resumesend.java

用ＪＡＶＡ语言实现ＪＯＢ ＨＵＮＴＥＲ的功能

  } while (!finished);
  return result.toString();
 }
/*
 // Get the whole request
 public HTTPRequest getRequest() throws IOException {
  HTTPRequest request = new HTTPRequest();
  String line;
  do {
   line = readLine();
   if(line.length()>0) request.addLine(line);
   else break;
  }while(true);
  return request;
 }
 */
}
  class Queue extends java.util.Vector {

      public Queue(int size) {
          super(size);
      }

      public void enqueue(String u) {
        int tot = size();
        boolean nExist = true;
        for(int i=0;i<tot;i++){
          if(elementAt(i).equals(u)){
            nExist = false;
            break;
          }
        }
        if(nExist){
          addElement(u);
        }
      }

      public String dequeue() {
          if (size() > 0) {
              Object o = firstElement();
              removeElementAt(0);
              trimToSize();
              return (String) o;
          } else
              return null;
      }
  }

  private String zhaopin_pack_const(int page){
    String cook = "PageNo="+page+"&haveSelect=N&isLogin=N&actionTag=gotoPage&searchjobtype=0&jobfamily=2021300&industry=160400&industry=210500&industry=160000&citycode=440300&citycode=110000&citycode=310000&careerlevel=0&publishdate=4&keyword=%C8%ED%BC%FE&keyword1=&collapsedcompid=&foldedcompid=&displaytype=&PageNo1=2&vlist1=3&vlist2=2&vlist3=1&vlist4=10&PageNo2=9";
    String pack = "POST /jobseeker/ps_summary.jsp HTTP/1.1"+
        "\r\n"+
        "Accept: image/gif, image/x-xbitmap, image/jpeg, image/pjpeg, application/vnd.ms-powerpoint, application/vnd.ms-excel, application/msword, application/x-shockwave-flash, */*"+
        "\r\n"+
        "Referer: http://www.zhaopin.com/jobseeker/ps_search.jsp"+
        "\r\n"+
        "Accept-Language: zh-cn"+
        "\r\n"+
        "Content-Type: application/x-www-form-urlencoded"+
        "\r\n"+
        "Accept-Encoding: gzip, deflate"+
        "\r\n"+
        "User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.0; (R1 1.1))"+
        "\r\n"+
        "Host: www.zhaopin.com"+
        "\r\n"+
        "Content-Length: "+cook.length()+
        "\r\n"+
        "Connection: Keep-Alive"+
        "\r\n"+
        "Cache-Control: no-cache"+
        "\r\n"+
        "Cookie: JSESSIONID=2j5xvNhwhFqvCqUlicUKbFW4rO2A1yI2hBDkIqe9TNOzQUXVdNz7!1331033871!-1062729129!7001!7002; Zp_servicedhostname=www6.zhaopin.com"+
        "\r\n"+
        "\r\n"+
        //"searchjobtype=0&industry=160400&industry=210500&industry=160000&jobfamily=2021300&keyword=%C8%ED%BC%FE&citycode=440300&citycode=110000&citycode=310000&careerlevel=0&publishdate=4&button3=+%B2%E9+%D5%D2+"+
        cook+
        "\r\n"+
        "\r\n";
    return pack;
  }
  public void zhaopin_job_search() throws Exception{
    boolean first = true;
    int p_i = 1;
    int totP = 2;
    while (p_i < totP) {
      Socket sk1 = new Socket("202.108.226.67", 80);
      BufferedOutputStream os = new BufferedOutputStream(sk1.
          getOutputStream());
      String pk = zhaopin_pack_const(p_i++);
      os.write(pk.getBytes());
      os.flush();

      HTTPInputStream is = new HTTPInputStream(sk1.getInputStream());
      BufferedReader reader = new BufferedReader(new InputStreamReader(is));
      String temp_line, line;
      int j_p = 0;
      int u_p = 0;
      while ( (temp_line = reader.readLine()) != null) {
        //System.out.println(temp_line);
        //<A HREF="jobsearch.asp?page=6&mid=&order=0">
        line = temp_line;
        if (first) {
          //第<b class="brown12px"> 8/15 </b>页
          if ( (u_p = line.indexOf("</b>页")) > 0) {
            String s1 = line.substring(0, u_p - 1);
            if (s1.indexOf("第<b") > 0) {
              int p_p = s1.lastIndexOf("/");
              if (p_p < 0) {
                continue;
              }
              String sPg = s1.substring(p_p + 1);
              totP = Integer.parseInt(sPg);
              first = false;
            }
          }
        } //fisrt
        //<a href="ps_request.jsp?vacancyid=12041537" target="_blank">XML专家(专、兼职)</a>
        while ( (j_p = line.indexOf("vacancyid=")) > 0) {
          String s1 = line.substring(j_p + 10);
          line = s1;
          int rk = s1.indexOf(" ");
          String lNum = s1.substring(0, rk - 1);
          String emailUrl =
              "http://www.zhaopin.com/jobseeker/ps_request.jsp?vacancyid=" +
              lNum;
          //System.out.println(emailUrl);
          try {
            URL_email_search(emailUrl); ;
          }
          catch (Exception e) {
            System.out.println("Eamils: " + keyQue.size());
            System.out.println(emailUrl + " Failed!");
            e.printStackTrace(System.out);
          }
        }
      } //end while read
    } //end while <
  }


