Functional Programming in Go
A library of iterators for use with iter.Seq. Requires Go 1.23+.
// The first 5 natural numbers
numbers := slices.Collect(
it.Take(it.NaturalNumbers[int](), 5),
)
// All even numbers
evens := it.Filter(it.NaturalNumbers[int](), filter.IsEven)
// String representations of integers
numbers := it.Map(it.NaturalNumbers[int](), strconv.Itoa)
Installation
go get github.com/BooleanCat/go-functional/v2@latestOverview
Most functions offered by this package are either consumers or iterators.
Consumers will iterate over an iterator and completely or partially drain them of values and (in most cases) collect the values into a data type.
Iterators are functions that yield new values and can be ranged over. See Go's documentation for iterators for more details.
Consumers
The standard libary provides functions to collect iterators in the slices and maps packages that should satisfy most cases where collection is needed.
This package provides additional collection methods and makes existing consumers from the standard library chainable.
[!WARNING] Attempting to collect infinite iterators will cause an infinite loop and likely deadlock. Consider bounding infinite iterators before collect (for example using Take).
Collect
In most cases slices.Collect from the standard library may be used to collect items from an iterator into a slice. There are several other variants of collect available for use for different use cases.
// Chainable
numbers := itx.NaturalNumbers[int]().Take(5).Collect()
// Collect an iter.Seq2[V, W] into two slices
keys, values := it.Collect2(maps.All(map[string]int{"one": 1, "two": 2}))
// As above, but chainable
keys, values := itx.FromMap(map[string]int{"one": 1, "two": 2}).Collect()TryCollect & MustCollect
Dealing with iterators that return T, error can involve the boilerplate of checking that the
returned slice of errors only contains nil. TryCollect solves this by collecting all values into
a slice and returning a single error: the first one encountered.
text := strings.NewReader("one\ntwo\nthree\n")
if lines, err := it.TryCollect(it.LinesString(text)); err != nil {
fmt.Println(lines)
}MustCollect is similar except that if an error is encountered then a panic will occur.
text := strings.NewReader("one\ntwo\nthree\n")
lines := it.MustCollect(it.LinesString(text))[!TIP] Use
MustCollectwhen you can guarantee that no error will occur (such as with strings.Reader).
[!NOTE] If an error is encountered, collection stops. This means the iterator being collected may not be fully drained.
[!NOTE] The
itxpackage does not containTryCollectorMustCollectdue to limitations with Go's type system.
ForEach
ForEach consumes an iterator and applies a function to each value yielded.
it.ForEach(slices.Values([]int{1, 2, 3}), func(number int) {
fmt.Println(number)
})
// Chainable
itx.FromSlice([]int{1, 2, 3}).ForEach(func(number int) {
fmt.Println(number)
})
// For each member of an iter.Seq2
it.ForEach2(slices.All([]int{1, 2, 3}), func(index int, number int) {
fmt.Println(index, number)
})
// As above, but chainable
itx.FromSlice([]int{1, 2, 3}).Enumerate().ForEach(func(index int, number int) {
fmt.Println(index, number)
})Fold
Fold every element into an accumulator by applying a function and passing an initial value.
it.Fold(slices.Values([]int{1, 2, 3}), op.Add, 0)
// Fold an iter.Seq2
it.Fold2(slices.All([]int{1, 2, 3}), func(i, a, b int) int {
return i + 1
}, 0)[!TIP] The op package contains some simple, pre-defined operation functions.
[!NOTE] The
itxpackage does not containFolddue to limitations with Go's type system.
Max & Min
Max and Min consume an iterator and return the maximum or minimum value yielded and true if the
iterator contained at least one value, or the zero value and false if the iterator was empty.
The type of the value yielded by the iterator must be comparable.
max, ok := it.Max(slices.Values([]int{1, 2, 3}))
min, ok := it.Min(slices.Values([]int{1, 2, 3}))[!NOTE] The
itxpackage does not containMaxorMindue to limitations with Go's type system.
