weatherextractor.java~14~

本程序用JAVA编制

import java.util.*;
import java.io.*;
import java.sql.Timestamp;
import org.w3c.dom.*;

/**
 * The WeatherExtractor class extracts weather information from
 * a web page and saves it in XML format. If a weather XML file
 * all ready exists, the data is merged with that file, otherwise
 * a new XML data file is created
 */

public class WeatherExtractor {

  // The location of the various resources and files
  private static final String source =
      "http://weather.yahoo.com/forecast/UKXX0085_c.html?force_units=1";
  private static final String xslFile = "new-xml/london.xsl";
  private static final String xmlFile = "new-xml/london-result.xml";
  private static Calendar date = Calendar.getInstance();

  public static void main(String args[]) {
    try {
      // Retrieve the source document as XML
      Document xhtml = XMLHelper.tidyHTML(source);
      // Parse the XSL document
      Document xsl   = XMLHelper.parseXMLFromFile(xslFile);
      // Get the current data by applying the XSL transform
      Document data  = XMLHelper.transformXML(xhtml, xsl);
      // For convenience, add a timestamp to what we have done

      //Timestamp time = new Timestamp(now.getTime());

      Element root = data.getDocumentElement();
      NodeList nl = root.getElementsByTagName("WEATHER");
      Element weather = (Element)nl.item(0);
      int year=2003;
      int month=date.getTime().getMonth()+1;
      int day=date.getTime().getDate();
      int hours=date.getTime().getHours()-7;

      //System.out.println("hours= "+hours);
      //System.out.println("date="+date);
      if (hours<0){
        day=day-1;
        //hours=date.getTime().getHours()+24-7;
      }
      String calendar=year+"-"+month+"-"+day;
      weather.setAttribute("Retreived",calendar);
      String value="";
      System.out.print("length="+weather.getChildNodes().getLength());
      for(int i=0;i<weather.getChildNodes().getLength();i++){
        value=weather.getChildNodes().item(i).getChildNodes().item(0).getNodeValue();
        weather.getChildNodes().item(i).getChildNodes().item(0).setNodeValue(value.trim());
        //System.out.println("value="+value);
      }
      String time=weather.getChildNodes().item(0).getChildNodes().item(0).getNodeValue();
      if(time.endsWith("pm")){
        System.out.print("this is pm");
      }else{
        System.out.println("this is am ,time="+time.substring(0,2));
        //char hour=time.substring(0,1);
        //time.replace(time.substring(0,1).toCharArray(),(time.substring(0,1)+12).toCharArray());

      }
      //System.out.println("date="+(day.getTime().getMonth()+1)+"-"+day.getTime().getDate());

      // Write the output file, merging if necessary
      File dataFile  = new File(xmlFile);
      if (dataFile.exists()) {
        // If we have extracted before, merge the data and write the file
        Document oldData = XMLHelper.parseXMLFromFile(dataFile);
        XMLHelper.mergeXML(oldData.getDocumentElement(), weather, false);
        XMLHelper.outputXMLToFile(oldData, xmlFile);
      } else {
        // If this is our first extraction, just write the file
        XMLHelper.outputXMLToFile(data, xmlFile);
      }
    } catch (XMLHelperException xmlhe) {
      System.err.println(
          "There was an error in the extraction process:\n" +
          xmlhe.getMessage());
    }
  }
}