merge multiple observables to an observable array - f#

Hi I am trying to merge a number of observables to an observable array. Here an example that works in fsi. (sorry that it is lengthy)
#r "./bin/Debug/System.Reactive.dll"
open System
open System.Reactive.Linq
/// Subscribes to the Observable with all 3 callbacks.
let subscribeComplete next error completed (observable: IObservable<'T>) =
observable.Subscribe(
(fun x -> next x),
(fun e -> error e),
(fun () -> completed()))
/// Subscribes to the Observable with a next and an error-function.
let subscribeWithError next error observable =
subscribeComplete next error (fun () -> ()) observable
/// Subscribes to the Observable with a next-function
let subscribe (next: 'T -> unit) (observable: IObservable<'T>) : IDisposable =
subscribeWithError next ignore observable
/// Static method to generate observable from input functions
let ObsGenerate (initState: 'TS) (termCond: 'TS -> bool) (iterStep: 'TS -> 'TS)
(resSelect: 'TS -> 'TR) (timeSelect : 'TS -> System.TimeSpan) =
Observable.Generate(initState, termCond, iterStep, resSelect, timeSelect)
//maps the given observable with the given function
let obsMap (f: 'T -> 'U) (observable : IObservable<'T>) : IObservable<'U> =
Observable.Select(observable, Func<_,_>(f))
/// Merges two observable sequences into one observable sequence whenever one of the observable sequences has a new value.
let combineLatest (obs1: IObservable<'T>) (obs2: IObservable<'U>) : IObservable<'T * 'U> =
Observable.CombineLatest(
obs1, obs2, Func<_,_,_>(fun a b -> a, b))
/// Merges three observable sequences into one observable sequence whenever one of the observable sequences has a new value.
let combineLatest3 (obs1: IObservable<'T>) (obs2: IObservable<'U>) (obs3: IObservable<'V>) : IObservable<'T * 'U * 'V> =
let obs12 =obs1.CombineLatest(obs2, Func<_,_,_>(fun a b -> a, b))
obs12.CombineLatest(obs3, Func<_,_,_>(fun (a,b) c -> a, b, c))
/// Merges four observable sequences into one observable sequence whenever one of the observable sequences has a new value.
let combineLatest4 (obs1: IObservable<'T>) (obs2: IObservable<'U>) (obs3: IObservable<'V>) (obs4: IObservable<'W>) : IObservable<'T * 'U * 'V * 'W> =
let obsNew = combineLatest3 obs1 obs2 obs3
obsNew.CombineLatest(obs4, Func<_,_,_>(fun (a,b,c) d -> a, b, c, d))
// second section generating arrays
let combineLatestArray (obs1: IObservable<'T>) (obs2: IObservable<'T>) =
combineLatest obs1 obs2
|> obsMap (fun (a, b) -> [a; b] |> List.toArray)
let combineLatest3Array (obs1: IObservable<'T>) (obs2: IObservable<'T>) (obs3: IObservable<'T>) =
combineLatest3 obs1 obs2 obs3
|> obsMap (fun (a, b, c) -> [a; b; c] |> List.toArray)
let combineLatest4Array (obs1: IObservable<'T>) (obs2: IObservable<'T>) (obs3: IObservable<'T>) (obs4: IObservable<'T>) =
combineLatest4 obs1 obs2 obs3 obs4
|> obsMap (fun (a, b, c, d) -> [a; b; c; d] |> List.toArray)
let combineLatestListToArray (list: IObservable<'T> List) =
match list.Length with
| 2 -> combineLatestArray list.[0] list.[1]
| 3 -> combineLatest3Array list.[0] list.[1] list.[2]
| 4 -> combineLatest4Array list.[0] list.[1] list.[2] list.[3]
| _ -> failwith "combine latest on unsupported list size"
type FooType =
{ NameVal : string
IdVal : int
RetVal : float }
member x.StringKey() =
x.NameVal.ToString() + ";" + x.IdVal.ToString()
// example code starts here
let rnd = System.Random()
let fooListeners = Collections.Generic.Dictionary()
let AddAFoo (foo : FooType) =
let fooId = foo.StringKey()
if fooListeners.ContainsKey(fooId)
then fooListeners.[fooId]
else
let myObs = ObsGenerate {NameVal = foo.NameVal; IdVal = foo.IdVal; RetVal = foo.RetVal} (fun x -> true) (fun x -> {NameVal = (x.NameVal); IdVal = (x.IdVal); RetVal = (x.RetVal + rnd.NextDouble() - 0.5)}) (fun x -> x) (fun x -> System.TimeSpan.FromMilliseconds(rnd.NextDouble() * 2000.0))
fooListeners.Add(fooId,myObs)
myObs
let fooInit = [6..9]
|> List.map (fun index -> {NameVal = (string index + "st"); IdVal = index; RetVal = (float index + 1.0)})
|> List.map (fun foo -> AddAFoo foo)
let fooValuesArray = fooInit
|> List.map(fun x -> (x |> obsMap (fun x -> x.RetVal)))
|> combineLatestListToArray
let mySub =
fooValuesArray
|> subscribe (fun fooVals -> printfn "fooArray: %A" fooVals)
//execute until here to start example
// execute this last line to unsubscribe
mySub.Dispose()
I have two questions about this code:
Is there a smarter way of merging the observables to arrays? (it gets very lengthy as I need to merge larger arrays)
I want to throttle the updates. What I mean by that is that I want all updates that occur within (say) the same half a second window to be handled as one update on the array. Ideally, I want this window to open only when a first update comes in, i.e if no updates arrive in 2 seconds, then one update arrives, then we wait and include further updates for 0.5 seconds and then trigger the observable. I don't want it to publish periodically every 0.5 seconds although no observables are triggered. I hope this description is clear enough.
update: I have decided to accept one of the F# answers, but I haven't done the C# answers justice yet. I hope to be able to check them out properly soon.

For 1, an application of List.fold and List.toArray and a few Observable operators should work nicely. Something like:
let combineLatest observables =
Observable.Select(
(observables
|> List.fold (fun ol o
-> Observable.CombineLatest(o, ol, (fun t tl -> t :: tl))
) (Observable.Return<_>([]))
),
List.toArray)
Due to the nesting, you may end up with performance issues if you have a large list of Observables, but it's worth at least trying before you resort to writing it by hand.
For 2, I would agree with the other answers to apply Throttling to the result.

I'm sorry this isn't F# - I wish I had time to learn it - but here's a possible answer in C#.
Here are a set of extension methods that will combine the latest from an IEnumerable<IObservable<T>> to an IObservable<IEnumerable<T>>:
public static IObservable<IEnumerable<T>> CombineLatest<T>(this IObservable<T> first, IObservable<T> second)
{
if (first == null) throw new ArgumentNullException("first");
if (second == null) throw new ArgumentNullException("second");
return first.CombineLatest(second, (t0, t1) => EnumerableEx.Return(t0).Concat(EnumerableEx.Return(t1)));
}
public static IObservable<IEnumerable<T>> CombineLatest<T>(this IObservable<IEnumerable<T>> firsts, IObservable<T> second)
{
if (firsts == null) throw new ArgumentNullException("firsts");
if (second == null) throw new ArgumentNullException("second");
return firsts.CombineLatest(second, (t0s, t1) => t0s.Concat(EnumerableEx.Return(t1)));
}
public static IObservable<IEnumerable<T>> CombineLatest<T>(this IEnumerable<IObservable<T>> sources)
{
if (sources == null) throw new ArgumentNullException("sources");
return sources.CombineLatest(() => sources.First().CombineLatest(sources.Skip(1)), () => Observable.Empty<IEnumerable<T>>());
}
public static IObservable<IEnumerable<T>> CombineLatest<T>(this IObservable<T> first, IEnumerable<IObservable<T>> seconds)
{
if (first == null) throw new ArgumentNullException("first");
if (seconds == null) throw new ArgumentNullException("seconds");
return seconds.CombineLatest(() => first.CombineLatest(seconds.First()).CombineLatest(seconds.Skip(1)), () => first.Select(t => EnumerableEx.Return(t)));
}
public static IObservable<IEnumerable<T>> CombineLatest<T>(this IObservable<IEnumerable<T>> firsts, IEnumerable<IObservable<T>> seconds)
{
if (firsts == null) throw new ArgumentNullException("firsts");
if (seconds == null) throw new ArgumentNullException("seconds");
return seconds.CombineLatest(() => firsts.CombineLatest(seconds.First()).CombineLatest(seconds.Skip(1)), () => firsts);
}
private static IObservable<IEnumerable<T>> CombineLatest<T>(this IEnumerable<IObservable<T>> sources, Func<IObservable<IEnumerable<T>>> any, Func<IObservable<IEnumerable<T>>> none)
{
if (sources == null) throw new ArgumentNullException("sources");
if (any == null) throw new ArgumentNullException("any");
if (none == null) throw new ArgumentNullException("none");
return Observable.Defer(() => sources.Any() ? any() : none());
}
They may not be very efficient, but they do handle any number of observables that need to be combined.
I'd be keen to see these methods converted to F#.
As for your second question, I'm not sure I can answer with what you've said so far because CombineLatest and Throttle both lose values so it is probably prudent to understand your use case in more detail before attempting an answer.

