I want to create a series of possible equations based on a general specification:
test = ["12", "34=", "56=", "78"]
Each string (e.g. "12") represents a possible character at that location, in this case '1' or '2'.)
So possible equations from test would be "13=7" or "1=68".
I know the examples I give are not balanced but that's because I'm deliberately giving a simplified short string.
(I also know that I could use 'sequence' to search all possibilities but I want to be more intelligent so I need a different approach explained below.)
What I want is to try fixing each of the equals in turn and then removing all other equals in the equation. So I want:
[["12","=","56","78"],["12","34","=","78”]]
I've written this nested list comprehension:
(it needs: {-# LANGUAGE ParallelListComp #-} )
fixEquals :: [String] -> [[String]]
fixEquals re
= [
[
if index == outerIndex then equals else remain
| equals <- map (filter (== '=')) re
| remain <- map (filter (/= '=')) re
| index <- [1..]
]
| outerIndex <- [1..length re]
]
This produces:
[["","34","56","78"],["12","=","56","78"],["12","34","=","78"],["12","34","56","”]]
but I want to filter out any with empty lists within them. i.e. in this case, the first and last.
I can do:
countOfEmpty :: (Eq a) => [[a]] -> Int
countOfEmpty = length . filter (== [])
fixEqualsFiltered :: [String] -> [[String]]
fixEqualsFiltered re = filter (\x -> countOfEmpty x == 0) (fixEquals re)
so that "fixEqualsFiltered test" gives:
[["12","=","56","78"],["12","34","=","78”]]
which is what I want but it doesn’t seem elegant.
I can’t help thinking there’s another way to filter these out.
After all, it’s whenever "equals" is used in the if statement and is empty that we want to drop the equals so it seems a waste to build the list (e.g. ["","34","56","78”] and then ditch it.)
Any thoughts appreciated.
I don't know if this is any cleaner than your code, but it might be a bit more clear and maybe more efficient using a recursion:
fixEquals = init . f
f :: [String] -> [[String]]
f [] = [[]]
f (x:xs) | '=' `elem` x = ("=":removeEq xs) : map (removeEq [x] ++) (f xs)
| otherwise = map (x:) (f xs)
removeEq :: [String] -> [String]
removeEq = map (filter (/= '='))
The way it works is that, if there's an '=' in the current string, then it splits the return into two, if not just calls recursively. The init is needed as in the last element returned there's no equal in any string.
Finally, I believe you can probably find a better data structure to do what you need to achieve instead of using list of strings
Let
xs = [["","34","56","78"],["12","=","56","78"],["12","34","=","78"],["12","34","56",""]]
in
filter (not . any null) xs
will give
[["12","=","56","78"],["12","34","=","78"]]
If you want list comprehension then do
[x | x <- xs, and [not $ null y | y <- x]]
I think I'd probably do it this way. First, a preliminary that I've written so many times it's practically burned into my fingers by now:
zippers :: [a] -> [([a], a, [a])]
zippers = go [] where
go _ [] = []
go b (h:e) = (b,h,e):go (h:b) e
Probably running it once or twice in ghci will be a more clear explanation of what this does than any English writing I could do:
> zippers "abcd"
[("",'a',"bcd"),("a",'b',"cd"),("ba",'c',"d"),("cba",'d',"")]
In other words, it gives a way of selecting each element of a list in turn, giving the "leftovers" of what was before and after the selection point. Given that tool, here's our plan: we'll nondeterministically choose a String to serve as our equals sign, double-check that we've got an equals sign in the first place, and then clear out the equals from the others. So:
fixEquals ss = do
(prefix, s, suffix) <- zippers ss
guard ('=' `elem` s)
return (reverse (deleteEquals prefix) ++ ["="] ++ deleteEquals suffix)
deleteEquals = map (filter ('='/=))
Let's try it:
> fixEquals ["12", "34=", "56=", "78"]
[["12","=","56","78"],["12","34","=","78"]]
Perfect! But this is just a stepping-stone to actually generating the equations, right? It turns out to be not that hard to go all the way in one step, skipping this intermediate. Let's do that:
equations ss = do
(prefixes, s, suffixes) <- zippers ss
guard ('=' `elem` s)
prefix <- mapM (filter ('='/=)) (reverse prefixes)
suffix <- mapM (filter ('='/=)) suffixes
return (prefix ++ "=" ++ suffix)
