Imagine taking a book (say, Tom Sawyer) and determining the probability with which each character occurs. You'd probably find that spaces are the most common, that the character 'e' is fairly common, and that the character 'q' is rather uncommon. After completing this level 0 analysis, you'd be able to produce random Tom Sawyer text based on character probabilities. It wouldn't have much in common with the real thing, but at least the characters would tend to occur in the proper proportion. In fact, here's an example of what you might produce:
Level 0
rla bsht eS ststofo hhfosdsdewno oe wee h .mr ae irii ela iad o r te u t mnyto onmalysnce, ifu en c fDwn oee iteo
Now imagine doing a slightly more sophisticated level 1 analysis by determining the probability with which each character follows every other character. You would probably discover that 'h' follows 't' more frequently than 'x' does, and you would probably discover that a space follows '.' more frequently that ',' does. You could now produce some randomly generated Tom Sawyer by picking a character to begin with and then always choosing the next character based on the previous one and the probabilities revealed by the analysis. Here's an example:
Level 1
"Shand tucthiney m?" le ollds mind Theybooure He, he s whit Pereg lenigabo Jodind alllld ashanthe ainofevids tre lin--p asto oun theanthadomoere
Now imagine doing a level k analysis by determining the probability with which each character follows every possible sequence of characters of length k. A level 5 analysis of Tom Sawyer for example, would reveal that 'r' follows ''Sawye'' more frequently than any other character. After a level k analysis, you'd be able to produce random Tom Sawyer by always choosing the next character based on the previous k characters (the seed) and the probabilities revealed by the analysis.
At only a moderate level of analysis (say, levels 5--7), the randomly generated text begins to take on many of the characteristics of the source text. It probably won't make complete sense, but you'll be able to tell that it was derived from Tom Sawyer as opposed to, say, The Sound and the Fury. Here are some more examples:
Level 2
"Yess been." for gothin, Tome oso; ing, in to weliss of an'te cle -- armit. Papper a comeasione, and smomenty, fropeck hinticer, sid, a was Tom, be suck tied. He sis tred a youck to themen
Level 4
en themself, Mr. Welshman, but him awoke, the balmy shore. I'll give him that he couple overy because in the slated snufflindeed structure's kind was rath. She said that the wound the door a fever eyes that WITH him.
Level 6
people had eaten, leaving. Come -- didn't stand it better judgment; His hands and bury it again, tramped herself! She'd never would be. He found her spite of anything the one was a prime feature sunset, and hit upon that of the forever.
Level 8
look-a-here -- I told you before, Joe. I've heard a pin drop. The stillness was complete, how- ever, this is awful crime, beyond the village was sufficient. He would be a good enough to get that night, Tom and Becky.
Level 10
you understanding that they don't come around in the cave should get the word "beauteous" was over-fondled, and that together" and decided that he might as we used to do -- it's nobby fun. I'll learn you."
Assignment
You are to implement a Java public class RandomWriter that provides a random writing application. Your class should have a public main method that takes the following four command line arguments:
* A non-negative integer k
* A non-negative integer length.
* The name of an input file source that contains more than k characters.
* The name of an output file result.
Your program should validate the command line arguments by making sure that k and length are non-negative, that source contains more than k characters and can be opened for reading, and that result can be opened for writing. If any of the command line arguments are invalid, your program should write an informative error message to System.err and terminate.
Otherwise, your program should pick k consecutive characters at random from source and use them as the initial seed in the random writing process outlined above. Your program should then write length characters to result. Each of these additional characters should be chosen based on the current seed. (Each time a character c is written to result, the seed is updated by removing its first character and appending c to the end.)
For example, suppose that k = 2 and the source file contains
the three pirates charted that course the other day
Here is how the first three characters might be chosen:
* A two-character seed is chosen at random to become the initial seed. Let's suppose that "th" is chosen.
* The first character must be chosen based on the probability that it follows the seed (currently "th") in the source. The source contains five occurrences of "th". Three times it is followed by 'e', once it is followed by 'r', and once it is followed by 'a'. Thus, the next character must be chosen so that there is a 3/5 chance that an 'e' will be chosen, a 1/5 chance that an 'r' will be chosen, and a 1/5 chance that an 'a' will be chosen. Let's suppose that we choose an 'e' this time.
* The next character must be chosen based on the probability that it follows the seed (currently "he") in the source. The source contains three occurrences of "he". Twice it is followed by a space and once it is followed by 'r'. Thus, the next character must be chosen so that there is a 2/3 chance that a space will be chosen and a 1/3 chance that an 'r' will be chosen. Let's suppose that we choose an 'r' this time.
* The next character must be chosen based on the probability that it follows the seed (currently "er") in the source. The source contains only one occurrence of "er", and it is followed by a space. Thus, the next character must be a space.
If your program ever gets into a situation in which there are no characters to choose from (which can happen if the only occurrence of the current seed is at the exact end of the source), your program should pick a new random seed and continue.
Approach
There is a simple way to attack this problem. Create a String object that contains all of the characters from the source file. To choose the next character, find each occurrence of the seed in the source and store the character that follows it into an ArrayList. When you have found all occurrences, choose a character at random from the ArrayList.
Here are some useful things to know about Java. You'll need to read the Java documentation for more details.
* A java.util.ArrayList can contain only objects; it cannot contain a scalar values such as a char. To store a char into an ArrayList, you need to wrap it in a java.lang.Character object.
* You can use a java.util.Random object to generate random integers in a specified range.
* A java.io.FileWriter object is useful for writing strings and characters to a file.
* A java.io.FileReader object is useful for reading one character at a time from a file.
* A java.lang.StringBuffer object is useful for efficiently composing characters into a string.
* A java.lang.String object contains several member functions for searching for substrings.
Project Gutenberg maintains a huge library of public domain books that you can use as source texts. If your program generates something that is particularly amusing, please send it to me so I can share it with the class. Be sure to identify the source text and the level of the analysis.