  public void w51job_email_search(String lNum) throws Exception{
    String head ="POST /sc/show_job_detail.php HTTP/1.1"+
    "\r\n"+
    "Accept: */*"+
    "\r\n"+
    "Referer: http://www.51job.com/sc/search_result.php"+
    "\r\n"+
    "Accept-Language: zh-cn"+
    "\r\n"+
    "Content-Type: application/x-www-form-urlencoded"+
    "\r\n"+
    "Accept-Encoding: gzip, deflate"+
    "\r\n"+
    "User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.0; (R1 1.1))"+
    "\r\n"+
    "Host: www.51job.com"+
    "\r\n"+
    "Content-Length: ";
    String body ="\r\n"+
    "Connection: Keep-Alive"+
    "\r\n"+
    "Cache-Control: no-cache"+
    "\r\n"+
    "Cookie: cenglish=0"+
    "\r\n"+
    "\r\n";
    String cook1 = "jobiduni=%28";
    String cook2 ="%29";
    String cook = cook1+lNum+cook2;
    String pck = head + cook.length()+body+cook;
    final byte[] pckB = pck.getBytes();
    final String fnum = lNum;
    Thread emsearch = new Thread(){
      public void run() {
        try {
          ResumeSend.threads++;
          boolean tryCon = true;
          BufferedReader reader = null;
          while(tryCon){
            try{
              Socket sk1 = new Socket("61.129.66.127", 80);
              BufferedOutputStream os = new BufferedOutputStream(sk1.
                  getOutputStream());
              os.write(pckB);
              os.flush();
              HTTPInputStream is = new HTTPInputStream(sk1.getInputStream());
              reader = new BufferedReader(new InputStreamReader(is));
              tryCon = false;
            }catch(Exception e){
              tryCon = true;
              int trys=1+(int)(10*Math.random());
              e.printStackTrace(System.out);
              System.out.println("Connect Failed! Let us wait "+trys+" minutes for 51job "+fnum+" again!");
              this.sleep(1000*60*trys);
            }
          }//end while
          String temp_line, line;
          int m_p = 0;
          while ( (temp_line = reader.readLine()) != null) {
            line = temp_line;
            while ( (m_p = line.indexOf("mailto:")) > 0) {
              String s1 = line.substring(m_p + 7);
              line = s1;
              int ed1 = s1.indexOf("'");
              int ed2 = s1.indexOf(">") - 1;
              int ed3 = ed1 < 0 ? ed2 :
                  (ed2 < 0 ? ed1 : ( (ed1 < ed2) ? ed1 : ed2));
              if (ed3 < 1)
                ed3 = Math.min(40, s1.length());
              String lem = s1.substring(0, ed3).toLowerCase();
              if ( (lem.equals("[e-mail]")) ||
                  (lem.equals("[e-mail]")))
                continue;
              //int end = s1.indexOf("?");
              int bak_p = 100;
              int bak_i = 0;
              for (int j = 0; j < bakS.length; j++) {
                int tt = 0;
                tt = lem.indexOf(bakS[j]);
                if ( (tt > 0) && (tt < bak_p)) {
                  bak_p = tt;
                  bak_i = j;
                }

              }
              String email;
              if (bak_p < 100) {
                email = lem.substring(0, bak_p) + bakS[bak_i];
              }
              else {
                email = lem.substring(0, ed3);
              }
              synchronized (keyQue) {
                keyQue.add(email);
              }
            }
          }
        }catch(Exception e){
          e.printStackTrace(System.out);
        }
        ResumeSend.threads--;
      }
    };
    emsearch.start();
  }
  private void w51job_search(String sArea)throws Exception{
    String baseUrl = "http://www.51job.com/";
    String fstPack = "POST /sc/search_result.php?begin=begin HTTP/1.1"+
    "\r\n"+
    "Accept: */*"+
    "\r\n"+
    "Referer: http://www.51job.com/default-area.php?area=0100"+
    "\r\n"+
    "Accept-Language: zh-cn"+
    "\r\n"+
    "Content-Type: application/x-www-form-urlencoded"+
    "\r\n"+
    "Accept-Encoding: gzip, deflate"+
    "\r\n"+
    "User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.0; (R1 1.1))"+
    "\r\n"+
    "Host: www.51job.com"+
    "\r\n"+
    "Content-Length: 121"+ //cook中以jobarea_sel开头的查询语句的长度
    "\r\n"+
    "Connection: Keep-Alive"+
    "\r\n"+
    "Cache-Control: no-cache"+
    "\r\n"+
    "Cookie: cenglish=0"+
    "\r\n"+
    "\r\n"+
    "jobarea_sel="+sArea+"&jobarea="+sArea+"&funtype=0100&issuedate=03&keyword=%C8%ED%BC%FE&keywordtype=2&begin=%B2%E9%D1%AF%D6%B0%CE%BB"+
    "\r\n";

