1
0
mirror of https://github.com/IBM/fp-go.git synced 2025-11-23 22:14:53 +02:00
Files
fp-go/v2/io/io.go
Dr. Carsten Leue aa5e908810 fix: introduce Kleisli type
Signed-off-by: Dr. Carsten Leue <carsten.leue@de.ibm.com>
2025-11-07 14:35:46 +01:00

392 lines
12 KiB
Go

// Copyright (c) 2023 - 2025 IBM Corp.
// All rights reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package io
import (
"time"
F "github.com/IBM/fp-go/v2/function"
"github.com/IBM/fp-go/v2/internal/apply"
"github.com/IBM/fp-go/v2/internal/chain"
"github.com/IBM/fp-go/v2/internal/functor"
INTL "github.com/IBM/fp-go/v2/internal/lazy"
M "github.com/IBM/fp-go/v2/monoid"
R "github.com/IBM/fp-go/v2/reader"
S "github.com/IBM/fp-go/v2/semigroup"
T "github.com/IBM/fp-go/v2/tuple"
)
const (
// useParallel is the feature flag to control if we use the parallel or the sequential implementation of ap
useParallel = true
)
var (
// undefined represents an undefined value
undefined = struct{}{}
)
type (
// IO represents a synchronous computation that cannot fail
// refer to [https://andywhite.xyz/posts/2021-01-27-rte-foundations/#ioltagt] for more details
IO[A any] = func() A
Kleisli[A, B any] = R.Reader[A, IO[B]]
Operator[A, B any] = Kleisli[IO[A], B]
Monoid[A any] = M.Monoid[IO[A]]
Semigroup[A any] = S.Semigroup[IO[A]]
)
// Of wraps a pure value in an IO context, creating a computation that returns that value.
// This is the monadic return operation for IO.
//
// Example:
//
// greeting := io.Of("Hello, World!")
// result := greeting() // returns "Hello, World!"
func Of[A any](a A) IO[A] {
return F.Constant(a)
}
// FromIO is an identity function that returns the IO value unchanged.
// Useful for type conversions and maintaining consistency with other monad packages.
func FromIO[A any](a IO[A]) IO[A] {
return a
}
// FromImpure converts a side effect without a return value into a side effect that returns any
func FromImpure[ANY ~func()](f ANY) IO[any] {
return func() any {
f()
return undefined
}
}
// MonadOf wraps a pure value in an IO context.
// This is an alias for Of, following the monadic naming convention.
func MonadOf[A any](a A) IO[A] {
return F.Constant(a)
}
// MonadMap transforms the result of an IO computation by applying a function to it.
// The function is only applied when the IO is executed.
//
// Example:
//
// doubled := io.MonadMap(io.Of(21), func(n int) int { return n * 2 })
// result := doubled() // returns 42
func MonadMap[A, B any](fa IO[A], f func(A) B) IO[B] {
return func() B {
return f(fa())
}
}
// Map returns an operator that transforms the result of an IO computation.
// This is the curried version of MonadMap.
//
// Example:
//
// double := io.Map(func(n int) int { return n * 2 })
// doubled := double(io.Of(21))
func Map[A, B any](f func(A) B) Operator[A, B] {
return F.Bind2nd(MonadMap[A, B], f)
}
// MonadMapTo replaces the result of an IO computation with a constant value.
// The original computation is still executed, but its result is discarded.
//
// Example:
//
// always42 := io.MonadMapTo(sideEffect, 42)
func MonadMapTo[A, B any](fa IO[A], b B) IO[B] {
return MonadMap(fa, F.Constant1[A](b))
}
// MapTo returns an operator that replaces the result with a constant value.