Although Gideon Engelberth has answered your question with one of the possible way to solve the problem. Other possible way could be something like below, it doesn't use nesting.
let combineLatestToArray (list : IObservable<'T> list) =
let s = new Subject<'T array>()
let arr = Array.init list.Length (fun _ -> Unchecked.defaultof<'T>)
let cb (i:int,v:'T) =
arr.[i] <- v
s.OnNext(arr |> Array.toList |> List.toArray)
let main = list |> List.mapi (fun i o -> o.Select(fun t -> (i,t)))
|> Observable.Merge
main.Subscribe(new Action<int * 'T>(cb)
,new Action<exn>(fun e -> s.OnError(e))
,new Action(fun () -> s.OnCompleted()) ) |> ignore
s :> IObservable<'T array>
Let me know if this solved your problem as I haven't testing it much :)
NOTE: This is for the first part, for second part everyone has already mentioned what you need to do
UPDATE:
Another implementation :
let combineLatestToArray (list : IObservable<'T> list) =
let s = new Subject<'T array>()
let arr = Array.init list.Length (fun _ -> Unchecked.defaultof<'T>)
let main = list |> List.mapi (fun i o -> o.Select(fun t -> (i,t)))
|> Observable.Merge
async {
try
let se = main.ToEnumerable() |> Seq.scan (fun ar (i,t) -> Array.set ar i t; ar) arr
for i in se do
s.OnNext(i |> Array.toList |> List.toArray)
s.OnCompleted()
with
| :? Exception as ex -> s.OnError(ex)
} |> Async.Start
s :> IObservable<'T array>

Seems that Observable.Merge() which has overloads for variable number of IObservables is closer to what you want.
Observable.Buffer() with the time overloads will do what you want here. In the "no events" situation, Buffer will still OnNext() an empty list, letting you react to that stiuation.

This is the best I could come up with. I've been wanting to solve this for a while.
public static class Extensions
{
public static IObservable<IEnumerable<T>> CombineLatest<T>(this Observable observable, IEnumerable<IObservable<T>> observableCollection)
{
return observableCollection.CombineLatest();
}
public static IObservable<IEnumerable<T>> CombineLatest<T>(this IEnumerable<IObservable<T>> observables)
{
return observables.Aggregate<IObservable<T>, IObservable<IEnumerable<T>>>
(
Observable.Return(Enumerable.Empty<T>()),
(o, n) => o.CombineLatest
(
n,
(list, t) => list.Concat(EnumerableEx.Return(t))
)
);
}
}
So an example usage would be:
var obs = new List<IObservable<bool>>
{
Observable.Return(true),
Observable.Return(false),
Observable.Return(true)
};
var result = obs.CombineLatest().Select(list => list.All(x => x));
result.Subscribe(Console.WriteLine);
Console.ReadKey();
You would have to operate on the knowledge, though, that the resulting IObservable<IEnumerable<T>> will not fire until all observables have yielded a value. This is what I needed in my scenarios, but might not be appropriate for your scenario.
My worry with this is the performance of all of the .Concats. Might be more performant to deal in a mutable collection in the extension method. Not sure.
Sorry, I don't know F#. I'll get around to it one of these days.
Throttling is just done with the .Throttle operator after you get your final observable.
Edit: This answer is the iterative Ying to Enigmativity's recursive Yang.

Related

F# List.exists on two lists

I have two lists listA and listB where I want to return true if listB contains any element also in listA.
let listA = ["A";"B";"C"]
let listB = ["D";"E";"A"]
Should return true in this case. I feel like this should be easy to solve and I'm missing something fundamental somewhere.
For example, why can't I do like this?
let testIntersect = for elem in listA do List.exists (fun x -> x = elem) listB
You can't write something like your example code because a plain for doesn't return a result, it just evaluates an expression for its side-effects. You could write the code in a for comprehension:
let testIntersect listA listB =
[for elem in listA do yield List.exists (fun x -> x = elem) listB]
Of course, this then returns a bool list rather than a single bool.
val testIntersect :
listA:seq<'a> -> listB:'a list -> bool list when 'a : equality
let listA = ["A";"B";"C"]
let listB = ["D";"E";"A"]
testIntersect listA listB
val it : bool list = [true; false; false]
So, we can use the List.exists function to ensure that a true occurs at least once:
let testIntersect listA listB =
[for elem in listA do yield List.exists (fun x -> x = elem) listB]
|> List.exists id
val testIntersect :
listA:seq<'a> -> listB:'a list -> bool list when 'a : equality
val listA : string list = ["A"; "B"; "C"]
val listB : string list = ["D"; "E"; "A"]
val it : bool = false
It's pretty inefficient to solve this problem using List though, it's better to use Set. With Set, you can calculate intersection in O(log N * log M) time rather than O(N*M).
let testSetIntersect listA listB =
Set.intersect (Set.ofList listA) (Set.ofList listB)
|> Set.isEmpty
|> not
One function that you could use is List.except, which is not yet documented (!) but can be seen in this pull request that was merged a couple of years ago. You'd probably use it like this:
let testIntersect a b =
let b' = b |> List.except a
// If b' is shorter than b, then b contained at least one element of a
List.length b' < List.length b
However, this runs through list B about three times, once to do the except algorithm and once each to do both the length calls. So another approach might be to do what you did, but turn list A into a set so that the exists call won't be O(N):
let testIntersect a b =
let setA = a |> Set.ofList
match b |> List.tryFind (fun x -> setA |> Set.contains x) with
| Some _ -> true
| None -> false
The reason I used tryFind is because List.find would throw an exception if the predicate didn't match any items of the list.
Edit: An even better approach is to use List.exists, which I temporarily forgot about (thanks to Honza Brestan for reminding me about it):
let testIntersect a b =
let setA = a |> Set.ofList
b |> List.exists (fun x -> setA |> Set.contains x)
Which, of course, is pretty much what you were originally wanting to do in your testIntersect code sample. The only difference is that you were using the for ... in syntax in your code sample, which wouldn't work. In F#, the for loop is exclusively for expressions that return unit (and thus, probably have side effects). If you want to return a value, the for loop won't do that. So using the functions that do return value, like List.exists, is the approach you want to take.
let testIntersect listA listB =
(Set.ofList listA) - (Set.ofList listB) |> Set.isEmpty |> not

(How) can I make this monadic bind tail-recursive?

