1
1
mirror of https://github.com/github/semantic.git synced 2024-11-23 16:37:50 +03:00

Merge branch 'master' into syntax-redux

This commit is contained in:
Rick Winfrey 2016-07-12 17:03:35 -05:00
commit 177af03bd3
13 changed files with 192 additions and 110 deletions

View File

@ -14,4 +14,6 @@ error "Avoid return" =
where note = "return is obsolete as of GHC 7.10"
error "use pure" = free . Pure ==> pure
error "use wrap" = free . Free ==> wrap
error "use extract" = headF . runCofree ==> extract

View File

@ -1,25 +1,41 @@
{-# LANGUAGE DeriveAnyClass, DeriveGeneric, FlexibleInstances, StandaloneDeriving #-}
{-# OPTIONS_GHC -fno-warn-orphans #-}
module Main where
import Criterion.Main
import Data.Function
import Data.List (genericLength)
import Data.String
import Patch
import Prologue
import SES
import Test.QuickCheck hiding (Fixed)
main :: IO ()
main = do
benchmarks <- sequenceA []
benchmarks <- sequenceA [ generativeBenchmarkWith "ses" 10 arbitrarySESInputs (uncurry ((*) `on` length)) (nf (uncurry benchmarkSES)) ]
defaultMain benchmarks
where arbitrarySESInputs = (,) <$> sized (`vectorOf` arbitrary) <*> sized (`vectorOf` arbitrary)
benchmarkSES :: [String] -> [String] -> [Either String (Patch String)]
benchmarkSES as bs = ses compare cost as bs
where compare a b = if a == b then Just (Left a) else Nothing
cost = either (const 0) (sum . fmap genericLength)
instance NFData a => NFData (Patch a)
-- | Defines a named group of n benchmarks over inputs generated by an `Arbitrary` instance.
-- |
-- | The inputs sizes at each iteration are measured by a metric function, which gives the name of the benchmark. This makes it convenient to correlate a benchmark of some function over lists with e.g. input `length`.
generativeBenchmark :: (Arbitrary a, Show m, Ord m) => String -> Int -> (a -> m) -> (a -> Benchmarkable) -> IO Benchmark
generativeBenchmark name n metric benchmark = do
benchmarks <- traverse measure (replicate n defaultSize)
generativeBenchmark name n metric benchmark = generativeBenchmarkWith name n arbitrary metric benchmark
generativeBenchmarkWith :: (Show m, Ord m) => String -> Int -> Gen a -> (a -> m) -> (a -> Benchmarkable) -> IO Benchmark
generativeBenchmarkWith name n generator metric benchmark = do
benchmarks <- traverse measure (take n [0,(defaultSize `div` n)..defaultSize])
pure $! bgroup name (snd <$> (sortOn fst benchmarks))
where measure n = do
input <- generate (resize n arbitrary)
input <- generate (resize n generator)
let measurement = metric input
pure $! (measurement, bench (show measurement) (benchmark input))
defaultSize = 100

View File

@ -16,6 +16,7 @@ library
exposed-modules: Algorithm
, Alignment
, Category
, Data.Align.Generic
, Data.Bifunctor.Join.Arbitrary
, Data.Functor.Both
, Data.RandomWalkSimilarity

