/**
 * Some utility methods to help work with URLs.<p>
 * 
 * @author Xiannong Meng
 * @date April 2005, revised October 2006 for CSCI 335.
 */
public class UrlUtil {

    // Data members.
    private String thisHost;
    private int thisPort;
    private String thisDir;
    private String thisProt;
    private String prefix;
    private String thisFile;


    /**
     * Default constructor.<p>
     */
    public UrlUtil()    {

	thisHost = thisDir = null;
	thisPort = 80;
	thisProt = "http";
	thisDir  = "/";
	thisFile = "index-test.html";
    }

    /**
     * Alternate constructor.<p>
     */
    public UrlUtil(String host,
		   int    port,
		   String currentDir,
		   String prot,
		   String fileName)    {

	thisHost = host;
	thisPort = port;
	thisDir  = currentDir;
	thisProt = prot;
	thisFile = fileName;
    }


    // access methods
    public String getProtocol() {

	return thisProt;
    }

    public String getHost() {

	return thisHost;
    }

    public String getDir() {

	return thisDir;
    }

    public int getPort()  {

	return thisPort;
    }

    public String getFileName() {

	return thisFile;
    }

    // check to see if this is a valid url format
    // we don't care about the specific protocols, hostnames, port numbers and
    // directories, as long as the format is of
    // proto://host:port/dir/file   or
    // mailto:someone@somewhere     or
    // somedir/somefile             (without protocol, port, host)
    static public boolean isValidUrl(String inUrl)    {

	if (inUrl.length() < 10)
	    return false;

	// first to see if it is a mailto
	if ("mailto".compareToIgnoreCase(inUrl.substring(0,6)) == 0
	    && inUrl.indexOf("@") > 6)
	    return true;

	// then check the rest
	// if not relative, has to have protocol and host
	int colonLoc = inUrl.indexOf(":");
	int slashLoc  = inUrl.lastIndexOf("//");
	if (colonLoc > 0 && colonLoc < slashLoc
	    && slashLoc < inUrl.length() - 2)
	    return true;

	if (colonLoc < 0 && slashLoc < 0) // relative url
	    return true;

	return false;
    }

    /**
     * Some URLs come with extra slashes e.g. <br>
     * http://www.host.com//dir/file<br>
     * remove them.<p>
     */
    static public String removeExtraSlash(String inStr)  {

	int slashLoc = inStr.indexOf("://");
	int l = inStr.length();
	int hold = slashLoc + 3;
	slashLoc = inStr.indexOf("//", hold);
	while (slashLoc > 0 && slashLoc < l) {
	    hold = slashLoc;
	    while (hold < l && inStr.charAt(hold) == '/')
		hold ++;
	    if (hold >= l) {
		inStr = inStr.substring(0, slashLoc+1);
		break;
	    }
	    else
		inStr = inStr.substring(0, slashLoc+1) 
		    + inStr.substring(hold);
	    slashLoc = inStr.indexOf("//", hold);
	    l = inStr.length();
	}

	return inStr;
    }

    // given an incomplete url (partial, relative ...), construct
    // a complete one from the given url information in urlInfo
    // urlInfor has: host, port, and dir
    static public String fixUrl(String inUrl, UrlUtil thisUrlInfor)  {

	String path = "";
	String thisProto = extractProtocol(inUrl);
	if (thisProto != null) 	{
	    // this is not a relative url
	    if (isValidProtocol(thisProto) == true)  {
		// revise thisUrlInfor
		thisUrlInfor = UrlUtil.extractUrlInfor(inUrl);
		path = thisUrlInfor.getDir() + thisUrlInfor.getFileName();
	    }
	    else
		return null;
	}
	else
	    path = inUrl;

	String completedUrl = thisUrlInfor.getProtocol() + "://";
	completedUrl += thisUrlInfor.getHost();
	if (thisUrlInfor.getPort() != 80)
	    completedUrl += ":" + thisUrlInfor.getPort();
	String newDir = convertRelativePath(path, thisUrlInfor);
	completedUrl += newDir;

	// remove the section tag '#'
	int sectionLoc = completedUrl.lastIndexOf("#");
	if (sectionLoc > 0)
	    completedUrl = completedUrl.substring(0, sectionLoc);
	
	// attach 'index.html' if the filename ends with '/'
	if (completedUrl.endsWith("/"))
	    completedUrl += "index-test.html";

	// remove the annoying '//' in the middle of a url
	// I am sure there are better ways of doing it, but ...
	completedUrl = removeExtraSlash(completedUrl);

	return completedUrl;
    }

    // precondition: the rPath IS a relative path that doesn't have a protocol
    // and a host!
    static public String convertRelativePath(String rPath, 
					     UrlUtil thisUrlInfor)  {

	if (rPath.startsWith("/") == false) { // a relative path
	    String tempPath = thisUrlInfor.getDir() + rPath;
	    rPath = tempPath;
	}
	String completePath = rPath;

	// keep track of whether we change anything
	boolean changed = false;     