I have this monad called Desync -
[<AutoOpen>]
module DesyncModule =
/// The Desync monad. Allows the user to define in a sequential style an operation that spans
/// across a bounded number of events. Span is bounded because I've yet to figure out how to
/// make Desync implementation tail-recursive (see note about unbounded recursion in bind). And
/// frankly, I'm not sure if there is a tail-recursive implementation of it...
type [<NoComparison; NoEquality>] Desync<'e, 's, 'a> =
Desync of ('s -> 's * Either<'e -> Desync<'e, 's, 'a>, 'a>)
/// Monadic return for the Desync monad.
let internal returnM (a : 'a) : Desync<'e, 's, 'a> =
Desync (fun s -> (s, Right a))
/// Monadic bind for the Desync monad.
let rec internal bind (m : Desync<'e, 's, 'a>) (cont : 'a -> Desync<'e, 's, 'b>) : Desync<'e, 's, 'b> =
Desync (fun s ->
match (match m with Desync f -> f s) with
// ^--- NOTE: unbounded recursion here
| (s', Left m') -> (s', Left (fun e -> bind (m' e) cont))
| (s', Right v) -> match cont v with Desync f -> f s')
/// Builds the Desync monad.
type DesyncBuilder () =
member this.Return op = returnM op
member this.Bind (m, cont) = bind m cont
/// The Desync builder.
let desync = DesyncBuilder ()
It allows the implementation of game logic that executes across several game ticks to written in a seemingly sequential style using computation expressions.
Unfortunately, when used for tasks that last for an unbounded number of game ticks, it crashes with StackOverflowException. And even when it's not crashing, it's ending up with unwieldy stack traces like this -
InfinityRpg.exe!InfinityRpg.GameplayDispatcherModule.desync#525-20.Invoke(Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen> _arg10) Line 530 F#
Prime.exe!Prime.DesyncModule.bind#20<Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>,Nu.SimulationModule.World,Microsoft.FSharp.Core.Unit,Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>>.Invoke(Nu.SimulationModule.World s) Line 24 F#
Prime.exe!Prime.DesyncModule.bind#20<Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>,Nu.SimulationModule.World,Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>.Invoke(Nu.SimulationModule.World s) Line 21 F#
Prime.exe!Prime.DesyncModule.bind#20<Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>,Nu.SimulationModule.World,Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>.Invoke(Nu.SimulationModule.World s) Line 21 F#
Prime.exe!Prime.DesyncModule.bind#20<Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>,Nu.SimulationModule.World,Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>.Invoke(Nu.SimulationModule.World s) Line 21 F#
Prime.exe!Prime.DesyncModule.bind#20<Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>,Nu.SimulationModule.World,Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>.Invoke(Nu.SimulationModule.World s) Line 21 F#
Prime.exe!Prime.DesyncModule.bind#20<Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>,Nu.SimulationModule.World,Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>.Invoke(Nu.SimulationModule.World s) Line 21 F#
Prime.exe!Prime.DesyncModule.bind#20<Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>,Nu.SimulationModule.World,Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>.Invoke(Nu.SimulationModule.World s) Line 21 F#
Prime.exe!Prime.DesyncModule.bind#20<Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>,Nu.SimulationModule.World,Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>.Invoke(Nu.SimulationModule.World s) Line 21 F#
Prime.exe!Prime.Desync.step<Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>,Nu.SimulationModule.World,Microsoft.FSharp.Core.Unit>(Prime.DesyncModule.Desync<Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>,Nu.SimulationModule.World,Microsoft.FSharp.Core.Unit> m, Nu.SimulationModule.World s) Line 71 F#
Prime.exe!Prime.Desync.advanceDesync<Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>,Nu.SimulationModule.World,Microsoft.FSharp.Core.Unit>(Microsoft.FSharp.Core.FSharpFunc<Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>,Prime.DesyncModule.Desync<Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>,Nu.SimulationModule.World,Microsoft.FSharp.Core.Unit>> m, Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen> e, Nu.SimulationModule.World s) Line 75 F#
Nu.exe!Nu.Desync.advance#98<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>.Invoke(Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen> event, Nu.SimulationModule.World world) Line 100 F#
Nu.exe!Nu.Desync.subscription#104-16<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>.Invoke(Nu.SimulationModule.Event<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen> event, Nu.SimulationModule.World world) Line 105 F#
Nu.exe!Nu.World.boxableSubscription#165<Prime.EitherModule.Either<Microsoft.FSharp.Core.Unit,Microsoft.FSharp.Core.Unit>,Nu.SimulationModule.Screen>.Invoke(object event, Nu.SimulationModule.World world) Line 166 F#
I am hoping to solve the problem by making the Left case of the bind function tail-recursive. However, I'm not sure of two things -
1) if it can be done at all, and
2) how it would actually be done.
If it's impossible to make bind tail-recursive here, is there some way to restructure my monad to allow it to become tail-recursive?
EDIT 3 (subsumes previous edits): Here is additional code that implements the desync combinators I will use to demonstrate the stack overflow -
module Desync =
/// Get the state.
let get : Desync<'e, 's, 's> =
Desync (fun s -> (s, Right s))
/// Set the state.
let set s : Desync<'e, 's, unit> =
Desync (fun _ -> (s, Right ()))
/// Loop in a desynchronous context while 'pred' evaluate to true.
let rec loop (i : 'i) (next : 'i -> 'i) (pred : 'i -> 's -> bool) (m : 'i -> Desync<'e, 's, unit>) =
desync {
let! s = get
do! if pred i s then
desync {
do! m i
let i = next i
do! loop i next pred m }
else returnM () }
/// Loop in a desynchronous context while 'pred' evaluates to true.
let during (pred : 's -> bool) (m : Desync<'e, 's, unit>) =
loop () id (fun _ -> pred) (fun _ -> m)
/// Step once into a desync.
let step (m : Desync<'e, 's, 'a>) (s : 's) : 's * Either<'e -> Desync<'e, 's, 'a>, 'a> =
match m with Desync f -> f s
/// Run a desync to its end, providing e for all its steps.
let rec runDesync (m : Desync<'e, 's, 'a>) (e : 'e) (s : 's) : ('s * 'a) =
match step m s with
| (s', Left m') -> runDesync (m' e) e s'
| (s', Right v) -> (s', v)
Here is the Either implementation -
[<AutoOpen>]
module EitherModule =
/// Haskell-style Either type.
type Either<'l, 'r> =
| Right of 'r
| Left of 'l
And finally, here's simple a line of code that will yield a stack overflow -
open Desync
ignore <| runDesync (desync { do! during (fun _ -> true) (returnM ()) }) () ()
It seems to me your monad is a State with error handling.
It's basically ErrorT< State<'s,Either<'e,'a>>> but the error branch binds again which is not very clear to me why.
Anyway I was able to reproduce your Stack Overflow with a basic State monad:
type State<'S,'A> = State of ('S->('A * 'S))
module State =
let run (State x) = x :'s->_
let get() = State (fun s -> (s , s)) :State<'s,_>
let put x = State (fun _ -> ((), x)) :State<'s,_>
let result a = State(fun s -> (a, s))
let bind (State m) k = State(fun s ->
let (a, s') = m s
let (State u) = (k a)
u s') :State<'s,'b>
type StateBuilder() =
member this.Return op = result op
member this.Bind (m, cont) = bind m cont
let state = StateBuilder()
let rec loop (i: 'i) (next: 'i -> 'i) (pred: 'i -> 's -> bool) (m: 'i -> State<'s, unit>) =
state {
let! s = get()
do! if pred i s then
state {
do! m i
let i = next i
do! loop i next pred m }
else result () }
let during (pred : 's -> bool) (m : State<'s, unit>) =
loop () id (fun _ -> pred) (fun _ -> m)
// test
open State
ignore <| run (state { do! during (fun c -> true) (result ()) }) () // boom
As stated in the comments one way to solve this is to use a StateT<'s,Cont<'r,'a>>.
Here's an example of the solution. At the end there is a test with the zipIndex function which blows the stack as well when defined with a normal State monad.
Note you don't need to use the Monad Transformers from FsControl (now FSharpPlus), I use them because it's easier for me since I write less code but you can always create your transformed monad by hand.

How do I write a computation expression builder that accumulates a value and also allows standard language constructs?

