Copyright | (c) Adam Conner-Sax 2019 |
---|---|

License | BSD-3-Clause |

Maintainer | adam_conner_sax@yahoo.com |

Stability | experimental |

Safe Haskell | Safe |

Language | Haskell2010 |

MapReduce as folds This is all just wrapping around Control.Foldl so that it's easier to see the map-reduce structure The Mapping step is broken into 2 parts:

- unpacking, which could include "melting" or filtering,
- assigning, which assigns a group to each unpacked item. Could just be choosing a key column(s)

The items are then grouped by key and "reduced"

The reduce step is conceptually simpler, just requiring a function from the (key, grouped data) pair to the result.

Reduce could be as simple as combining the key with a single data row or some very complex function of the grouped data.
E.g., reduce could itself be a map-reduce on the grouped data.
Since these are folds, we can share work by using the Applicative instance of MapStep (just the Applicative instance of Control.Foldl.Fold)
and we will loop over the data only once.
The Reduce type is also Applicative so there could be work sharing there as well, especially if you
specify your reduce as a Fold.
e.g., if your `reduce :: (k -> h c -> d)`

has the form `reduce :: k -> FL.Fold c d`

We combine these steps with an Engine, resulting in a fold from a container of `x`

to some container of `d`

.
The Engine amounts to a choice of grouping algorithm (usually using `Data.Map`

or `Data.HashMap`

) and a choice of result container type.
The result container type is used for the intermediate steps as well.

The goal is to make assembling a large family of common map/reduce patterns in a straightforward way. At some level of complication, you may as well write them by hand. An in-between case would be writing the unpack function as a complex hand written filter

## Synopsis

- data Unpack x y where
- data Assign k y c where
- data Reduce k x d where
- data UnpackM m x y where
- data AssignM m k y c where
- data ReduceM m k x d where
- generalizeUnpack :: Monad m => Unpack x y -> UnpackM m x y
- generalizeAssign :: Monad m => Assign k y c -> AssignM m k y c
- generalizeReduce :: Monad m => Reduce k x d -> ReduceM m k x d
- functionToFold :: (forall h. Foldable h => h x -> a) -> Fold x a
- functionToFoldM :: Monad m => (forall h. Foldable h => h x -> m a) -> FoldM m x a
- postMapM :: Monad m => (a -> m b) -> FoldM m x a -> FoldM m x b
- data Fold a b
- data FoldM (m :: Type -> Type) a b
- fold :: Foldable f => Fold a b -> f a -> b
- foldM :: (Foldable f, Monad m) => FoldM m a b -> f a -> m b

# Basic Types for map reduce

## Non-Monadic

data Unpack x y where Source #

`Unpack`

is for filtering rows or "melting" them, that is doing something which turns each `x`

into several `y`

.
Filter is a special case because it can often be done faster directly than via folding over `Maybe`

## Instances

Profunctor Unpack Source # | |

Defined in Control.MapReduce.Core | |

Functor (Unpack x) Source # | |

data Assign k y c where Source #

map `y`

into a `(k,c)`

pair for grouping

## Instances

Profunctor (Assign k) Source # | |

Defined in Control.MapReduce.Core | |

Functor (Assign k y) Source # | |

data Reduce k x d where Source #

Wrapper for functions to reduce keyed and grouped data to the result type. It is *strongly* suggested that you use Folds for this step. Type-inference and applicative optimization are more straightforward that way. The non-Fold contructors are there in order to retro-fit existing functions.

Reduce step for non-effectful reductions

Reduce :: (k -> forall h. (Foldable h, Functor h) => h x -> d) -> Reduce k x d | |

ReduceFold :: (k -> Fold x d) -> Reduce k x d |

## Instances

Profunctor (Reduce k) Source # | |

Defined in Control.MapReduce.Core | |

Functor (Reduce k x) Source # | |

Applicative (Reduce k x) Source # | |

Defined in Control.MapReduce.Core |

## Monadic

data UnpackM m x y where Source #

`UnpackM`

is for filtering or melting the input type. This version has a monadic result type to
accomodate unpacking that might require randomness or logging during unpacking.
filter is a special case since filtering can be often be done faster. So we single it out.