And we can try it in ghci:
> equations ["12", "34=", "56=", "78"]
["1=57","1=58","1=67","1=68","2=57","2=58","2=67","2=68","13=7","13=8","14=7","14=8","23=7","23=8","24=7","24=8"]
The easiest waty to achieve what you want is to create all the combinations and to filter the ones that have a meaning:
Prelude> test = ["12", "34=", "56=", "78"]
Prelude> sequence test
["1357","1358","1367","1368","13=7","13=8","1457","1458","1467","1468","14=7","14=8","1=57","1=58","1=67","1=68","1==7","1==8","2357","2358","2367","2368","23=7","23=8","2457","2458","2467","2468","24=7","24=8"
Prelude> filter ((1==).length.filter('='==)) $ sequence test
["13=7","13=8","14=7","14=8","1=57","1=58","1=67","1=68","23=7","23=8","24=7","24=8","2=57","2=58","2=67","2=68"]
You pointed the drawback: imagine we have the followig list of strings: ["=", "=", "0123456789", "0123456789"]. We will generate 100 combinations and drop them all.
You can look at the combinations as a tree. For the ["12", "34"], you have:
/ \
1 2
/ \ / \
3 4 3 4
You can prune the tree: just ignore the subtrees when you have two = on the path.
Let's try to do it. First, a simple combinations function:
Prelude> :set +m
Prelude> let combinations :: [String] -> [String]
Prelude| combinations [] = [""]
Prelude| combinations (cs:ts) = [c:t | c<-cs, t<-combinations ts]
Prelude|
Prelude> combinations test
["1357","1358","1367","1368","13=7","13=8","1457","1458","1467","1468","14=7","14=8","1=57","1=58","1=67","1=68","1==7","1==8","2357","2358","2367","2368","23=7","23=8","2457","2458","2467","2468","24=7","24=8", ...]
Second, we need a variable to store the current number of = signs met:
if we find a second = sign, just drop the subtree
if we reach the end of a combination with no =, drop the combination
That is:
Prelude> let combinations' :: [String] -> Int -> [String]
Prelude| combinations' [] n= if n==1 then [""] else []
Prelude| combinations' (cs:ts) n = [c:t | c<-cs, let p = n+(fromEnum $ c=='='), p <= 1, t<-combinations' ts p]
Prelude|
Prelude> combinations' test 0
["13=7","13=8","14=7","14=8","1=57","1=58","1=67","1=68","23=7","23=8","24=7","24=8","2=57","2=58","2=67","2=68"]
We use p as the new number of = sign on the path: if p>1, drop the subtree.
If n is zero, we don't have any = sign in the path, drop the combination.
You may use the variable n to store more information, eg type of the last char (to avoid +* sequences).
I am trying to write a Haskell function that takes in a list of strings, compares all the strings in the list, and outputs a list of strings that are of the longest length. I want to do this without any predefined functions or imports, I want to try and figure it out all recursively. For example:
longeststrings["meow","cats","dog","woof"] -> ["meow","cats","woof"]
I know it is a silly example, but I think it proves the point.
I want to do something like
longeststrings:: [string] -> [string]
longeststrings [] = []
longeststrings [x:xs] = if (x > xs) x:longeststrings[xs]
But I don't know how to only take the largest size strings out of the list, or remove the smallest ones. I would appreciate any help.
you could trivially keep track of the longest length string and an accumulator of values of that length.
longestStrings :: [String] -> [String]
longestStrings = go [] 0
where
go acc _ [] = acc -- base case
go acc n (x:xs)
| length x > n = go [x] (length x) xs
-- if x is longer than the previously-seen longest string, then set
-- accumulator to the list containing only x, set the longest length
-- to length x, and keep looking
| length x == n = go (x:acc) n xs
-- if x is the same length as the previously-seen longest string, then
-- add it to the accumulator and keep looking
| otherwise = go acc n xs
-- otherwise, ignore it
or, as Davislor rightly mentions in the comments, this can be implemented as a fold by teaching the helper function to determine its own longest length
longestStrings :: [String] -> [String]
longestStrings = foldr f []
where
f x [] = [x]
f x yss#(y:_) =
case compare (length x) (length y) of
GT -> [x]
EQ -> x : yss
LT -> yss
As requested, here’s a version with and without the use of where. I think this is a good demonstration of why the advice not to use where is bad advice. I think the first version is a lot easier to understand.