I have a computation expression builder that builds up a value as you go, and has many custom operations. However, it does not allow for standard F# language constructs, and I'm having a lot of trouble figuring out how to add this support.
To give a stand-alone example, here's a dead-simple and fairly pointless computation expression that builds F# lists:
type Items<'a> = Items of 'a list
type ListBuilder() =
member x.Yield(()) = Items []
[<CustomOperation("add")>]
member x.Add(Items current, item:'a) =
Items [ yield! current; yield item ]
[<CustomOperation("addMany")>]
member x.AddMany(Items current, items: seq<'a>) =
Items [ yield! current; yield! items ]
let listBuilder = ListBuilder()
let build (Items items) = items
I can use this to build lists just fine:
let stuff =
listBuilder {
add 1
add 5
add 7
addMany [ 1..10 ]
add 42
}
|> build
However, this is a compiler error:
listBuilder {
let x = 5 * 39
add x
}
// This expression was expected to have type unit, but
// here has type int.
And so is this:
listBuilder {
for x = 1 to 50 do
add x
}
// This control construct may only be used if the computation expression builder
// defines a For method.
I've read all the documentation and examples I can find, but there's something I'm just not getting. Every .Bind() or .For() method signature I try just leads to more and more confusing compiler errors. Most of the examples I can find either build up a value as you go along, or allow for regular F# language constructs, but I haven't been able to find one that does both.
If someone could point me in the right direction by showing me how to take this example and add support in the builder for let bindings and for loops (at minimum - using, while and try/catch would be great, but I can probably figure those out if someone gets me started) then I'll be able to gratefully apply the lesson to my actual problem.
The best place to look is the spec. For example,
b {
let x = e
op x
}
gets translated to
T(let x = e in op x, [], fun v -> v, true)
=> T(op x, {x}, fun v -> let x = e in v, true)
=> [| op x, let x = e in b.Yield(x) |]{x}
=> b.Op(let x = e in in b.Yield(x), x)
So this shows where things have gone wrong, though it doesn't present an obvious solution. Clearly, Yield needs to be generalized since it needs to take arbitrary tuples (based on how many variables are in scope). Perhaps more subtly, it also shows that x is not in scope in the call to add (see that unbound x as the second argument to b.Op?). To allow your custom operators to use bound variables, their arguments need to have the [<ProjectionParameter>] attribute (and take functions from arbitrary variables as arguments), and you'll also need to set MaintainsVariableSpace to true if you want bound variables to be available to later operators. This will change the final translation to:
b.Op(let x = e in b.Yield(x), fun x -> x)
Building up from this, it seems that there's no way to avoid passing the set of bound values along to and from each operation (though I'd love to be proven wrong) - this will require you to add a Run method to strip those values back off at the end. Putting it all together, you'll get a builder which looks like this:
type ListBuilder() =
member x.Yield(vars) = Items [],vars
[<CustomOperation("add",MaintainsVariableSpace=true)>]
member x.Add((Items current,vars), [<ProjectionParameter>]f) =
Items (current # [f vars]),vars
[<CustomOperation("addMany",MaintainsVariableSpace=true)>]
member x.AddMany((Items current, vars), [<ProjectionParameter>]f) =
Items (current # f vars),vars
member x.Run(l,_) = l
The most complete examples I've seen are in §6.3.10 of the spec, especially this one:
/// Computations that can cooperatively yield by returning a continuation
type Eventually<'T> =
| Done of 'T
| NotYetDone of (unit -> Eventually<'T>)
[<CompilationRepresentation(CompilationRepresentationFlags.ModuleSuffix)>]
module Eventually =
/// The bind for the computations. Stitch 'k' on to the end of the computation.
/// Note combinators like this are usually written in the reverse way,
/// for example,
/// e |> bind k
let rec bind k e =
match e with
| Done x -> NotYetDone (fun () -> k x)
| NotYetDone work -> NotYetDone (fun () -> bind k (work()))
/// The return for the computations.
let result x = Done x
type OkOrException<'T> =
| Ok of 'T
| Exception of System.Exception
/// The catch for the computations. Stitch try/with throughout
/// the computation and return the overall result as an OkOrException.
let rec catch e =
match e with
| Done x -> result (Ok x)
| NotYetDone work ->
NotYetDone (fun () ->
let res = try Ok(work()) with | e -> Exception e
match res with
| Ok cont -> catch cont // note, a tailcall
| Exception e -> result (Exception e))
/// The delay operator.
let delay f = NotYetDone (fun () -> f())
/// The stepping action for the computations.
let step c =
match c with
| Done _ -> c
| NotYetDone f -> f ()
// The rest of the operations are boilerplate.
/// The tryFinally operator.
/// This is boilerplate in terms of "result", "catch" and "bind".
let tryFinally e compensation =
catch (e)
|> bind (fun res -> compensation();
match res with
| Ok v -> result v
| Exception e -> raise e)
/// The tryWith operator.
/// This is boilerplate in terms of "result", "catch" and "bind".
let tryWith e handler =
catch e
|> bind (function Ok v -> result v | Exception e -> handler e)
/// The whileLoop operator.
/// This is boilerplate in terms of "result" and "bind".
let rec whileLoop gd body =
if gd() then body |> bind (fun v -> whileLoop gd body)
else result ()
/// The sequential composition operator
/// This is boilerplate in terms of "result" and "bind".
let combine e1 e2 =
e1 |> bind (fun () -> e2)
/// The using operator.
let using (resource: #System.IDisposable) f =
tryFinally (f resource) (fun () -> resource.Dispose())
/// The forLoop operator.
/// This is boilerplate in terms of "catch", "result" and "bind".
let forLoop (e:seq<_>) f =
let ie = e.GetEnumerator()
tryFinally (whileLoop (fun () -> ie.MoveNext())
(delay (fun () -> let v = ie.Current in f v)))
(fun () -> ie.Dispose())
// Give the mapping for F# computation expressions.
type EventuallyBuilder() =
member x.Bind(e,k) = Eventually.bind k e
member x.Return(v) = Eventually.result v
member x.ReturnFrom(v) = v
member x.Combine(e1,e2) = Eventually.combine e1 e2
member x.Delay(f) = Eventually.delay f
member x.Zero() = Eventually.result ()
member x.TryWith(e,handler) = Eventually.tryWith e handler
member x.TryFinally(e,compensation) = Eventually.tryFinally e compensation
member x.For(e:seq<_>,f) = Eventually.forLoop e f
member x.Using(resource,e) = Eventually.using resource e
The tutorial at "F# for fun and profit" is first class in this regard.
http://fsharpforfunandprofit.com/posts/computation-expressions-intro/
Following a similar struggle to Joel's (and not finding §6.3.10 of the spec that helpful) my issue with getting the For construct to generate a list came down to getting types to line up properly (no special attributes required). In particular I was slow to realise that For would build a list of lists, and therefore need flattening, despite the best efforts of the compiler to put me right. Examples that I found on the web were always wrappers around seq{}, using the yield keyword, repeated use of which invokes a call to Combine, which does the flattening. In case a concrete example helps, the following excerpt uses for to build a list of integers - my ultimate aim being to create lists of components for rendering in a GUI (with some additional laziness thrown in). Also In depth talk on CE here which elaborates on kvb's points above.
module scratch
type Dispatcher = unit -> unit
type viewElement = int
type lazyViews = Lazy<list<viewElement>>
type ViewElementsBuilder() =
member x.Return(views: lazyViews) : list<viewElement> = views.Value
member x.Yield(v: viewElement) : list<viewElement> = [v]
member x.ReturnFrom(viewElements: list<viewElement>) = viewElements
member x.Zero() = list<viewElement>.Empty
member x.Combine(listA:list<viewElement>, listB: list<viewElement>) = List.concat [listA; listB]
member x.Delay(f) = f()
member x.For(coll:seq<'a>, forBody: 'a -> list<viewElement>) : list<viewElement> =
// seq {for v in coll do yield! f v} |> List.ofSeq
Seq.map forBody coll |> Seq.collect id |> List.ofSeq
let ve = new ViewElementsBuilder()
let makeComponent(m: int, dispatch: Dispatcher) : viewElement = m
let makeComponents() : list<viewElement> = [77; 33]
let makeViewElements() : list<viewElement> =
let model = {| Scores = [33;23;22;43;] |> Seq.ofList; Trainer = "John" |}
let d:Dispatcher = fun() -> () // Does nothing here, but will be used to raise messages from UI
ve {
for score in model.Scores do
yield makeComponent (score, d)
yield makeComponent (score * 100 / 50 , d)
if model.Trainer = "John" then
return lazy
[ makeComponent (12, d)
makeComponent (13, d)
]
else
return lazy
[ makeComponent (14, d)
makeComponent (15, d)
]
yield makeComponent (33, d)
return! makeComponents()
}