Len
Len consumes an iterator and returns the number of values yielded.
it.Len(slices.Values([]int{1, 2, 3}))
// Chainable
itx.FromSlice([]int{1, 2, 3}).Len()
// Len of an iter.Seq2
it.Len2(slices.All([]int{1, 2, 3}))
// As above, but chainable
itx.FromSlice([]int{1, 2, 3}).Enumerate().Len()Find
Find consumes an iterator until a value is found that satisfies a predicate. It returns the value and true if one was found, or the zero value and false if the iterator was exhausted before a value was found.
found, ok := it.Find(slices.Values([]int{1, 2, 3}), func(i int) bool {
return i == 2
})
// Chainable
value, ok := itx.FromSlice([]int{1, 2, 3}).Find(func(number int) bool {
return number == 2
})
// Finding within an iter.Seq2
index, value, ok := it.Find2(slices.All([]int{1, 2, 3}), func(i, v int) bool {
return i == 2
})
// As above, but chainable
index, value, ok := itx.FromSlice([]int{1, 2, 3}).Enumerate().Find(func(index int, number int) bool {
return index == 1
})[!TIP] The filter package contains some simple, pre-defined predicate functions.
Contains
Contains consumes an iterator until the provided value is found, returning true if the value was found or false if the iterator was exhausted.
ok := it.Contains(slices.Values([]int{1, 2, 3}), 2)[!NOTE] The
itxpackage does not containContainsdue to limitations with Go's type system.
From, FromSlice, FromMap & Seq
The itx package contains some helper functions to convert iterators, slices or maps directly into chainable iterators to avoid some boilerplate.
itx.From(slices.Values([]int{1, 2, 3})).Collect()
itx.From2(maps.All(map[int]string{1: "one"}))
itx.FromSlice([]int{1, 2, 3}).Collect()
itx.FromMap(map[int]int{1: 2}).Collect()The itx package also contains a helper function Seq that will convert a chainable iterator into an
iter.Seq so that it can be used in functions that accept that type
(such as the standard library).
The standard library functions that work with iterators (such as
slices.Collect) accept the
iter.Seq family of types. This precludes those functions from
accepting types with another alias (such as itx.Iterator) with the same type definition. This
means it is necessary to "covert" an itx.Iterator into an iter.Seq
before passing the iterator into those functions.
slices.Collect(itx.NaturalNumbers[int]().Take(3).Seq())[!TIP] go-functional's functions that accept iterators always accept
func(func(V) bool)orfunc(func(V, W) bool)rather than any specific type alias so that they can accept any type alias with the definitions iter.Seq, iter.Seq2,itx.Iterator,itx.Iterator2or any other third-party types aliased to the same type.
ToChannel
ToChannel sends yielded values to a channel.
The channel is closed when the iterator is exhausted. Beware of leaked go routines when using this function with an infinite iterator.
channel := it.ToChannel(slices.Values([]int{1, 2, 3}))
for number := range channel {
fmt.Println(number)
}
// Chainable
channel := itx.FromSlice([]int{1, 2, 3}).ToChannel()
for number := range channel {
fmt.Println(number)
}[!NOTE] Unlike most consumers, the iterator is not immediately consumed by ToChannel. Instead is it consumed as values are pulled from the channel.
Drain
Drain consumes an iterator's values and drops them.
printValue := func(n int) int {
fmt.Println(n)
return n
}
it.Drain(it.Map(slices.Values([]int{1, 2, 3}), printValue))
// Chainable
itx.From(it.Map(slices.Values([]int{1, 2, 3}), printValue)).Drain()
// Drain an iter.Seq2
printValue2 := func(i, n int) (int, int) {
fmt.Println(n)
return i, n
}
it.Drain2(it.Map2(slices.All([]int{1, 2, 3}), printValue2))
// As above, but chainable
itx.From2(it.Map2(slices.All([]int{1, 2, 3}), printValue2)).Drain()[!TIP] Use
Drainto consume an iterator to invoke any side effects when you don't need to collect the values.
Iterators
This library contains two kinds of iterators in the it and itx packages. In most cases you'll
find the same iterators in each package, the difference between them being that the iterators in the
itx package can be "dot-chained" (e.g. iter.Filter(...).Take(3).Collect()) and those in it
cannot.