// This is the curried version of MonadMapTo.
func MapTo[A, B any](b B) Operator[A, B] {
return Map(F.Constant1[A](b))
}
// MonadChain composes computations in sequence, using the return value of one computation to determine the next computation.
func MonadChain[A, B any](fa IO[A], f Kleisli[A, B]) IO[B] {
return func() B {
return f(fa())()
}
}
// Chain composes computations in sequence, using the return value of one computation to determine the next computation.
func Chain[A, B any](f Kleisli[A, B]) Operator[A, B] {
return F.Bind2nd(MonadChain[A, B], f)
}
// MonadApSeq implements the applicative on a single thread by first executing mab and the ma
func MonadApSeq[A, B any](mab IO[func(A) B], ma IO[A]) IO[B] {
return MonadChain(mab, F.Bind1st(MonadMap[A, B], ma))
}
// MonadApPar implements the applicative on two threads, the main thread executes mab and the actuall
// apply operation and the second thread computes ma. Communication between the threads happens via a channel
func MonadApPar[A, B any](mab IO[func(A) B], ma IO[A]) IO[B] {
return func() B {
c := make(chan A, 1)
go func() {
c <- ma()
close(c)
}()
return mab()(<-c)
}
}
// MonadAp implements the `ap` operation. Depending on a feature flag this will be sequential or parallel, the preferred implementation
// is parallel
func MonadAp[A, B any](mab IO[func(A) B], ma IO[A]) IO[B] {
if useParallel {
return MonadApPar(mab, ma)
}
return MonadApSeq(mab, ma)
}
// Ap returns an operator that applies a function wrapped in IO to a value wrapped in IO.
// This is the curried version of MonadAp and uses parallel execution by default.
//
// Example:
//
// add := func(a int) func(int) int { return func(b int) int { return a + b } }
// result := io.Ap(io.Of(2))(io.Of(add(3))) // parallel execution
func Ap[B, A any](ma IO[A]) Operator[func(A) B, B] {
return F.Bind2nd(MonadAp[A, B], ma)
}
// ApSeq returns an operator that applies a function wrapped in IO to a value wrapped in IO sequentially.
// Unlike Ap, this executes the function and value computations in sequence.
func ApSeq[B, A any](ma IO[A]) Operator[func(A) B, B] {
return Chain(F.Bind1st(MonadMap[A, B], ma))
}
// ApPar returns an operator that applies a function wrapped in IO to a value wrapped in IO in parallel.
// This explicitly uses parallel execution (same as Ap when useParallel is true).
func ApPar[B, A any](ma IO[A]) Operator[func(A) B, B] {
return F.Bind2nd(MonadApPar[A, B], ma)
}
// Flatten removes one level of nesting from a nested IO computation.
// Converts IO[IO[A]] to IO[A].
//
// Example:
//
// nested := io.Of(io.Of(42))
// flattened := io.Flatten(nested)
// result := flattened() // returns 42
func Flatten[A any](mma IO[IO[A]]) IO[A] {
return MonadChain(mma, F.Identity)
}
// Memoize computes the value of the provided [IO] monad lazily but exactly once
func Memoize[A any](ma IO[A]) IO[A] {
return INTL.Memoize(ma)
}
// MonadChainFirst composes computations in sequence, using the return value of one computation to determine the next computation and
// keeping only the result of the first.
func MonadChainFirst[A, B any](fa IO[A], f Kleisli[A, B]) IO[A] {
return chain.MonadChainFirst(MonadChain[A, A], MonadMap[B, A], fa, f)
}
// ChainFirst composes computations in sequence, using the return value of one computation to determine the next computation and
// keeping only the result of the first.
func ChainFirst[A, B any](f Kleisli[A, B]) Operator[A, A] {
return chain.ChainFirst(
Chain[A, A],
Map[B, A],
f,
)
}
// MonadApFirst combines two effectful actions, keeping only the result of the first.
func MonadApFirst[A, B any](first IO[A], second IO[B]) IO[A] {
return apply.MonadApFirst(
MonadAp[B, A],
MonadMap[A, func(B) A],
first,
second,
)
}
// ApFirst combines two effectful actions, keeping only the result of the first.
func ApFirst[A, B any](second IO[B]) Operator[A, A] {
return apply.ApFirst(
MonadAp[B, A],
MonadMap[A, func(B) A],
second,
)
}
// MonadApSecond combines two effectful actions, keeping only the result of the second.
func MonadApSecond[A, B any](first IO[A], second IO[B]) IO[B] {
return apply.MonadApSecond(
MonadAp[B, B],
MonadMap[A, func(B) B],
first,
second,
)
}
// ApSecond combines two effectful actions, keeping only the result of the second.
