HORIZON HASKELLDocslts/ghc-10.0.x8fef28d2026-11-02Search packagesCtrl K
Theme: |

GHC 10.0.0.20260917 · lts/ghc-10.0.x · 8fef28d · 2026-11-02

Moduleghc-lib-parser-9.14.1.20251220GHC2021

GHC.Utils.Misc

Highly random utility functions

Contents
  1. Miscellaneous higher-order functions
  2. General list processing
  3. Tuples
  4. List operations controlled by another list
  5. Sorting
  6. Comparisons
  7. Edit distance
  8. Transitive closures
  9. Strictness
  10. Module names
  11. Integers
  12. Floating point
  13. IO-ish utilities
  14. Filenames and paths
  15. Utils for defining Data instances
  16. Utils for printing C code
  17. Hashing
  18. Call stacks
Synopsis

Miscellaneous higher-order functions

valuenTimes :: Int -> (a -> a) -> a -> a

Apply a function n times to a given value.

valueconst2 :: a -> b -> c -> a
#

General list processing

valuestretchZipWith :: (a -> Bool) -> b -> (a -> b -> c) -> [a] -> [b] -> [c]

stretchZipWith p z f xs ys stretches ys by inserting z in the places where p returns True

valuezipAndUnzip :: [a] -> [b] -> ([a], [b])

This has the effect of making the two lists have equal length by dropping the tail of the longer one.

valuefilterByList :: [Bool] -> [a] -> [a]

filterByList takes a list of Bools and a list of some elements and filters out these elements for which the corresponding value in the list of Bools is False. This function does not check whether the lists have equal length.

valuefilterByLists :: [Bool] -> [a] -> [a] -> [a]

filterByLists takes a list of Bools and two lists as input, and outputs a new list consisting of elements from the last two input lists. For each Bool in the list, if it is True, then it takes an element from the former list. If it is False, it takes an element from the latter list. The elements taken correspond to the index of the Bool in its list. For example:

filterByLists [True, False, True, False] "abcd" "wxyz" = "axcz"

This function does not check whether the lists have equal length.

valuepartitionByList :: [Bool] -> [a] -> ([a], [a])

partitionByList takes a list of Bools and a list of some elements and partitions the list according to the list of Bools. Elements corresponding to True go to the left; elements corresponding to False go to the right. For example, partitionByList [True, False, True] [1,2,3] == ([1,3], [2]) This function does not check whether the lists have equal length; when one list runs out, the function stops.

valueunzipWith :: (a -> b -> c) -> [(a, b)] -> [c]
#
valuemapAndUnzip :: (a -> (b, c)) -> [a] -> ([b], [c])
#
valuemapAndUnzip3 :: (a -> (b, c, d)) -> [a] -> ([b], [c], [d])
#
valuemapAndUnzip4 :: (a -> (b, c, d, e)) -> [a] -> ([b], [c], [d], [e])
#
valuefilterOut :: (a -> Bool) -> [a] -> [a]

Like filter, only it reverses the sense of the test

valuepartitionWith :: (a -> Either b c) -> [a] -> ([b], [c])

Uses a function to determine which of two output lists an input element should join

valuespanEnd :: (a -> Bool) -> [a] -> ([a], [a])

spanEnd p l == reverse (span p (reverse l)). The first list returns actually comes after the second list (when you look at the input list).

valueonJust :: b -> Maybe a -> (a -> b) -> b

onJust x m f applies f to the value inside the Just or returns the default.

valuefoldl2 :: (acc -> a -> b -> acc) -> acc -> [a] -> [b] -> acc
#
valueall2Prefix :: (a -> b -> Bool) -> [a] -> [b] -> Bool

`all2Prefix p xs ys` is a fused version of `and $ zipWith2 p xs ys`. (It generates good code nonetheless.) So if one list is shorter than the other, p is assumed to be True for the suffix.

valueall3Prefix :: (a -> b -> c -> Bool) -> [a] -> [b] -> [c] -> Bool

`all3Prefix p xs ys zs` is a fused version of `and $ zipWith3 p xs ys zs`. (It generates good code nonetheless.) So if one list is shorter than the others, p is assumed to be True for the suffix.

valueatLength :: ([a] -> b) -> b -> [a] -> Int -> b

atLength atLen atEnd ls n unravels list ls to position n. Precisely:

 atLength atLenPred atEndPred ls n
  | n < 0         = atLenPred ls
  | length ls < n = atEndPred (n - length ls)
  | otherwise     = atLenPred (drop n ls)
valueonly :: [a] -> a

Utility function to go from a singleton list to it's element.

