如何自动追踪 eBay 售价？Python 爬虫实战解析

一、业务与环境概述

电商价格监控核心诉求为周期性采集 eBay 商品售价、留存价格快照、触发降价告警。eBay 页面以服务端渲染为主，无需解析 JS，但存在 URL 结构复杂、区间价、严格访问频率限制等问题；长期批量爬取品类数据时，单 IP 高频请求极易触发限流封禁。

本文基于 Python 实现一套完整 eBay 价格监控体系，涵盖商品检索、价格解析、数据持久化、价格异动检测、降价通知全链路，并集成亿牛云隧道代理规避 IP 封禁风险，代码可直接部署运行，仅作技术演示用途。

页面数据定位

依托浏览器开发者工具（F12）解析 DOM 节点，下表为当前可用 CSS 选择器（受平台 A/B 测试影响，Class 名称会动态变更，选择器失效时需重新校验）：

表格

数据项	CSS 选择器
商品标题	#itemTitle / h1.x-item-title__mainTitle
在售价格	.x-price-primary .ux-textspans
划线原价	.x-price-approx__price
运费	.ux-labels-values__values .ux-textspans
搜索列表项	.srp-results .s-item
列表页价格	.s-item__price

二、搜索列表采集与代理方案

eBay 对搜索接口管控严格，单 IP 连续请求短时间内即触发限流。方案采用亿牛云隧道代理 ，统一请求入口为<font style="color:rgb(0, 0, 0);background-color:rgba(0, 0, 0, 0);">t.16yun.cn:31111</font>，由云端自动完成 IP 轮换，标准版 IP 池体量超 30 万，网络延迟低至 100ms。

代理接入核心逻辑

通过请求头<font style="color:rgb(0, 0, 0);background-color:rgba(0, 0, 0, 0);">Proxy-Tunnel</font>传递随机数值实现 IP 动态切换，该方式相比 TCP 长连接模式稳定性更强，连接中断也可正常切换出口 IP。

python

运行

import requests
import random

# 隧道代理基础配置
proxyHost = "t.16yun.cn"
proxyPort = "31111"
proxyUser = "your_username"
proxyPass = "your_password"

proxyMeta = f"http://{proxyUser}:{proxyPass}@{proxyHost}:{proxyPort}"
proxies = {"http": proxyMeta, "https": proxyMeta}

搜索采集器完整实现

基于<font style="color:rgb(0, 0, 0);background-color:rgba(0, 0, 0, 0);">requests</font>会话维持请求上下文，结合<font style="color:rgb(0, 0, 0);background-color:rgba(0, 0, 0, 0);">BeautifulSoup</font>实现页面解析，通过数据类结构化存储商品信息，内置随机请求间隔降低风控概率：

python

运行

from bs4 import BeautifulSoup
from urllib.parse import quote
from dataclasses import dataclass
from typing import Optional, List
import time
import random
import sqlite3
import hashlib

@dataclass
class Product:
    """商品数据实体类"""
    ebay_id: str = ""
    title: str = ""
    url: str = ""
    price: float = 0.0
    original_price: float = 0.0
    currency: str = "USD"
    shipping: str = ""
    seller: str = ""
    condition: str = ""
    image_url: str = ""
    uid: str = ""

    def __post_init__(self):
        if not self.uid and self.ebay_id:
            self.uid = hashlib.md5(self.ebay_id.encode()).hexdigest()[:12]

class EbaySearchScraper:
    """eBay搜索列表采集器（集成隧道代理）"""
    BASE_URL = "https://www.ebay.com/sch/i.html"

    def __init__(self, proxy_user: str = "", proxy_pass: str = ""):
        self.session = requests.Session()
        # 模拟标准浏览器请求头
        self.session.headers.update({
            "User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 Chrome/125.0.0.0 Safari/537.36",
            "Accept-Language": "en-US,en;q=0.9",
        })
        self.use_proxy = bool(proxy_user and proxy_pass)
        self.proxies = None
        if self.use_proxy:
            proxy_meta = f"http://{proxy_user}:{proxy_pass}@t.16yun.cn:31111"
            self.proxies = {"http": proxy_meta, "https": proxy_meta}

    def _get(self, url, **kwargs):
        """封装请求方法，动态切换出口IP"""
        headers = {}
        if self.use_proxy:
            headers["Proxy-Tunnel"] = str(random.randint(1, 10000))
        return self.session.get(url, proxies=self.proxies, headers=headers, timeout=15,** kwargs)

    def search(self, keyword: str, max_pages: int = 3, delay: float = 3.0) -> List[Product]:
        """分页检索商品列表"""
        all_products = []
        for page in range(1, max_pages + 1):
            params = {"_nkw": keyword, "_pgn": page, "_ipg": 60}
            try:
                resp = self._get(self.BASE_URL, params=params)
                if resp.status_code != 200:
                    break
                products = self._parse_search(resp.text)
                all_products.extend(products)
                time.sleep(delay + random.uniform(0.5, 2.0))
            except Exception:
                break
        return all_products