70
src/Data/Align/Generic.hs Normal file
View File

@ -0,0 +1,70 @@
{-# LANGUAGE DefaultSignatures, TypeOperators #-}
module Data.Align.Generic where
import Control.Monad
import Data.Align
import Data.These
import GHC.Generics
import Data.OrderedMap
import Prologue
import Syntax
-- | Functors which can be aligned (structure-unioning-ly zipped). The default implementation will operate generically over the constructors in the aligning type.
class Functor f => GAlign f where
galign :: f a -> f b -> Maybe (f (These a b))
default galign :: (Generic1 f, GAlign (Rep1 f)) => f a -> f b -> Maybe (f (These a b))
galign a b = to1 <$> galign (from1 a) (from1 b)
-- Generically-derived instances
instance Eq a => GAlign (Syntax a)
-- 'Data.Align.Align' instances
instance GAlign [] where galign = galignAlign
instance GAlign Maybe where galign = galignAlign
instance Eq key => GAlign (OrderedMap key) where galign = galignAlign
-- | Implements a function suitable for use as the definition of 'galign' for 'Align'able functors.
galignAlign :: Align f => f a -> f b -> Maybe (f (These a b))
galignAlign a = Just . align a
-- Generics
-- | 'GAlign' over unit constructors.
instance GAlign U1 where
galign _ _ = Just U1
-- | 'GAlign' over parameters.
instance GAlign Par1 where
galign (Par1 a) (Par1 b) = Just (Par1 (These a b))
-- | 'GAlign' over non-parameter fields. Only equal values are aligned.
instance Eq c => GAlign (K1 i c) where
galign (K1 a) (K1 b) = guard (a == b) >> Just (K1 b)
-- | 'GAlign' over applications over parameters.
instance GAlign f => GAlign (Rec1 f) where
galign (Rec1 a) (Rec1 b) = Rec1 <$> galign a b
-- | 'GAlign' over metainformation (constructor names, etc).
instance GAlign f => GAlign (M1 i c f) where
galign (M1 a) (M1 b) = M1 <$> galign a b
-- | 'GAlign' over sums. Returns 'Nothing' for disjoint constructors.
instance (GAlign f, GAlign g) => GAlign (f :+: g) where
galign a b = case (a, b) of
(L1 a, L1 b) -> L1 <$> galign a b
(R1 a, R1 b) -> R1 <$> galign a b
_ -> Nothing
-- | 'GAlign' over products.
instance (GAlign f, GAlign g) => GAlign (f :*: g) where
galign (a1 :*: b1) (a2 :*: b2) = (:*:) <$> galign a1 a2 <*> galign b1 b2
-- | 'GAlign' over type compositions.
instance (Traversable f, Applicative f, GAlign g) => GAlign (f :.: g) where
galign (Comp1 a) (Comp1 b) = Comp1 <$> sequenceA (galign <$> a <*> b)

View File

@ -1,4 +1,5 @@
{-# LANGUAGE GeneralizedNewtypeDeriving #-}
{-# OPTIONS_GHC -fno-warn-warnings-deprecations #-}
module Data.OrderedMap (
OrderedMap
, fromList
@ -14,12 +15,15 @@ module Data.OrderedMap (
, difference
) where
import Data.Align
import Data.These
import GHC.Generics
import Prologue hiding (toList, empty)
import Test.QuickCheck
-- | An ordered map of keys and values.
newtype OrderedMap key value = OrderedMap { toList :: [(key, value)] }
deriving (Eq, Foldable, Functor, Generic, Ord, Show, Traversable)
deriving (Eq, Foldable, Functor, Generic, Generic1, Ord, Show, Traversable)
-- | Construct an ordered map from a list of pairs of keys and values.
fromList :: [(key, value)] -> OrderedMap key value
@ -33,7 +37,7 @@ infixl 9 !
-- | Look up a value in the map by key, erroring if it doesn't exist.
(!) :: Eq key => OrderedMap key value -> key -> value
map ! key = fromMaybe (error "no value found for key") $ Data.OrderedMap.lookup key map
m ! key = fromMaybe (error "no value found for key") $ Data.OrderedMap.lookup key m
-- | Look up a value in the map by key, returning Nothing if it doesn't exist.
lookup :: Eq key => key -> OrderedMap key value -> Maybe value
@ -49,7 +53,7 @@ empty = OrderedMap []
-- | Combine `a` and `b`, picking the values from `a` when keys overlap.
union :: Eq key => OrderedMap key value -> OrderedMap key value -> OrderedMap key value
union a b = OrderedMap $ toList a ++ toList (difference b a)
union a b = OrderedMap $ toList a <> toList (difference b a)
-- | Union a list of ordered maps.
unions :: Eq key => [OrderedMap key value] -> OrderedMap key value
@ -75,3 +79,7 @@ instance Eq key => Monoid (OrderedMap key value) where
instance (Arbitrary key, Arbitrary value) => Arbitrary (OrderedMap key value) where
arbitrary = fromList <$> arbitrary
shrink = genericShrink
instance Eq key => Align (OrderedMap key) where
nil = fromList []
align a b = intersectionWith These a b <> (This <$> difference a b) <> (That <$> difference b a)