FilterM :: Monad m => (x -> m Bool) -> UnpackM m x x | |

UnpackM :: (Monad m, Traversable g) => (x -> m (g y)) -> UnpackM m x y |

## Instances

Profunctor (UnpackM m) Source # | |

Defined in Control.MapReduce.Core dimap :: (a -> b) -> (c -> d) -> UnpackM m b c -> UnpackM m a d # lmap :: (a -> b) -> UnpackM m b c -> UnpackM m a c # rmap :: (b -> c) -> UnpackM m a b -> UnpackM m a c # (#.) :: Coercible c b => q b c -> UnpackM m a b -> UnpackM m a c # (.#) :: Coercible b a => UnpackM m b c -> q a b -> UnpackM m a c # | |

Functor (UnpackM m x) Source # | |

data AssignM m k y c where Source #

Effectfully map `y`

into a `(k,c)`

pair for grouping

## Instances

Profunctor (AssignM m k) Source # | |

Defined in Control.MapReduce.Core dimap :: (a -> b) -> (c -> d) -> AssignM m k b c -> AssignM m k a d # lmap :: (a -> b) -> AssignM m k b c -> AssignM m k a c # rmap :: (b -> c) -> AssignM m k a b -> AssignM m k a c # (#.) :: Coercible c b => q b c -> AssignM m k a b -> AssignM m k a c # (.#) :: Coercible b a => AssignM m k b c -> q a b -> AssignM m k a c # | |

Functor (AssignM m k y) Source # | |

data ReduceM m k x d where Source #

Reduce step for effectful reductions. It is *strongly* suggested that you use Folds for this step. Type-inference and applicative optimization are more straightforward that way. The non-Fold contructors are there in order to retro-fit existing functions.

ReduceM :: Monad m => (k -> forall h. (Foldable h, Functor h) => h x -> m d) -> ReduceM m k x d | |

ReduceFoldM :: Monad m => (k -> FoldM m x d) -> ReduceM m k x d |

## Instances

Profunctor (ReduceM m k) Source # | |

Defined in Control.MapReduce.Core dimap :: (a -> b) -> (c -> d) -> ReduceM m k b c -> ReduceM m k a d # lmap :: (a -> b) -> ReduceM m k b c -> ReduceM m k a c # rmap :: (b -> c) -> ReduceM m k a b -> ReduceM m k a c # (#.) :: Coercible c b => q b c -> ReduceM m k a b -> ReduceM m k a c # (.#) :: Coercible b a => ReduceM m k b c -> q a b -> ReduceM m k a c # | |

Functor (ReduceM m k x) Source # | |

Monad m => Applicative (ReduceM m k x) Source # | |

Defined in Control.MapReduce.Core pure :: a -> ReduceM m k x a # (<*>) :: ReduceM m k x (a -> b) -> ReduceM m k x a -> ReduceM m k x b # liftA2 :: (a -> b -> c) -> ReduceM m k x a -> ReduceM m k x b -> ReduceM m k x c # (*>) :: ReduceM m k x a -> ReduceM m k x b -> ReduceM m k x b # (<*) :: ReduceM m k x a -> ReduceM m k x b -> ReduceM m k x a # |

## Non-Monadic -> Monadic

generalizeUnpack :: Monad m => Unpack x y -> UnpackM m x y Source #

lift a non-monadic Unpack to a monadic one for any monad m

generalizeAssign :: Monad m => Assign k y c -> AssignM m k y c Source #

lift a non-monadic Assign to a monadic one for any monad m

generalizeReduce :: Monad m => Reduce k x d -> ReduceM m k x d Source #

Make a non-monadic reduce monadic.

# Foldl helpers

functionToFold :: (forall h. Foldable h => h x -> a) -> Fold x a Source #

convert a function which takes a foldable container of x and produces an a into a Fold x a. uses Data.Sequence.Seq as an intermediate type because it should behave well whether f is a left or right fold. This can be helpful in putting an existing function into a Reduce.

functionToFoldM :: Monad m => (forall h. Foldable h => h x -> m a) -> FoldM m x a Source #

convert a function which takes a foldable container of x and produces an (m a) into a FoldM m x a. uses Data.Sequence.Seq as an intermediate type because it should behave well whether f is a left or right fold. This can be helpful in putting an existing function into a ReduceM.

postMapM :: Monad m => (a -> m b) -> FoldM m x a -> FoldM m x b Source #

Given an effectful (monadic) Control.Foldl fold, a `FoldM m x a`

, we can use its `Functor`

instance
to apply a non-effectful `(a -> b)`

to its result type. To apply `(a -> m b)`

, we need this combinator.

