAGE_HOME,
PAGE_LOGIN_REQUIRED,
PAGE_NETWORK_ERROR,
PAGE_PAYMENT,
PAGE_RISK_CONTROL,
GoodsOpenTracker,
PddPageObservation,
classify_pdd_page,
)
from .pdd_purchase_adapter import (
PddLivePurchaseAdapter,
PddPurchaseAdapter,
PddPurchaseError,
PurchasePageState,
)
from .util.get_size_panle_coord import get_size_panel_coord
from .util.select_color_size import select_color, select_size
Bounds = tuple[int, int, int, int]
_BOUNDS_PATTERN = re.compile(r"^\[(\d+),(\d+)\]\[(\d+),(\d+)\]$")
_PRICE_PATTERN = re.compile(r"[¥¥]\s*(\d+(?:\.\d{1,2})?)")
_READY_MARKERS = ("发起拼单", "立即购买", "单独购买", "免拼购买", "快要抢光")
_LOGIN_MARKERS = ("手机号登录", "登录后继续", "验证码登录", "账号登录")
_CAPTCHA_MARKERS = ("请完成验证", "拖动滑块", "安全验证", "点击图中")
_RISK_MARKERS = ("操作频繁", "异常请求", "风险提示", "账号异常")
_PAYMENT_MARKERS = ("输入支付密码", "立即支付", "支付成功", "支付失败")
_FINAL_SUBMIT_MARKERS = ("提交订单", "现在买,仅", "确认购买")
_OUT_OF_STOCK_MARKERS = ("已售罄", "暂时缺货", "库存不足", "该商品已售罄")
_SUPPORTED_OPTION_KEYS = frozenset({"color", "size"})
def _parse_xml(xml_data: str | bytes) -> ET.Element:
try:
return ET.fromstring(xml_data)
except (ET.ParseError, TypeError) as exc:
raise PddPurchaseError(
"PDD_DATA_XML_INVALID",
"PDD 返回的无障碍控件树不是有效 XML",
step="purchase_page_check",
) from exc
def _label(node: ET.Element) -> str:
return " ".join(
value.strip()
for value in (node.get("text", ""), node.get("content-desc", ""))
if value.strip()
)
def _labels(root: ET.Element) -> list[str]:
return [value for node in root.iter("node") if (value := _label(node))]
def _parse_bounds(value: str) -> Optional[Bounds]:
match = _BOUNDS_PATTERN.fullmatch((value or "").strip())
if not match:
return None
left, top, right, bottom = map(int, match.groups())
if right <= left or bottom <= top:
return None
return left, top, right, bottom
def _goods_id_from_url(goods_url: str) -> str:
parsed = urlparse(goods_url)
host = (parsed.hostname or "").lower()
supported_host = (
host == "yangkeduo.com"
or host.endswith(".yangkeduo.com")
or host == "pinduoduo.com"
or host.endswith(".pinduoduo.com")
)
goods_id = parse_qs(parsed.query).get("goods_id", [""])[0].strip()
if parsed.scheme not in {"http", "https"} or not supported_host:
raise PddPurchaseError(
"PURCHASE_GOODS_URL_INVALID",
"采购商品链接不是受支持的 PDD 链接",
step="purchase_open_goods",
)
if not goods_id.isdigit():
raise PddPurchaseError(
"PURCHASE_GOODS_ID_MISSING",
"采购商品链接缺少有效 goods_id",
step="purchase_open_goods",
)
return goods_id
def _page_kind(root: ET.Element, current_package: str) -> str:
return classify_pdd_page(root, current_package).kind
def _selected(root: ET.Element, target: str) -> bool:
"""从最新控件树确认一个规格已选中。"""
checked_target = target.strip().casefold()
parents = {
child: parent
for parent in root.iter()
for child in parent
if child.tag == "node"
}
for node in root.iter("node"):
labels = {
node.get("text", "").strip().casefold(),
node.get("content-desc", "").strip().casefold(),
}
if checked_target not in labels:
continue
current: Optional[ET.Element] = node
while current is not None:
if (
current.get("selected") == "true"
or current.get("checked") == "true"
):
return True
current = parents.get(current)
# 选中节点滑出屏幕后,PDD 仍会在顶部“已选”摘要中保留完整文字。
return any(
label.startswith("已选") and target.strip() in label
for label in _labels(root)
)
def _quantity(root: ET.Element) -> int:
values = []
for node in root.iter("node"):
if node.get("class") != "android.widget.EditText":
continue
text = node.get("text", "").strip()
if text.isdigit() and int(text) > 0:
values.append(int(text))
return values[0] if len(values) == 1 else 0
def _price_cent(root: ET.Element) -> int:
"""优先读取页面上半部的当前单价,不把划线价当成当前价。"""