View File

@ -3,7 +3,6 @@ module Data.RandomWalkSimilarity where
import Control.Arrow ((&&&))
import Control.Monad.Random
import Control.Monad.State
import Data.Bifunctor.Join
import qualified Data.DList as DList
import Data.Functor.Foldable as Foldable
import Data.Hashable
@ -20,35 +19,34 @@ import Test.QuickCheck hiding (Fixed)
import Test.QuickCheck.Random
-- | Given a function comparing two terms recursively, and a function to compute a Hashable label from an annotation, compute the diff of a pair of lists of terms using a random walk similarity metric, which completes in log-linear time. This implementation is based on the paper [_RWS-Diff—Flexible and Efficient Change Detection in Hierarchical Data_](https://github.com/github/semantic-diff/files/325837/RWS-Diff.Flexible.and.Efficient.Change.Detection.in.Hierarchical.Data.pdf).
rws :: (Hashable label, Hashable leaf, Eq leaf, Ord annotation) => (Term leaf annotation -> Term leaf annotation -> Maybe (Diff leaf annotation)) -> (annotation -> label) -> [Term leaf annotation] -> [Term leaf annotation] -> [Diff leaf annotation]
rws :: (Hashable label, Hashable leaf, Eq leaf, Eq annotation) => (Term leaf annotation -> Term leaf annotation -> Maybe (Diff leaf annotation)) -> (annotation -> label) -> [Term leaf annotation] -> [Term leaf annotation] -> [Diff leaf annotation]
rws compare getLabel as bs
| null as, null bs = []
| null as = insert <$> bs
| null bs = delete <$> as
| otherwise = uncurry deleteRemaining . (`runState` fas) $ traverse findNearestNeighbourTo fbs
| otherwise = fmap snd . uncurry deleteRemaining . (`runState` (negate 1, fas)) $ traverse findNearestNeighbourTo fbs
where insert = pure . Insert
delete = pure . Delete
replace = (pure .) . Replace
(p, q, d) = (2, 2, 15)
fas = featurize <$> as
fbs = featurize <$> bs
kdas = KdTree.build (Vector.toList . fst) fas
featurize = featureVector d . pqGrams p q getLabel &&& identity
findNearestNeighbourTo kv@(_, v) = do
unmapped <- get
let (k, _) = KdTree.nearest kdas kv
case k `List.lookup` unmapped of
Nothing -> pure $! insert v
Just found -> do
put (List.delete (k, found) unmapped)
pure $! fromMaybe (replace found v) (compare found v)
deleteRemaining diffs unmapped = foldl' (flip (List.insertBy (comparing firstAnnotation))) diffs (delete . snd <$> unmapped)
fas = zipWith featurize [0..] as
fbs = zipWith featurize [0..] bs
kdas = KdTree.build (Vector.toList . feature) fas
featurize index term = UnmappedTerm index (featureVector d (pqGrams p q getLabel term)) term
findNearestNeighbourTo kv@(UnmappedTerm _ _ v) = do
(previous, unmapped) <- get
let (UnmappedTerm i _ _) = KdTree.nearest kdas kv
fromMaybe (pure (negate 1, insert v)) $ do
found <- find ((== i) . termIndex) unmapped
guard (i >= previous)
compared <- compare (term found) v
pure $! do
put (i, List.delete found unmapped)
pure (i, compared)
deleteRemaining diffs (_, unmapped) = foldl' (flip (List.insertBy (comparing fst))) diffs ((termIndex &&& delete . term) <$> unmapped)
-- | Extract the annotation for the before state of a diff node. This is returned in `Maybe` because e.g. an `Insert` patch does not have an annotation for the before state.
firstAnnotation :: Diff leaf annotation -> Maybe annotation
firstAnnotation diff = case runFree diff of
Free (annotations :< _) -> Just (fst (runJoin annotations))
Pure patch -> maybeFst (unPatch $ extract <$> patch)
-- | A term which has not yet been mapped by `rws`, along with its feature vector summary & index.
data UnmappedTerm leaf annotation = UnmappedTerm { termIndex :: {-# UNPACK #-} !Int, feature :: !(Vector.Vector Double), term :: !(Term leaf annotation) }
deriving Eq
-- | A `Gram` is a fixed-size view of some portion of a tree, consisting of a `stem` of _p_ labels for parent nodes, and a `base` of _q_ labels of sibling nodes. Collectively, the bag of `Gram`s for each node of a tree (e.g. as computed by `pqGrams`) form a summary of the tree.
data Gram label = Gram { stem :: [Maybe label], base :: [Maybe label] }

