基于spark的网络爬虫实现

爬虫是获取网络大数据的重要手段，爬虫是一种非常成熟的技术了，然而想着在spark环境下测试一下效果．

还是非常简单的，利用JavaSparkContext来构建，就可以采用原来java中的网页获取那一套来实现．

首先给定几个初始种子，生成一个JavaRDD对象即可

JavaRDD<String> rdd = sc.parallelize("urllist");

JavaRDD<String> content = rdd.map(new Function<String, String>() {

public String call(String url) throws Exception {

System.out.println(url);

CloseableHttpClient client = null;

HttpGet get = null;

CloseableHttpResponse response = null;

try {

//## 创建默认连接

client = HttpClients.createDefault();

get = new HttpGet(url);

response = client.execute(get);

HttpEntity entity = response.getEntity();

//## 获得输出字节流

ByteArrayOutputStream byteArrayOutputStream = new ByteArrayOutputStream();

entity.writeTo(byteArrayOutputStream);

//## 转化为文档

String html = new String(byteArrayOutputStream.toByteArray(), Charsets.UTF_8);

Document document = Jsoup.parse(html);

return html;

} catch (Exception ex) {

ex.printStackTrace();

return "";

} finally {

if (response != null) {

response.close();

}

if (client != null) {

client.close();

}

}

}

});

当然可以从ＨＴＭＬ再找到子页连接，继续以深度或者广度进行优先爬虫．

如输出http://docs.opencv.org/的文档如下：