View source for User:Mwn3d/Seasoning Sandwich Caesar Cipher

I was reading about the [[Caesar cipher]] and it was mentioned that an easy way for a person to decode the message would be to find the most common character and consider it to be "E" (or the most common character in whatever language). Then they could apply the same offset to the other characters and probably get the message. I wanted to make a version that would make that strategy impossible (don't worry, the Caesar cipher is still really easy to beat no matter what you do). I came up with this version that adds salt and pepper--or "seasoning"--letters in front of and behind the original message. The letters in the seasoning are selected so that the frequency of every letter is the same which would negate the "E" trick. The seasoning also has spaces so that it looks sort of natural. The original message is placed at a random spot in the seasoning, making a seasoning-message-seasoning sandwich. Then the whole sandwich is encoded. When the message is decoded on the other end the recipient just has to scan until they run into readable text to find the message. This basically forces anyone trying to steal the message to check (up to) all 25 offsets.
<lang java5>import static java.util.Collections.nCopies;

import java.util.Random;
import java.security.SecureRandom;

public class Cipher{
	public static void main(String[] args){
		String original = "The quick brown fox Jumped over the lazy Dog and then he landed on a cat and it hurt";
		String enc = encode(original, 12);
		System.out.println(original);
		System.out.println(enc);
		System.out.println(decode(enc, 12));
	}
 
	public static String decode(String enc, int offset){
		return encode(enc, -offset, false);
	}
	
	public static String encode(String enc, int offset){
		return encode(enc, offset, true);
	}
 
	private static String encode(String enc, int offset, boolean addNoise){
		offset = offset % 26 + 26; //make the offset a valid number
		StringBuilder encoded = new StringBuilder();
		if(addNoise){ //if we are encoding
			StringBuilder noise = genNoise(enc); //get noise based on the original string ignoring non-letter characters
			
			noise.insert((int)(Math.random() * noise.length()), " " + enc + " "); //insert the message into a random place in the noise
			enc = noise.toString().replaceAll("[ ]{2,}"," ").trim(); //get rid of double spaces and spaces at the start and end
		}
		
		//do the regular encoding using the original string with noise around it
		for(char i : enc.toLowerCase().toCharArray()){
			if(Character.isLetter(i)){
				int j = (i - 'a' + offset) % 26;
				encoded.append((char)(j + 'a'));
			}else{
				encoded.append(i);
			}
		}
		return encoded.toString();
	}
	
	/*
	 * Generate noise to surround the actual message
	 */
	private static StringBuilder genNoise(String orig){
		//remove all non-word chars for figuring out which letters to include in noise
		String enc = orig.replaceAll("\\W", "");
		
		//get the number of each letter needed to bring them all up to the same frequency
		int[] invertedFreq = invert(letterFreq(enc));
		
		StringBuilder rawNoise = new StringBuilder();
		
		//start with all of those letters in organized groups
		for(int i = 0; i < invertedFreq.length; i++){
			//string multiplication would be nice here
			rawNoise.append(nCopies(invertedFreq[i], (char)('a' + i)).toString().replaceAll("\\W", ""));
		}
		
		StringBuilder noise = new StringBuilder();
		Random rand = new SecureRandom();
		double meanLen = getMeanLen(orig.split("\\b++"));
		double std = getStdDev(orig.split("\\b++"));
		
		int waitForSpace = (int)(rand.nextGaussian() * std + meanLen) + 1; //wait this many characters before adding a space
		
		while(rawNoise.length() > 0){ //while we still have letters left to add
			int randIdx = rand.nextInt(rawNoise.length()); //pick a random letter
			char ch = rawNoise.charAt(randIdx);
			if(noise.length() >= 3 && noise.substring(noise.length() - 3).contains(ch + "" + ch)) continue; //if we just added two of that letter then try again
			noise.append(ch); //add that letter
			rawNoise.deleteCharAt(randIdx); //remove it from our list of letters to add
			waitForSpace--; //we added another character so we decrease the wait time for spaces
			if(waitForSpace == 0){ //if we hit the wait time
				noise.append(' '); //add a space
				do{
					waitForSpace = (int)(rand.nextGaussian() * std + meanLen) + 1; //select a new wait time
				}while(waitForSpace == 0);
			}
		}
		
		return noise;
	}
	
	/*
	 * Count up the frequency of letters in a given string (case-insensitive, ignores non-letter chars)
	 */
	private static int[] letterFreq(String msg){
		int[] freq = new int[26];
		
		for(char ch : msg.toLowerCase().toCharArray()){
			freq[ch - 'a']++;
		}
		
		return freq;
	}
	
	/*
	 * Calculate the number of each character needed to bring each one up to the same frequency
	 */
	private static int[] invert(int[] arr){
		int max = 0;
		for(int i:arr){
			if(i > max) max = i;
		}
		
		int[] freqLet = new int[arr.length];
		for(int i = 0; i < arr.length; i++){
			freqLet[i] = max - arr[i];
		}
		
		return freqLet;
	}
	
	private static double getMeanLen(String[] words){
		int total = 0;
		for(String word:words){
			total += word.length();
		}
		return (double)total / words.length;
	}
	
	private static double getStdDev(String[] words){
		double mean = getMeanLen(words);
		int totalDiff = 0;
		for(String word:words){
			totalDiff += (int)Math.pow(word.length() - mean, 2);
		}
		return Math.sqrt(totalDiff / words.length);
	}
}</lang>
{{out}}
<pre>The quick brown fox Jumped over the lazy Dog and then he landed on a cat and it hurt
trnifd lc d dh o ftq cguow ndaiz raj vgybqp ahqd ftq xmlk pas mzp ftqz tq xmzpqp az m omf mzp uf tgdf ove edh u lc vox wjb hl uc giv lrxr uxwgh beyw ab scj a knhst ngso l byc yjiy ejkb iwp sns ins ie mvclru eknv yowr ju ywbkgre kkj v xhz x
hfbwtr zq r rv c the quick brown fox jumped over the lazy dog and then he landed on a cat and it hurt cjs srv i zq jcl kxp vz iq uwj zflf ilkuv psmk op gqx o ybvgh bugc z pmq mxwm sxyp wkd gbg wbg ws ajqzfi sybj mckf xi mkpyufs yyx j lvn l</pre>