// Copyright (c) 2001 Hursh Jain (http://www.mollypages.org) 
// The Molly framework is freely distributable under the terms of an
// MIT-style license. For details, see the molly pages web site at:
// http://www.mollypages.org/. Use, modify, have fun !

package fc.util;

import java.util.*;
import java.util.regex.*;
import java.security.*;

/* 
UUID and token utilities

@author hursh jain
*/
public final class UUIDUtil
{
static MessageDigest digester = null;

static {
  try {
    digester = MessageDigest.getInstance("SHA-1");
    }
  catch (Exception e) {
    e.printStackTrace();
    }
  }

/**
Returns a ID suitable for a session/cookie identifier.
<pre>
See: cookies.lcs.mit.edu
See: www.across.si

There are 2 issues with generating sessionid's. 

1) uniqueness - 2 or more sessionid's should not end up being 
   the same.
2) hard-to-guess - For example, sequential values like 
   1, 2, 3 are unique but easy to guess and therefore easy
   to session hijack.

Our sessionid's have 2 parts:
a) a timestamp for guaranteed uniqueness  (easy to guess)
b) random data (hard to guess)
</pre>
*/
public static String newSessionID() 
  {
  //although this function has been written for some time,
  //i recently looked this over with webscarab (around aug 2007)
  //and it visually appears fairly scattered/random. (9999 sid's)
  //dunno, maybe hotbits retrieved/cached every minute via a background
  //thread is good if one wants to be hardcore but it's pretty good
  //as-is 
  
  //there is also no visual difference between nanos & millis. intuitively
  //nanos should have more differences in the lower bits but after
  //all the bits get churned/digested, there is no diff that i can tell,
  //at least visually via webscarab. so i've left this as millis.
  
  final long time = System.currentTimeMillis();   

  Random random = new java.util.Random();
  final int rand = random.nextInt(); 
  final String cookie = String.valueOf(time) + rand;

  byte[] digestbuf = null;
  
  //digests always returns 16 bytes         
  synchronized (UUIDUtil.class) {   
    digestbuf = digester.digest(cookie.getBytes());
      }
    
  /*
  we simply convert bytes toHex. note, we don't use Base64
  because '/' and '+' are valid Base64 chars and they can
  mess up url's. 
  */
    
    char[] encoded = null;
    encoded = encode(digestbuf);
    return new String(encoded);
    
    //note, we don't ever need to use the decode() method
    //but it's there for testing foo->encode->decode->foo
    }


//well, i was very...young...when I wrote this... :-]
//
//all right, as of 2014, i took out the original string, a bit too risque
//new lookup string is less embarassing
//                1..............16 [or 0-15, 16 things]  
//                0123456789ABCDEF
static final char[] lookup="wInteEMubY2169x0".toCharArray();

/*
base16 encoding
     0 -> some ascii code
       1 -> some ascii code
       ...
       15-> some ascii code
       
when ascii codes are for 0...F it's called "hex" encoding. 
*/

//we need a reverse lookup table since are ain't
//using straight hexadecimal to encode
//encode: half byte [0-16] -> char value from lookup string
//decode: char value from lookup string (0-128) -> half byte
static final int[] reverselookup = new int[128];
  
static  
  {
  for (int n = 0; n < 16; n++) {
    int x =  lookup[n] & 0xFF; //x ='c'('c'&0xff->99), 'n'(110) etc
    reverselookup[x] = n;  //this is why reverselookup[128] above, not 16
    }
  }
  
private static final char[] encode(byte[] buf) 
  {
  final int len = buf.length;

  char[] tmp = new char[len * 2];  
  
  for (int n = 0, i = 0; n < len ; n++, i+=2)
    {
    int high4 =  (buf[n] >> 4) & 0x0F;
    int low4  =  buf[n] & 0x0F;
    
    tmp[i]   = lookup [high4];
    tmp[i+1] = lookup [low4];
    }
  return tmp;
  }

private static final byte[] decode(char[] buf) 
  {
  final int len = buf.length;

