This assignment concerns the game of Scrabble. You may know the game of Scrabble better as Words with Friends. If you want to try out Words with Friends yourself you can download the free app for your smartphone. However, the programming assignment is not to create the game itself, but to write a console-based program that finds all possible words that can be made from a rack of Scrabble tiles (so it could help someone playing Scrabble). We'll elaborate on the exact requirements of this assignment in the section on the assignment below.
The starter files we are providing for you on Vocareum are listed here. The files in bold below are ones you create and/or modify and submit. The ones not in bold are ones that you will use, but not modify. More details about the java classes below are in the section on class design. The files are:
"I certify that the work submitted for this assignment does not violate USC's student conduct code. In particular, the work is my own, not a collaboration, and does not involve code created by other people or AI software, with the exception of the resources explicitly mentioned in the CS 455 Course Syllabus. And I did not share my solution or parts of it with other students in the course."
For example, if your rack had the letters c m a l you could rearrange the letters to form the words calm or clam, but you could also form shorter words from a subset of the letters, e.g., lam or ma. It's generally difficult to figure out all such sequences of the letters that form real words (unless you are a tournament Scrabble competitor who knows the Scrabble dictionary very well).
For your program, you will display all such words for a rack, with the corresponding Scrabble score for each word, in decreasing order by score. Each letter has a score associated with it, the score for a word is the sum of the scores of each letter in that word. For words with the same scrabble score, the words must appear in alphabetical order. Here are the results for a rack consisting of "cmal" (using the sowpods dictionary) shown in the output format you will be using for your program (user input is shown in italics):
Rack? cmal We can make 11 words from "cmal" All of the words with their scores (sorted by score): 8: calm 8: clam 7: cam 7: mac 5: lac 5: lam 5: mal 4: am 4: ma 2: al 2: la
We'll provide you the Scrabble score for each letter later in this document.
Here's more about exactly how to run your program and what happens:
Your program will take an optional command-line argument for the dictionary file name. If that argument is left off, it will use the Scrabble dictionary file sowpods.txt (see assignment files) from the same directory as you are running your program. (Note: Required error-checking related to the dictionary file is described in the following section.)
Once the program starts it will print the message:
Type . to quit.Then the program will run in a loop on the console, printing the prompt "Rack? " (as seen in the earlier example) and reading and processing each rack you enter, until you tell it to exit. The user tells the program to exit by typing in "." at the prompt (i.e., a period). We aren't use a command such as "quit" as the sentinel, since that could be a legal rack.
We have provided you a few sample data files, and corresponding correct reference output from running those on the sowpods.txt (the Scrabble dictionary given) in the testFiles directory. Also in that directory are a few other smaller dictionaries and sample input and output for them. Please see the README.txt in that directory for guide to the sample files and how to use them. Your output must match the reference output character by character.
The real game of Scrabble has only upper-case letters on tiles, but for our program we'll accept any sequence of non-whitespace characters as a legal "rack." However, words will only be able to be formed from actual letters if that's what's in the given dictionary. E.g., if the rack given is "abc@" you will report the words such as "cab", but there will be no words containing "@", since @ doesn't appear in any dictionary words. If there are such characters in the rack, they also get printed out in the initial message displayed about the rack. E.g, We can make 11 words from "cm!a#l"
The program will work on both lower-and-upper case versions of dictionaries, but all processing will be case-sensitive. E.g., if the dictionary given has only upper-case versions of words, it will find words from a rack such as "CMAL", but won't be able to find any words from the rack "cmal".
Some other differences between this program and Scrabble:
java WordFinder [dictionaryFile]Note: in this common format for showing Unix command-line syntax the square brackets (i.e., []) are not part of the command that is typed: it is just a notation indicating that the command line argument shown is optional.
Additional program requirements are described in the following sections and summarized here:
Exiting program.and then exit immediately.
Suppose the dictionary file given was testFiles/foobar.txt
Error message:
ERROR: Dictionary file "testFiles/foobar.txt" does not exist.
Suppose the dictionary file contained the word "cat" in two places. E.g., dictionary file contents:
house cat the dog cat doggy
Error message:
ERROR: Illegal dictionary: dictionary file has a duplicate word: catYour program does not have to report all duplicate words, just the first one it detects. The example given above doesn't have any other words duplicated ("dog" and "doggy" are two different words).
The second approach, which is the one you will be using for the assignment, involves preprocessing the dictionary so that you organize the words by the set of letters each one contains (this set is actually a multiset, because letters can appear more than once in a word; the rack itself is also a multiset). Then for each rack you'll generate all the subsets of that multiset of letters, and for each subset add all the words from the dictionary that have exactly the same elements as that subset. This is slower to process the dictionary, but once we do this processing, it's faster to process each rack than the first approach. This approach is explained in more detail in the following two sections.
It's a little complicated to describe in big-O terms the time for each approach, but what makes the first approach slower for processing one rack is traversing the whole dictionary (which will typically be large) for each rack. For the second approach, the slow part of processing a rack is creating all the subsets. The worst case for creating the subsets is if there are no repeated letters in the rack (i.e., largest number of subsets created). Even though generating the subsets for such a rack would take O(n * 2n) for a rack of n unique characters (because there are 2n subsets when there are no repeat characters, and n n steps to form each subset), n will typically be small: for a 7-tile rack: 27 is only 128, times 7 is 896). In an instrumented solution we wrote using this approach, processing the sowpods dictionary took under half a second, and processing a 7-character rack with no repeating characters, and consisting of the most commonly occurring letters in English took under 15 milliseconds (this is test file testFiles/aestnlr.in). (Commonly occurring letters will result in a larger resulting word list.) These runs were done on Vocareum.
