获取单个页面的所有信息

package cn.edu.hfut.dmic.contentextractor;

import java.io.File;

import java.io.IOException;

import javax.lang.model.util.Elements;

import org.jsoup.Jsoup;

import org.jsoup.nodes.Document;

import org.jsoup.nodes.Element;

import jxl.*;

import jxl.write.*;

public class addToExcel {

public static void main(String[] args) throws IOException {

try

{

//生成excel

WritableWorkbook book=Workbook.createWorkbook(new File("c:/wexcel/noah-fund.xls"));

//生成名为“第一页”的工作表，参数0表示这是第一页

WritableSheet sheet=book.createSheet("第一页",0);

Label label1=new Label(0,0,"代码");

sheet.addCell(label1);

Label label2=new Label(1,0,"名称");

sheet.addCell(label2);

Label label3=new Label(2,0,"单位净值");

sheet.addCell(label3);

Label label4=new Label(3,0,"累计净值");

sheet.addCell(label4);

Label label5=new Label(4,0,"单位净值");

sheet.addCell(label5);

Label label6=new Label(5,0,"累计净值");

sheet.addCell(label6);

Label label7=new Label(6,0,"日增长值");

sheet.addCell(label7);

Label label8=new Label(7,0,"日增长率");

sheet.addCell(label8);

Label label9=new Label(8,0,"基金类型");

sheet.addCell(label9);

Label label10=new Label(9,0,"购买状态");

sheet.addCell(label10);

Label label120=new Label(10,0,"赎回状态");

sheet.addCell(label120);

Label label11=new Label(11,0,"基金经理");

sheet.addCell(label11);

Label label12=new Label(12,0,"成立时间");

sheet.addCell(label12);

Label label13=new Label(13,0,"基金状态");

sheet.addCell(label13);

/*——————————————————————————————————————————获取基金数据——————————————————————————————————————————*/

Document doc = Jsoup.connect("http://www.noah-fund.com/html/fundtools/initDaQuan.html").get();

// String script1=doc.select("script").get(15).data();

// System.out.println("script1:"+script1);

/*统计指定元素的个数*/

int num = doc.select("td[bgcolor=#FFFFFF]").size();

System.out.println("num:"+num);

/*统计当前页面基金个数*/

int fundnum=doc.select("a[href*=/fundtools/FundInfoAction.jsp?symbol=]").size();

System.out.println("fundnum:"+fundnum);

/*直接获取每一行的基金数据数据*/

for(int i=0;i<num;i++){

String tags=doc.select("td[bgcolor=#FFFFFF]").get(i).text();

System.out.print(tags+" ");

if((i+1)%11==0)

System.out.println();

}

/*如何获取所有基金的数据？*/

// Element pagehref=doc.select("a[href*=javascript:jumpPage]").last();

// String pagelink=pagehref.attr("abs:href");

// System.out.println("最后一页："+pagelink);

/*获取所有基金的绝对路径*/

for(int i=0;i<fundnum;i++){

Element links=doc.select("a[href*=/fundtools/FundInfoAction.jsp?symbol=]").get(i);

String fundString=links.attr("abs:href");

// System.out.println(fundString);

/*获取子页面基金主页数据*/

Document document =Jsoup.connect(fundString).get();

Element ifr=document.select("iframe").first();

String linkString=ifr.attr("abs:src");

// System.out.println(linkString);

/*获取基金份额+此展规模*/

// String fundpercent=document.select("div.pflo-info-box").first().select("p").text();

// System.out.println(fundpercent);

/*获取单个基金的详细信息*/

Document fundDocument=Jsoup.connect(linkString).get();

String fundtype=fundDocument.select("td[width=160]").first().text().substring(5);

Label label14=new Label(8,i+1,fundtype);

sheet.addCell(label14);

String isbuy=fundDocument.select("td[width=160]").get(1).text().substring(3, 5);

Label label15=new Label(9,i+1,isbuy);

sheet.addCell(label15);

String issell=fundDocument.select("td[width=160]").get(1).text().substring(9, 11);

Label label19=new Label(10,i+1,issell);

sheet.addCell(label19);

String manager=fundDocument.select("tr[class=row1]").get(1).children().get(1).text().substring(5);

Label label16=new Label(11,i+1,manager);

sheet.addCell(label16);

String setdate=fundDocument.select("tr[class=row1]").get(1).children().get(2).text().substring(5);

Label label17=new Label(12,i+1,setdate);

sheet.addCell(label17);

String status=fundDocument.select("div.btn-box").text();

Label label18=new Label(13,i+1,status);

sheet.addCell(label18);

System.out.println(fundtype+" "+isbuy+" "+manager+" "+setdate+" "+status);

}

/*基金详细信息：代码~累计净值*/

for(int i=0;i<num;i++){

int x=i/11+1;

int y=(i-1)%11;

if((i+1)%11==2){

String fundcode=doc.select("td[bgcolor=#FFFFFF]").get(i).text();

Label label=new Label(y,x,fundcode);

sheet.addCell(label);

System.out.print("fundcode:"+fundcode);

}

if((i+1)%11==3){

String fundname=doc.select("td[bgcolor=#FFFFFF]").get(i).text();

Label label=new Label(y,x,fundname);

sheet.addCell(label);

System.out.print("fundname:"+fundname);

}

if((i+1)%11==4){

String price=doc.select("td[bgcolor=#FFFFFF]").get(i).text();

Label label=new Label(y,x,price);

sheet.addCell(label);

System.out.print("price:"+price);

}

if((i+1)%11==5){

String pricesum=doc.select("td[bgcolor=#FFFFFF]").get(i).text();

Label label=new Label(y,x,pricesum);

sheet.addCell(label);

System.out.print(" pricesum:"+pricesum);

}

if((i+1)%11==6){

String price1=doc.select("td[bgcolor=#FFFFFF]").get(i).text();

Label label=new Label(y,x,price1);

sheet.addCell(label);

System.out.print(" price1:"+price1);

}

if((i+1)%11==7){

String pricesum1=doc.select("td[bgcolor=#FFFFFF]").get(i).text();

Label label=new Label(y,x,pricesum1);

sheet.addCell(label);

System.out.print(" pricesum1:"+pricesum1);

}

if((i+1)%11==8){

String dayplus=doc.select("td[bgcolor=#FFFFFF]").get(i).text();

Label label=new Label(y,x,dayplus);

sheet.addCell(label);

System.out.print(" dayplus:"+dayplus);

}

if((i+1)%11==9){

String dayrate=doc.select("td[bgcolor=#FFFFFF]").get(i).text();

Label label=new Label(y,x,dayrate);

sheet.addCell(label);

System.out.println(" dayrate:"+dayrate);

}

}

book.write();

book.close();

System.out.println("成功将数据写入excel");

}catch(Exception e)

{

System.out.println(e);

}

}

}