package task;

import java.io.File;
import java.io.FileNotFoundException;
import java.io.IOException;
import java.io.PrintWriter;
import java.util.ArrayList;

import org.jsoup.Jsoup;
import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.select.Elements;

public class Crawler {
    
    //1.成员变量
    private static String outputfile = "output\\2018年世界各国GDP数据.txt";
    private static String sourcefile = "resource\\2018年世界各国GDP数据 .htm";
    
    
    //2.解析本地网页，返回Document
    public static Document getHtml() throws IOException {
        File f=new File(sourcefile);  //创建文件对象
        //解析本地文件，第一个参数是文件对象
        //第二个参数是网页的字符编码格式
        Document doc=Jsoup.parse(f,"UTF-8");  
        return doc;
    }
    
    //3.提取数据，返回集合
            public static ArrayList<EntityBean> getData(Document doc) throws IOException {
                //利用selector选择器，提取32行数据
                Elements trs=doc.select("body > div.container > div.container > div:nth-child(5) > div.col-md-9 > div > div > div > table > tbody > tr");
                ArrayList<EntityBean> list=new ArrayList<EntityBean>();//创建集合对象
                for(Element tr : trs) {
                    if(tr.childNodeSize()<5) {
                        continue;
                    }
                    //获取tr，每一行的前6列数据
                    String pm=tr.child(0).text();
                    String gj=tr.child(1).text();
                    String szz=tr.child(2).text();
                    String GDP=tr.child(3).text();
                    String zb=tr.child(4).text();
                    //每行数据创建一个对应的实例化对象ed
                    EntityBean eb=new EntityBean(pm,gj,szz,GDP,zb);
                    //System.out.println(eb.toString());
                    list.add(eb);//将每一行数据对象放进集合
                }
                return list;
            }
    
    //4.写出到本地文件系统
            public static void wirteToFile(ArrayList<EntityBean>  ebs) throws FileNotFoundException {
                PrintWriter pw=new PrintWriter(outputfile);//创建字节打印流对象
                for(EntityBean eb : ebs) {
                    pw.println(eb.getPm()+"\t"+eb.getGj()+"\t"+eb.getSzz()+
                            "\t"+eb.getGDP()+"\t"+eb.getZb());
                }
                pw.close();
            }
    //5.主方法
    public static void main(String[] args) throws IOException {
        Document doc = Crawler.getHtml();//解析网页，得到doc对象
        System.out.println("网页标题："+doc.title());
        ArrayList<EntityBean> entitys = Crawler.getData(doc);  //提取数据
        Crawler.wirteToFile(entitys);
    }
    
}

package task;

public class EntityBean {
    //成员变量列表
    private String pm;
    private String gj;
    private String szz;
    private String GDP;
    private String zb;
    
    //get()方法和set()方法
    public String getPm() {
        return pm;
    }
    public void setPm(String pm) {
        this.pm = pm;
    }
    public String getGj() {
        return gj;
    }
    public void setGj(String gj) {
        this.gj = gj;
    }
    public String getSzz() {
        return szz;
    }
    public void setSzz(String szz) {
        this.szz = szz;
    }
    public String getGDP() {
        return GDP;
    }
    public void setGDP(String gDP) {
        GDP = gDP;
    }
    public String getZb() {
        return zb;
    }
    public void setZb(String zb) {
        this.zb = zb;
    }
    
    //重写toString()方法
    @Override
    public String toString() {
        return "EntityBean [pm=" + pm + ", gj=" + gj + ", szz=" + szz + ", GDP=" + GDP + ", zb=" + zb + "]";
    }
    
    //无参构造方法
    public EntityBean() {
        super();
    }
    
    //有参构造方法
    public EntityBean(String pm, String gj, String szz, String gDP, String zb) {
        super();
        this.pm = pm;
        this.gj = gj;
        this.szz = szz;
        GDP = gDP;
        this.zb = zb;
    }
}