yellowpage.java

数字图书馆的互操作接口

package dli2fe.sample;

/**
 * Title:        Digial Library Interoperable Interface Fudan Edition
 * Description:  This project contains all the classes required for DLI2FE interface. Developers use these classes to implement the wrapper and client side codes. The basic functions of DLI2FE is as follows:
 * Search: Search a digital library source site.
 * Metadata: Fetch metadata for a site.
 * ResultAccess: Get results for a given query.
 * DLI2FE uses Dublin Core as the basic attribute model, DAV/DASL as the general XML-based query language and CORBA as distributed object transportation mechanism.
 * Copyright:    Copyright (c) 2001
 * Company:      Fudan University
 * @author Carl Tao
 * @version 1.0
 */

/**
 * this is the internal implementation class for YellowpageLSP.
 * this class will search yellowpage.com.cn for enterprise name/address/phone
 * information from http://www.yellowpage.com.cn/
 *
 */
import dli2fe.DLI2FE;
import dli2fe.DLI2FEException;
import java.io.*;
import dli2fe.xml.XMLObject;
import java.util.*;
import java.net.*;
import org.w3c.tools.codec.Base64Encoder;

//import HTTPClient.*;

public class Yellowpage {

  public static final String Namespace = "http://www.yellowpage.com.cn/#";
  public static final String NAME = Namespace + "name";
  public static final String ADDRESS = Namespace + "address";
  public static final String PHONE = Namespace + "phone";

  // search URL and parameters
  public static final String ypHOST = "www.yellowpage.com.cn";
  public static final String ypURL = "http://www.yellowpage.com.cn/";
  public static final String queryPATH = ypURL + "Prog/Busi.asp?";
  public static final String paramNAME = "NameText";
  public static final String paramADDRESS = "AddrText";
  public static final String paramPHONE = "TeleText";
  public static final String paramAREA = "AreaText";
  public static final String paramPAGE = "p";

  // patterns for search results
  public static final String NOT_FOUND = "未查到符合条件的企事业单位";
  public static final String FOUND1 = "共查到";
  public static final String FOUND2 = "个企事业";
  public static final String JS_DETAIL = "javascript:Detail";

  // patterns for detail
  public static final String detailNAME = "名　称：";
  public static final String detailADDR = "地　址：";
  public static final String detailPOST = "邮　编：";
  public static final String detailPHONE = "电　话：";

  public static final int resultsPerPage = 20;
  public static final String detailURL = "http://www.yellowpage.com.cn/Prog/Trade/Detail.asp?";

  public static boolean useProxy = true;
  String name;
  String addr;
  String phone;
  String city;
  //HTTPConnection con;
  Vector result;
  int numResults;

  public class ResultDoc {
    public String name;
    public String url;
  }

  public Yellowpage() {
    //HTTPConnection.setProxyServer("202.120.224.4", 8080);
    //AuthorizationInfo.addBasicAuthorization("202.120.224.4", 8080, "unspecified", "xh_xh", "xh_xh");
    result = new Vector();
  }

  public Yellowpage(String name0, String addr0, String phone0, String city0) {
    this();
    name = name0 == null ? "" : name0.trim();
    addr = addr0 == null ? "" : addr0.trim();
    phone = phone0 == null ? "" : phone0.trim();
    city = city0 == null ? "上海" : city0.trim();
  }

  public ResultDoc getResult(int index) throws DLI2FEException {
    if (index < 0 || index >= result.size())
      throw new DLI2FEException(DLI2FEException.INVALID_REQUEST_EXC, "Non existent result:" + index);
    return (ResultDoc)result.elementAt(index);
  }

  public String getResultString(int index) throws DLI2FEException {
    if (index < 0 || index >= result.size())
      throw new DLI2FEException(DLI2FEException.INVALID_REQUEST_EXC, "Non existent result" + index);
    ResultDoc resultDoc = (ResultDoc)result.elementAt(index);
    return "<doc> <DID>" + (index + 1) + "</DID> <propList> <dc:Title>" + resultDoc.name +
      "</dc:Title> <dc:Identifier>" + URLEncoder.encode(resultDoc.url) + "</dc:Identifier> </propList> </doc>";
  }

  public int getNumDocs() {
    return numResults;
  }

  public void searchYellowpage() throws DLI2FEException {
    searchYellowpage(name, addr, phone, city);
  }

  public void searchYellowpage(String name0, String addr0, String phone0, String city0) throws DLI2FEException {
    String resultHTML = "";
    int i, j;
    name = name0 == null ? "" : name0.trim();
    addr = addr0 == null ? "" : addr0.trim();
    phone = phone0 == null ? "" : phone0.trim();
    city = city0 == null ? "上海" : city0.trim();

    resultHTML = getResultPage(1);
    if (resultHTML.indexOf(NOT_FOUND) != -1) {
      result = new Vector();
      throw new DLI2FEException(DLI2FEException.NOT_FOUND_EXC, "No document found");
    }
    i = resultHTML.indexOf(FOUND1);
    j = resultHTML.indexOf(FOUND2);
    if (i == -1 || j == -1)
      throw new DLI2FEException(DLI2FEException.NOT_FOUND_EXC, "Error parsing documents");
    numResults = Integer.parseInt(resultHTML.substring(i + 3, j));
    result = new Vector();
    parsePage(resultHTML);  // parse this page and get the results to result Vector
    if (numResults > resultsPerPage)
      for (i = 2; i <= numResults / resultsPerPage + 1; i++) {
        resultHTML = getResultPage(i);
        parsePage(resultHTML);
      }
    numResults = result.size();
  }

