256 lines
8.5 KiB
Python
256 lines
8.5 KiB
Python
"""PDD 页面分类和商品深链跳转状态。
|
|
|
|
本模块只处理脱敏的页面结构，不依赖 Qt、SQLite 或具体业务服务。采集和采购必须
|
|
共用这里的分类，避免一边把首页当商品页、另一边只能等待超时。
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import hashlib
|
|
import re
|
|
import xml.etree.ElementTree as ET
|
|
from dataclasses import dataclass
|
|
|
|
from .pdd_device_service import PDD_PACKAGE_NAME
|
|
|
|
|
|
PAGE_EXTERNAL = "external"
|
|
PAGE_LOADING = "loading"
|
|
PAGE_GOODS = "goods"
|
|
PAGE_HOME = "home"
|
|
PAGE_NETWORK_ERROR = "network_error"
|
|
PAGE_LOGIN_REQUIRED = "login_required"
|
|
PAGE_CAPTCHA = "captcha"
|
|
PAGE_RISK_CONTROL = "risk_control"
|
|
PAGE_PAYMENT = "payment"
|
|
PAGE_ORDER_CONFIRMATION = "order_confirmation"
|
|
PAGE_UNKNOWN = "unknown"
|
|
|
|
ACTION_WAIT = "wait"
|
|
ACTION_READY = "ready"
|
|
ACTION_REOPEN = "reopen"
|
|
ACTION_UNAVAILABLE = "unavailable"
|
|
ACTION_NETWORK_ERROR = "network_error"
|
|
|
|
_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")
|
|
_NETWORK_MARKERS = (
|
|
"网络不给力",
|
|
"网络异常",
|
|
"加载失败",
|
|
"连接失败",
|
|
"服务异常",
|
|
"请检查网络",
|
|
)
|
|
_LOADING_MARKERS = ("加载中", "正在加载", "努力加载")
|
|
_HOME_NAVIGATION = frozenset({"首页", "聊天", "个人中心"})
|
|
_BOUNDS_PATTERN = re.compile(r"^\[(\d+),(\d+)\]\[(\d+),(\d+)\]$")
|
|
_VOLATILE_NUMBER = re.compile(r"\d+(?:\.\d+)?")
|
|
|
|
|
|
@dataclass(frozen=True)
|
|
class PddPageObservation:
|
|
"""一次页面读取的分类和仅供当前任务比较的脱敏签名。"""
|
|
|
|
kind: str
|
|
signature: str
|
|
pdd_hierarchy: bool
|
|
|
|
|
|
@dataclass(frozen=True)
|
|
class GoodsOpenDecision:
|
|
"""商品深链观察器给调用方的下一步动作。"""
|
|
|
|
action: str
|
|
attempt: int
|
|
|
|
|
|
def classify_pdd_page(
|
|
root: ET.Element,
|
|
current_package: str,
|
|
) -> PddPageObservation:
|
|
"""使用包名和组合控件特征分类当前页面。"""
|
|
|
|
nodes = list(root.iter("node"))
|
|
labels = [_node_label(node) for node in nodes]
|
|
labels = [label for label in labels if label]
|
|
combined = " ".join(labels)
|
|
pdd_node_count = sum(
|
|
1 for node in nodes if node.get("package") == PDD_PACKAGE_NAME
|
|
)
|
|
pdd_hierarchy = pdd_node_count >= 3
|
|
is_pdd = current_package == PDD_PACKAGE_NAME or pdd_hierarchy
|
|
|
|
if any(marker in combined for marker in _CAPTCHA_MARKERS):
|
|
kind = PAGE_CAPTCHA
|
|
elif any(marker in combined for marker in _LOGIN_MARKERS):
|
|
kind = PAGE_LOGIN_REQUIRED
|
|
elif any(marker in combined for marker in _RISK_MARKERS):
|
|
kind = PAGE_RISK_CONTROL
|
|
elif any(marker in combined for marker in _PAYMENT_MARKERS):
|
|
kind = PAGE_PAYMENT
|
|
elif any(marker in combined for marker in _FINAL_SUBMIT_MARKERS):
|
|
kind = PAGE_ORDER_CONFIRMATION
|
|
elif is_pdd and any(marker in combined for marker in _NETWORK_MARKERS):
|
|
kind = PAGE_NETWORK_ERROR
|
|
elif is_pdd and any(marker in combined for marker in _LOADING_MARKERS):
|
|
kind = PAGE_LOADING
|
|
elif is_pdd and any(marker in combined for marker in _READY_MARKERS):
|
|
kind = PAGE_GOODS
|
|
elif is_pdd and _is_loaded_home(nodes, labels):
|
|
kind = PAGE_HOME
|
|
elif not is_pdd:
|
|
kind = PAGE_EXTERNAL
|
|
else:
|
|
kind = PAGE_UNKNOWN
|
|
|
|
return PddPageObservation(
|
|
kind=kind,
|
|
signature=_page_signature(nodes, kind),
|
|
pdd_hierarchy=pdd_hierarchy,
|
|
)
|
|
|
|
|
|
class GoodsOpenTracker:
|
|
"""确认商品深链已离开旧页，并控制首页场景只重开一次。"""
|
|
|
|
def __init__(
|
|
self,
|
|
before_open: PddPageObservation | None,
|
|
opened_at: float,
|
|
*,
|
|
stable_home_reads: int = 3,
|
|
home_grace_seconds: float = 2.0,
|
|
) -> None:
|
|
if stable_home_reads < 2:
|
|
raise ValueError("首页稳定读取次数至少为 2")
|
|
if home_grace_seconds < 0:
|
|
raise ValueError("首页跳转宽限时间不能为负数")
|
|
self._before_open = before_open
|
|
self._opened_at = float(opened_at)
