Jeason
2ebdaec965
- cart_service: 拦截购物车API提取真实itemID映射 - cart_service: 从Vue组件/data属性/window全局变量多路提取itemID - tasks: 区分item_id和cart_item_id,只有真实itemID才拼URL - snatcher: 增加商品不存在/已下架检测,增加空URL检测
211 lines
7.9 KiB
Python
211 lines
7.9 KiB
Python
"""
|
||
购物车预售商品抓取服务
|
||
通过 Playwright 打开购物车页面，拦截 API + DOM 提取商品信息
|
||
"""
|
||
import asyncio
|
||
import json
|
||
from playwright.async_api import async_playwright
|
||
from utils.stealth import stealth_async
|
||
from server.services.auth_service import get_browser_context, has_auth
|
||
|
||
CART_URL = "https://weidian.com/new-cart/index.php"
|
||
|
||
# 从 DOM 提取购物车商品（含尝试从 Vue 组件获取 itemID）
|
||
EXTRACT_JS = """() => {
|
||
const R = [];
|
||
const sws = document.querySelectorAll(
|
||
'div.shop_info.cart_content div.shop_warp'
|
||
);
|
||
for (const sw of sws) {
|
||
const sn = (sw.querySelector('.shop_name') || {}).textContent || '';
|
||
const iws = sw.querySelectorAll('.item_warp');
|
||
for (const iw of iws) {
|
||
const o = {
|
||
shop_name: sn.trim(),
|
||
cart_item_id: iw.id,
|
||
item_id: '',
|
||
title: '', sku_name: '', price: '',
|
||
is_presale: false, countdown_text: '',
|
||
sale_time: '', presale_type: ''
|
||
};
|
||
|
||
// 尝试从 Vue 组件数据中提取 itemID
|
||
try {
|
||
const vue = iw.__vue__ || (iw.__vue_app__ && iw.__vue_app__._instance);
|
||
if (vue) {
|
||
const d = vue.$data || vue.data || vue;
|
||
o.item_id = String(d.itemID || d.itemId || d.item_id || '');
|
||
}
|
||
} catch(e) {}
|
||
|
||
// 尝试从 data-* 属性提取
|
||
if (!o.item_id) {
|
||
o.item_id = iw.dataset.itemId || iw.dataset.itemid || '';
|
||
}
|
||
|
||
// 尝试从内部链接提取
|
||
if (!o.item_id) {
|
||
const a = iw.querySelector('a[href*="itemID"]');
|
||
if (a) {
|
||
const m = a.href.match(/itemID=(\\d+)/);
|
||
if (m) o.item_id = m[1];
|
||
}
|
||
}
|
||
|
||
const te = iw.querySelector('.item_title');
|
||
if (te) o.title = te.textContent.trim();
|
||
const sk = iw.querySelector('.item_sku');
|
||
if (sk) o.sku_name = sk.textContent.trim();
|
||
const pr = iw.querySelector('.item_prices');
|
||
if (pr) o.price = pr.textContent.replace(/[^\\d.]/g, '');
|
||
const de = iw.querySelector('.item_desc');
|
||
if (de) {
|
||
const dt = de.querySelector('.title');
|
||
const dd = de.querySelector('.desc');
|
||
const wm = de.querySelector('.warn_msg');
|
||
if (dt && /\\u5b9a\\u65f6\\s*\\u5f00\\u552e/.test(dt.textContent)) {
|
||
o.is_presale = true;
|
||
const d = dd ? dd.textContent.trim() : '';
|
||
const w = wm ? wm.textContent.trim() : '';
|
||
if (d.includes('\\u8ddd\\u79bb\\u5f00\\u552e\\u8fd8\\u5269')) {
|
||
o.presale_type = 'countdown';
|
||
o.countdown_text = w;
|
||
} else if (d.includes('\\u5f00\\u552e\\u65f6\\u95f4')) {
|
||
o.presale_type = 'scheduled';
|
||
o.sale_time = w;
|
||
} else {
|
||
o.presale_type = 'unknown';
|
||
o.countdown_text = w;
|
||
}
|
||
}
|
||
}
|
||
R.push(o);
|
||
}
|
||
}
|
||
return R;
|
||
}"""
|
||
|
||
|
||
async def fetch_cart_presale_items(account_id):
|
||
"""
|
||
获取指定账号购物车中的预售商品列表。
|
||
双重提取：拦截购物车 API 获取 itemID 映射 + DOM 提取预售信息。
|
||
返回: (success, items_or_msg)
|
||
"""
|
||
if not has_auth(account_id):
|
||
return False, "账号未登录"
|
||
|
||
# 用于存储 API 返回的 cart_item_id -> itemID 映射
|
||
api_item_map = {}
|
||
|
||
async with async_playwright() as p:
|
||
browser, context = await get_browser_context(