  String getResultPage(int pageNo) throws DLI2FEException {
    String dbg;
    String page = "";
    if (name.equals("") && addr.equals("") && phone.equals(""))
      throw new DLI2FEException(DLI2FEException.BAD_QUERY_EXC, "Query for nothing");
    try {
      if (!useProxy) {
        BufferedReader din = new BufferedReader(new InputStreamReader(
          new URL(dbg = makeRequestWithPage(pageNo)).openStream()));
        System.out.println(dbg);
        while ((dbg = din.readLine()) != null) {
          page += dbg + "\n";
        }
        din.close();
      } else {
        System.getProperties().put("proxySet", "false");
        System.getProperties().put("proxyHost", "202.120.224.104");
        System.getProperties().put("proxyPort", "8080" );
        URL url = new URL(dbg = makeRequestWithPage(pageNo));
        System.out.println(dbg);
        URLConnection con = url.openConnection();

        Base64Encoder b = new Base64Encoder("xh_ns:xh_ns");
        con.setRequestProperty("Proxy-Authorization", "Basic " + b.processString());
        con.setDoInput(true);
        con.setDoOutput(false);
        BufferedReader din = new BufferedReader(new InputStreamReader(con.getInputStream()));
        while ((dbg = din.readLine()) != null) {
          page += dbg + "\n";
        }
        din.close();
        System.getProperties().put("proxySet", "false");
      }
    } catch (Exception any) {
      throw new DLI2FEException(DLI2FEException.SERVICE_UNAVAILABLE_EXC, "Failed to contact server");
    }
    return page;
  }

  void parsePage(String page) {
    int i, j, k, did;
    String detailURL, unitName;
    i = page.indexOf(JS_DETAIL);
    while (i != -1) {
      ResultDoc doc = new ResultDoc();
      String dbg;
      j = page.indexOf(")", i);
      doc.url = detailURL = makeDetailURL(page.substring(i, j + 1));
      j = page.indexOf(">", j);
      j = page.indexOf(">", j + 1);
      k = page.indexOf("<", j);
      doc.name = unitName = page.substring(j + 1, k);
      result.addElement(doc);
      i = page.indexOf(JS_DETAIL, i + 5);
    }
  }

  String makeRequestWithPage(int page) {
    return queryPATH + paramNAME + "=" + name + "&" + paramADDRESS + "=" + addr + "&" + paramPHONE
      + "=" + phone + "&" + paramAREA + "=" + getAreaText() + "&" + paramPAGE + "=" + String.valueOf(page);
  }

  // detailJSString example: javascript:Detail ( 'B310000', '0795', '0000219998', '' )
  // corresponding detailURL: http://www.yellowpage.com.cn/Prog/Trade/Detail.asp?r=B3100000&c=0795&u=00002199998&a=
  String makeDetailURL(String detailJSString) {
    String request = detailURL;
    detailJSString = detailJSString.substring(detailJSString.indexOf("(") + 1, detailJSString.indexOf(")"));
    StringTokenizer stk = new StringTokenizer(detailJSString, ",");
    request += "r=" + trimQuotation(stk.nextToken());
    request += "&c=" + trimQuotation(stk.nextToken());
    request += "&u=" + trimQuotation(stk.nextToken());
    request += "&a=" + trimQuotation(stk.nextToken());
    return request;
  }

  String trimQuotation(String quotedStr) {
    int first = quotedStr.indexOf("'");
    int last = quotedStr.lastIndexOf("'");
    if (last == first + 1)
      return "";
    else
      return quotedStr.substring(first + 1, last);
  }

  String getAreaText() {
    if (city == null)
      return "";
    if (city.equals("北京"))
      return "B110000BJ-北京";
    else if (city.equals("武汉"))
      return "B420100WH-武汉";
    return "B310000SH-上海";
  }


  public static void main(String[] args) {
    int i;
    Yellowpage yp = new Yellowpage(null, "天山", null, "上海");
    try {
      yp.searchYellowpage();
      for (i = 0; i < yp.getNumDocs(); i++) {
        System.out.println("NAME: " + yp.getResult(i).name + "\tURL: " + yp.getResult(i).url);
      }
    } catch (DLI2FEException dli2fe) {
      System.err.println("DLI2FE exception (" + dli2fe.getCode() + "): " + dli2fe.getReason());
    }
  }
}