func ApSecond[A, B any](second IO[B]) Operator[A, B] {
return apply.ApSecond(
MonadAp[B, B],
MonadMap[A, func(B) B],
second,
)
}
// MonadChainTo composes computations in sequence, ignoring the return value of the first computation
func MonadChainTo[A, B any](fa IO[A], fb IO[B]) IO[B] {
return MonadChain(fa, F.Constant1[A](fb))
}
// ChainTo composes computations in sequence, ignoring the return value of the first computation
func ChainTo[A, B any](fb IO[B]) Operator[A, B] {
return Chain(F.Constant1[A](fb))
}
// Now is an IO computation that returns the current timestamp when executed.
// Each execution returns the current time at that moment.
//
// Example:
//
// timestamp := io.Now()
var Now IO[time.Time] = time.Now
// Defer creates an IO by creating a brand new IO via a generator function each time.
// This allows for dynamic creation of IO computations based on runtime conditions.
//
// Example:
//
// deferred := io.Defer(func() io.IO[int] {
// if someCondition() {
// return io.Of(1)
// }
// return io.Of(2)
// })
func Defer[A any](gen func() IO[A]) IO[A] {
return func() A {
return gen()()
}
}
// MonadFlap applies a value to a function wrapped in IO.
// This is the reverse of Ap - instead of applying IO[func] to IO[value],
// it applies a pure value to IO[func].
//
// Example:
//
// addFive := io.Of(func(n int) int { return n + 5 })
// result := io.MonadFlap(addFive, 10) // returns IO[15]
func MonadFlap[B, A any](fab IO[func(A) B], a A) IO[B] {
return functor.MonadFlap(MonadMap[func(A) B, B], fab, a)
}
// Flap returns an operator that applies a pure value to a function wrapped in IO.
// This is the curried version of MonadFlap.
func Flap[B, A any](a A) Operator[func(A) B, B] {
return functor.Flap(Map[func(A) B, B], a)
}
// Delay creates an operator that delays execution by the specified duration.
// The delay occurs before executing the wrapped computation.
//
// Example:
//
// delayed := io.Delay(time.Second)(io.Of(42))
// result := delayed() // waits 1 second, then returns 42
func Delay[A any](delay time.Duration) Operator[A, A] {
return func(ga IO[A]) IO[A] {
return func() A {
time.Sleep(delay)
return ga()
}
}
}
func after(timestamp time.Time) func() {
return func() {
// check if we need to wait
current := time.Now()
if current.Before(timestamp) {
time.Sleep(timestamp.Sub(current))
}
}
}
// After creates an operator that delays execution until after the given timestamp.
// If the timestamp is in the past, the computation executes immediately.
//
// Example:
//
// future := time.Now().Add(5 * time.Second)
// scheduled := io.After(future)(io.Of(42))
// result := scheduled() // waits until future time, then returns 42
func After[A any](timestamp time.Time) Operator[A, A] {
aft := after(timestamp)
return func(ga IO[A]) IO[A] {
return func() A {
// wait as long as necessary
aft()
// execute after wait
return ga()
}
}
}
// WithTime returns an IO that measures the start and end time.Time of the operation.
// Returns a tuple containing the result, start time, and end time.
//
// Example:
//
// timed := io.WithTime(expensiveComputation)
// result, start, end := timed()
func WithTime[A any](a IO[A]) IO[T.Tuple3[A, time.Time, time.Time]] {
return func() T.Tuple3[A, time.Time, time.Time] {
t0 := time.Now()
res := a()
t1 := time.Now()
return T.MakeTuple3(res, t0, t1)
}
}
// WithDuration returns an IO that measures the execution time.Duration of the operation.
// Returns a tuple containing the result and the duration.
//
// Example:
//
// timed := io.WithDuration(expensiveComputation)
// result, duration := timed()
// fmt.Printf("Took %v\n", duration)
func WithDuration[A any](a IO[A]) IO[T.Tuple2[A, time.Duration]] {
return func() T.Tuple2[A, time.Duration] {
t0 := time.Now()
res := a()
t1 := time.Now()
return T.MakeTuple2(res, t1.Sub(t0))
}
}