程序代写代做代考 Haskell C Java G6021: Comparative Programming

G6021: Comparative Programming
Exercise Sheet 9
A: Logic Programming
We will implement a small fragment of a logic programming language. First we explain the language. Programs are collections of:
• facts: things that are known to be true,
• and rules: ways to generate new facts.
In addition, we have a query which is a question about the facts that we have (the result of a query will be either True or False). We write facts and rules as shown in the following example:
likePasta.
likeRissoto.
likePizza.
likeItalianFood :- likePizza, likePasta, likeRissoto.
The first three lines are facts, the fourth is a rule. We read the rule as: someone likes Italian food if they like pizza, pasta and risotto. (i.e., we read :- as “if”, and comma as “and”.)
A query is a question about the facts. For example: ?likeRisotto is true, because it is a fact. When we write queries, we put a question mark at the start. A more complicated query is ?likeRisotto,likePasta. This is again true, because both facts are true, however, ?likeHaggis is false, because there is no fact about this in our program.
An even more complicated query: ?likeItalianFood is true. In this case, we have use the rule, and now we have a new query: ?likePizza, likePasta, likeRisotto, and now we need to check each of these, etc. Computation is searching: either the fact is there, or we have to apply a rule and we get some new facts to check. When we have no more facts to check, then the query is true. If we cannot find a fact, then it is false.
Facts and rules are in fact the same, it’s just that there is nothing else to check for a fact. Facts and rules are called Clauses (specifically Horn Clauses). In a Clause such as likeItalianFood :- likePizza, likePasta, likeRisotto, we call likeItalianFood the head of the Clause, and the list: likePizza, likePasta, likeRisotto is called the tail of the Clause. Note that for a fact, the tail is empty.
We can represent Clauses as a pair: the head and the tail of the Clause. Here is one way to represent the above example in Haskell:
(“likeItalianFood”,[“likeRissoto”,”likePasta”,”likePizza”])
(“likeRissoto”,[])
(“likePasta”,[])
(“likePizza”,[])
Using this representation, we can write some functions to implement this form of computation.
1. Write a function to look-up a query in the list of clauses. If it is there, return the pair (True, t) where t are the new queries (the tail of the Clause, or the empty list if it was a fact). If the goal is not in the list of clauses, then return (False,[]).
1

Answer:
lookup h [] = (False,[])
lookup h ((p,t):r) = if h==p then (True,t) else lookup h r
2. Write a function called solution that takes a list of queries and a list of clauses, and returns True if the queries are satisfied by the program clauses, False otherwise.
Answer:
solution [] prog = True
solution (h:t) prog =
let (a,b) = (lookup h prog) in
if a then solution (b++t) prog else False
3. Test the above example with:
solution [“likeItalianFood”]
[ (“likeItalianFood”,[“likeRissoto”,”likePasta”,”likePizza”]),
(“likeRissoto”,[]), (“likePasta”,[]), (“likePizza”,[])]
Make up some other example programs and queries to test your functions.
B: Input/Output
There are a number of problems with user interaction in Haskell programs:
• Reading from the keyboard and writing to the screen are actions with side effects,
• Side effects change the state of the environment the function runs in and cannot be undone,
• Pure functional programming has no side effects.
However, there are some solutions to this, that allow impure functions to be introduced in a controlled way (through types). Without going into the details, the following examples illustrate how things work.
The standard library provides a number of actions, including:
• getChar :: IO Char reads a char from the keyboard, echoes it to the screen and returns the read value. getChar has type IO Char, which means that the return type is a Char, but some side effects might have taken place.
• putChar :: Char -> IO () writes a char to the screen and returns (). (() is analogous to void in Java.)
• return :: a -> IO a returns its argument with no interaction.
A sequence of actions can be combined into a single composite action using the keyword do,
and we use the operator <- to bind the result of these kinds of functions: getTwo :: IO (Char, Char) getTwo = do x <- getChar y <- getChar return (x,y) Here is another example, that reads a String from the Keyboard: 2 getChars getChars :: IO String = do c <- getChar if c == ’\n’ then return "" else do l <- getChars return (c:l) Finally, here is an example where we write a String to the Screen: putString :: String -> IO ()
putString [] = return ()
putString (x:xs) = do putChar x
putStr xs
putStringLn :: String -> IO ()
putStringLn xs = do putString xs
putChar ’\n’
1. Test the above functions to make sure you understand how these work.
2. Write a function len :: IO () that will prompt the user to enter a string, will read the string, and then calculate and print the length of the string.
Answer:
len :: IO ()
len = do putStr “Enter a string: ”
xs <- getLine putStr "The string has " putStr (show (length xs)) putStrLn " characters" 3