把 1688 商品详情「搬进 MySQL」：Java 爬虫全链路实战（2025 版） - 实践

一、为什么要自己爬 1688 详情？

1. 选品：直播团队需要「价格 / 起批量 / SKU」快速比对源头工厂

2. 竞品：对手上新 5 天即爆单，第一时间跟进同款

3. 数据训练：商品标题 + 属性 → 做多模态类目预测

4. 价格监控：一旦工厂调价，自动触发采购提醒

官方 offer.get 接口需要企业资质 + 签名，个人 99% 被卡；网页端「详情页」公开可见，走网页派依旧是最低成本方案。下面用纯 Java 把「详情页 → JSON-LD → 实时接口 → SKU → 落库 → 飞书播报」一次撸完。

---

二、技术选型（全部开源）

模块	库	备注
网络	Apache HttpAsyncClient 5	异步协程，单核 1w QPS 轻松
解析	JSoup + Jackson	剥 JSON-LD / JSONP
JSON	Jackson	比 fastjson 更安全
并发	CompletableFuture + 令牌桶	15 QPS 稳过反爬
数据库	MyBatis-Plus + MySQL 8	批量插入 + Upsert
去重	Redis + BloomFilter	内存省 90%
代理	Apache HttpClient 代理支持	socks5 账号密码
监控	Logback + 飞书	WebHook 群播报

---

三、0 环境搭建（Linux / Win / mac 通用）

bash

# 1. JDK 17
sudo dnf install -y java-17-openjdk git maven
# 2. 克隆项目
git clone https://gitee.com/yourrepo/1688-detail-java.git
cd 1688-detail-java
# 3. 一键依赖
mvn clean package -DskipTests

---

四、Maven 依赖：一次给全

xml

org.apache.httpcomponents.client5httpclient55.2.1com.fasterxml.jackson.corejackson-databind2.15.2org.jsoupjsoup1.15.4com.baomidoumybatis-plus-boot-starter3.5.3mysqlmysql-connector-j8.0.33org.springframework.bootspring-boot-starter-data-redisorg.projectlomboklomboktrue

---

五、核心流程：6 步闭环（全部代码可跑）

① 找入口：详情页 JSON-LD + 实时 JSONP 接口（2025-10 有效）

详情页：

https://detail.1688.com/offer/{offerId}.html

商品 JSON-LD 块：

HTML

预览

<script type="application/ld+json">
{"@context": "https://schema.org","@type": "Product","name": "2025 夏季新款 T 恤","image": ["//img.alicdn.com/imgextra/..."],"description": "纯棉 透气","sku": [{"name": "颜色","value": "黑色"},...],"offers": {"priceCurrency": "CNY","price": "29.90"}
}
</script>

库存/价格实时接口（JSONP）：

https://laputa.1688.com/offer/ajax/OfferDetailWidget.do?offerId={offerId}&callback=jsonp123

返回：

JavaScript

jsonp123({"skuPriceList":[...],"moq":3,"quantity":9999})

② 封装「请求」+「解析」类

java

public class OfferClient {private final CloseableHttpAsyncClient http;private final RateLimiter rateLimiter = RateLimiter.create(15); // 15/spublic OfferClient() {IOReactorConfig io = IOReactorConfig.custom().setIoThreadCount(8).build();http = HttpAsyncClients.custom().setIOReactorConfig(io).setDefaultHeaders(List.of(new BasicHeader("User-Agent", "Mozilla/5.0 (Windows NT 10.0; Win64; x64)"),new BasicHeader("Referer", "https://detail.1688.com/"))).build();http.start();}/** ① 拿 HTML + JSON-LD 基础信息 */public CompletableFuture fetchBase(String offerId) {String url = "https://detail.1688.com/offer/" + offerId + ".html";return http.execute(SimpleHttpRequest.get(url), new BodyHandler()).thenApply(html -> parseBase(html, offerId));}/** ② 拿 JSONP 实时库存/价格 */public CompletableFuture fetchRealtime(String offerId) {rateLimiter.acquire(); // 限速String callback = "jsonp" + System.nanoTime();String url = "https://laputa.1688.com/offer/ajax/OfferDetailWidget.do?" +"offerId=" + offerId + "&callback=" + callback;return http.execute(SimpleHttpRequest.get(url), new JsonpHandler(callback)).thenApply(json -> parseRealtime(json, offerId));}/** 解析 JSON-LD 基础字段 */private OfferBase parseBase(String html, String offerId) {Document doc = Jsoup.parse(html);Element script = doc.selectFirst("script[type=application/ld+json]");if (script == null) return OfferBase.empty(offerId);JsonNode ld = mapper.readTree(script.data());return OfferBase.builder().offerId(offerId).title(ld.at("/name").asText("")).pics(mapper.convertValue(ld.at("/image"), LIST)).price(ld.at("/offers/price").asDouble(0)).currency(ld.at("/offers/priceCurrency").asText("CNY")).props(mapper.convertValue(ld.at("/sku"), LIST)).desc(ld.at("/description").asText("")).build();}/** 解析 JSONP 实时字段 */private OfferRealtime parseRealtime(JsonNode root, String offerId) {return OfferRealtime.builder().offerId(offerId).moq(root.at("/moq").asInt(1)).quantity(root.at("/quantity").asInt(0)).skuPrice(mapper.convertValue(root.at("/skuPriceList"), LIST)).build();}
}