Some of the time spent for processing a rack in the second approach is to get the list of anagrams for each subset; we'll discuss that further in the next section.
The rest of the time spent processing a rack is to sort the resulting word list.
For full credit on this assignment you'll need to use this second approach for the assignment; we'll go into further details about it in the following sections.
You are required to create an AnagramDictionary class to handle this. It will have a getAnagramsOf method that finds all anagrams of a particular string efficiently. For, example, suppose we have a variable, dictionary, of type AnagramDictionary, that contains data from the sowpods dictionary. If we did the call
dictionary.getAnagramsOf("rlee")
it would return an ArrayList of the following dictionary words:
How to do this efficiently? One insight is that if we put two words into some kind of canonical form, then we could figure out if they are anagrams of each other by just comparing the canonical versions of them for equality. This canonical form will be a sorted version of the characters in the word. In the earlier example given the rack contained "cmal". The sorted version of this rack is "aclm". The first two words listed in the output are "calm" and "clam", anagrams of "aclm", or put another way, these first two words are the only dictionary words we can make using all the letters on the rack, and all the other words listed are anagrams of subsets of "cmal".
For full credit your AnagramDictionary is required to find all the anagrams of one String in time linear in the size of the output set (not including the time to sort the letters in the String given to put it into canonical form.)
The allSubsets method uses a particular representation for the rack which we'll explain with an example here. Earlier we mentioned that a rack is a multiset of letters (set because we don't care about the order of the letters, and multiset because letters can appear more than once). Suppose our rack is:
a b a d b b
Gathering together the like letters, we could rewrite this as "aabbbd". We could also say that 'a' appears with multiplicity 2, 'b' appears with multiplicity 3, and that 'd' appears with multiplicity 1. allSubsets expects the rack information to be in two parallel arrays: one has the unique letters, and the other has the multiplicity of that letter at the same array index. The array of unique letters is actually a String, so we can do String operations on it. For the example given, we could create this rack representation as follows:
// create variables for the rack "aabbbd"
String unique = "abd";
int[] mult = {2, 3, 1};
// example to show relation between values in unique and mult:
for (int i = 0; i < unique.length(); i++) {
System.out.prinln(unique.charAt(i) + " appears " + multi[i] + " times in the rack");
}
Like other examples of recursion over an array that we've seen,
allSubsets will take a third argument, k, which
is the starting position of the part of the array that this recursive call will
process. So for this code, it's the starting postion from which to
find the subsets. So, for example, if we called
allSubsets(unique, mult, 1); // starts at position 1 in unique and multit would find all the subsets of the rack "bbbd" (i.e., it wouldn't consider the subsets that included any 'a's in it).
When doing an object-oriented design, you first come up with a candidate set of classes, choosing a name for each, and identifying the responsibilities of each in the context of the larger program overall. We have done that step for you here. We are requiring you to have at least the following five classes in your solution, with the responsibilities described. You are allowed to add more classes to your design as you see fit. The five, with their overall responsibilities described, are:
One thing to keep in mind is you want the code that operates on some data to be in the same class that contains that data. One sign that your design doesn't have that feature is if your classes tend to have a lot of get and set methods and not much else. That would indicate that all the code operating on this data is outside of the class itself.
Hopefully we've made clear the importance of making all instance variables private. But even if you make your data private there are other ways to expose the implementation of your objects. For example, if you have a class that contains an ArrayList, and also provide an accessor method for this ArrayList, it gives clients the ability to change the contents of that arraylist from outside of the object methods, possibly invalidating the object. (We discussed these types of issues and how to cope with them in the material on side effects in week 6.)
You are welcome to add additional classes as part of your design. These ones would be designed and implemented by you, of course. If you have more classes, just make sure the additional .java files are in your Vocareum home directory when you submit the assignment. If a class is just used by one other class, you could put it in the same file as that class, or a separate file. If it is used by multiple classes, it should be in its own file. Make sure you discuss these additional classes in your design write-up in your README (including telling us where to find them).
You'll want to test your complete program (and your AnagramDictionary) on a small dictionary file before subjecting it to sowpods.txt. We provided a sample small dictionary and input and corresponding output for some racks in the testFiles directory (more about that in the next paragraph). If you find AnagramDictionary-related bugs, you may want to use an even tinier dictionary for when you are single-stepping, etc.
Once you have all your modules working, you can also check if your program produces the right answers for sowpods.txt with the other test input files and corresponding output in the testFiles directory. Note: The testFiles/README.txt file describes what's what that directory.
Your README file must document known bugs in your program, contain the signed certification shown near the top of this document, and contain any special instructions or information for the grader.
In addition, for this assignment, your README must also document your design. This includes the approach you took to solving the problem (i.e., description of the data structures and algorithms involved). One part of this was discussed in the section on approach. You will also include there information about how your class design relates to this approach, including what data structures and algorithms are encapsulated in which of your classes.
When you are ready to submit the assignment press the big "Submit" button in your PA4 Vocareum work area. Because you may have additional files in your program, it will try to compile all files in your work area, and test the resulting program on the small dictionary data we gave you in testFiles (not on sowpods). As usual, you will want to submit for the first time well before the final deadline, so you have time to fix any errors you get on the submit script.
Passing these submit checks is not necessary or sufficient to submit your code (the graders will get a copy of what you submitted either way). (It would be necessary but not sufficient for getting full credit.) However, if your code does not pass all the tests we would expect that you would include some explanation of that in your README. One situation where it might fail would be if you only completed a subset of the assignment (and your README would document what subset you completed.)