Iterators within the it package are of the type iter.Seq or
iter.Seq2 (from the standard library). Iterators within the itx
package are of the type itx.Iterator[V] or itx.Iterator2[V, W].
There are two important factors to consider when using iterators:
- Some iterators yield an infinite number of values and care should be taken to avoid consuming (using functions such as slices.Collect) otherwise it's likely to cause an infinite while loop.
- Many iterators take another iterator as an argument (such as Filter or Map). Avoid using an iterator after it has been passed to another iterator otherwise you'll risk multiple functions consuming a single (likely not thread-safe) iterator and causing confusing and difficult to debug behaviour.
Chain
Chain yields values from multiple iterators in the sequence they are provided in. Think of it as glueing many iterators together.
When provided zero iterators it will behave like Exhausted.
numbers := it.Chain(slices.Values([]int{1, 2}), slices.Values([]int{3, 4}))
pairs := itx.FromSlice([]int{1, 2}).Chain(slices.Values([]int{3, 4}))
pairs := it.Chain2(maps.All(map[string]int{"a": 1}), maps.All(map[string]int{"b": 2}))
pairs := itx.FromMap(map[string]int{"a": 1}).Chain(maps.All(map[string]int{"b": 2}))FromChannel
FromChannel pulls values from a channel and yields them via an iterator. The usual concerns around channel deadlocks apply here.
The iterator is exhausted when the channel is closed and it is the responsibility of the caller to close the channel.
items := make(chan int)
go func() {
defer close(items)
items <- 1
items <- 2
}()
for number := range it.FromChannel(items) {
fmt.Println(number)
}
// Chainable
items := make(chan int)
go func() {
defer close(items)
items <- 1
items <- 0
}()
for number := range itx.FromChannel(items).Exclude(filter.IsZero) {
fmt.Println(number)
}[!WARNING] In order to prevent a deadlock, the channel must be closed before attemping to stop the iterator when it's used in a pull style. See iter.Pull.
Cycle
Cycle yields all values from an iterator before returning to the beginning and yielding all values again (indefinitely).
numbers := it.Take(it.Cycle(slices.Values([]int{1, 2})), 5)
// Chainable
numbers := itx.FromSlice([]int{1, 2}).Cycle().Take(5)
// Cycling an iter.Seq2
numbers := it.Take2(it.Cycle2(maps.All(map[int]string{1: "one"})), 5)
// As above, but chainable
numbers := itx.FromMap(map[int]string{1: "one"}).Cycle().Take(5)[!NOTE] Since cycle needs to store all values yielded in memory, its memory usage will grow as the first cycle is consumed and remain at a constant size on subsequent cycles.
[!WARNING] This iterator yields an infinite number of values and care should be taken when consuming it otherwise it's likely to result in an infinite while loop. Consider bounding the size of the iterator before consuming (e.g. using Take).
Drop & DropWhile
Drop yields all values from a delegate iterator after dropping a number of values from the beginning. Values are not dropped immediately, but when consumption begins.
When dropping a number of values larger than the length of the iterator, it behaves like Exhausted.
numbers := it.Drop(slices.Values([]int{1, 2, 3, 4, 5}), 2)
// Chainable
numbers := itx.FromSlice([]int{1, 2, 3, 4, 5}).Drop(2)
// Dropping on iter.Seq2
numbers := it.Drop2(maps.All(map[int]string{1: "one", 2: "two", 3: "three"}), 1)
// As above, but chainable
numbers := itx.FromMap(map[int]string{1: "one", 2: "two", 3: "three"}).Drop(1)DropWhile drops values from the provided iterator whilst the predicate returns true for each value. After the first value results in the predicate returning false, the iterator resumes as normal.
slices.Collect(it.DropWhile(slices.Values([]int{1, 2, 3, 4, 5}), filter.LessThan(3)))
// Chainable
itx.FromSlice([]int{1, 2, 3, 4, 5}).DropWhile(filter.LessThan(3)).Collect()
lessThanThree := func(string, number int) { return number < 3 }
// Taking from iter.Seq2
maps.Collect(it.DropWhile2(maps.All(map[string]int{"one": 1, "four": 4}), lessThanThree))
// As above, but chainable
itx.FromMap(map[string]int{"one": 1, "four": 4}).DropWhile(lessThanThree).Collect()Enumerate
Enumerating an iter.Seq like iterator returns an iter.Seq2 like iterator yielding the index of each value and the value.
indexedValues := it.Enumerate(slices.Values([]int{1, 2, 3}))
// Chainable
indexedValues := itx.FromSlice([]int{1, 2, 3}).Enumerate()[!TIP] When iterating over a slice and immediately enumerating, consider instead using the standard library's slices.All function rather than this.