parsed_nodes: list[tuple[int, int, str]] = []
screen_bottom = 0
for node in root.iter("node"):
bounds = _parse_bounds(node.get("bounds", ""))
if bounds is not None:
screen_bottom = max(screen_bottom, bounds[3])
if screen_bottom <= 0:
return 0
for node in root.iter("node"):
label = _label(node)
match = _PRICE_PATTERN.search(label)
bounds = _parse_bounds(node.get("bounds", ""))
if match is None or bounds is None:
continue
center_y = (bounds[1] + bounds[3]) // 2
if center_y >= screen_bottom * 0.75:
continue
# “折后/到手/快卖完”比单独的划线价更可信。
semantic_score = 1 if any(
word in label for word in ("折后", "到手", "快卖完", "现价")
) else 0
try:
cents = int(
(Decimal(match.group(1)) * 100).quantize(
Decimal("1"), rounding=ROUND_HALF_UP
)
)
except (InvalidOperation, ValueError):
continue
if cents > 0:
parsed_nodes.append((semantic_score, -center_y, cents))
if not parsed_nodes:
return 0
return max(parsed_nodes)[2]
def _final_submit_targets(root: ET.Element) -> list[Bounds]:
"""返回底部可见、启用且文字明确的唯一提交按钮坐标。"""
screen_bottom = 0
for node in root.iter("node"):
bounds = _parse_bounds(node.get("bounds", ""))
if bounds is not None:
screen_bottom = max(screen_bottom, bounds[3])
if screen_bottom <= 0:
return []
targets = set()
for node in root.iter("node"):
label = _label(node)
bounds = _parse_bounds(node.get("bounds", ""))
if not label or bounds is None:
continue
if not any(marker in label for marker in _FINAL_SUBMIT_MARKERS):
continue
if node.get("visible-to-user") != "true" or node.get("enabled") != "true":
continue
if (bounds[1] + bounds[3]) // 2 < screen_bottom * 0.6:
continue
targets.add(bounds)
return sorted(targets)
class U2PddPurchaseAdapter(PddPurchaseAdapter):
"""PDD 真机采购演练会话,不提供真实下单能力。"""
def __init__(
self,
device_address: str,
*,
device_service: Optional[PddDeviceService] = None,
cancelled: Callable[[], bool] = lambda: False,
sleeper: Callable[[float], None] = time.sleep,
monotonic: Callable[[], float] = time.monotonic,
page_timeout: float = 30.0,
panel_timeout: float = 10.0,
select_color_fn: Callable[..., bool] = select_color,
select_size_fn: Callable[..., bool] = select_size,
) -> None:
self._device_address = str(device_address or "").strip()
self._device_service = device_service or PddDeviceService()
self._cancelled = cancelled
self._sleep = sleeper
self._monotonic = monotonic
self._page_timeout = page_timeout
self._panel_timeout = panel_timeout
self._select_color = select_color_fn
self._select_size = select_size_fn
self._session = None
self._device = None
self._goods_id = ""
self._requested_options: dict[str, str] = {}
self._submit_attempted = False
def open_goods(self, goods_url: str) -> None:
self._goods_id = _goods_id_from_url(goods_url)
self._check_cancelled("purchase_open_goods")
try:
self._session = self._device_service.connect(self._device_address)
self._device = self._session.__enter__()
current = self._session.initial_app_state
before_open = self._read_observation(current)
trace = current_performance_trace()
if current.get("package") != PDD_PACKAGE_NAME:
stage = trace.stage("pdd_start_or_wait") if trace else nullcontext()
with stage:
self._device.app_start(PDD_PACKAGE_NAME)
if not self._device.app_wait(PDD_PACKAGE_NAME, timeout=10):
raise PddPurchaseError(
"DEVICE_APP_START_FAILED",
"PDD 应用启动失败",
step="purchase_open_goods",
retryable=True,
)
elif trace is not None:
trace.record("pdd_start_or_wait", 0, "already_foreground")
stage = trace.stage("open_url") if trace else nullcontext()
with stage:
self._device.open_url(goods_url)
self._wait_for_goods_page(goods_url, before_open)
except PddPurchaseError:
raise