|
|
self._stable_home_reads = stable_home_reads
|
|
self._home_grace_seconds = float(home_grace_seconds)
|
|
self._attempt = 1
|
|
self._home_reads = 0
|
|
self._stale_goods_seen = False
|
|
|
|
@property
|
|
def attempt(self) -> int:
|
|
return self._attempt
|
|
|
|
@property
|
|
def stale_goods_seen(self) -> bool:
|
|
return self._stale_goods_seen
|
|
|
|
def reopened(self, opened_at: float) -> None:
|
|
"""调用方重新打开同一 URL 后开始第二次、也是最后一次观察。"""
|
|
|
|
if self._attempt != 1:
|
|
raise RuntimeError("商品链接最多只允许受控重开一次")
|
|
self._attempt = 2
|
|
self._opened_at = float(opened_at)
|
|
self._home_reads = 0
|
|
|
|
def observe(
|
|
self,
|
|
observation: PddPageObservation,
|
|
observed_at: float,
|
|
) -> GoodsOpenDecision:
|
|
"""读取一个新页面分类，并返回等待、成功、重开或失败。"""
|
|
|
|
if observation.kind == PAGE_GOODS:
|
|
self._home_reads = 0
|
|
before = self._before_open
|
|
if (
|
|
before is not None
|
|
and before.kind == PAGE_GOODS
|
|
and before.signature == observation.signature
|
|
):
|
|
self._stale_goods_seen = True
|
|
return GoodsOpenDecision(ACTION_WAIT, self._attempt)
|
|
return GoodsOpenDecision(ACTION_READY, self._attempt)
|
|
|
|
if observation.kind == PAGE_NETWORK_ERROR:
|
|
self._home_reads = 0
|
|
return GoodsOpenDecision(ACTION_NETWORK_ERROR, self._attempt)
|
|
|
|
if observation.kind != PAGE_HOME:
|
|
self._home_reads = 0
|
|
return GoodsOpenDecision(ACTION_WAIT, self._attempt)
|
|
|
|
self._home_reads += 1
|
|
stable = self._home_reads >= self._stable_home_reads
|
|
grace_elapsed = (
|
|
float(observed_at) - self._opened_at >= self._home_grace_seconds
|
|
)
|
|
if not stable or not grace_elapsed:
|
|
return GoodsOpenDecision(ACTION_WAIT, self._attempt)
|
|
if self._attempt == 1:
|
|
return GoodsOpenDecision(ACTION_REOPEN, self._attempt)
|
|
return GoodsOpenDecision(ACTION_UNAVAILABLE, self._attempt)
|
|
|
|
|
|
def _is_loaded_home(nodes: list[ET.Element], labels: list[str]) -> bool:
|
|
exact_labels = set(labels)
|
|
if not _HOME_NAVIGATION.issubset(exact_labels):
|
|
return False
|
|
home_selected = any(
|
|
_node_label(node) == "首页" and node.get("selected") == "true"
|
|
for node in nodes
|
|
)
|
|
has_clickable_home = any(
|
|
node.get("content-desc", "").strip() == "首页"
|
|
and node.get("clickable") == "true"
|
|
for node in nodes
|
|
)
|
|
return home_selected and has_clickable_home
|
|
|
|
|
|
def _page_signature(nodes: list[ET.Element], kind: str) -> str:
|
|
"""对页面上半部结构做哈希；不把原始文字写入日志或跨任务保存。"""
|
|
|
|
screen_bottom = 0
|
|
parsed_bounds: dict[int, tuple[int, int, int, int]] = {}
|
|
for node in nodes:
|
|
match = _BOUNDS_PATTERN.fullmatch(node.get("bounds", ""))
|
|
if match is None:
|
|
continue
|
|
bounds = tuple(map(int, match.groups()))
|
|
parsed_bounds[id(node)] = bounds
|
|
screen_bottom = max(screen_bottom, bounds[3])
|
|
|
|
parts = [kind]
|
|
for node in nodes:
|
|
label = _node_label(node)
|
|
if not label or node.get("visible-to-user") == "false":
|
|
continue
|
|
bounds = parsed_bounds.get(id(node))
|
|
if bounds is not None and screen_bottom and bounds[1] > screen_bottom * 0.72:
|
|
continue
|
|
normalized = _VOLATILE_NUMBER.sub("#", " ".join(label.split()))
|
|
parts.append(
|
|
"|".join(
|
|
(
|
|
node.get("class", ""),
|
|
normalized[:120],
|
|
node.get("selected", ""),
|
|
node.get("checked", ""),
|
|
)
|
|
)
|
|
)
|
|
raw = "\n".join(parts).encode("utf-8", errors="replace")
|
|
return hashlib.sha256(raw).hexdigest()[:24]
|
|
|
|
|
|
def _node_label(node: ET.Element) -> str:
|
|
values = []
|
|
for key in ("text", "content-desc"):
|
|
value = node.get(key, "").strip()
|
|
if value and value not in values:
|
|
values.append(value)
|
|
return " ".join(values)
|