Recursively update a State Monad

this question is related to this question
I have a state monad. An object provides an update function as in the OOD strategy pattern.
The choice of having a object is that in real, production code, the
class provides an array of operations, all sharing state through the
monad. Inheritance helped me extend the basic functionality and
further customizing the class providing the operations.
The choice of having a monad instead of a mutable property within the class is that the monad, through proper use of generics, is helping me abstracting and being more flexible on what variables/information must be carried along the computation as "state".
I have a simple toy example:
/////////////////////////////////////////////////////////////////////////////////////
// Definition of the state
/////////////////////////////////////////////////////////////////////////////////////
type StateFunc<'State, 'T> = 'State -> 'T * 'State
/////////////////////////////////////////////////////////////////////////////////////
// Definition of the State monad type
/////////////////////////////////////////////////////////////////////////////////////
type StateMonadBuilder<'State>() =
// M<'T> -> M<'T>
member b.ReturnFrom a : StateFunc<'State, 'T> = a
// 'T -> M<'T>
member b.Return a : StateFunc<'State, 'T> = ( fun s -> a, s)
// M<'T> * ('T -> M<'U>) -> M<'U>
member b.Bind(p : StateFunc<_, 'T>, rest : 'T -> StateFunc<_,_>) : StateFunc<'State, 'U> =
(fun s ->
let a, s' = p s
rest a s')
// Getter for the whole state, this type signature is because it passes along the state & returns the state
member b.getState : StateFunc<'State, _> = (fun s -> s, s)
// Setter for the state
member b.putState (s:'State) : StateFunc<'State, _> = (fun _ -> (), s)
let runState f init = f init
/////////////////////////////////////////////////////////////////////////////////////
// STRATEGY PATTERN
/////////////////////////////////////////////////////////////////////////////////////
let state = StateMonadBuilder<int> ()
// DoubleFunctOne defines standard operations that remain always the same
type Strategy (aFunction) =
member this.Update (x: int) = state {
let! currState = state.getState
let processedx = aFunction x
do! state.putState (currState + x) }
// Create a function that customizes the strategy
let myFunction x =
2 * x
// Customize the strategy with the desired function:
let strategy = Strategy (myFunction)
/////////////////////////////////////////////////////////////////////////////////////////////////////////
// Update recursively
/////////////////////////////////////////////////////////////////////////////////////////////////////////
// ?? How to run update recursively ??
let result initialCondition =
initialCondition
|> (for i = 10 to 100 do
yield state { do! strategy.Update i } )
My goal is to apply the initial conditions, fetch data and launch recursively (within a for or a while loop or even some functional operation) the functions provided by strategy. Working with the monad, I am not sure how to do this.
Thank you.
Computational Expression For
Inspired by #kvb answer, I have added a for method to the computational expression.
// Loops through seqnc of numbers that constitute an input to func
member b.For (seqnc:_ List, func) =
seqnc
|> List.map (fun item -> func item)
|> List.reduce (fun acc item ->
(fun s ->
let _, s' = acc s
item s' ) )
I run a few tests and I have the impression that this one works.
Thanks.
Something like this?
let result initialCondition =
let rec loop = function
| 101 -> state { return () }
| i ->
state {
do! strategy.Update i
do! loop (i+1)
}
initialCondition
|> runState (loop 10)
Alternatively, define a For member on your builder and write it the more imperative way:
let result initialCondition =
let f = state {
for i in 10 to 100 do
do! strategy.Update i
}
initialCondition
|> runState f
Also, note that there is likely a bug in your definition of Strategy.Update: processedx is bound but unused.

Handy F# snippets [closed]

