1 -----------------------------------------------------------------------------
3 -- Module : Data.Traversable
4 -- Copyright : Conor McBride and Ross Paterson 2005
5 -- License : BSD-style (see the LICENSE file in the distribution)
7 -- Maintainer : libraries@haskell.org
8 -- Stability : experimental
9 -- Portability : portable
11 -- Class of data structures that can be traversed from left to right,
12 -- performing an action on each element.
16 -- * /Applicative Programming with Effects/,
17 -- by Conor McBride and Ross Paterson, online at
18 -- <http://www.soi.city.ac.uk/~ross/papers/Applicative.html>.
20 -- * /The Essence of the Iterator Pattern/,
21 -- by Jeremy Gibbons and Bruno Oliveira,
22 -- in /Mathematically-Structured Functional Programming/, 2006, and online at
23 -- <http://web.comlab.ox.ac.uk/oucl/work/jeremy.gibbons/publications/#iterator>.
25 -- Note that the functions 'mapM' and 'sequence' generalize "Prelude"
26 -- functions of the same names from lists to any 'Traversable' functor.
27 -- To avoid ambiguity, either import the "Prelude" hiding these names
28 -- or qualify uses of these function names with an alias for this module.
30 module Data.Traversable (
40 import Prelude hiding (mapM, sequence, foldr)
41 import qualified Prelude (mapM, foldr)
42 import Control.Applicative
43 import Data.Foldable (Foldable())
44 import Data.Monoid (Monoid)
46 -- | Functors representing data structures that can be traversed from
49 -- Minimal complete definition: 'traverse' or 'sequenceA'.
51 -- Instances are similar to 'Functor', e.g. given a data type
53 -- > data Tree a = Empty | Leaf a | Node (Tree a) a (Tree a)
55 -- a suitable instance would be
57 -- > instance Traversable Tree
58 -- > traverse f Empty = pure Empty
59 -- > traverse f (Leaf x) = Leaf <$> f x
60 -- > traverse f (Node l k r) = Node <$> traverse f l <*> f k <*> traverse f r
62 -- This is suitable even for abstract types, as the laws for '<*>'
63 -- imply a form of associativity.
65 -- The superclass instances should satisfy the following:
67 -- * In the 'Functor' instance, 'fmap' should be equivalent to traversal
68 -- with the identity applicative functor ('fmapDefault').
70 -- * In the 'Foldable' instance, 'Data.Foldable.foldMap' should be
71 -- equivalent to traversal with a constant applicative functor
72 -- ('foldMapDefault').
74 class (Functor t, Foldable t) => Traversable t where
75 -- | Map each element of a structure to an action, evaluate
76 -- these actions from left to right, and collect the results.
77 traverse :: Applicative f => (a -> f b) -> t a -> f (t b)
78 traverse f = sequenceA . fmap f
80 -- | Evaluate each action in the structure from left to right,
81 -- and collect the results.
82 sequenceA :: Applicative f => t (f a) -> f (t a)
83 sequenceA = traverse id
85 -- | Map each element of a structure to a monadic action, evaluate
86 -- these actions from left to right, and collect the results.
87 mapM :: Monad m => (a -> m b) -> t a -> m (t b)
88 mapM f = unwrapMonad . traverse (WrapMonad . f)
90 -- | Evaluate each monadic action in the structure from left to right,
91 -- and collect the results.
92 sequence :: Monad m => t (m a) -> m (t a)
95 -- instances for Prelude types
97 instance Traversable Maybe where
98 traverse f Nothing = pure Nothing
99 traverse f (Just x) = Just <$> f x
101 instance Traversable [] where
102 traverse f = Prelude.foldr cons_f (pure [])
103 where cons_f x ys = (:) <$> f x <*> ys
109 -- | 'for' is 'traverse' with its arguments flipped.
110 for :: (Traversable t, Applicative f) => t a -> (a -> f b) -> f (t b)
114 -- | 'forM' is 'mapM' with its arguments flipped.
115 forM :: (Traversable t, Monad m) => t a -> (a -> m b) -> m (t b)
119 -- left-to-right state transformer
120 newtype StateL s a = StateL { runStateL :: s -> (s, a) }
122 instance Functor (StateL s) where
123 fmap f (StateL k) = StateL $ \ s ->
124 let (s', v) = k s in (s', f v)
126 instance Applicative (StateL s) where
127 pure x = StateL (\ s -> (s, x))
128 StateL kf <*> StateL kv = StateL $ \ s ->
133 -- |The 'mapAccumL' function behaves like a combination of 'fmap'
134 -- and 'foldl'; it applies a function to each element of a structure,
135 -- passing an accumulating parameter from left to right, and returning
136 -- a final value of this accumulator together with the new structure.
137 mapAccumL :: Traversable t => (a -> b -> (a, c)) -> a -> t b -> (a, t c)
138 mapAccumL f s t = runStateL (traverse (StateL . flip f) t) s
140 -- right-to-left state transformer
141 newtype StateR s a = StateR { runStateR :: s -> (s, a) }
143 instance Functor (StateR s) where
144 fmap f (StateR k) = StateR $ \ s ->
145 let (s', v) = k s in (s', f v)
147 instance Applicative (StateR s) where
148 pure x = StateR (\ s -> (s, x))
149 StateR kf <*> StateR kv = StateR $ \ s ->
154 -- |The 'mapAccumR' function behaves like a combination of 'fmap'
155 -- and 'foldr'; it applies a function to each element of a structure,
156 -- passing an accumulating parameter from right to left, and returning
157 -- a final value of this accumulator together with the new structure.
158 mapAccumR :: Traversable t => (a -> b -> (a, c)) -> a -> t b -> (a, t c)
159 mapAccumR f s t = runStateR (traverse (StateR . flip f) t) s
161 -- | This function may be used as a value for `fmap` in a `Functor` instance.
162 fmapDefault :: Traversable t => (a -> b) -> t a -> t b
163 fmapDefault f = getId . traverse (Id . f)
165 -- | This function may be used as a value for `Data.Foldable.foldMap`
166 -- in a `Foldable` instance.
167 foldMapDefault :: (Traversable t, Monoid m) => (a -> m) -> t a -> m
168 foldMapDefault f = getConst . traverse (Const . f)
172 newtype Id a = Id { getId :: a }
174 instance Functor Id where
175 fmap f (Id x) = Id (f x)
177 instance Applicative Id where
179 Id f <*> Id x = Id (f x)