webmagic 爬虫学习代码。

webmagic 爬虫学习代码。

import us.codecraft.webmagic.Page;
import us.codecraft.webmagic.Site;
import us.codecraft.webmagic.Spider;
import us.codecraft.webmagic.pipeline.FilePipeline;
import us.codecraft.webmagic.processor.PageProcessor;

public class JobProcessor implements PageProcessor {
    //这个方法解析页面
    @Override
    public void process(Page page) {
        //解析返回的数据page,并且把解析的结果放到ResultItems中
//        page.putField("div",page.getHtml().css("head"));
        System.out.println("=============================");

//        page.putField("div2",page.getHtml().css("head"));
//        page.putField("div3",page.getHtml().css("html"));
        page.addTargetRequests(page.getHtml().css("div.album_detail_wrap a").links().all());
        page.putField("html",page.getHtml().css("div.album_detail_wrap dl dt a").links().all());
        page.putField("url",page.getHtml().css("div.album_detail_wrap a").links().all());

    }
    private Site site =Site.me()
            .setCharset("utf8")
            .setTimeOut(10000)
            .setRetrySleepTime(3000)//设置重试的时间间隔
            .setSleepTime(3);
    @Override
    public Site getSite() {
        return site;
    }

    public static void main(String[] args) {
        Spider.create(new JobProcessor())
                .addUrl("https://download.csdn.net/download/weixin_45026040/11260253")
                .addPipeline(new FilePipeline("C:\\Users\\Administrator\\Desktop\\yyycsdn"))
                .thread(5)
                .run();
    }
}