As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 10 years ago.
There are already two questions about F#/functional snippets.
However what I'm looking for here are useful snippets, little 'helper' functions that are reusable. Or obscure but nifty patterns that you can never quite remember.
Something like:
open System.IO
let rec visitor dir filter=
seq { yield! Directory.GetFiles(dir, filter)
for subdir in Directory.GetDirectories(dir) do
yield! visitor subdir filter}
I'd like to make this a kind of handy reference page. As such there will be no right answer, but hopefully lots of good ones.
EDIT Tomas Petricek has created a site specifically for F# snippets http://fssnip.net/.
Perl style regex matching
let (=~) input pattern =
System.Text.RegularExpressions.Regex.IsMatch(input, pattern)
It lets you match text using let test = "monkey" =~ "monk.+" notation.
Infix Operator
I got this from http://sandersn.com/blog//index.php/2009/10/22/infix-function-trick-for-f go to that page for more details.
If you know Haskell, you might find yourself missing infix sugar in F#:
// standard Haskell call has function first, then args just like F#. So obviously
// here there is a function that takes two strings: string -> string -> string
startsWith "kevin" "k"
//Haskell infix operator via backQuotes. Sometimes makes a function read better.
"kevin" `startsWith` "K"
While F# doesn't have a true 'infix' operator, the same thing can be accomplished almost as elegantly via a pipeline and a 'backpipeline' (who knew of such a thing??)
// F# 'infix' trick via pipelines
"kevin" |> startsWith <| "K"
Multi-Line Strings
This is pretty trivial, but it seems to be a feature of F# strings that is not widely known.
let sql = "select a,b,c \
from table \
where a = 1"
This produces:
val sql : string = "select a,b,c from table where a = 1"
When the F# compiler sees a back-slash followed by a carriage return inside a string literal, it will remove everything from the back-slash to the first non-space character on the next line. This allows you to have multi-line string literals that line up, without using a bunch of string concatenation.
Generic memoization, courtesy of the man himself
let memoize f =
let cache = System.Collections.Generic.Dictionary<_,_>(HashIdentity.Structural)
fun x ->
let ok, res = cache.TryGetValue(x)
if ok then res
else let res = f x
cache.[x] <- res
res
Using this, you could do a cached reader like so:
let cachedReader = memoize reader
Simple read-write to text files
These are trivial, but make file access pipeable:
open System.IO
let fileread f = File.ReadAllText(f)
let filewrite f s = File.WriteAllText(f, s)
let filereadlines f = File.ReadAllLines(f)
let filewritelines f ar = File.WriteAllLines(f, ar)
So
let replace f (r:string) (s:string) = s.Replace(f, r)
"C:\\Test.txt" |>
fileread |>
replace "teh" "the" |>
filewrite "C:\\Test.txt"
And combining that with the visitor quoted in the question:
let filereplace find repl path =
path |> fileread |> replace find repl |> filewrite path
let recurseReplace root filter find repl =
visitor root filter |> Seq.iter (filereplace find repl)
Update Slight improvement if you want to be able to read 'locked' files (e.g. csv files which are already open in Excel...):
let safereadall f =
use fs = new FileStream(f, FileMode.Open, FileAccess.Read, FileShare.ReadWrite)
use sr = new StreamReader(fs, System.Text.Encoding.Default)
sr.ReadToEnd()
let split sep (s:string) = System.Text.RegularExpressions.Regex.Split(s, sep)
let fileread f = safereadall f
let filereadlines f = f |> safereadall |> split System.Environment.NewLine
For performance intensive stuff where you need to check for null
let inline isNull o = System.Object.ReferenceEquals(o, null)
if isNull o then ... else ...
Is about 20x faster then
if o = null then ... else ...
Active Patterns, aka "Banana Splits", are a very handy construct that let one match against multiple regular expression patterns. This is much like AWK, but without the high performance of DFA's because the patterns are matched in sequence until one succeeds.
#light
open System
open System.Text.RegularExpressions
let (|Test|_|) pat s =
if (new Regex(pat)).IsMatch(s)
then Some()
else None
let (|Match|_|) pat s =
let opt = RegexOptions.None
let re = new Regex(pat,opt)
let m = re.Match(s)
if m.Success
then Some(m.Groups)
else None
Some examples of use:
let HasIndefiniteArticle = function
| Test "(?: |^)(a|an)(?: |$)" _ -> true
| _ -> false
type Ast =
| IntVal of string * int
| StringVal of string * string
| LineNo of int
| Goto of int
let Parse = function
| Match "^LET\s+([A-Z])\s*=\s*(\d+)$" g ->
IntVal( g.[1].Value, Int32.Parse(g.[2].Value) )
| Match "^LET\s+([A-Z]\$)\s*=\s*(.*)$" g ->
StringVal( g.[1].Value, g.[2].Value )
| Match "^(\d+)\s*:$" g ->
LineNo( Int32.Parse(g.[1].Value) )
| Match "^GOTO \s*(\d+)$" g ->
Goto( Int32.Parse(g.[1].Value) )
| s -> failwithf "Unexpected statement: %s" s
Maybe monad
type maybeBuilder() =
member this.Bind(v, f) =
match v with
| None -> None
| Some(x) -> f x
member this.Delay(f) = f()
member this.Return(v) = Some v
let maybe = maybeBuilder()
Here's a brief intro to monads for the uninitiated.
Option-coalescing operators
I wanted a version of the defaultArg function that had a syntax closer to the C# null-coalescing operator, ??. This lets me get the value from an Option while providing a default value, using a very concise syntax.
/// Option-coalescing operator - this is like the C# ?? operator, but works with
/// the Option type.
/// Warning: Unlike the C# ?? operator, the second parameter will always be
/// evaluated.
/// Example: let foo = someOption |? default
let inline (|?) value defaultValue =
defaultArg value defaultValue
/// Option-coalescing operator with delayed evaluation. The other version of
/// this operator always evaluates the default value expression. If you only
/// want to create the default value when needed, use this operator and pass
/// in a function that creates the default.
/// Example: let foo = someOption |?! (fun () -> new Default())
let inline (|?!) value f =
match value with Some x -> x | None -> f()
'Unitize' a function which doesn't handle units
Using the FloatWithMeasure function http://msdn.microsoft.com/en-us/library/ee806527(VS.100).aspx.
let unitize (f:float -> float) (v:float<'u>) =
LanguagePrimitives.FloatWithMeasure<'u> (f (float v))
Example:
[<Measure>] type m
[<Measure>] type kg
let unitize (f:float -> float) (v:float<'u>) =
LanguagePrimitives.FloatWithMeasure<'u> (f (float v))
//this function doesn't take units
let badinc a = a + 1.
//this one does!
let goodinc v = unitize badinc v
goodinc 3.<m>
goodinc 3.<kg>
OLD version:
let unitize (f:float -> float) (v:float<'u>) =
let unit = box 1. :?> float<'u>
unit * (f (v/unit))
Kudos to kvb
Scale/Ratio function builder
Again, trivial, but handy.
//returns a function which will convert from a1-a2 range to b1-b2 range
let scale (a1:float<'u>, a2:float<'u>) (b1:float<'v>,b2:float<'v>) =
let m = (b2 - b1)/(a2 - a1) //gradient of line (evaluated once only..)
(fun a -> b1 + m * (a - a1))
Example:
[<Measure>] type m
[<Measure>] type px
let screenSize = (0.<px>, 300.<px>)
let displayRange = (100.<m>, 200.<m>)
let scaleToScreen = scale displayRange screenSize
scaleToScreen 120.<m> //-> 60.<px>
Transposing a list (seen on Jomo Fisher's blog)
///Given list of 'rows', returns list of 'columns'
let rec transpose lst =
match lst with
| (_::_)::_ -> List.map List.head lst :: transpose (List.map List.tail lst)
| _ -> []
transpose [[1;2;3];[4;5;6];[7;8;9]] // returns [[1;4;7];[2;5;8];[3;6;9]]
And here is a tail-recursive version which (from my sketchy profiling) is mildly slower, but has the advantage of not throwing a stack overflow when the inner lists are longer than 10000 elements (on my machine):
let transposeTR lst =
let rec inner acc lst =
match lst with
| (_::_)::_ -> inner (List.map List.head lst :: acc) (List.map List.tail lst)
| _ -> List.rev acc
inner [] lst
If I was clever, I'd try and parallelise it with async...
F# Map <-> C# Dictionary
(I know, I know, System.Collections.Generic.Dictionary isn't really a 'C#' dictionary)
C# to F#
(dic :> seq<_>) //cast to seq of KeyValuePair
|> Seq.map (|KeyValue|) //convert KeyValuePairs to tuples
|> Map.ofSeq //convert to Map
(From Brian, here, with improvement proposed by Mauricio in comment below. (|KeyValue|) is an active pattern for matching KeyValuePair - from FSharp.Core - equivalent to (fun kvp -> kvp.Key, kvp.Value))
Interesting alternative
To get all of the immutable goodness, but with the O(1) lookup speed of Dictionary, you can use the dict operator, which returns an immutable IDictionary (see this question).
I currently can't see a way to directly convert a Dictionary using this method, other than
(dic :> seq<_>) //cast to seq of KeyValuePair
|> (fun kvp -> kvp.Key, kvp.Value) //convert KeyValuePairs to tuples
|> dict //convert to immutable IDictionary
F# to C#
let dic = Dictionary()
map |> Map.iter (fun k t -> dic.Add(k, t))
dic
What is weird here is that FSI will report the type as (for example):
val it : Dictionary<string,int> = dict [("a",1);("b",2)]
but if you feed dict [("a",1);("b",2)] back in, FSI reports
IDictionary<string,int> = seq[[a,1] {Key = "a"; Value = 1; } ...
Tree-sort / Flatten a tree into a list
I have the following binary tree:
___ 77 _
/ \
______ 47 __ 99
/ \
21 _ 54
\ / \
43 53 74
/