Exhausted
Exhausted is an iterator that yields no values.
slices.Collect(it.Exhausted[int]())
// Chainable
it.Exhausted[int]().Collect()
// Exhausted iter.Seq2
maps.Collect(it.Exhausted2[int, string]())
// As above, but chainable
itx.Exhausted2[int, string]().Collect()Filter & Exclude
Filter yields values from an iterator that satisfy a predicate. Likewise, exclude yields values from an iterator that do not satisfy a predicate.
it.Filter(slices.Values([]int{1, 2, 3, 4, 5}), filter.IsEven)
it.Exclude(slices.Values([]int{1, 2, 3, 4, 5}), filter.IsEven)
// Chainable
itx.FromSlice([]int{1, 2, 3, 4, 5}).Filter(filter.IsEven)
itx.FromSlice([]int{1, 2, 3, 4, 5}).Exclude(filter.IsEven)
// Filtering an iter.Seq2
isOne := func(n int, _ string) bool { return n == 1 }
it.Filter2(maps.All(map[int]string{1: "one", 2: "two", 3: "three"}), isOne)
it.Exclude2(maps.All(map[int]string{1: "one", 2: "two", 3: "three"}), isOne)
// As above, but chainable
itx.FromMap(map[int]string{1: "one", 2: "two", 3: "three"}).Filter(isOne)
itx.FromMap(map[int]string{1: "one", 2: "two", 3: "three"}).Exclude(isOne)[!TIP] The filter package contains some simple, pre-defined predicate functions.
FilterError & ExcludeError
Similar to Filter and Exclude, these functions filter values from an iterator using a predicate that may return an error.
isFoo := func(s string) (bool, error) { return s == "foo", nil }
values, err := it.TryCollect(it.FilterError(slices.Values([]string{"foo", "bar"}), isFoo))
values, err := it.TryCollect(it.ExcludeError(slices.Values([]string{"foo", "bar"}), isFoo))
// Chainable
values, err := it.TryCollect(itx.FromSlice([]int{"foo", "bar"}).FilterError(isFoo))
values, err := it.TryCollect(itx.FromSlice([]int{"foo", "bar"}).ExcludeError(isFoo))Integers
Integers yields all integers in the range [start, stop) with the given step.
for i := range it.Integers[uint](0, 3, 1) {
fmt.Println(i)
}
// Chainable
for i := range itx.Integers[uint](0, 3, 1).Drop(1) {
fmt.Println(i)
}Lines & LinesString
Lines yields lines from an io.Reader. Lines are split using the standard library's bufio.Scanner. Each value yielded from the iterator is a line from the provided reader. Empty lines will result in empty values.
Since reading from an io.Reader can fail, each line is returned with
a corresponding error value.
LinesString behaves exactly like Lines except that its value are strings rather than byte slices.
buffer := strings.NewReader("one\ntwo\nthree\n")
for line, err := range it.Lines(buffer) {
if err != nil {
fmt.Println(err)
break
}
fmt.Println(string(line))
}
for line, err := range it.LinesString(buffer) {
if err != nil {
fmt.Println(err)
break
}
fmt.Println(line)
}[!TIP] Consider using TryCollect in conjunction with Lines to make error collection less cumbersome:
lines, err := it.TryCollect(it.LinesString(buffer))
[!TIP] For cases where errors will never result from reading, such as with bytes.Buffer, consider dropping the error value before collection:
lines := itx.LinesString(buffer).Left().Collect()
[!WARNING] As with bufio.Scanner, there is a maximum line length per line of 65536 characters.