except PddDeviceError as exc:
raise PddPurchaseError(
exc.code,
exc.message,
step="purchase_open_goods",
retryable=True,
) from exc
except Exception as exc:
self._raise_device_or_page_error(exc, "purchase_open_goods")
def read_state(self) -> PurchasePageState:
device = self._require_device()
self._check_cancelled("purchase_page_check")
try:
current = device.app_current()
xml_data = self._dump_hierarchy()
root = _parse_xml(xml_data)
kind = _page_kind(root, str(current.get("package") or ""))
labels = _labels(root)
submit_targets = _final_submit_targets(root)
selected = {
key: value
for key, value in self._requested_options.items()
if _selected(root, value)
}
return PurchasePageState(
page_kind=kind,
goods_id=self._goods_id,
selected_options=selected,
quantity=_quantity(root),
price_cent=_price_cent(root),
candidate_count=1 if kind != "unknown" else 0,
in_stock=not any(
marker in " ".join(labels)
for marker in _OUT_OF_STOCK_MARKERS
),
submit_candidate_count=len(submit_targets),
)
except PddPurchaseError:
raise
except Exception as exc:
self._raise_device_or_page_error(exc, "purchase_page_check")
def select_options(self, options: Mapping[str, str]) -> None:
device = self._require_device()
checked = {
str(key).strip(): str(value).strip()
for key, value in options.items()
}
unsupported = sorted(set(checked) - _SUPPORTED_OPTION_KEYS)
if unsupported or not checked or any(not value for value in checked.values()):
names = "、".join(unsupported) if unsupported else "空规格"
raise PddPurchaseError(
"PURCHASE_OPTIONS_UNSUPPORTED",
f"当前真机 Adapter 不支持规格:{names}",
step="purchase_select_options",
)
self._requested_options = checked
self._check_cancelled("purchase_select_options")
try:
home_xml = self._dump_hierarchy()
coord = get_size_panel_coord(home_xml)
if coord is None:
raise PddPurchaseError(
"PURCHASE_PANEL_ENTRY_MISSING",
"商品页没有可靠的采购规格入口",
step="purchase_select_options",
)
device.click(*coord)
panel_xml = self._wait_for_confirmation_panel()
color = checked.get("color")
if color and not self._select_color(
device, panel_xml, color, action_delay=0.2
):
raise PddPurchaseError(
"PURCHASE_OPTIONS_MISMATCH",
f"没有精确选中颜色:{color}",
step="purchase_select_options",
)
size = checked.get("size")
if size:
latest_xml = self._dump_hierarchy()
if not self._select_size(
device, latest_xml, size, action_delay=0.2
):
raise PddPurchaseError(
"PURCHASE_OPTIONS_MISMATCH",
f"没有精确选中尺码:{size}",
step="purchase_select_options",
)
except PddPurchaseError:
raise
except Exception as exc:
self._raise_device_or_page_error(exc, "purchase_select_options")
def set_quantity(self, quantity: int) -> None:
device = self._require_device()
if isinstance(quantity, bool) or not isinstance(quantity, int) or quantity <= 0:
raise PddPurchaseError(
"PURCHASE_QUANTITY_INVALID",
"采购数量必须是大于 0 的整数",
step="purchase_set_quantity",
)
self._check_cancelled("purchase_set_quantity")
try:
editor = device(className="android.widget.EditText")
if int(getattr(editor, "count", 0)) != 1:
raise PddPurchaseError(
"PURCHASE_QUANTITY_CONTROL_MISSING",
"页面没有唯一可靠的采购数量输入框",
step="purchase_set_quantity",
)
editor.set_text(str(quantity))
self._sleep(0.2)
if _quantity(_parse_xml(self._dump_hierarchy())) != quantity:
raise PddPurchaseError(
"PURCHASE_QUANTITY_MISMATCH",
"PDD 页面没有保存目标采购数量",
step="purchase_set_quantity",
)
except PddPurchaseError:
raise
except Exception as exc:
self._raise_device_or_page_error(exc, "purchase_set_quantity")
def enter_confirmation(self) -> None:
"""只确认已到最终提交前页面,不点击底部提交按钮。"""
state = self.read_state()
if state.page_kind != "order_confirmation":
raise PddPurchaseError(
