diff --git a/Data/SearchEngine/DocIdSet.hs b/Data/SearchEngine/DocIdSet.hs --- a/Data/SearchEngine/DocIdSet.hs +++ b/Data/SearchEngine/DocIdSet.hs @@ -1,6 +1,7 @@ -{-# LANGUAGE BangPatterns, GeneralizedNewtypeDeriving #-} +{-# LANGUAGE BangPatterns, GeneralizedNewtypeDeriving, MultiParamTypeClasses, + TypeFamilies #-} module Data.SearchEngine.DocIdSet ( - DocId(..), + DocId(DocId), DocIdSet(..), null, size, @@ -19,10 +20,10 @@ import Data.Word import qualified Data.Vector.Unboxed as Vec import qualified Data.Vector.Unboxed.Mutable as MVec -import qualified Data.Vector.Generic.Base as VecGen -import qualified Data.Vector.Unboxed.Base as VecBase -import qualified Data.Vector.Generic.Mutable as VecMut +import qualified Data.Vector.Generic as GVec +import qualified Data.Vector.Generic.Mutable as GMVec import Control.Monad.ST +import Control.Monad (liftM) import qualified Data.Set as Set import Data.List (foldl', sortBy) import Data.Function (on) @@ -30,10 +31,8 @@ import Prelude hiding (null) -newtype DocId = DocId Word32 - deriving (Eq, Ord, Show, Enum, Bounded, Vec.Unbox, - VecGen.Vector VecBase.Vector, - VecMut.MVector VecBase.MVector) +newtype DocId = DocId { unDocId :: Word32 } + deriving (Eq, Ord, Show, Enum, Bounded) newtype DocIdSet = DocIdSet (Vec.Vector DocId) deriving (Eq, Show) @@ -147,4 +146,57 @@ EQ -> do MVec.write out i x go (Vec.tail xs) (Vec.tail ys) (i+1) LT -> go (Vec.tail xs) ys i + +------------------------------------------------------------------------------ +-- verbose Unbox instances +-- + +instance MVec.Unbox DocId + +newtype instance MVec.MVector s DocId = MV_DocId (MVec.MVector s Word32) + +instance GMVec.MVector MVec.MVector DocId where + basicLength (MV_DocId v) = GMVec.basicLength v + basicUnsafeSlice i l (MV_DocId v) = MV_DocId (GMVec.basicUnsafeSlice i l v) + basicUnsafeNew l = MV_DocId `liftM` GMVec.basicUnsafeNew l + basicUnsafeReplicate l x = MV_DocId `liftM` GMVec.basicUnsafeReplicate l (unDocId x) + basicUnsafeRead (MV_DocId v) i = DocId `liftM` GMVec.basicUnsafeRead v i + basicUnsafeWrite (MV_DocId v) i x = GMVec.basicUnsafeWrite v i (unDocId x) + basicClear (MV_DocId v) = GMVec.basicClear v + basicSet (MV_DocId v) x = GMVec.basicSet v (unDocId x) + basicUnsafeGrow (MV_DocId v) l = MV_DocId `liftM` GMVec.basicUnsafeGrow v l + basicUnsafeCopy (MV_DocId v) (MV_DocId v') = GMVec.basicUnsafeCopy v v' + basicUnsafeMove (MV_DocId v) (MV_DocId v') = GMVec.basicUnsafeMove v v' + basicOverlaps (MV_DocId v) (MV_DocId v') = GMVec.basicOverlaps v v' + {-# INLINE basicLength #-} + {-# INLINE basicUnsafeSlice #-} + {-# INLINE basicOverlaps #-} + {-# INLINE basicUnsafeNew #-} + {-# INLINE basicUnsafeReplicate #-} + {-# INLINE basicUnsafeRead #-} + {-# INLINE basicUnsafeWrite #-} + {-# INLINE basicClear #-} + {-# INLINE basicSet #-} + {-# INLINE basicUnsafeCopy #-} + {-# INLINE