39
/
32
Which is represented as follows:
type 'a tree =
| Node of 'a tree * 'a * 'a tree
| Nil
let myTree =
Node
(Node
(Node (Nil,21,Node (Node (Node (Nil,32,Nil),39,Nil),43,Nil)),47,
Node (Node (Nil,53,Nil),54,Node (Nil,74,Nil))),77,Node (Nil,99,Nil))
A straightforward method to flatten the tree is:
let rec flatten = function
| Nil -> []
| Node(l, a, r) -> flatten l # a::flatten r
This isn't tail-recursive, and I believe the # operator causes it to be O(n log n) or O(n^2) with unbalanced binary trees. With a little tweaking, I came up with this tail-recursive O(n) version:
let flatten2 t =
let rec loop acc c = function
| Nil -> c acc
| Node(l, a, r) ->
loop acc (fun acc' -> loop (a::acc') c l) r
loop [] (fun x -> x) t
Here's the output in fsi:
> flatten2 myTree;;
val it : int list = [21; 32; 39; 43; 47; 53; 54; 74; 77; 99]
LINQ-to-XML helpers
namespace System.Xml.Linq
// hide warning about op_Explicit
#nowarn "77"
[<AutoOpen>]
module XmlUtils =
/// Converts a string to an XName.
let xn = XName.op_Implicit
/// Converts a string to an XNamespace.
let xmlns = XNamespace.op_Implicit
/// Gets the string value of any XObject subclass that has a Value property.
let inline xstr (x : ^a when ^a :> XObject) =
(^a : (member get_Value : unit -> string) x)
/// Gets a strongly-typed value from any XObject subclass, provided that
/// an explicit conversion to the output type has been defined.
/// (Many explicit conversions are defined on XElement and XAttribute)
/// Example: let value:int = xval foo
let inline xval (x : ^a when ^a :> XObject) : ^b =
((^a or ^b) : (static member op_Explicit : ^a -> ^b) x)
/// Dynamic lookup operator for getting an attribute value from an XElement.
/// Returns a string option, set to None if the attribute was not present.
/// Example: let value = foo?href
/// Example with default: let value = defaultArg foo?Name "<Unknown>"
let (?) (el:XElement) (name:string) =
match el.Attribute(xn name) with
| null -> None
| att -> Some(att.Value)
/// Dynamic operator for setting an attribute on an XElement.
/// Example: foo?href <- "http://www.foo.com/"
let (?<-) (el:XElement) (name:string) (value:obj) =
el.SetAttributeValue(xn name, value)
OK, this has nothing to do with snippets, but I keep forgetting this:
If you are in the interactive window, you hit F7 to jump back to the code window (without deselecting the code which you just ran...)
Going from code window to F# window (and also to open the F# window) is Ctrl Alt F
(unless CodeRush has stolen your bindings...)
Weighted sum of arrays
Calculating a weighted [n-array] sum of a [k-array of n-arrays] of numbers, based on a [k-array] of weights
(Copied from this question, and kvb's answer)
Given these arrays
let weights = [|0.6;0.3;0.1|]
let arrs = [| [|0.0453;0.065345;0.07566;1.562;356.6|] ;
[|0.0873;0.075565;0.07666;1.562222;3.66|] ;
[|0.06753;0.075675;0.04566;1.452;3.4556|] |]
We want a weighted sum (by column), given that both dimensions of the arrays can be variable.
Array.map2 (fun w -> Array.map ((*) w)) weights arrs
|> Array.reduce (Array.map2 (+))
First line: Partial application of the first Array.map2 function to weights yields a new function (Array.map ((*) weight) which is applied (for each weight) to each array in arr.
Second line: Array.reduce is like fold, except it starts on the second value and uses the first as the initial 'state'. In this case each value is a 'line' of our array of arrays. So applying an Array.map2 (+) on the first two lines means that we sum the first two arrays, which leaves us with a new array, which we then (Array.reduce) sum again onto the next (in this case last) array.
Result:
[|0.060123; 0.069444; 0.07296; 1.5510666; 215.40356|]
Performance testing
(Found here and updated for latest release of F#)
open System
open System.Diagnostics
module PerformanceTesting =
let Time func =
let stopwatch = new Stopwatch()
stopwatch.Start()
func()
stopwatch.Stop()
stopwatch.Elapsed.TotalMilliseconds
let GetAverageTime timesToRun func =
Seq.initInfinite (fun _ -> (Time func))
|> Seq.take timesToRun
|> Seq.average
let TimeOperation timesToRun =
GC.Collect()
GetAverageTime timesToRun
let TimeOperations funcsWithName =
let randomizer = new Random(int DateTime.Now.Ticks)
funcsWithName
|> Seq.sortBy (fun _ -> randomizer.Next())
|> Seq.map (fun (name, func) -> name, (TimeOperation 100000 func))
let TimeOperationsAFewTimes funcsWithName =
Seq.initInfinite (fun _ -> (TimeOperations funcsWithName))
|> Seq.take 50
|> Seq.concat
|> Seq.groupBy fst
|> Seq.map (fun (name, individualResults) -> name, (individualResults |> Seq.map snd |> Seq.average))
DataSetExtensions for F#, DataReaders
System.Data.DataSetExtensions.dll adds the ability to treat a DataTable as an IEnumerable<DataRow> as well as unboxing the values of individual cells in a way that gracefully handles DBNull by supporting System.Nullable. For example, in C# we can get the value of an integer column that contains nulls, and specify that DBNull should default to zero with a very concise syntax:
var total = myDataTable.AsEnumerable()
.Select(row => row.Field<int?>("MyColumn") ?? 0)
.Sum();
There are two areas where DataSetExtensions are lacking, however. First, it doesn't support IDataReader and second, it doesn't support the F# option type. The following code does both - it allows an IDataReader to be treated as a seq<IDataRecord>, and it can unbox values from either a reader or a dataset, with support for F# options or System.Nullable. Combined with the option-coalescing operator in another answer, this allows for code such as the following when working with a DataReader:
let total =
myReader.AsSeq
|> Seq.map (fun row -> row.Field<int option>("MyColumn") |? 0)
|> Seq.sum
Perhaps a more idiomatic F# way of ignoring database nulls would be...
let total =
myReader.AsSeq
|> Seq.choose (fun row -> row.Field<int option>("MyColumn"))
|> Seq.sum
Further, the extension methods defined below are usable from both F# and from C#/VB.
open System
open System.Data
open System.Reflection
open System.Runtime.CompilerServices
open Microsoft.FSharp.Collections
/// Ported from System.Data.DatasetExtensions.dll to add support for the Option type.
[<AbstractClass; Sealed>]
type private UnboxT<'a> private () =
// This class generates a converter function based on the desired output type,
// and then re-uses the converter function forever. Because the class itself is generic,
// different output types get different cached converter functions.
static let referenceField (value:obj) =
if value = null || DBNull.Value.Equals(value) then
Unchecked.defaultof<'a>
else
unbox value
static let valueField (value:obj) =
if value = null || DBNull.Value.Equals(value) then
raise <| InvalidCastException("Null cannot be converted to " + typeof<'a>.Name)
else
unbox value
static let makeConverter (target:Type) methodName =
Delegate.CreateDelegate(typeof<Converter<obj,'a>>,
typeof<UnboxT<'a>>
.GetMethod(methodName, BindingFlags.NonPublic ||| BindingFlags.Static)
.MakeGenericMethod([| target.GetGenericArguments().[0] |]))
|> unbox<Converter<obj,'a>>
|> FSharpFunc.FromConverter
static let unboxFn =
let theType = typeof<'a>
if theType.IsGenericType && not theType.IsGenericTypeDefinition then
let genericType = theType.GetGenericTypeDefinition()
if typedefof<Nullable<_>> = genericType then
makeConverter theType "NullableField"
elif typedefof<option<_>> = genericType then
makeConverter theType "OptionField"
else
invalidOp "The only generic types supported are Option<T> and Nullable<T>."
elif theType.IsValueType then
valueField
else
referenceField
static member private NullableField<'b when 'b : struct and 'b :> ValueType and 'b:(new:unit -> 'b)> (value:obj) =
if value = null || DBNull.Value.Equals(value) then
Nullable<_>()
else
Nullable<_>(unbox<'b> value)
static member private OptionField<'b> (value:obj) =
if value = null || DBNull.Value.Equals(value) then
None
else
Some(unbox<'b> value)
static member inline Unbox =
unboxFn
/// F# data-related extension methods.
[<AutoOpen>]
module FsDataEx =
type System.Data.IDataReader with
/// Exposes a reader's current result set as seq<IDataRecord>.
/// Reader is closed when sequence is fully enumerated.
member this.AsSeq =
seq { use reader = this
while reader.Read() do yield reader :> IDataRecord }
/// Exposes all result sets in a reader as seq<seq<IDataRecord>>.
/// Reader is closed when sequence is fully enumerated.
member this.AsMultiSeq =
let rowSeq (reader:IDataReader) =
seq { while reader.Read() do yield reader :> IDataRecord }
seq {
use reader = this
yield rowSeq reader
while reader.NextResult() do
yield rowSeq reader
}
/// Populates a new DataSet with the contents of the reader. Closes the reader after completion.
member this.ToDataSet () =
use reader = this
let dataSet = new DataSet(RemotingFormat=SerializationFormat.Binary, EnforceConstraints=false)
dataSet.Load(reader, LoadOption.OverwriteChanges, [| "" |])
dataSet
type System.Data.IDataRecord with
/// Gets a value from the record by name.
/// DBNull and null are returned as the default value for the type.
/// Supports both nullable and option types.
member this.Field<'a> (fieldName:string) =
this.[fieldName] |> UnboxT<'a>.Unbox
/// Gets a value from the record by column index.
/// DBNull and null are returned as the default value for the type.
/// Supports both nullable and option types.
member this.Field<'a> (ordinal:int) =