Wether or not the argument is a singleton list is only checked in debug builds.

valueexpectOnly :: HasCallStack => [a] -> a

Extract the single element of a list and panic with the given message if there are more elements or the list was empty. Like expectJust, but for lists.

valuesnocView :: [a] -> Maybe ([a], a)

Split a list into its last element and the initial part of the list. snocView xs = Just (init xs, last xs) for non-empty lists. snocView xs = Nothing otherwise. Unless both parts of the result are guaranteed to be used prefer separate calls to last + init. If you are guaranteed to use both, this will be more efficient.

valueholes :: [a] -> [(a, [a])]

Compute all the ways of removing a single element from a list.

holes [1,2,3] = [(1, [2,3]), (2, [1,3]), (3, [1,2])]
valuechangeLast :: [a] -> a -> [a]

Replace the last element of a list with another element.

valuemergeListsBy :: (a -> a -> Ordering) -> [[a]] -> [a]

Merge an unsorted list of sorted lists, for example:

mergeListsBy compare [ [2,5,15], [1,10,100] ] = [1,2,5,10,15,100]

O(n \log{} k)

Tuples

valuefstOf3 :: (a, b, c) -> a
#
valuesndOf3 :: (a, b, c) -> b
#
valuethdOf3 :: (a, b, c) -> c
#
valuefstOf4 :: (a, b, c, d) -> a
#
valuesndOf4 :: (a, b, c, d) -> b
#
valuefst3 :: (a -> d) -> (a, b, c) -> (d, b, c)
#
valuesnd3 :: (b -> d) -> (a, b, c) -> (a, d, c)
#
valuethird3 :: (c -> d) -> (a, b, c) -> (a, b, d)
#
valueuncurry3 :: (a -> b -> c -> d) -> (a, b, c) -> d
#

List operations controlled by another list

valuesplitAtList :: [b] -> [a] -> ([a], [a])

Given two lists xs and ys, return `splitAt (length xs) ys`.

Sorting

valuesortWith :: Ord b => (a -> b) -> [a] -> [a]

The sortWith function sorts a list of elements using the user supplied function to project something out of each element

In general if the user supplied function is expensive to compute then you should probably be using sortOn, as it only needs to compute it once for each element. sortWith, on the other hand must compute the mapping function for every comparison that it performs.

valueordNub :: Ord a => [a] -> [a]

Remove duplicates but keep elements in order. O(n * log n)

valueordNubOn :: Ord b => (a -> b) -> [a] -> [a]

Remove duplicates but keep elements in order. O(n * log n)

Comparisons

Edit distance

valuefuzzyLookup :: String -> [(String, a)] -> [a]

Search for possible matches to the users input in the given list, returning a small number of ranked results

Transitive closures

Strictness

valuestrictZipWith3 :: (a -> b -> c -> d) -> [a] -> [b] -> [c] -> [d]
#

Module names

Integers

Floating point

Parse a string into a significand and exponent. A trivial example might be: ghci> readSignificandExponentPair "1E2" (1,2) In a more complex case we might return a exponent different than that which the user wrote. This is needed in order to use a Integer significand. ghci> readSignificandExponentPair "-1.11E5" (-111,3)

Parse a string into a significand and exponent according to the "Hexadecimal Floats in Haskell" proposal. A trivial example might be: ghci> readHexSignificandExponentPair "0x1p+1" (1,1) Behaves similar to readSignificandExponentPair but the base is 16 and numbers are given in hexadecimal: ghci> readHexSignificandExponentPair "0xAp-4" (10,-4) ghci> readHexSignificandExponentPair "0x1.2p3" (18,-1)

IO-ish utilities

Filenames and paths

Utils for defining Data instances

Utils for printing C code

Hashing

A sample hash function for Strings. We keep multiplying by the golden ratio and adding. The implementation is:

hashString = foldl' f golden
  where f m c = fromIntegral (ord c) * magic + hashInt32 m
        magic = 0xdeadbeef

Where hashInt32 works just as hashInt shown above.

Knuth argues that repeated multiplication by the golden ratio will minimize gaps in the hash space, and thus it's a good choice for combining together multiple keys to form one.

Here we know that individual characters c are often small, and this produces frequent collisions if we use ord c alone. A particular problem are the shorter low ASCII and ISO-8859-1 character strings. We pre-multiply by a magic twiddle factor to obtain a good distribution. In fact, given the following test:

testp :: Int32 -> Int
testp k = (n - ) . length . group . sort . map hs . take n $ ls
  where ls = [] : [c : l | l <- ls, c <- ['\0'..'\xff']]
        hs = foldl' f golden
        f m c = fromIntegral (ord c) * k + hashInt32 m
        n = 100000

We discover that testp magic = 0.

Call stacks

typetype HasCallStack = IP "callStack" CallStack

Request a CallStack.

NOTE: The implicit parameter ?callStack :: CallStack is an implementation detail and should not be considered part of the CallStack API, we may decide to change the implementation in the future.