    def _parse_search(self, html: str) -> List[Product]:
        """解析搜索页HTML"""
        soup = BeautifulSoup(html, "html.parser")
        products = []
        for item in soup.select(".srp-results .s-item"):
            try:
                title_el = item.select_one(".s-item__title")
                link_el = item.select_one(".s-item__link")
                if not all([title_el, link_el]):
                    continue
                title = title_el.get_text(strip=True)
                url = link_el.get("href", "")
                ebay_id = url.split("/itm/")[1].split("?")[0] if "/itm/" in url else ""
                price = self._parse_price(item.select_one(".s-item__price"))
                shipping = item.select_one(".s-item__shipping").get_text(strip=True) if item.select_one(".s-item__shipping") else ""
                image_url = item.select_one(".s-item__image-img").get("src", "") if item.select_one(".s-item__image-img") else ""
                if ebay_id and price > 0:
                    products.append(Product(ebay_id=ebay_id, title=title, url=url, price=price, shipping=shipping, image_url=image_url))
            except Exception:
                continue
        return products

    @staticmethod
    def _parse_price(el) -> float:
        """价格文本清洗与类型转换，区间价取首值"""
        if not el:
            return 0.0
        text = el.get_text(strip=True).split("to")[0].replace("$", "").replace("C", "").replace(",", "")
        try:
            return float(text)
        except ValueError:
            return 0.0

三、商品详情页数据解析

复用采集器的会话与代理配置，拉取详情页完整数据，提取原价、精准运费、卖家信息等维度数据：

python

运行

class EbayDetailScraper:
    """商品详情页采集解析器"""
    def __init__(self, scraper: EbaySearchScraper):
        self._get = scraper._get

    def fetch_detail(self, product: Product) -> Product:
        try:
            resp = self._get(product.url)
            if resp.status_code != 200:
                return product
            soup = BeautifulSoup(resp.text, "html.parser")
            # 解析在售价格
            price_el = soup.select_one(".x-price-primary .ux-textspans")
            if price_el:
                product.price = self._parse(price_el.get_text(strip=True))
            # 解析划线原价
            orig_el = soup.select_one(".x-price-approx__price .ux-textspans--STRIKETHROUGH")
            if orig_el:
                product.original_price = self._parse(orig_el.get_text(strip=True))
            # 解析运费
            for sec in soup.select(".ux-labels-values__values-content .ux-textspans"):
                text = sec.get_text(strip=True).lower()
                if "free" in text or "$" in text:
                    product.shipping = text
                    break
            # 解析卖家名称
            seller_el = soup.select_one(".x-seller-info__name")
            if seller_el:
                product.seller = seller_el.get_text(strip=True)
        except Exception:
            pass
        return product

    @staticmethod
    def _parse(text: str) -> float:
        """详情页价格清洗"""
        cleaned = text.replace("$", "").replace(",", "").strip()
        try:
            return float(cleaned.split()[0])
        except (ValueError, IndexError):
            return 0.0

四、历史价格数据持久化

采用轻量级 SQLite 构建数据存储层，分商品基础表 与价格历史表做数据隔离，建立索引提升查询效率，支持数据更新、历史记录写入、价格比对查询：

python

运行

class PriceDatabase:
    def __init__(self, db_path: str = "ebay_prices.db"):
        self.conn = sqlite3.connect(db_path)
        # 商品基础信息表
        self.conn.execute("""
            CREATE TABLE IF NOT EXISTS products (
                ebay_id TEXT PRIMARY KEY,
                title TEXT, url TEXT, image_url TEXT, seller TEXT,
                created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
            )
        """)
        # 价格历史记录表
        self.conn.execute("""
            CREATE TABLE IF NOT EXISTS price_history (
                id INTEGER PRIMARY KEY AUTOINCREMENT,
                ebay_id TEXT, price REAL, original_price REAL, shipping TEXT,
                recorded_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
                FOREIGN KEY (ebay_id) REFERENCES products(ebay_id)
            )
        """)
        self.conn.execute("CREATE INDEX IF NOT EXISTS idx_hist ON price_history(ebay_id)")
        self.conn.commit()

    def upsert_product(self, p: Product):
        """新增/更新商品基础信息"""
        self.conn.execute(
            "INSERT OR REPLACE INTO products (ebay_id, title, url, image_url, seller) VALUES (?,?,?,?,?)",
            (p.ebay_id, p.title, p.url, p.image_url, p.seller)
        )
        self.conn.commit()

    def record_price(self, p: Product):
        """写入价格快照"""
        self.conn.execute(
            "INSERT INTO price_history (ebay_id, price, original_price, shipping) VALUES (?,?,?,?)",
            (p.ebay_id, p.price, p.original_price, p.shipping)
        )
        self.conn.commit()

    def get_latest(self, ebay_id: str) -> Optional[dict]:
        """查询单商品最新价格"""
        row = self.conn.execute(
            "SELECT price, recorded_at FROM price_history WHERE ebay_id=? ORDER BY recorded_at DESC LIMIT 1",
            (ebay_id,)
        ).fetchone()
        return {"price": row[0], "date": row[1]} if row else None

