Форум программистов [Powered by Invision Power Board]

Код


import java.io.*;
import java.net.*;
public class ReadTag {
    /** The URL that this ReadTag object is reading */
    protected URL myURL = null;
    /** The Reader for this object */
    protected BufferedReader inrdr = null;
  
    /* Simple main showing one way of using the ReadTag class. */
    public static void main(String[] args) throws MalformedURLException, IOException {
        if (args.length == 0) {
            System.err.println("Usage: ReadTag URL [...]");
            return;
        }

        for (int i=0; i<args.length; i++) {
            ReadTag rt = new ReadTag(args[0]);
            String tag;
            while ((tag = rt.nextTag()) != null) {
                System.out.println(tag);
            }
            rt.close();
        }
    }
  
    /** Construct a ReadTag given a URL String */
    public ReadTag(String theURLString) throws 
            IOException, MalformedURLException {

        this(new URL(theURLString));
    }

    /** Construct a ReadTag given a URL */
    public ReadTag(URL theURL) throws IOException {
        myURL = theURL;
        // Open the URL for reading
        inrdr = new BufferedReader(new InputStreamReader(myURL.openStream()));
    }

    /** Read the next tag.  */
    public String nextTag() throws IOException {
        int i;
        while ((i = inrdr.read()) != -1) {
            char thisChar = (char)i;
            if (thisChar == '<') {
                String tag = readTag();
                return tag;
            }
        }
        return null;
    }

    public void close() throws IOException {
        inrdr.close();
    }

    /** Read one tag. Adapted from code by Elliotte Rusty Harold */
    protected String readTag() throws IOException {
        StringBuffer theTag = new StringBuffer("<");
        int i = '<';
      
        while (i != '>' && (i = inrdr.read()) != -1) {
                theTag.append((char)i);
        }     
        return theTag.toString();
    }

    /* Return a String representation of this object */
    public String toString() {
        return "ReadTag[" + myURL.toString() + "]";
    }
}

Код


package geturls;
import java.io.*;
import java.net.*;
import java.util.*;

public class GetURLs {
    /** The tag reader */
    ReadTag reader;

    public GetURLs(URL theURL) throws IOException {
        reader = new ReadTag(theURL);
    }

    public GetURLs(String theURL) throws MalformedURLException, IOException {
        reader = new ReadTag(theURL);
    }

    /* The tags we want to look at */
    public final static String[] wantTags = {
        "<a ", "<A "
        
    };

    public ArrayList getURLs() throws IOException {
        ArrayList al = new ArrayList();
        String tag;
        while ((tag = reader.nextTag()) != null) {
            for (int i=0; i<wantTags.length; i++) {
                if (tag.startsWith(wantTags[i])) {
                    al.add(tag);
                    continue;        // optimization
                }
            }
        }
        return al;
    }

    public void close() throws IOException {
        if (reader != null) 
            reader.close();
    }
    public static void main(String[] argv) throws 
            MalformedURLException, IOException {
        String theURL = argv.length == 0 ?
            "http://www.google.com/search?hl=en&q=%22google.com%22+-inurl%3Agoogle.com&btnG=Search" : argv[0];
        GetURLs gu = new GetURLs(theURL);
        ArrayList urls = gu.getURLs();
        Iterator urlIterator = urls.iterator();
        while (urlIterator.hasNext()) {
            System.out.println(urlIterator.next());
        }
    }
}