View File

@ -3,6 +3,7 @@ module Interpreter (Comparable, DiffConstructor, diffTerms) where
import Algorithm
import qualified Category as C
import Data.Align
import Data.Align.Generic
import Data.Functor.Foldable
import Data.Functor.Both hiding (fst, snd)
import Data.Hashable
@ -34,16 +35,14 @@ diffTerms construct comparable cost a b = fromMaybe (pure $ Replace a b) $ const
-- | Constructs an algorithm and runs it
constructAndRun :: (Show leaf, Show (Record fields), Eq leaf, Hashable leaf, Ord (Record fields), HasField fields C.Category) => DiffConstructor leaf (Record fields) -> Comparable leaf (Record fields) -> SES.Cost (Diff leaf (Record fields)) -> Term leaf (Record fields) -> Term leaf (Record fields) -> Maybe (Diff leaf (Record fields))
constructAndRun _ comparable _ a b | not $ comparable a b = Nothing
constructAndRun construct _ _ a b | (() <$ a) == (() <$ b) = hylo construct runCofree <$> zipTerms a b
constructAndRun construct comparable cost t1 t2 =
run construct comparable cost $ algorithm a b where
algorithm (Indexed a') (Indexed b') = free . Free $ ByIndex a' b' (annotate . Indexed)
algorithm (Keyed a') (Keyed b') = free . Free $ ByKey a' b' (annotate . Keyed)
constructAndRun construct comparable cost t1 t2
| not $ comparable t1 t2 = Nothing
| (() <$ t1) == (() <$ t2) = hylo construct runCofree <$> zipTerms t1 t2
| otherwise = run construct comparable cost $ algorithm a b where
algorithm (Indexed a') (Indexed b') = wrap $! ByIndex a' b' (annotate . Indexed)
algorithm (Keyed a') (Keyed b') = wrap $! ByKey a' b' (annotate . Keyed)
algorithm (Leaf a') (Leaf b') | a' == b' = annotate $ Leaf b'
algorithm a' b' = free . Free $ Recursive (cofree (annotation1 :< a')) (cofree (annotation2 :< b')) pure
algorithm a' b' = wrap $! Recursive (cofree (annotation1 :< a')) (cofree (annotation2 :< b')) pure
(annotation1 :< a, annotation2 :< b) = (runCofree t1, runCofree t2)
annotate = pure . construct . (both annotation1 annotation2 :<)
@ -55,36 +54,8 @@ run construct comparable cost algorithm = case runFree algorithm of
(annotation1 :< a, annotation2 :< b) = (runCofree t1, runCofree t2)
annotate = construct . (both annotation1 annotation2 :<)
diffTerms' = diffTerms construct comparable cost
recur (Indexed a') (Indexed b') = annotate . Indexed $ alignWith diffThese a' b'
recur (Fixed a') (Fixed b') = annotate . Fixed $ alignWith diffThese a' b'
recur (Keyed a') (Keyed b') | Map.keys a' == bKeys = annotate . Keyed . Map.fromList . fmap repack $ bKeys where
bKeys = Map.keys b'
repack key = (key, interpretInBoth key a' b')
interpretInBoth key x y = diffTerms' (x ! key) (y ! key)