|
||
p, account_id, headless=True
|
||
)
|
||
page = await context.new_page()
|
||
await stealth_async(page)
|
||
|
||
# 拦截购物车相关 API，提取 itemID
|
||
async def on_response(response):
|
||
url = response.url
|
||
# 购物车 API 通常包含 cart 相关路径
|
||
if any(k in url for k in ['cart/list', 'cart/query', 'cartList', 'getCart',
|
||
'cart-server', 'newcart']):
|
||
try:
|
||
data = await response.json()
|
||
_extract_item_ids(data, api_item_map)
|
||
except Exception:
|
||
pass
|
||
|
||
page.on("response", on_response)
|
||
|
||
try:
|
||
await page.goto(
|
||
CART_URL, wait_until="networkidle", timeout=20000
|
||
)
|
||
await asyncio.sleep(3)
|
||
|
||
if "login" in page.url.lower():
|
||
await browser.close()
|
||
return False, "登录态已过期，请重新登录"
|
||
|
||
if "error" in page.url.lower():
|
||
await browser.close()
|
||
return False, "购物车页面加载失败"
|
||
|
||
except Exception as e:
|
||
await browser.close()
|
||
return False, f"打开购物车失败: {e}"
|
||
|
||
# 也尝试从页面内嵌的 JS 变量/window 对象提取
|
||
try:
|
||
extra_map = await page.evaluate("""() => {
|
||
const m = {};
|
||
// 尝试从 window.__INITIAL_STATE__ 或类似全局变量提取
|
||
const sources = [
|
||
window.__INITIAL_STATE__,
|
||
window.__NUXT__,
|
||
window.__APP_DATA__,
|
||
window.cartData,
|
||
window.__data__,
|
||
];
|
||
function walk(obj, depth) {
|
||
if (!obj || depth > 5) return;
|
||
if (Array.isArray(obj)) {
|
||
for (const item of obj) walk(item, depth + 1);
|
||
} else if (typeof obj === 'object') {
|
||
const cid = String(obj.cartItemId || obj.cart_item_id || obj.cartId || '');
|
||
const iid = String(obj.itemID || obj.itemId || obj.item_id || obj.goodsId || '');
|
||
if (cid && iid && iid !== cid) m[cid] = iid;
|
||
// 也存 itemUrl
|
||
if (cid && obj.itemUrl) m[cid + '_url'] = obj.itemUrl;
|
||
for (const v of Object.values(obj)) walk(v, depth + 1);
|
||
}
|
||
}
|
||
for (const s of sources) { if (s) walk(s, 0); }
|
||
return m;
|
||
}""")
|
||
if extra_map:
|
||
api_item_map.update(extra_map)
|
||
except Exception:
|
||
pass
|
||
|
||
raw_items = await page.evaluate(EXTRACT_JS)
|
||
await browser.close()
|
||
|
||
# 合并 API 数据到 DOM 提取结果
|
||
for item in raw_items:
|
||
cid = item.get('cart_item_id', '')
|
||
if not item.get('item_id') and cid in api_item_map:
|
||
item['item_id'] = api_item_map[cid]
|
||
# 检查是否有 URL
|
||
url_key = cid + '_url'
|
||
if url_key in api_item_map:
|
||
item['url'] = api_item_map[url_key]
|
||
|
||
# 只返回预售商品
|
||
presale = [it for it in raw_items if it.get("is_presale")]
|
||
return True, presale
|
||
|
||
|
||
def _extract_item_ids(data, result_map):
|
||
"""递归遍历 API 响应 JSON，提取 cart_item_id -> itemID 映射"""
|
||
if isinstance(data, list):
|
||
for item in data:
|
||
_extract_item_ids(item, result_map)
|
||
elif isinstance(data, dict):
|
||
# 常见字段名
|
||
cid = str(data.get('cartItemId', data.get('cart_item_id',
|
||
data.get('cartId', ''))))
|
||
iid = str(data.get('itemID', data.get('itemId',
|
||
data.get('item_id', data.get('goodsId', '')))))
|
||
if cid and iid and cid != iid and iid != 'None':
|
||
result_map[cid] = iid
|
||
# 也提取 URL
|
||
item_url = data.get('itemUrl', data.get('item_url', ''))
|
||
if cid and item_url:
|
||
result_map[cid + '_url'] = item_url
|
||
for v in data.values():
|
||
if isinstance(v, (dict, list)):
|
||
_extract_item_ids(v, result_map)
|