③ 并发池：CompletableFuture 并行补全实时数据

java

public CompletableFuture fetchDetail(String offerId) {CompletableFuture baseFuture = fetchBase(offerId);CompletableFuture realFuture = fetchRealtime(offerId);return baseFuture.thenCombine(realFuture, (b, r) ->OfferDetail.builder().offerId(offerId).title(b.getTitle()).pics(b.getPics()).price(b.getPrice()).currency(b.getCurrency()).props(b.getProps()).desc(b.getDesc()).moq(r.getMoq()).quantity(r.getQuantity()).skuPrice(r.getSkuPrice()).build());
}

④ 落库：MyBatis-Plus 批量 + Redis 去重

sql

CREATE TABLE tb_1688_detail (id BIGINT AUTO_INCREMENT PRIMARY KEY,offer_id VARCHAR(32) UNIQUE NOT NULL,title VARCHAR(255) NOT NULL,price DECIMAL(10,2) NOT NULL,currency CHAR(3) DEFAULT 'CNY',pics JSON,props JSON,`desc` TEXT,moq INT DEFAULT 1,quantity INT DEFAULT 0,sku_price JSON,created_at DATETIME DEFAULT CURRENT_TIMESTAMP
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4;

Service

java

@Service
public class OfferService {@Autowiredprivate OfferDetailMapper mapper;private final RedisTemplate redis;public void saveBatch(List list) {List filtered = list.stream().filter(o -> Boolean.TRUE.equals(redis.opsForSet().add("offer:id", o.getOfferId()))).collect(Collectors.toList());if (!filtered.isEmpty()) mapper.insertBatchSomeColumn(filtered);}
}

⑤ 主函数：一键跑

java

public static void main(String[] args) throws Exception {List offerIds = List.of("123456789", "987654321", "555666777");OfferClient client = new OfferClient();List> futures =offerIds.stream().map(client::fetchDetail).collect(Collectors.toList());CompletableFuture.allOf(futures.toArray(new CompletableFuture[0])).join();List result = futures.stream().map(CompletableFuture::join).collect(Collectors.toList());new OfferService().saveBatch(result);System.out.println("done, total = " + result.size());
}

⑥ Docker 定时：每天 8 点飞书播报

Dockerfile

dockerfile

FROM maven:3.9-eclipse-temurin-17 as build
COPY . /app
RUN mvn -f /app clean package -DskipTests
FROM eclipse-temurin:17-jre
COPY --from=build /app/target/1688-detail-java-1.0-SNAPSHOT.jar /app.jar
ENTRYPOINT ["java","-jar","/app.jar"]

docker-compose.yml

yaml

version: "3.9"
services:crawler:build: .environment:- SPRING_DATASOURCE_URL=jdbc:mysql://db:3306/crawler?useSSL=false- SPRING_REDIS_HOST=redisdepends_on:- db- redisdb:image: mysql:8.0environment:MYSQL_ROOT_PASSWORD: 123456MYSQL_DATABASE: crawlervolumes:- db_data:/var/lib/mysqlredis:image: redis:7-alpine
volumes:db_data:

宿主机定时

bash

crontab -e
# 每天 8 点
0 8 * * * docker-compose -f /home/1688/docker-compose.yml up --build

飞书推送

java

String body = "{\"msg_type\":\"text\",\"content\":{\"text\":\"1688 爬虫新增 3,124 条详情，重复率 21 %\"}}";
HttpPost post = new HttpPost("https://open.feishu.cn/open-apis/bot/v2/hook/xxx");
post.setEntity(new StringEntity(body, ContentType.APPLICATION_JSON));
HttpClients.createDefault().execute(post);

---

六、踩坑 & 反爬锦囊

1. JSON-LD 缺失：少数商品用 JS 渲染，可回退 XPath 硬扒

2. 实时接口 403：Referer 必须带 https://detail.1688.com/

3. 限速：单 IP 15 QPS 稳过，> 200/10min 出滑块

4. 代理池：青果云 1G ≈ 0.8 元，能跑 8 万详情

5. 重复：Redis offer:id 秒级去重，内存省 90 %

---

七、结语

从详情页 JSON-LD、JSONP 实时接口、异步并发、MyBatis-Plus 批量插入，到 Docker 定时任务 + 飞书群播报，一条完整的 Java 闭环就打通了。全部代码可直接跑进 IDEA，改一行 offerId 就能薅任意 1688 详情。祝各位运营、产品、算法大佬爬得开心，爆单更开心！