indexation 0.4.6 → 0.4.7

raw patch · 9 files changed

+64/−3 lines, 9 filesdep +cereal-vectordep +foldldep +vector-algorithmsPVP ok

version bump matches the API change (PVP)

Dependencies added: cereal-vector, foldl, vector-algorithms

API changes (from Hackage documentation)

Files

• indexation.cabal +7/−2
• library/Indexation/Data.hs +1/−1
• library/Indexation/FoldM/Basic.hs +14/−0
• library/Indexation/FoldM/Index.hs +12/−0
• library/Indexation/Functions.hs +18/−0
• library/Indexation/Instances.hs +4/−0
• library/Indexation/Instances/Cereal.hs +2/−0
• library/Indexation/Prelude.hs +4/−0
• library/Indexation/Types.hs +2/−0

@@ -1,5 +1,5 @@ name: indexation-version: 0.4.6+version: 0.4.7 category: Data synopsis: Tools for entity indexation description: A set of tools for indexing entities@@ -23,6 +23,7 @@     Indexation.Predicate     Indexation.Potoki.Produce     Indexation.Potoki.Transform+    Indexation.FoldM.Index   other-modules:     Indexation.Cereal.Get     Indexation.Cereal.Put@@ -32,6 +33,7 @@     Indexation.Prelude     Indexation.Types     Indexation.Functions+    Indexation.FoldM.Basic     Indexation.Utils.Vector     Indexation.Utils.UnboxedVector     Indexation.Utils.Unfoldr@@ -39,9 +41,11 @@     base >=4.11 && <5,     bytestring >=0.10.8 && <0.11,     cereal >=0.5.5 && <0.6,+    cereal-vector >=0.2.0.1 && <0.3,     contravariant >=1.4 && <2,     deferred-folds >=0.9 && <0.10,     focus >=1.0.1 && <1.1,+    foldl >=1 && <2,     hashable >=1 && <2,     list-t >=1 && <1.1,     mmorph >=1 && <2,@@ -52,4 +56,5 @@     text >=1 && <2,     transformers >=0.5 && <0.6,     unordered-containers >=0.2.9 && <0.3,-    vector >=0.12 && <0.13+    vector >=0.12 && <0.13,+    vector-algorithms >=0.7.0.4 && <0.8

@@ -4,6 +4,6 @@ ) where -import Indexation.Types as Exports (Index(..), EntityTable, IndexTable, Indexer, IndexSet)+import Indexation.Types as Exports (Index(..), EntityTable, IndexTable, Indexer, IndexSet, IndexCounts) import Indexation.Instances () import Indexation.Functions as Exports

@@ -0,0 +1,14 @@+module Indexation.FoldM.Basic+where++import Indexation.Prelude+import Control.Foldl+import qualified Data.Vector.Unboxed as UnboxedVector+import qualified Data.Vector.Unboxed.Mutable as MutableUnboxedVector+++countIndices :: Int -> FoldM IO Int (UnboxedVector.Vector Word32)+countIndices amount = FoldM step init extract where+  init = MutableUnboxedVector.new amount+  step mv index = MutableUnboxedVector.modify mv succ index $> mv+  extract = UnboxedVector.unsafeFreeze

@@ -0,0 +1,12 @@+module Indexation.FoldM.Index+where++import Indexation.Prelude+import Indexation.Types+import Indexation.Data+import Control.Foldl+import qualified Indexation.FoldM.Basic as Basic+++indexCounts :: Int -> FoldM IO (Index a) (IndexCounts a)+indexCounts amount = dimap (\ (Index x) -> x) IndexCounts (Basic.countIndices amount)

@@ -7,6 +7,8 @@ import qualified Data.Vector as Vector import qualified Data.HashMap.Strict as HashMap import qualified Data.Vector.Unboxed as UnboxedVector+import qualified Data.Vector.Unboxed.Mutable as MutableUnboxedVector+import qualified Data.Vector.Algorithms.Intro as IntroVectorAlgorithm import qualified Indexation.Utils.UnboxedVector as UnboxedVector import qualified Indexation.Utils.Unfoldr as Unfoldr @@ -28,3 +30,19 @@  mergeIndexSets :: IndexSet entity -> IndexSet entity -> IndexSet entity mergeIndexSets (IndexSet vec1) (IndexSet vec2) = IndexSet $ UnboxedVector.zipWith (||) vec1 vec2++topCountedIndexSet :: Int -> IndexCounts a -> IndexSet a+topCountedIndexSet amount (IndexCounts countVec) = let+  countVecLength = UnboxedVector.length countVec+  limitedAmount = min amount countVecLength+  in runST $ do+    pairMVec <- UnboxedVector.unsafeThaw (UnboxedVector.imap (\ index count -> (count, index)) countVec)+    IntroVectorAlgorithm.selectBy (\ a b -> compare (fst b) (fst a)) pairMVec limitedAmount+    indexSetMVec <- MutableUnboxedVector.new countVecLength+    forM_ [0..(pred limitedAmount)] $ \ pairIndex -> do+      (_, index) <- MutableUnboxedVector.unsafeRead pairMVec pairIndex+      MutableUnboxedVector.write indexSetMVec index True+    IndexSet <$> UnboxedVector.unsafeFreeze indexSetMVec++indexSetByMinCount :: Word32 -> IndexCounts a -> IndexSet a+indexSetByMinCount min (IndexCounts countVec) = IndexSet (UnboxedVector.map (>= min) countVec)

@@ -4,9 +4,13 @@ import Indexation.Prelude import Indexation.Types import Indexation.Instances.Cereal ()+import qualified Data.Vector.Unboxed as UnboxedVector   instance Show (Index a) where show (Index int) = show int deriving instance Eq (Index a) deriving instance Ord (Index a) deriving instance Hashable (Index a)++instance Show (IndexSet a) where+  show = show . UnboxedVector.ifoldr (\ i -> bool id (i :)) [] . (\ (IndexSet x) -> x)

@@ -24,3 +24,5 @@ instance Serialize (IndexSet a) where   get = IndexSet <$> Get.getVector get   put (IndexSet vector) = Put.putVector put vector++deriving instance Serialize (IndexCounts a)

@@ -107,6 +107,10 @@ ------------------------- import Data.Serialize as Exports (Serialize) +-- cereal-vector+-------------------------+import Data.Vector.Serialize ()+ -- bytestring ------------------------- import Data.ByteString as Exports (ByteString)

@@ -19,3 +19,5 @@ A more efficient alternative to @HashSet (Index entity)@. -} newtype IndexSet entity = IndexSet (UnboxedVector.Vector Bool)++newtype IndexCounts entity = IndexCounts (UnboxedVector.Vector Word32)