recur (FunctionCall a' as') (FunctionCall b' bs') | length as' == length bs' = annotate $ FunctionCall (diffTerms' a' b') (zipWith diffTerms' as' bs')
recur (Function a' as' aExprs') (Function b' bs' bExprs') = annotate $ Function (liftA2 diffTerms' a' b') (liftA2 diffTerms' as' bs') (diffTerms' aExprs' bExprs')
recur (Assignment a' as') (Assignment b' bs') = annotate $ Assignment (diffTerms' a' b') (diffTerms' as' bs')
recur (MathAssignment a' as') (MathAssignment b' bs') = annotate $ MathAssignment (diffTerms' a' b') (diffTerms' as' bs')
recur (MemberAccess a' as') (MemberAccess b' bs') = annotate $ MemberAccess (diffTerms' a' b') (diffTerms' as' bs')
recur (SubscriptAccess a' as') (SubscriptAccess b' bs') = annotate $ SubscriptAccess (diffTerms' a' b') (diffTerms' as' bs')
recur (MethodCall a' as' aParams') (MethodCall b' bs' bParams') = annotate $ MethodCall (diffTerms' a' b') (diffTerms' as' bs') (diffTerms' aParams' bParams')
recur (Ternary aCondition' aCases') (Ternary bCondition' bCases') = annotate $ Ternary (diffTerms' aCondition' bCondition') (alignWith diffThese aCases' bCases')
recur (Args as') (Args bs') = annotate . Args $ alignWith diffThese as' bs'
recur (VarDecl a') (VarDecl b') = annotate . VarDecl $ diffTerms' a' b'
recur (VarAssignment a' as') (VarAssignment b' bs') = annotate $ VarAssignment (diffTerms' a' b') (diffTerms' as' bs')
recur (Operator a') (Operator b') = annotate $ Operator (alignWith diffThese a' b')
recur (Switch a' as') (Switch b' bs') = annotate $ Switch (diffTerms' a' b') (alignWith (these (pure . Delete) (pure . Insert) diffTerms') as' bs')
recur (Case a' as') (Case b' bs') = annotate $ Case (diffTerms' a' b') (diffTerms' as' bs')
recur (Object as') (Object bs') = annotate $ Object (alignWith diffThese as' bs')
recur (Pair a1 a2) (Pair b1 b2) = annotate $ Pair (diffTerms' a1 b1) (diffTerms' a2 b2)
recur (Commented a1 a2) (Commented b1 b2) = annotate $ Commented (alignWith diffThese a1 b1) (diffTerms' <$> a2 <*> b2)
recur (Comment _) (Comment _) = pure $ Replace (cofree (annotation1 :< a)) (cofree (annotation2 :< b))
recur (Leaf _) (Leaf _) = pure $ Replace (cofree (annotation1 :< a)) (cofree (annotation2 :< b))
recur _ _ = pure $ Replace (cofree (annotation1 :< a)) (cofree (annotation2 :< b))
recur a b = maybe (pure (Replace t1 t2)) (annotate . fmap diffThese) (galign a b)
diffThese = these (pure . Delete) (pure . Insert) (diffTerms construct comparable cost)
Free (ByKey a b f) -> run construct comparable cost $ f byKey where
byKey = Map.fromList $ toKeyValue <$> List.union aKeys bKeys
toKeyValue key | key `List.elem` deleted = (key, pure . Delete $ a ! key)