basicUnsafeMove #-} + {-# INLINE basicUnsafeGrow #-} + +newtype instance Vec.Vector DocId = V_DocId (Vec.Vector Word32) + +instance GVec.Vector Vec.Vector DocId where + basicUnsafeFreeze (MV_DocId mv) = V_DocId `liftM` GVec.basicUnsafeFreeze mv + basicUnsafeThaw (V_DocId v) = MV_DocId `liftM` GVec.basicUnsafeThaw v + basicLength (V_DocId v) = GVec.basicLength v + basicUnsafeSlice i l (V_DocId v) = V_DocId (GVec.basicUnsafeSlice i l v) + basicUnsafeIndexM (V_DocId v) i = DocId `liftM` GVec.basicUnsafeIndexM v i + basicUnsafeCopy (MV_DocId mv) + (V_DocId v) = GVec.basicUnsafeCopy mv v + elemseq (V_DocId v) x = GVec.elemseq v (unDocId x) + {-# INLINE basicUnsafeFreeze #-} + {-# INLINE basicUnsafeThaw #-} + {-# INLINE basicLength #-} + {-# INLINE basicUnsafeSlice #-} + {-# INLINE basicUnsafeIndexM #-} + {-# INLINE basicUnsafeCopy #-} + {-# INLINE elemseq #-} diff --git a/Data/SearchEngine/TermBag.hs b/Data/SearchEngine/TermBag.hs --- a/Data/SearchEngine/TermBag.hs +++ b/Data/SearchEngine/TermBag.hs @@ -1,6 +1,7 @@ -{-# LANGUAGE BangPatterns, GeneralizedNewtypeDeriving #-} +{-# LANGUAGE BangPatterns, GeneralizedNewtypeDeriving, MultiParamTypeClasses, + TypeFamilies #-} module Data.SearchEngine.TermBag ( - TermId(..), TermCount, + TermId(TermId), TermCount, TermBag, size, fromList, @@ -13,20 +14,18 @@ import qualified Data.Vector.Unboxed as Vec import qualified Data.Vector.Unboxed.Mutable as MVec -import qualified Data.Vector.Generic.Base as VecGen -import qualified Data.Vector.Unboxed.Base as VecBase -import qualified Data.Vector.Generic.Mutable as VecMut +import qualified Data.Vector.Generic as GVec +import qualified Data.Vector.Generic.Mutable as GMVec import Control.Monad.ST +import Control.Monad (liftM) import qualified Data.Map as Map import Data.Word (Word32, Word8) import Data.Bits import Data.List (sortBy, foldl') import Data.Function (on) -newtype TermId = TermId Word32 - deriving (Eq, Ord, Show, Enum, - Vec.Unbox, VecGen.Vector VecBase.Vector, - VecMut.MVector VecBase.MVector) +newtype TermId = TermId { unTermId :: Word32 } + deriving (Eq, Ord, Show, Enum) instance Bounded TermId where minBound = TermId 0 @@ -207,3 +206,57 @@ go (Vec.tail xs) (Vec.tail ys) (i+1) LT -> do MVec.write out i x go (Vec.tail xs) ys (i+1) + +------------------------------------------------------------------------------ +-- verbose Unbox instances +-- + +instance MVec.Unbox TermId + +newtype instance MVec.MVector s TermId = MV_TermId (MVec.MVector s Word32) + +instance GMVec.MVector MVec.MVector TermId where + basicLength (MV_TermId v) = GMVec.basicLength v + basicUnsafeSlice i l (MV_TermId v) = MV_TermId (GMVec.basicUnsafeSlice i l v) + basicUnsafeNew l = MV_TermId `liftM` GMVec.basicUnsafeNew l + basicUnsafeReplicate l x = MV_TermId `liftM` GMVec.basicUnsafeReplicate l (unTermId x) + basicUnsafeRead (MV_TermId v) i = TermId `liftM` GMVec.basicUnsafeRead v i + basicUnsafeWrite (MV_TermId v) i x = GMVec.basicUnsafeWrite v i (unTermId x) + basicClear (MV_TermId v) = GMVec.basicClear v + basicSet (MV_TermId v) x = GMVec.basicSet v (unTermId x) + basicUnsafeGrow (MV_TermId v) l = MV_TermId `liftM` GMVec.basicUnsafeGrow v l + basicUnsafeCopy (MV_TermId v) (MV_TermId v') = GMVec.basicUnsafeCopy v v' + basicUnsafeMove (MV_TermId v) (MV_TermId v') = GMVec.basicUnsafeMove v v' + basicOverlaps (MV_TermId v) (MV_TermId v') = GMVec.basicOverlaps v v' + {-# INLINE basicLength #-} + {-# INLINE basicUnsafeSlice #-} + {-# INLINE basicOverlaps #-} + {-# INLINE basicUnsafeNew #-} + {-# INLINE basicUnsafeReplicate #-} + {-# INLINE basicUnsafeRead #-} + {-# INLINE basicUnsafeWrite #-} + {-# INLINE basicClear #-} + {-# INLINE basicSet #-} + {-# INLINE basicUnsafeCopy #-} + {-# INLINE basicUnsafeMove #-} + {-# INLINE basicUnsafeGrow #-} + +newtype instance Vec.Vector TermId = V_TermId (Vec.Vector Word32) + +instance GVec.Vector Vec.Vector TermId where + basicUnsafeFreeze (MV_TermId mv) = V_TermId `liftM` GVec.basicUnsafeFreeze mv + basicUnsafeThaw (V_TermId v) = MV_TermId `liftM` GVec.basicUnsafeThaw v + basicLength (V_TermId v) = GVec.basicLength v + basicUnsafeSlice i l (V_TermId v) = V_TermId (GVec.basicUnsafeSlice i l v) + basicUnsafeIndexM (V_TermId v) i = TermId `liftM` GVec.basicUnsafeIndexM v i + basicUnsafeCopy (MV_TermId mv) + (V_TermId v) = GVec.basicUnsafeCopy mv v + elemseq (V_TermId v) x = GVec.elemseq v (unTermId x) + {-# INLINE basicUnsafeFreeze #-} + {-# INLINE basicUnsafeThaw #-} + {-# INLINE basicLength #-} + {-# INLINE basicUnsafeSlice #-} + {-# INLINE basicUnsafeIndexM #-} + {-# INLINE basicUnsafeCopy #-} + {-# INLINE elemseq #-} + diff --git a/changelog b/changelog --- a/changelog +++ b/changelog @@ -1,3 +1,9 @@ +0.2.1.3 Duncan Coutts May 2015 + * Compatibility with GHC 7.10 (no API changes) + +0.2.1.2 Adam Gundry July 2014 + * Compatibility with GHC 7.8 (no API changes) + 0.2.1.1 Duncan Coutts May 2014 * Fix for NaN scores with a doc field that is empty for all docs * Add instance Show NoFeatures diff --git a/full-text-search.cabal b/full-text-search.cabal --- a/full-text-search.cabal +++ b/full-text-search.cabal @@ -1,5 +1,5 @@ name: full-text-search -version: 0.2.1.1 +version: 0.2.1.3 synopsis: In-memory full text search engine description: An in-memory full text search engine library. It lets you run full-text queries on a collection of your documents. @@ -77,11 +77,11 @@ RecordWildCards, GeneralizedNewtypeDeriving, ScopedTypeVariables - build-depends: base >=4.5 && <4.7, - array >=0.4 && <0.5, + build-depends: base >=4.5 && <4.9, + array >=0.4 && <0.6, vector >=0.10 && <0.11, containers >=0.4 && <0.6, - text >=0.11 && <1.2 + text >=0.11 && <1.3 default-language: Haskell2010 ghc-options: -Wall -funbox-strict-fields