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：