  byte[] tmp = new byte[(len/2) + (len%2)];

  for (int n = 0, tmp_n = 0; n < len; n+=2, tmp_n++) 
    {
    //0xFF makes it an int. we could cast to (int)
    //too but then that would sign extend the char
    //in the general case [but since all are chars
    //are < 127], it would be ok to do so in this
    //case. Used 0xFF because that's the right thing
    //generally speaking

    int c1 = buf[n] & 0xFF;   //'w', 'I', 'n', 't' etc.. 
    int c2 = buf[n+1] & 0xFF;

    if ( c1 > 127 || c2 > 127)
      throw new IllegalArgumentException("malformed bufing, encoded with characters I don't understand [" + c1 + "] and [" + c2 + "]");
        
    int i = (reverselookup[c1] << 4  | reverselookup [c2]); 
    //System.out.println( (reverselookup[c1] << 4) + ", " + reverselookup[c2] + ",->" + i);

    tmp[tmp_n] = (byte) i;
    }
  return tmp;
  }
  

static char[] URL_safe_chars = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ123456789_-.~".toCharArray();

/**
Returns a N-char string suitable for tracking, affiliate, short codes, etc. Uses
characters safe for all URL's.

@param  len the length of the returned string
*/
public static String newID(int len)
  {
  final Random r = new Random();
  final char[] buf = new char[len];
  for (int n = 0; n < len; n++) {
    buf[n] = URL_safe_chars[r.nextInt(URL_safe_chars.length)];
    }
  
  return new String(buf);
  }

/**
Returns a 8 character string suitable for tracking, affiliate, short codes, etc. Uses
characters safe for all URL's.
*/
public static String newID()
  {
  return newID(8);
  }
    
public static void main (String args[])
  {
  Args myargs = new Args(args);
  int LOOP = myargs.getInt("loop", 100000);
  myargs.setUsage("java fc.util.UUIDUtil [-loop <int, default 100,000>]");
  
  if (myargs.flagExists("help")) {
    myargs.showError();
    return;
    }
  System.out.println("Session ID: " + newSessionID());
  System.out.println("Generating " + LOOP + " test session ids...");

  fc.util.Watch w = new fc.util.Watch();
  Map map = null;
  int count = -1;
  
  w.start();  
  for (int n = 0; n < LOOP; n++) {
    String s = newSessionID();
    }
  System.out.println("Time for " + LOOP + ": " + w.time() + " ms");
  
  System.out.println("Now testing for collisions...");
  w.restart();

  map = new HashMap();
  count = 0;
  for (int n = 0; n < LOOP; n++) 
    {
    String s = newSessionID();
    if (map.containsKey(s)) {
      System.out.println("collision found with key: " + s);
      count++;
      }
    if (n > 0 && n % 20000 == 0) {
      w.stop();
      System.out.println("generated: " + n + " (+ " + w.time() + " ms)");
      w.restart();
      }
    map.put(s, null);
    }
  if (count == 0) {
    System.out.println("No collisions found...");
    }
  else{
    System.out.println(count + " collisions found...");
    }

  System.out.println("------------------------------");
  w.restart();  
  System.out.println("newID (8): " + newID(8));
  System.out.println("Generating " + LOOP + " test session ids...");

  for (int n = 0; n < LOOP; n++) {
    String s = newID(8);
    }
  System.out.println("Time for " + LOOP + ": " + w.time() + " ms");

  System.out.println("Now testing for collisions...");
  w.restart();

  map = new HashMap();
  count = 0;
  for (int n = 0; n < LOOP; n++) 
    {
    String s = newID(8);
    if (map.containsKey(s)) {
      System.out.println("collision found with key: " + s);
      count++;
      }
    if (n > 0 && n % 20000 == 0) {
      w.stop();
      System.out.println("generated: " + n + " (+ " + w.time() + " ms)");
      w.restart();
      }
    map.put(s, null);
    }
  if (count == 0) {
    System.out.println("No collisions found...");
    }
  else{
    System.out.println(count + " collisions found...");
    }
  }
  

}