"PURCHASE_CONFIRMATION_NOT_REACHED",
"当前没有到达最终提交前确认页",
step="purchase_enter_confirmation",
)
def stop_before_submit(self) -> None:
"""再次确认最终提交按钮存在,然后不做任何点击。"""
state = self.read_state()
if state.page_kind != "order_confirmation":
raise PddPurchaseError(
"PURCHASE_CONFIRMATION_LOST",
"最终提交前确认页已变化,已停止演练",
step="purchase_dry_run_stopped",
)
def close(self) -> None:
session = self._session
self._session = None
self._device = None
if session is not None:
session.__exit__(None, None, None)
def _wait_for_goods_page(
self,
goods_url: str,
before_open: Optional[PddPageObservation],
) -> str:
"""确认本次深链进入新商品页;稳定首页时只重开一次。"""
trace = current_performance_trace()
ready_stage = trace.stage("goods_page_ready") if trace else nullcontext()
with ready_stage:
opened_at = self._monotonic()
deadline = opened_at + self._page_timeout
tracker = GoodsOpenTracker(before_open, opened_at)
last_kind = "unknown"
last_recorded = ""
first_dump = True
while self._monotonic() < deadline:
self._check_cancelled("purchase_open_goods")
device = self._require_device()
current = device.app_current()
if first_dump and trace is not None:
with trace.stage("first_dump_hierarchy"):
xml_data = self._dump_hierarchy()
first_dump = False
else:
xml_data = self._dump_hierarchy()
root = _parse_xml(xml_data)
observation = classify_pdd_page(
root, str(current.get("package") or "")
)
last_kind = observation.kind
if trace is not None and last_kind != last_recorded:
trace.record(
"pdd_page_transition",
0,
f"attempt_{tracker.attempt}_{last_kind}",
)
last_recorded = last_kind
if last_kind in {
PAGE_CAPTCHA,
PAGE_LOGIN_REQUIRED,
PAGE_RISK_CONTROL,
PAGE_PAYMENT,
}:
self._raise_special_page(last_kind)
decision = tracker.observe(observation, self._monotonic())
if decision.action == ACTION_READY:
if trace is not None:
trace.checkpoint("end_to_end_total")
return xml_data
if decision.action == ACTION_REOPEN:
with (
trace.stage("open_url_retry")
if trace is not None
else nullcontext()
):
device.open_url(goods_url)
tracker.reopened(self._monotonic())
last_recorded = ""
continue
if decision.action == ACTION_UNAVAILABLE:
raise PddPurchaseError(
"PDD_GOODS_UNAVAILABLE",
"商品链接已失效,PDD 无法打开商品详情页并返回了首页",
step="purchase_open_goods",
retryable=False,
diagnostics={
"page_kind": PAGE_HOME,
"open_attempts": 2,
},
)
if decision.action == ACTION_NETWORK_ERROR:
raise PddPurchaseError(
"PDD_PAGE_NETWORK_ERROR",
"PDD 商品页网络或服务异常,请稍后重试",
step="purchase_open_goods",
retryable=True,
diagnostics={"page_kind": PAGE_NETWORK_ERROR},
)
self._sleep(0.25)
if tracker.stale_goods_seen:
raise PddPurchaseError(
"PDD_GOODS_IDENTITY_UNCONFIRMED",
"打开商品链接后仍停留在原商品页,无法确认本次目标商品",
step="purchase_open_goods",
retryable=True,
diagnostics={
"page_kind": PAGE_GOODS,
"open_attempts": tracker.attempt,
},
)
raise PddPurchaseError(
"PDD_PAGE_TIMEOUT",
f"等待 PDD 商品详情页超时,最后页面为 {last_kind}",
step="purchase_open_goods",
retryable=True,
diagnostics={
"page_kind": last_kind,
"open_attempts": tracker.attempt,
},
)
def _read_observation(
self, current: Mapping[str, Any]
) -> Optional[PddPageObservation]:
"""读取打开前页面;失败时不阻止后续按新页面重新判断。"""
try:
root = _parse_xml(self._dump_hierarchy())
return classify_pdd_page(
root, str(current.get("package") or "")
)
except (PddPurchaseError, RuntimeError, OSError):
return None
def _wait_for_confirmation_panel(self) -> str:
deadline = self._monotonic() + self._panel_timeout
while self._monotonic() < deadline:
self._check_cancelled("purchase_select_options")
xml_data = self._dump_hierarchy()
root = _parse_xml(xml_data)
current = self._require_device().app_current()
kind = _page_kind(root, str(current.get("package") or ""))