Keep in mind, functional programming isn’t a monastic vow to forswear certain keywords out of masochism. Nor is it a checklist of fashion tips (where is so last season!). “You should avoid that construct arbitrarily because it’s not the ‘functional’ thing to do” really is not how it works. So you shouldn’t uglify your code for the sake of a tip like that.
It is often a good idea to follow the same coding style as other programmers so they will find it easier to understand you. (For example, Adam Smith was subtly trying to train you that acc is a common name for an accumulator and go a common name for a recursive helper function, and they help other programmers figure out the pattern he’s using.) But in fact Haskell programmers do use where, a lot.
Anyway, the code:
longeststrings :: [String] -> [String]
{- Filters all strings as long as any other in the list. -}
longeststrings = foldr go []
where
go x [] = [x]
go x leaders | newlength > record = [x]
| newlength == record = x:leaders
| otherwise = leaders
where
record = (length . head) leaders
newlength = length x
longestStringsUsingLambda :: [String] -> [String]
longestStringsUsingLambda = foldr
(\x leaders ->
if leaders == [] then [x]
else case compare (length x) (length $ head leaders) of
GT -> [x]
EQ -> x:leaders
LT -> leaders )
[]
main :: IO ()
main = let testcases = [ ["meow","cats","dog","woof"],
["foo","bar","baz"],
["meep","foo","bar","baz","fritz","frotz"]
]
in foldMap print $
map longestStringsUsingLambda testcases
You can try eliminating the let testcases = ... and see if you consider that an improvement.
I'm trying to write a function which adds single characters from a string to a list of strings, for instance
combine ", !" ["Hello", "", "..."] = ["Hello,", " ", "...!"]
I've tried this:
combine :: String -> [String] -> [String]
combine (y:ys) (x:xs) =
[x:y, combine ys xs]
A simple one would be
combine :: [Char] -> [String] -> [String]
combine [] _ = []
combine _ [] = []
combine (c:cs) (x:xs) = x ++ [c] : combine cs xs
Or even more simply using zipWith
combine :: [Char] -> [String] -> [String]
combine = zipWith (\c x -> x ++ [c])
I had to do a bit extra to get this to work. I'll break it down for you.
First, I specified the type of the function as [Char] -> [String] -> [String]. I could have used String for the first argument, but what you're operating on conceptually is a list of characters and a list of strings, not a string and a list of strings.
Next, I had to specify the edge cases for this function. What happens when either argument is the empty list []? The easy answer is to just end the computation then, so we can write
combine [] _ = []
combine _ [] = []
Here the _ is matching anything, but throwing it away because it isn't used in the return value.
Next, for the actual body of the function We want to take the first character and the first string, then append that character to the end of the string:
combine (c:cs) (x:xs) = x ++ [c]
But this doesn't do anything with cs or xs, the rest of our lists (and won't even compile with the type signature above). We need to keep going, and since we're generating a list, this is normally done with the prepend operator :
combine (c:cs) (x:xs) = x ++ [c] : combine cs xs
However, this is such a common pattern that there is a helper function called zipWith that handles the edge cases for us. It's type signature is
zipWith :: (a -> b -> c) -> [a] -> [b] -> [c]
It walks down both input lists simultaneously, passing the corresponding elements into the provided function. Since the function we want to apply is \c x -> x ++ [c] (turned into a lambda function), we can drop it in to zipWith as
combine cs xs = zipWith (\c x -> x ++ [c]) cs xs
But Haskell will let us drop arguments when possible, so we can eta reduce this to
combine :: [Char] -> [String] -> [String]
combine = zipWith (\c x -> x ++ [c])
And that's it!