View File

@ -16,7 +16,7 @@ data Patch a
= Replace a a
| Insert a
| Delete a
deriving (Eq, Foldable, Functor, Ord, Show, Traversable)
deriving (Eq, Foldable, Functor, Generic, Ord, Show, Traversable)
-- | Return the item from the after side of the patch.
after :: Patch a -> Maybe a

View File

@ -1,8 +1,9 @@
module SES where
import Prologue
import Patch
import qualified Data.Map as Map
import Patch
import Prologue
-- | Edit constructor for two terms, if comparable. Otherwise returns Nothing.
type Compare term edit = term -> term -> Maybe edit
@ -18,9 +19,6 @@ ses diffTerms cost as bs = fst <$> evalState diffState Map.empty where
-- | Find the shortest edit script between two terms at a given vertex in the edit graph.
diffAt :: Applicative edit => Compare term (edit (Patch term)) -> Cost (edit (Patch term)) -> (Integer, Integer) -> [term] -> [term] -> State (Map.Map (Integer, Integer) [(edit (Patch term), Integer)]) [(edit (Patch term), Integer)]
diffAt diffTerms cost (i, j) as bs
| null as, null bs = pure []
| null as = pure $ foldr insert [] bs
| null bs = pure $ foldr delete [] as
| (a : as) <- as, (b : bs) <- bs = do
cachedDiffs <- get
case Map.lookup (i, j) cachedDiffs of
@ -28,7 +26,7 @@ diffAt diffTerms cost (i, j) as bs
Nothing -> do
down <- recur (i, succ j) as (b : bs)
right <- recur (succ i, j) (a : as) bs
nomination <- fmap best $ case diffTerms a b of
nomination <- best <$> case diffTerms a b of
Just diff -> do
diagonal <- recur (succ i, succ j) as bs
pure [ delete a down, insert b right, consWithCost cost diff diagonal ]
@ -36,6 +34,9 @@ diffAt diffTerms cost (i, j) as bs
cachedDiffs' <- get
put $ Map.insert (i, j) nomination cachedDiffs'
pure nomination
| null as = pure $ foldr insert [] bs
| null bs = pure $ foldr delete [] as
| otherwise = pure []
where
delete = consWithCost cost . pure . Delete
insert = consWithCost cost . pure . Insert
@ -44,6 +45,7 @@ diffAt diffTerms cost (i, j) as bs
best = minimumBy (comparing costOf)
recur = diffAt diffTerms cost
-- | Prepend an edit script and the cumulative cost onto the edit script.
consWithCost :: Cost edit -> edit -> [(edit, Integer)] -> [(edit, Integer)]
consWithCost cost edit rest = (edit, cost edit + maybe 0 snd (fst <$> uncons rest)) : rest

View File

@ -3,6 +3,7 @@ module Syntax where
import Prologue
import Data.OrderedMap as Map
import Data.Text.Arbitrary ()
import GHC.Generics
import qualified Data.Text as T
import Test.QuickCheck hiding (Fixed)
@ -53,7 +54,7 @@ data Syntax
| Pair f f
| Comment a
| Commented [f] (Maybe f)
deriving (Eq, Foldable, Functor, Generic, Ord, Show, Traversable)
deriving (Eq, Foldable, Functor, Generic, Generic1, Ord, Show, Traversable)
-- Instances