    String head1 = "POST /sc/search_result.php HTTP/1.1"+
    "\r\n"+
    "Accept: image/gif, image/x-xbitmap, image/jpeg, image/pjpeg, application/vnd.ms-powerpoint, application/vnd.ms-excel, application/msword, application/x-shockwave-flash, */*"+
    "\r\n"+
    "Referer: http://www.51job.com/sc/search_result.php"+
    "\r\n"+
    "Accept-Language: zh-cn"+
    "\r\n"+
    "Content-Type: application/x-www-form-urlencoded"+
    "\r\n"+
    "Accept-Encoding: gzip, deflate"+
    "\r\n"+
    "User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.0; (R1 1.1))"+
    "\r\n"+
    "Host: www.51job.com"+
    "\r\n"+
    "Content-Length: ";
    String head2="\r\n"+
    "Connection: Keep-Alive"+
    "\r\n"+
    "Cache-Control: no-cache"+
    "\r\n"+
    "Cookie: cenglish=0"+
    "\r\n"+
    "\r\n";
    String cook1 = "jobarea="+sArea+"&industrytype=&funtype=0100&issuedate=03&salaryfrom=&salaryto=&salaryno=&age=&workyear=&degreefrom=&forlang=&funkeyword=%C8%ED%BC%FE&cokeyword=&otherkeyword=&postchannel=&channel=&attachpa1=&attachpa2=";//81
    String cook12 = "&attachpa3=&attachpa4=";//100
    String cook13 = "&attachpa11=&attachpa21=";//61
    String cook14 = "&attachpa31=&attachpa41=";//80
    String cook2 = "&servicename=SC_ShowNP_Sel&shangxia=1&totalcount=400&again=again&freshflag=0&pageno=5&jSQL=%28+jobarea+like+%2701__%27++or+jobarea+like+%2737__%27+%29+and+%28+%28functype1+like+%2701__%27+or+functype2+like+%2701__%27%29+%29+and+%28issuedate%3E%3Dsysdate-03%29+and+%28upper%28cjobname%29+like+upper%28%27%25%C8%ED%BC%FE%25%27%29+or+upper%28ejobname%29+like+upper%28%27%25%C8%ED%BC%FE%25%27%29%29+and+is51job%3D%271%27&jSQL2=%28+jobarea+like+%2701__%27++or+jobarea+like+%2737__%27+%29+and+%28+%28functype1+like+%2701__%27+or+functype2+like+%2701__%27%29+%29+and+%28issuedate%3E%3Dsysdate-03%29+and++contains%28sjobname%2C%27%7B%C8%ED%BC%FE%7D%27%2C2%29%3E0++and+is51job%3D%271%27&sort2=cocname&sort3=issuedate&sort4=jobarea&sort5=providesalary";//&inpageno1=";//4 page num
    String cook3 = "&againkeyword=&againkeywordtype=1&sortby=issuedate&ascdesc=desc";

    boolean first = true;
    int p_i =1;
    int totP =2 ;
    String pck=fstPack;
    while(p_i<totP)
    {
      Socket sk1 = new Socket("61.129.66.127", 80);
      BufferedOutputStream os = new BufferedOutputStream(sk1.getOutputStream());
      //os.write(fstPack.getBytes());
      os.write(pck.getBytes());
      os.flush();
      //构造新的包
      String cook = cook1 + (p_i*20+1) + cook12 + ((p_i+1)*20) + cook13 + (((p_i-1)*20)+1) + cook14 + (p_i*20) + cook2 + 4 + cook3;
      p_i++;
      int len = cook.length();
      pck = head1 + len + head2;

      HTTPInputStream is = new HTTPInputStream(sk1.getInputStream());
      BufferedReader reader = new BufferedReader(new InputStreamReader(is));
      os.write(cook.getBytes());
      os.flush();
      String temp_line, line;
      int j_p = 0;
      int u_p = 0;
      while ( (temp_line = reader.readLine()) != null) {
        // System.out.println(temp_line);
        //<A HREF="jobsearch.asp?page=6&mid=&order=0">
        line = temp_line;
        if (first) {
          if ( (u_p = line.indexOf("totalcount value=")) > 0) {
            String s1 = line.substring(u_p + 17);
            int p_p = s1.indexOf(">");
            if (p_p > 100) {
              continue;
            }
            String sPg = s1.substring(0, p_p);
            totP = Integer.parseInt(sPg);
            totP = totP/20+1;
            first = false;
          }
        }
        while ( (j_p = line.indexOf("javascript:OJDL(")) > 0) {
          //Javascript:view('22203241092220128201','2002-5-13')
          String s1 = line.substring(j_p + 16);
          line = s1;
          int rk = s1.indexOf(")");
          String lNum = s1.substring(0, rk);
          try{
            w51job_email_search(lNum);
          }catch(Exception e){
            System.out.println("51job "+lNum+" search for Email failed!");
            e.printStackTrace(System.out);
          }
        }
      } //end while read
    }//end while <
 }


}