When you want to combine lists element by element, it is usually a zip you are looking at. In this case, you know exactly how you want to combine the elements – that makes it a zipWith.
zipWith takes a "combining function" and then creates a function that combines two lists using said combining function. Let's call your "combining" function append, because it adds a characters to the end of a string. You can define it like this:
append char string = string ++ [char]
Do you see how this works? For example,
append 'e' "nic" = "nice"
or
append '!' "Hello" = "Hello!"
Now that we have that, recall that zipWith takes a "combining function" and then creates a function that combines two lists using that function. So your function is then easily implemented as
combine = zipWith append
and it will do append on each of the elements in order in the lists you supply, like so:
combine ", !" ["Hello", "", "..."] = ["Hello,", " ", "...!"]
You are close. There are a couple issues with what you have.
y has type Char, and x has type String which is an alias for [Char]. This means that you can add y to the top of a list with y : x, but you can't add y to the end of a list using the same : operator. Instead, you make y into a list and join the lists.
x ++ [y]
There must also be a base case, or this recursion will continue until it has no elements in either list and crash. In this case, we likely don't have anything we want to add.
combine [] [] = []
Finally, once we create the element y ++ [x] we want to add it to the top of the rest of the items we have computed. So we use : to cons it to our list.
combine :: String -> [String] -> [String]
combine [] [] = []
combine (x : xs) (y : ys) = (y ++ [x]) : (combine xs ys)
One note about this code, if there is ever a point where the number of characters in your string is different from the number of strings in you list, then this will crash. You can handle that case in a number of ways, bheklilr's answer addresses this.
kqr's answer also works perfectly and is probably the best one to use in practice.
I've just started to learn Haskell and I am trying to write a simple function that takes a list of strings and reverses each string in the list:
revComp :: [String] -> [String]
revComp [] = []
revComp [x] = [] ++ [reverse x]
revComp (x:xs) = revComp [xs]
When I try to load my code in GHCI, I get an error:
Couldn't match expected type `Char' with actual type `[Char]'
Expected type: String
Actual type: [String]
Could anyone explain what and where my problem is? Thanks very much.
The first three lines are fine. Your type signature is correct, the second line is correct, and so is the third. (However, [] ++ [reverse x] is the same as [reverse x].)
The fourth line, however, is wrong. Not only do you not use x at all on the right-hand side, but you have a type error: revComp [xs] calls revComp with a single-element list that has xs as its only element. Here, x is the first element of the list, and xs is the rest of the list. So, since xs has the type [String], [xs] has the type [[String]], but revComp takes a [String]! You want to reverse x, and prepend that to the result of reversing the rest of the list.
You can use revComp xs to reverse each string in the rest of the list, and (:) to prepend a value to a list (with the same syntax seen in the x:xs pattern you used on the left-hand side). That should be enough information for you to fix the last line. This makes the third line redundant, by the way, since [x] is just x:[].
ehird's answer is really complete, but I wanted to point out -- as a reference -- the "good/shorter" way to implement that function in real code, since you're new.
It can be defined as the partial application of map to reverse:
Prelude> let revComp = map reverse
Prelude> revComp ["olleh", "dlrow"]
["hello","world"]
It results in a function revComp :: [[a]] -> [[a]] (and not [String] -> [String] since both map and reverse work on any type of lists) that maps reverse to every element of the input list, returning the list of results.
The third line might not be required even with your example. The edge case (terminating condition) can be only the empty list. You can skip the condition for the single element.
revComp :: [String] -> [String]
revComp [] = []
revComp (x:xs) = reverse x : revComp xs
Don't know if there is a CS term for this. Someone in the community can add that information.
A few more ways to do the same thing:
revstr [] = []
revstr (x:xs) = reverse x : revstr xs
revstr2 [] = []
revstr2 (x:xs) = [reverse x] ++ revstr2 xs
revstr3 xs = map reverse xs
revstr4 = map reverse -- map takes 2 arguments ;)
revstr5 xs = foldl step [] xs
where step acc x = acc ++ [reverse x]
revstr6 xs = foldr step [] xs
where step x acc = reverse x : acc