this.GetValue(ordinal) |> UnboxT<'a>.Unbox
type System.Data.DataRow with
/// Identical to the Field method from DatasetExtensions, but supports the F# Option type.
member this.Field2<'a> (columnName:string) =
this.[columnName] |> UnboxT<'a>.Unbox
/// Identical to the Field method from DatasetExtensions, but supports the F# Option type.
member this.Field2<'a> (columnIndex:int) =
this.[columnIndex] |> UnboxT<'a>.Unbox
/// Identical to the Field method from DatasetExtensions, but supports the F# Option type.
member this.Field2<'a> (column:DataColumn) =
this.[column] |> UnboxT<'a>.Unbox
/// Identical to the Field method from DatasetExtensions, but supports the F# Option type.
member this.Field2<'a> (columnName:string, version:DataRowVersion) =
this.[columnName, version] |> UnboxT<'a>.Unbox
/// Identical to the Field method from DatasetExtensions, but supports the F# Option type.
member this.Field2<'a> (columnIndex:int, version:DataRowVersion) =
this.[columnIndex, version] |> UnboxT<'a>.Unbox
/// Identical to the Field method from DatasetExtensions, but supports the F# Option type.
member this.Field2<'a> (column:DataColumn, version:DataRowVersion) =
this.[column, version] |> UnboxT<'a>.Unbox
/// C# data-related extension methods.
[<Extension; AbstractClass; Sealed>]
type CsDataEx private () =
/// Populates a new DataSet with the contents of the reader. Closes the reader after completion.
[<Extension>]
static member ToDataSet(this:IDataReader) =
this.ToDataSet()
/// Exposes a reader's current result set as IEnumerable{IDataRecord}.
/// Reader is closed when sequence is fully enumerated.
[<Extension>]
static member AsEnumerable(this:IDataReader) =
this.AsSeq
/// Exposes all result sets in a reader as IEnumerable{IEnumerable{IDataRecord}}.
/// Reader is closed when sequence is fully enumerated.
[<Extension>]
static member AsMultipleEnumerable(this:IDataReader) =
this.AsMultiSeq
/// Gets a value from the record by name.
/// DBNull and null are returned as the default value for the type.
/// Supports both nullable and option types.
[<Extension>]
static member Field<'T> (this:IDataRecord, fieldName:string) =
this.Field<'T>(fieldName)
/// Gets a value from the record by column index.
/// DBNull and null are returned as the default value for the type.
/// Supports both nullable and option types.
[<Extension>]
static member Field<'T> (this:IDataRecord, ordinal:int) =
this.Field<'T>(ordinal)
Handling arguments in a command line application:
//We assume that the actual meat is already defined in function
// DoStuff (string -> string -> string -> unit)
let defaultOutOption = "N"
let defaultUsageOption = "Y"
let usage =
"Scans a folder for and outputs results.\n" +
"Usage:\n\t MyApplication.exe FolderPath [IncludeSubfolders (Y/N) : default=" +
defaultUsageOption + "] [OutputToFile (Y/N): default=" + defaultOutOption + "]"
let HandlArgs arr =
match arr with
| [|d;u;o|] -> DoStuff d u o
| [|d;u|] -> DoStuff d u defaultOutOption
| [|d|] -> DoStuff d defaultUsageOption defaultOutOption
| _ ->
printf "%s" usage
Console.ReadLine() |> ignore
[<EntryPoint>]
let main (args : string array) =
args |> HandlArgs
0
(I had a vague memory of this technique being inspired by Robert Pickering, but can't find a reference now)
A handy cache function that keeps up to max (key,reader(key)) in a dictionary and use a SortedList to track the MRU keys
let Cache (reader: 'key -> 'value) max =
let cache = new Dictionary<'key,LinkedListNode<'key * 'value>>()
let keys = new LinkedList<'key * 'value>()
fun (key : 'key) -> (
let found, value = cache.TryGetValue key
match found with
|true ->
keys.Remove value
keys.AddFirst value |> ignore
(snd value.Value)
|false ->
let newValue = key,reader key
let node = keys.AddFirst newValue
cache.[key] <- node
if (keys.Count > max) then
let lastNode = keys.Last
cache.Remove (fst lastNode.Value) |> ignore
keys.RemoveLast() |> ignore
(snd newValue))
Creating XElements
Nothing amazing, but I keep getting caught out by the implicit conversion of XNames:
#r "System.Xml.Linq.dll"
open System.Xml.Linq
//No! ("type string not compatible with XName")
//let el = new XElement("MyElement", "text")
//better
let xn s = XName.op_Implicit s
let el = new XElement(xn "MyElement", "text")
//or even
let xEl s o = new XElement(xn s, o)
let el = xEl "MyElement" "text"
Pairwise and pairs
I always expect Seq.pairwise to give me [(1,2);(3;4)] and not [(1,2);(2,3);(3,4)]. Given that neither exist in List, and that I needed both, here's the code for future reference. I think they're tail recursive.
//converts to 'windowed tuples' ([1;2;3;4;5] -> [(1,2);(2,3);(3,4);(4,5)])
let pairwise lst =
let rec loop prev rem acc =
match rem with
| hd::tl -> loop hd tl ((prev,hd)::acc)
| _ -> List.rev acc
loop (List.head lst) (List.tail lst) []
//converts to 'paged tuples' ([1;2;3;4;5;6] -> [(1,2);(3,4);(5,6)])
let pairs lst =
let rec loop rem acc =
match rem with
| l::r::tl -> loop tl ((l,r)::acc)
| l::[] -> failwith "odd-numbered list"
| _ -> List.rev acc
loop lst []
Naive CSV reader (i.e., won't handle anything nasty)
(Using filereadlines and List.transpose from other answers here)
///Given a file path, returns a List of row lists
let ReadCSV =
filereadlines
>> Array.map ( fun line -> line.Split([|',';';'|]) |> List.ofArray )
>> Array.toList
///takes list of col ids and list of rows,
/// returns array of columns (in requested order)
let GetColumns cols rows =
//Create filter
let pick cols (row:list<'a>) = List.map (fun i -> row.[i]) cols
rows
|> transpose //change list of rows to list of columns
|> pick cols //pick out the columns we want
|> Array.ofList //an array output is easier to index for user
Example
"C:\MySampleCSV"
|> ReadCSV
|> List.tail //skip header line
|> GetColumns [0;3;1] //reorder columns as well, if needs be.
Date Range
simple but useful list of dates between fromDate and toDate
let getDateRange fromDate toDate =
let rec dates (fromDate:System.DateTime) (toDate:System.DateTime) =
seq {
if fromDate <= toDate then
yield fromDate
yield! dates (fromDate.AddDays(1.0)) toDate
}
dates fromDate toDate
|> List.ofSeq
toggle code to sql
More trivial than most on this list, but handy nonetheless:
I'm always taking sql in and out of code to move it to a sql environment during development. Example:
let sql = "select a,b,c "
+ "from table "
+ "where a = 1"
needs to be 'stripped' to:
select a,b,c
from table
where a = 1
keeping the formatting. It's a pain to strip out the code symbols for the sql editor, then put them back again by hand when I've got the sql worked out. These two functions toggle the sql back and forth from code to stripped:
// reads the file with the code quoted sql, strips code symbols, dumps to FSI
let stripForSql fileName =
File.ReadAllText(fileName)
|> (fun s -> Regex.Replace(s, "\+(\s*)\"", ""))
|> (fun s -> s.Replace("\"", ""))
|> (fun s -> Regex.Replace(s, ";$", "")) // end of line semicolons
|> (fun s -> Regex.Replace(s, "//.+", "")) // get rid of any comments
|> (fun s -> printfn "%s" s)
then when you are ready to put it back into your code source file:
let prepFromSql fileName =
File.ReadAllText(fileName)
|> (fun s -> Regex.Replace(s, #"\r\n", " \"\r\n+\"")) // matches newline
|> (fun s -> Regex.Replace(s, #"\A", " \""))
|> (fun s -> Regex.Replace(s, #"\z", " \""))
|> (fun s -> printfn "%s" s)
I'd love to get rid of the input file but can't even begin to grok how to make that happen. anyone?
edit:
I figured out how to eliminate the requirement of a file for these functions by adding a windows forms dialog input/output. Too much code to show, but for those who would like to do such a thing, that's how I solved it.
Pascal's Triangle (hey, someone might find it useful)
So we want to create a something like this:
1
1 1
1 2 1
1 3 3 1
1 4 6 4 1
Easy enough:
let rec next = function
| [] -> []
| x::y::xs -> (x + y)::next (y::xs)
| x::xs -> x::next xs
let pascal n =
seq { 1 .. n }
|> List.scan (fun acc _ -> next (0::acc) ) [1]
The next function returns a new list where each item[i] = item[i] + item[i + 1].
Here's the output in fsi:
> pascal 10 |> Seq.iter (printfn "%A");;
[1]
[1; 1]
[1; 2; 1]
[1; 3; 3; 1]
[1; 4; 6; 4; 1]
[1; 5; 10; 10; 5; 1]
[1; 6; 15; 20; 15; 6; 1]
[1; 7; 21; 35; 35; 21; 7; 1]
[1; 8; 28; 56; 70; 56; 28; 8; 1]
[1; 9; 36; 84; 126; 126; 84; 36; 9; 1]
[1; 10; 45; 120; 210; 252; 210; 120; 45; 10; 1]
For the adventurous, here's a tail-recursive version:
let rec next2 cont = function
| [] -> cont []
| x::y::xs -> next2 (fun l -> cont <| (x + y)::l ) <| y::xs
| x::xs -> next2 (fun l -> cont <| x::l ) <| xs
let pascal2 n =
set { 1 .. n }
|> Seq.scan (fun acc _ -> next2 id <| 0::acc)) [1]
Flatten a List
if you have something like this:
let listList = [[1;2;3;];[4;5;6]]
and want to 'flatten' it down to a singe list so the result is like this:
[1;2;3;4;5;6]
it can be done thusly:
let flatten (l: 'a list list) =
seq {
yield List.head (List.head l)
for a in l do yield! (Seq.skip 1 a)
}
|> List.ofSeq
List comprehensions for float
This [23.0 .. 1.0 .. 40.0] was marked as deprecated a few versions backed.
But apparently, this works:
let dl = 9.5 / 11.
let min = 21.5 + dl
let max = 40.5 - dl
let a = [ for z in min .. dl .. max -> z ]
let b = a.Length
(BTW, there's a floating point gotcha in there. Discovered at fssnip - the other place for F# snippets)
Parallel map
let pmap f s =
seq { for a in s -> async { return f s } }
|> Async.Parallel
|> Async.Run

Resources