# Re-Exports

Efficient representation of a left fold that preserves the fold's step function, initial accumulator, and extraction function

This allows the `Applicative`

instance to assemble derived folds that
traverse the container only once

A '`Fold`

a b' processes elements of type **a** and results in a
value of type **b**.

## Instances

Choice Fold | |

Profunctor Fold | |

Functor (Fold a) | |

Applicative (Fold a) | |

Comonad (Fold a) | |

Semigroupoid Fold | |

Floating b => Floating (Fold a b) | |

Defined in Control.Foldl sqrt :: Fold a b -> Fold a b # (**) :: Fold a b -> Fold a b -> Fold a b # logBase :: Fold a b -> Fold a b -> Fold a b # asin :: Fold a b -> Fold a b # acos :: Fold a b -> Fold a b # atan :: Fold a b -> Fold a b # sinh :: Fold a b -> Fold a b # cosh :: Fold a b -> Fold a b # tanh :: Fold a b -> Fold a b # asinh :: Fold a b -> Fold a b # acosh :: Fold a b -> Fold a b # atanh :: Fold a b -> Fold a b # log1p :: Fold a b -> Fold a b # expm1 :: Fold a b -> Fold a b # | |

Fractional b => Fractional (Fold a b) | |

Num b => Num (Fold a b) | |

Semigroup b => Semigroup (Fold a b) | |

Monoid b => Monoid (Fold a b) | |

data FoldM (m :: Type -> Type) a b #

Like `Fold`

, but monadic.

A '`FoldM`

m a b' processes elements of type **a** and
results in a monadic value of type **m b**.

## Instances

Functor m => Profunctor (FoldM m) | |

Defined in Control.Foldl | |

Functor m => Functor (FoldM m a) | |

Applicative m => Applicative (FoldM m a) | |

(Monad m, Floating b) => Floating (FoldM m a b) | |

Defined in Control.Foldl exp :: FoldM m a b -> FoldM m a b # log :: FoldM m a b -> FoldM m a b # sqrt :: FoldM m a b -> FoldM m a b # (**) :: FoldM m a b -> FoldM m a b -> FoldM m a b # logBase :: FoldM m a b -> FoldM m a b -> FoldM m a b # sin :: FoldM m a b -> FoldM m a b # cos :: FoldM m a b -> FoldM m a b # tan :: FoldM m a b -> FoldM m a b # asin :: FoldM m a b -> FoldM m a b # acos :: FoldM m a b -> FoldM m a b # atan :: FoldM m a b -> FoldM m a b # sinh :: FoldM m a b -> FoldM m a b # cosh :: FoldM m a b -> FoldM m a b # tanh :: FoldM m a b -> FoldM m a b # asinh :: FoldM m a b -> FoldM m a b # acosh :: FoldM m a b -> FoldM m a b # atanh :: FoldM m a b -> FoldM m a b # log1p :: FoldM m a b -> FoldM m a b # expm1 :: FoldM m a b -> FoldM m a b # | |

(Monad m, Fractional b) => Fractional (FoldM m a b) | |

(Monad m, Num b) => Num (FoldM m a b) | |

Defined in Control.Foldl (+) :: FoldM m a b -> FoldM m a b -> FoldM m a b # (-) :: FoldM m a b -> FoldM m a b -> FoldM m a b # (*) :: FoldM m a b -> FoldM m a b -> FoldM m a b # negate :: FoldM m a b -> FoldM m a b # abs :: FoldM m a b -> FoldM m a b # signum :: FoldM m a b -> FoldM m a b # fromInteger :: Integer -> FoldM m a b # | |

(Semigroup b, Monad m) => Semigroup (FoldM m a b) | |

(Monoid b, Monad m) => Monoid (FoldM m a b) | |