Idris2/libs/base/Data/SortedMap.idr
2023-09-23 22:47:05 +01:00

169 lines
4.6 KiB
Idris

module Data.SortedMap
import Data.SortedMap.Dependent
%hide Prelude.toList
export
record SortedMap k v where
constructor M
unM : SortedDMap k $ const v
-- Helper function
unDPair : (x : a ** const b x) -> (a, b)
unDPair (k ** v) = (k, v)
export
empty : Ord k => SortedMap k v
empty = M empty
export
lookup : k -> SortedMap k v -> Maybe v
lookup k = map snd . lookup k . unM
export
insert : k -> v -> SortedMap k v -> SortedMap k v
insert k v = M . insert k v . unM
export
singleton : Ord k => k -> v -> SortedMap k v
singleton = M .: singleton
export
insertFrom : Foldable f => f (k, v) -> SortedMap k v -> SortedMap k v
insertFrom = flip $ foldl $ flip $ uncurry insert
export
delete : k -> SortedMap k v -> SortedMap k v
delete k = M . delete k . unM
||| Updates or deletes a value based on the decision function
|||
||| The decision function takes information about the presence of the value,
||| and the value itself, if it is present.
||| It returns a new value or the fact that there should be no value as the result.
|||
||| The current implementation performs up to two traversals of the original map
export
update : (Maybe v -> Maybe v) -> k -> SortedMap k v -> SortedMap k v
update f k m = case f $ lookup k m of
Just v => insert k v m
Nothing => delete k m
||| Updates existing value, if it is present, and does nothing otherwise
|||
||| The current implementation performs up to two traversals of the original map
export
updateExisting : (v -> v) -> k -> SortedMap k v -> SortedMap k v
updateExisting f k m = case lookup k m of
Just v => insert k (f v) m
Nothing => m
export
fromList : Ord k => List (k, v) -> SortedMap k v
fromList = flip insertFrom empty
export
toList : SortedMap k v -> List (k, v)
toList = map unDPair . toList . unM
||| Gets the keys of the map.
export
keys : SortedMap k v -> List k
keys = map fst . toList
||| Gets the values of the map. Could contain duplicates.
export
values : SortedMap k v -> List v
values = map snd . toList
export
implementation Functor (SortedMap k) where
map f = M . map f . unM
export
implementation Foldable (SortedMap k) where
foldr f z = foldr f z . values
foldl f z = foldl f z . values
null = null . unM
foldMap f = foldMap f . values
export
implementation Traversable (SortedMap k) where
traverse f = map M . traverse f . unM
||| Merge two maps. When encountering duplicate keys, using a function to combine the values.
||| Uses the ordering of the first map given.
export
mergeWith : (v -> v -> v) -> SortedMap k v -> SortedMap k v -> SortedMap k v
mergeWith f x y = insertFrom inserted x where
inserted : List (k, v)
inserted = do
(k, v) <- toList y
let v' = (maybe id f $ lookup k x) v
pure (k, v')
||| Merge two maps using the Semigroup (and by extension, Monoid) operation.
||| Uses mergeWith internally, so the ordering of the left map is kept.
export
merge : Semigroup v => SortedMap k v -> SortedMap k v -> SortedMap k v
merge = mergeWith (<+>)
||| Left-biased merge, also keeps the ordering specified by the left map.
export
mergeLeft : SortedMap k v -> SortedMap k v -> SortedMap k v
mergeLeft = mergeWith const
||| looks up a key in map, returning the left and right closest values, so that
||| k1 <= k < k2. If at the end of the beginning and/or end of the sorted map, returns
||| nothing appropriately
export
lookupBetween : key -> SortedMap key val -> (Maybe (key,val), Maybe (key,val))
lookupBetween k = bimap (map unDPair) (map unDPair) . lookupBetween k . unM
||| Returns the leftmost (least) key and value
export
leftMost : SortedMap key val -> Maybe (key,val)
leftMost = map unDPair . leftMost . unM
||| Returns the rightmost (greatest) key and value
export
rightMost : SortedMap key val -> Maybe (key,val)
rightMost = map unDPair . rightMost . unM
export
(Show k, Show v) => Show (SortedMap k v) where
show m = "fromList " ++ (show $ toList m)
export
(Eq k, Eq v) => Eq (SortedMap k v) where
(==) = (==) `on` toList
-- TODO: is this the right variant of merge to use for this? I think it is, but
-- I could also see the advantages of using `mergeLeft`. The current approach is
-- strictly more powerful I believe, because `mergeLeft` can be emulated with
-- the `First` monoid. However, this does require more code to do the same
-- thing.
export
Semigroup v => Semigroup (SortedMap k v) where
(<+>) = merge
||| For `neutral <+> y`, y is rebuilt in `Ord k`, so this is not a "strict" Monoid.
||| However, semantically, it should be equal.
export
(Ord k, Semigroup v) => Monoid (SortedMap k v) where
neutral = empty
export %inline
Cast (SortedDMap k (const v)) (SortedMap k v) where
cast = M
export %inline
Cast (SortedMap k v) (SortedDMap k (const v)) where
cast = unM