    def find_drops(self, threshold: float = 0.1) -> List[dict]:
        """批量筛选降幅达标的降价商品"""
        rows = self.conn.execute("""
            SELECT h1.ebay_id, p.title, h1.price old_price, h2.price new_price,
            (h1.price - h2.price)/h1.price drop_ratio, h2.recorded_at
            FROM price_history h1
            JOIN price_history h2 ON h1.ebay_id = h2.ebay_id
            JOIN products p ON h1.ebay_id = p.ebay_id
            WHERE h2.recorded_at > h1.recorded_at
            AND h1.id = (SELECT MAX(id) FROM price_history WHERE ebay_id=h1.ebay_id AND id < h2.id)
            AND (h1.price - h2.price)/h1.price >= ?
            ORDER BY drop_ratio DESC
        """, (threshold,)).fetchall()
        return [{
            "ebay_id": r[0], "title": r[1], "old": r[2], "new": r[3],
            "drop_pct": round(r[4]*100,1), "date": r[5]
        } for r in rows]

五、价格异动检测与告警

对比当期价格与历史快照，设定降幅阈值触发告警，整合全模块实现任务调度：

python

运行

class PriceAlert:
    def __init__(self, db: PriceDatabase):
        self.db = db

    def check(self, products: List[Product]) -> List[dict]:
        """检测价格变动，生成降价告警"""
        alerts = []
        for p in products:
            self.db.upsert_product(p)
            self.db.record_price(p)
            latest = self.db.get_latest(p.ebay_id)
            if not latest or latest["price"] <= 0 or p.price <= 0:
                continue
            drop_rate = (latest["price"] - p.price) / latest["price"]
            # 降幅≥5%触发告警
            if drop_rate >= 0.05:
                alerts.append({
                    "type": "↓", "ebay_id": p.ebay_id, "title": p.title,
                    "old": latest["price"], "new": p.price,
                    "drop": round(drop_rate*100,1), "url": p.url
                })
        return alerts

# 主运行入口
def track(keywords: List[str], delay: float = 3.0):
    db = PriceDatabase()
    scraper = EbaySearchScraper(proxy_user="your_user", proxy_pass="your_password")
    detail = EbayDetailScraper(scraper)
    alert = PriceAlert(db)
    all_alerts = []

    for kw in keywords:
        print(f"\n=== 检索关键词：{kw} ===")
        products = scraper.search(kw, max_pages=2, delay=delay)
        for idx, p in enumerate(products[:10]):
            print(f"解析详情 {idx+1}/10：{p.title[:30]}...")
            detail.fetch_detail(p)
            time.sleep(delay + random.uniform(1, 3))
        all_alerts.extend(alert.check(products))

    if all_alerts:
        print(f"\n{'='*50}\n检测到{len(all_alerts)}条价格变动：")
        for item in all_alerts:
            print(f"{item['type']} 降幅{item['drop']}% | ${item['old']:.2f} → ${item['new']:.2f} | {item['title'][:35]}")
    else:
        print("\n未检测到显著价格变动")

# 启动监控任务
if __name__ == "__main__":
    track(["mechanical keyboard", "rtx 4070", "sony wh-1000xm5"])

六、代理选型与技术要点

1. 代理方案选型

eBay 不同采集场景适配不同代理模式：

• 搜索列表 / 常规详情页：动态转发标准版 ，依托大规模 IP 池实现请求隔离，配合<font style="color:rgb(0, 0, 0);background-color:rgba(0, 0, 0, 0);">Proxy-Tunnel</font>头部动态换 IP；
• 登录态 / 会话保持采集：固定转发版，短时锁定 IP 维持会话有效性。

IP 切换两种机制对比：

1. TCP Keep-Alive：复用会话连接，连接重建后自动换 IP；
2. Proxy-Tunnel 请求头：通过随机数值强制切换 IP，不受网络连接状态影响，本方案优先采用。

2. 常见故障排查

1. 搜索结果为空 ：请求头<font style="color:rgb(0, 0, 0);background-color:rgba(0, 0, 0, 0);">Accept-Language</font>需配置为<font style="color:rgb(0, 0, 0);background-color:rgba(0, 0, 0, 0);">en-US</font>，禁用原生<font style="color:rgb(0, 0, 0);background-color:rgba(0, 0, 0, 0);">requests</font>默认 UA，匹配浏览器标识；
2. 代理 407 错误：隧道代理账号、密码校验失败，核对平台凭据；
3. 价格解析异常：存在多币种、区间价场景，需扩展清洗逻辑适配不同格式；
4. CSS 选择器失效：平台前端迭代更新 Class，需重新通过 F12 抓取节点；

3. 方案边界与合规说明

1. 合规约束 ：遵循 eBay <font style="color:rgb(0, 0, 0);background-color:rgba(0, 0, 0, 0);">robots.txt</font> 规则，请求间隔不低于 3 秒，数据仅作个人价格分析，禁止商用分发；
2. 场景限制 ：本方案为定时批处理模式，不支持秒杀类实时监控；部分动态渲染页面需改用 Playwright 等无头浏览器；
3. 性能边界：批量高频轮询会提升风控概率，大规模监控建议拆分任务、分布式部署。