if kind == "order_confirmation" and any(
"增加数量" in label for label in _labels(root)
):
return xml_data
if kind in {"captcha", "login_required", "risk_control", "payment"}:
self._raise_special_page(kind)
self._sleep(0.2)
raise PddPurchaseError(
"PURCHASE_PANEL_TIMEOUT",
"点击采购入口后没有到达可靠的提交前确认页",
step="purchase_select_options",
retryable=True,
)
def _dump_hierarchy(self) -> str:
raw = self._require_device().dump_hierarchy()
return (
raw.decode("utf-8", errors="replace")
if isinstance(raw, bytes)
else str(raw)
)
def _require_device(self) -> Any:
if self._device is None:
raise PddPurchaseError(
"DEVICE_SESSION_CLOSED",
"采购演练设备会话未建立或已关闭",
step="purchase_page_check",
retryable=True,
)
return self._device
def _check_cancelled(self, step: str) -> None:
if self._cancelled():
raise PddPurchaseError(
"PURCHASE_CANCELLED",
"用户已请求停止采购演练",
step=step,
)
@staticmethod
def _raise_special_page(kind: str) -> None:
mapping = {
"captcha": ("PDD_PAGE_CAPTCHA", "PDD 出现安全验证,请人工处理"),
"login_required": (
"PDD_PAGE_LOGIN_REQUIRED",
"PDD 登录状态失效,请人工登录",
),
"risk_control": ("PDD_PAGE_RISK_CONTROL", "PDD 出现风控页,请人工处理"),
"payment": ("PDD_PAGE_PAYMENT", "PDD 已进入支付页,已立即停止"),
}
code, message = mapping[kind]
raise PddPurchaseError(code, message, step="purchase_page_check")
@staticmethod
def _raise_device_or_page_error(error: Exception, step: str) -> None:
details = str(error)
lowered = details.lower()
disconnected = any(
marker in lowered
for marker in ("offline", "device not found", "disconnected", "closed transport")
)
raise PddPurchaseError(
"DEVICE_DISCONNECTED" if disconnected else "PURCHASE_ADAPTER_ERROR",
f"采购演练在 {step} 失败:{details}",
step=step,
retryable=disconnected,
diagnostics={"exception_type": type(error).__name__},
) from error
def create_u2_purchase_adapter(
device_address: str, cancelled: Callable[[], bool]
) -> PddPurchaseAdapter:
"""为正式 Client 创建一次 dry-run 采购会话。"""
return U2PddPurchaseAdapter(
device_address,
device_service=current_thread_device_service() or PddDeviceService(),
cancelled=cancelled,
)
class U2PddLivePurchaseAdapter(U2PddPurchaseAdapter, PddLivePurchaseAdapter):
"""只允许一次最终提交点击的真机 Adapter,不包含付款路径。"""
def submit_order_once(self) -> None:
if self._submit_attempted:
raise PddPurchaseError(
"PURCHASE_SUBMIT_ALREADY_ATTEMPTED",
"本次采购已经尝试提交,禁止再次点击",
step="purchase_submit_once",
)
self._check_cancelled("purchase_submit_once")
device = self._require_device()
try:
current = device.app_current()
root = _parse_xml(self._dump_hierarchy())
kind = _page_kind(root, str(current.get("package") or ""))
if kind in {"captcha", "login_required", "risk_control", "payment"}:
self._raise_special_page(kind)
if kind != "order_confirmation":
raise PddPurchaseError(
"PURCHASE_CONFIRMATION_LOST",
"最终提交前页面已经变化,禁止提交订单",
step="purchase_submit_once",
)
targets = _final_submit_targets(root)
if len(targets) != 1:
raise PddPurchaseError(
"PURCHASE_SUBMIT_TARGET_AMBIGUOUS",
"最终提交按钮不是唯一可靠目标,禁止提交订单",
step="purchase_submit_once",
diagnostics={"candidate_count": len(targets)},
)
left, top, right, bottom = targets[0]
self._submit_attempted = True
device.click((left + right) // 2, (top + bottom) // 2)
except PddPurchaseError:
raise
except Exception as exc:
self._submit_attempted = True
self._raise_device_or_page_error(exc, "purchase_submit_once")
def create_u2_live_purchase_adapter(
device_address: str, cancelled: Callable[[], bool]
) -> PddLivePurchaseAdapter:
"""为已通过本地绑定授权的任务创建一次 live 会话。"""
return U2PddLivePurchaseAdapter(
device_address,
device_service=current_thread_device_service() or PddDeviceService(),
cancelled=cancelled,
)