View File

@ -1,11 +1,12 @@
{-# LANGUAGE TypeFamilies, TypeSynonymInstances, Unsafe #-}
{-# LANGUAGE RankNTypes, TypeFamilies, TypeSynonymInstances, Unsafe #-}
{-# OPTIONS_GHC -fno-warn-orphans #-}
module Term where
import Prologue
import Data.Align.Generic
import Data.Functor.Foldable as Foldable
import Data.Functor.Both
import Data.OrderedMap hiding (size)
import Data.These
import Syntax
import Unsafe
@ -19,38 +20,24 @@ instance Functor f => Foldable.Unfoldable (Cofree f a) where embed = cofree
-- | Zip two terms by combining their annotations into a pair of annotations.
-- | If the structure of the two terms don't match, then Nothing will be returned.
zipTerms :: (Eq a, Eq annotation) => Term a annotation -> Term a annotation -> Maybe (Term a (Both annotation))
zipTerms t1 t2 = annotate (zipUnwrap a b)
where
(annotation1 :< a, annotation2 :< b) = (runCofree t1, runCofree t2)
annotate = fmap (cofree . (both annotation1 annotation2 :<))
zipUnwrap (Leaf _) (Leaf b') = Just $ Leaf b'
zipUnwrap (Comment _) (Comment b) = Just $ Comment b
zipUnwrap (Indexed a') (Indexed b') = Indexed <$> zipWithM zipTerms a' b'
zipUnwrap (Fixed a') (Fixed b') = Fixed <$> zipWithM zipTerms a' b'
zipUnwrap (Keyed a') (Keyed b') | keys a' == keys b' = Just . Keyed . fromList . catMaybes $ zipUnwrapMaps a' b' <$> keys a'
zipUnwrap (FunctionCall idA' a') (FunctionCall idB' b') = FunctionCall <$> zipTerms idA' idB' <*> zipWithM zipTerms a' b'
zipUnwrap (Function idA' paramsA' exprsA') (Function idB' paramsB' exprsB') = case (zipTerms exprsA' exprsB') of
Just exprs' -> Just (Function (join $ liftA2 zipTerms idA' idB') (join $ liftA2 zipTerms paramsA' paramsB') exprs')
_ -> Nothing
zipUnwrap (Case eA' bodyA') (Case eB' bodyB') = Case <$> zipTerms eA' eB' <*> zipTerms bodyA' bodyB'
zipUnwrap (Switch a' as') (Switch b' bs') = Switch <$> (zipTerms a' b') <*> zipWithM zipTerms as' bs'
zipUnwrap (MethodCall iA mA paramsA) (MethodCall iB mB paramsB) = MethodCall <$> zipTerms iA iB <*> zipTerms mA mB <*> zipTerms paramsA paramsB
zipUnwrap (Args a1) (Args b1) = Args <$> (zipWithM zipTerms a1 b1)
zipUnwrap (MemberAccess a1 a2) (MemberAccess b1 b2) = MemberAccess <$> zipTerms a1 b1 <*> zipTerms a2 b2
zipUnwrap (MathAssignment a1 a2) (MathAssignment b1 b2) = MathAssignment <$> zipTerms a1 b1 <*> zipTerms a2 b2
zipUnwrap (Assignment a1 a2) (Assignment b1 b2) = Assignment <$> zipTerms a1 b1 <*> zipTerms a2 b2
zipUnwrap (Ternary a1 a2) (Ternary b1 b2) = Ternary <$> zipTerms a1 b1 <*> (zipWithM zipTerms a2 b2)
zipUnwrap (Object as') (Object bs') | as' == bs' = Object <$> zipWithM zipTerms as' bs'
zipUnwrap (Operator as) (Operator bs) = Operator <$> zipWithM zipTerms as bs
zipUnwrap (VarDecl a) (VarDecl b) = VarDecl <$> zipTerms a b
zipUnwrap (VarAssignment a1 a2) (VarAssignment b1 b2) = VarAssignment <$> zipTerms a1 b1 <*> zipTerms a2 b2
zipUnwrap (SubscriptAccess a1 a2) (SubscriptAccess b1 b2) = SubscriptAccess <$> zipTerms a1 b1 <*> zipTerms a2 b2
zipUnwrap (Pair a1' a2') (Pair b1' b2') = Pair <$> zipTerms a1' b1' <*> zipTerms a2' b2'
zipUnwrap (Commented cs1 a) (Commented cs2 b) = Commented <$> zipWithM zipTerms cs1 cs2 <*> (zipTerms <$> a <*> b)
zipUnwrapMaps a' b' key = (,) key <$> zipTerms (a' ! key) (b' ! key)
zipTerms t1 t2 = iter go (alignCofreeWith galign (const Nothing) both (These t1 t2))
where go (a :< s) = cofree . (a :<) <$> sequenceA s
-- | Return the node count of a term.
termSize :: Term a annotation -> Integer
termSize = cata size where
size (_ :< syntax) = 1 + sum syntax
-- | Aligns (zips, retaining non-overlapping portions of the structure) a pair of terms.
alignCofreeWith :: Functor f
=> (forall a b. f a -> f b -> Maybe (f (These a b))) -- ^ A function comparing a pair of structures, returning `Just` the combined structure if they are comparable (e.g. if they have the same constructor), and `Nothing` otherwise. The 'Data.Align.Generic.galign' function is usually what you want here.
-> (These (Cofree f a) (Cofree f b) -> contrasted) -- ^ A function mapping a 'These' of incomparable terms into 'Pure' values in the resulting tree.
-> (a -> b -> combined) -- ^ A function mapping the input terms annotations into annotations in the 'Free' values in the resulting tree.
-> These (Cofree f a) (Cofree f b) -- ^ The input terms.
-> Free (CofreeF f combined) contrasted
alignCofreeWith compare contrast combine = go
where go terms = fromMaybe (pure (contrast terms)) $ case terms of
These t1 t2 -> wrap . (combine (extract t1) (extract t2) :<) . fmap go <$> compare (unwrap t1) (unwrap t2)
_ -> Nothing