	// collapse all occurances of "/./"

	int i = rPath.indexOf("/./");
	while (i >= 0) 	{
	    changed = true;
	    rPath = rPath.substring(0, i) + rPath.substring(i+2);
	    i = rPath.indexOf("/./");
	}

	// collapse all occurances of "/../" (by removing preceding directory)

	i = rPath.indexOf("/../");
	while (i >= 0) 	{
	    changed = true;
	    int i2 = rPath.lastIndexOf('/', i-1);
	    if (i2 < 0) i2 = i;
	    rPath = rPath.substring(0, i2) + rPath.substring(i+3);
	    i = rPath.indexOf("/../");
	}

	if (changed == true)
	    completePath = rPath;

	return completePath;
    }

    /**
     * Extract the protocol part of a URL.<p>
     */
    static public String extractProtocol(String inUrl) {

	String proto = null;
	int loc = inUrl.indexOf(":");
	if (loc > 0)  // the url includes a protocol
	    proto = inUrl.substring(0, loc);

	return proto;
    }

    /**
     * Check to see if this is a valid protocol. This one only
     * responds to 'http'. It can easily add others.<p>
     */
    static public boolean isValidProtocol(String inProto) {

	if (inProto.compareToIgnoreCase("http") == 0)
	    return true;
	else
	    return false;
    }

    // the 'inUrl' is expected to have the form of
    // http://host.domain:port/dir. This method will extract
    // each of the components out of this string.<p>
    static public UrlUtil extractUrlInfor(String inUrl)  {

	int protLoc = inUrl.indexOf("://");
	int hostLoc = -1;
	int portLoc = -1;
	int dirLoc = -1;

	String localHost = "";
	String localDir  = "/";
	int    localPort = 80;
	String localProt = "http";
	String localFile = "/";

	if (protLoc >= 0) {
	    hostLoc = protLoc + 3;
	    localProt = inUrl.substring(0, protLoc);
	}
	else
	    hostLoc = 0;

	portLoc = inUrl.indexOf(':', hostLoc);
	
	dirLoc = inUrl.indexOf('/', hostLoc);

	//	System.err.println("portloc " + portLoc + " dirLoc " + dirLoc);
	if (portLoc >= 0 && dirLoc > portLoc && dirLoc <= portLoc + 6) 	{
	    //	    System.err.println("branch A");
	    localHost = inUrl.substring(hostLoc, portLoc);
	    localPort = Integer.parseInt
		((inUrl.substring(portLoc+1, dirLoc)).trim());
	}
	else {
	    //	    System.err.println("branch B");
	    if (dirLoc >= 0)
		localHost = inUrl.substring(hostLoc, dirLoc);
	    else
		localHost = inUrl.substring(hostLoc);
	}

	if (dirLoc >= 0) { // need to remove the trailing file name, /in.html
	    if (inUrl.endsWith("/") == true) {
		localDir = inUrl.substring(dirLoc);
		localFile = "/";
	    }
	    else {
		localDir = inUrl.substring(dirLoc, inUrl.lastIndexOf("/")+1);
		localFile = inUrl.substring(inUrl.lastIndexOf("/")+1);
	    }
	}
	
	return new UrlUtil(localHost, localPort, 
			   localDir, localProt, localFile);
    }

    /**
     * Print the URL in a decent format.<p>
     */
    public void print()  {

	System.out.println("prot: " + thisProt);
	System.out.println("host: " + thisHost);
	System.out.println("port: " + thisPort);
	System.out.println("dir : " + thisDir);
    }

    /**
     * Some simple sanity check.<p>
     */
    static public void main(String[] argvs) {

	String sampleUrl = "http://www.eg.bucknell.edu:1234/home/dir/index.html";
	
	UrlUtil infor = UrlUtil.extractUrlInfor(sampleUrl);
	infor.print();

	// examples of input urls
	// http://www.cnn.com/level1/level2/../index.html
	// http://www.cnn.com/level1/level2/../../index.html
	// http://www.cnn.com/../../index.html
	if (argvs.length != 1) {
	    System.err.println("usage: java UrlUtil url");
	    System.exit(1);
	}
	//	/**
	sampleUrl = "http://polaris.bucknell.edu/hello";
	infor = UrlUtil.extractUrlInfor(sampleUrl);
	infor.print();

	sampleUrl = "ftp://polaris.bucknell.edu";
	infor = UrlUtil.extractUrlInfor(sampleUrl);
	infor.print();

	sampleUrl = "http://www.eg.bucknell.edu:1234/home/dir/index.html";
	infor = UrlUtil.extractUrlInfor(sampleUrl);
	infor.print();

	String samplePath = "./level1/../level2/path3/../path4/./index.html";
	System.out.println("new url:" 
			   + convertRelativePath(samplePath, infor));
	samplePath = "level1/level2/../index.html";
	System.out.println("new url:" 
			   + convertRelativePath(samplePath, infor));
	samplePath = "/level1/../level2/index.html";
	System.out.println("new url:" 
			   + convertRelativePath(samplePath, infor));
	//	**/

	System.out.println("orig  url: " + argvs[0]);
	System.out.println("fixed url: " + fixUrl(argvs[0], infor));
    }
}