View File

@ -3,7 +3,6 @@ module Data.RandomWalkSimilarity.Spec where
import Category
import Data.DList as DList hiding (toList)
import Data.RandomWalkSimilarity
import qualified Data.Set as Set
import Diff
import Patch
import Prologue
@ -33,7 +32,5 @@ spec = parallel $ do
\ (as, bs) -> let tas = toTerm <$> as
tbs = toTerm <$> bs
diff = free (Free (pure Program :< Indexed (rws compare identity tas tbs :: [Diff Text Category]))) in
(childrenOf <$> beforeTerm diff, childrenOf <$> afterTerm diff) `shouldBe` (Just (Set.fromList tas), Just (Set.fromList tbs))
(beforeTerm diff, afterTerm diff) `shouldBe` (Just (cofree (Program :< Indexed tas)), Just (cofree (Program :< Indexed tbs)))
childrenOf :: (Ord leaf, Ord annotation) => Term leaf annotation -> Set.Set (Term leaf annotation)
childrenOf = Set.fromList . toList . unwrap

29
weekly/2016-07-12.md Normal file
View File

@ -0,0 +1,29 @@
# July 12th, 2016 weekly
## What went well?
@joshvera: Glad to see objects working, and the test harness was helpful for that.
@robrix: RWS work went well, specifically about ordering. Also confirming that RWS does not conflict with moves, although we do not support moves yet. Also generics with zipping turned out well!
@rewinfrey: The move went well, more or less.
## What went less well?
@joshvera: Was sick and would have liked to contribute to test harness earlier.
@robrix: Scrap your boilerplate style of generic programming in Haskell is tough.
@rewinfrey: Adding test cases manually is tough. Want to find a way to automate that process.
## What did you learn?
@joshvera: Learned a lot about RWS, and generics in Haskell.
@robrix: Learned about natural transformations. Understood RWS ordering problem. How to metaprogram in Haskell without headaches! Wrote a catamorphism that annotates every element in the structure with the result of the fold up to that element.
@rewinfrey: Learned how to install curtains, and visiting Ikea is ideal on the 4th of July.
## Anything else?
@robrix: Reminder about Q3 goals, will be the focus of 1:1's for the next couple weeks. Checkin with how weekly format is going (@joshvera & @rewinfrey no complaints). Also see everyone at Summit!