Iterant

Returns a computation that should be evaluated in case the streaming must stop before reaching the end.

This is useful to release any acquired resources, like opened file handles or network sockets.

Appends the given stream to the end of the source, effectively concatenating them.

Example:

// Yields 1, 2, 3, 4
Iterant[Task].of(1, 2) ++ Iterant[Task].of(3, 4)

Appends a stream to the end of the source, effectively concatenating them.

The right hand side is suspended in the F[_] data type, thus allowing for laziness.

Example:

// Yields 1, 2, 3, 4
Iterant[Task].of(1, 2) ++ Task.suspend {
  Iterant[Task].of(3, 4)
}

Prepends an element to the iterant, returning a new iterant that will start with the given head and then continue with the source.

Example:

// Yields 1, 2, 3, 4
1 +: Iterant[Task].of(2, 3, 4)

Appends the right hand side element to the end of this iterant.

Example:

// Yields 1, 2, 3, 4
Iterant[Task].of(1, 2, 3) :+ 4

Converts the source Iterant that emits A elements into an iterant that emits Either[Throwable, A], thus materializing whatever error that might interrupt the stream.

Example:

// Yields Right(1), Right(2), Right(3)
Iterant[Task].of(1, 2, 3).attempt


// Yields Right(1), Right(2), Left(DummyException())
(Iterant[Task].of(1, 2) ++
  Iterant[Task].raiseError(DummyException())).attempt

Optimizes the access to the source by periodically gathering items emitted into batches of the specified size and emitting NextBatch nodes.

For this operation we have this law:

source.batched(16) <-> source

This means that the result will emit exactly what the source emits, however the underlying representation will be different, the emitted notes being of type NextBatch, wrapping arrays with the length equal to the given count.

Very similar in behavior with bufferTumbling, however the batches are implicit, not explicit. Useful for optimization.

Returns an iterant that emits buffers of items it collects from the source iterant. The resulting iterant emits buffers every skip items, each containing count items.

If the source iterant completes, then the current buffer gets signaled downstream. If the source triggers an error then the current buffer is being dropped and the error gets propagated immediately.

For count and skip there are 3 possibilities:

1. in case skip == count, then there are no items dropped and no overlap, the call being equivalent to buffer(count)
2. in case skip < count, then overlap between buffers happens, with the number of elements being repeated being count - skip
3. in case skip > count, then skip - count elements start getting dropped between windows

Example:

val source = Iterant[Coeval].of(1, 2, 3, 4, 5, 6, 7)

// Yields Seq(1, 2, 3), Seq(4, 5, 6), Seq(7)
source.bufferSliding(3, 3)

// Yields Seq(1, 2, 3), Seq(5, 6, 7)
source.bufferSliding(3, 4)

// Yields Seq(1, 2, 3), Seq(3, 4, 5), Seq(5, 6, 7)
source.bufferSliding(3, 2)

Periodically gather items emitted by an iterant into bundles and emit these bundles rather than emitting the items one at a time. This version of buffer is emitting items once the internal buffer has reached the given count.

If the source iterant completes, then the current buffer gets signaled downstream. If the source triggers an error then the current buffer is being dropped and the error gets propagated immediately.

// Yields Seq(1, 2, 3), Seq(4, 5, 6), Seq(7)
Iterant[Coeval].of(1, 2, 3, 4, 5, 6, 7).bufferTumbling(3)

Builds a new iterant by applying a partial function to all elements of the source on which the function is defined.

Example:

// Yields 2, 4, 6
Iterant[Task].of(1, 2, 3, 4, 5, 6)
  .map { x => Option(x).filter(_ % 2 == 0) }
  .collect { case Some(x) => x }

Upon evaluation of the result, consumes this iterant to completion.

Example:

val onFinish: Task[Unit] =
  iterant.completeL >> Task.eval(println("Done!"))

Counts the total number of elements emitted by the source.

Example:

// Yields 100
Iterant[IO].range(0, 100).countL

// Yields 1
Iterant[IO].pure(1).countL

// Yields 0
Iterant[IO].empty[Int].countL

Suppress duplicate consecutive items emitted by the source.

Example:

// Yields 1, 2, 1, 3, 2, 4
Iterant[Coeval].of(1, 1, 1, 2, 2, 1, 1, 3, 3, 3, 2, 2, 4, 4, 4)
  .distinctUntilChanged

Duplication is detected by using the equality relationship provided by the cats.Eq type class. This allows one to override the equality operation being used (e.g. maybe the default .equals is badly defined, or maybe you want reference equality, so depending on use case).

In case type A is a primitive type and an Eq[A] instance is not in scope, then you probably need this import:

import cats.instances.all._

Or in case your type A does not have an Eq[A] instance defined for it, then you can quickly define one like this:

import cats.Eq

implicit val eqA = Eq.fromUniversalEquals[A]

Given a function that returns a key for each element emitted by the source, suppress consecutive duplicate items.

Example:

// Yields 1, 2, 3, 4
Iterant[Coeval].of(1, 3, 2, 4, 2, 3, 5, 7, 4)
  .distinctUntilChangedBy(_ % 2)

Duplication is detected by using the equality relationship provided by the cats.Eq type class. This allows one to override the equality operation being used (e.g. maybe the default .equals is badly defined, or maybe you want reference equality, so depending on use case).

In case type K is a primitive type and an Eq[K] instance is not in scope, then you probably need this import:

import cats.instances.all._

Or in case your type K does not have an Eq[K] instance defined for it, then you can quickly define one like this:

import cats.Eq

implicit val eqK = Eq.fromUniversalEquals[K]

Given a routine make sure to execute it whenever the consumer executes the current stop action.

Example:

iterant.doOnEarlyStop(Task.eval {
  println("Was stopped early!")
})

Returns a new enumerator in which f is scheduled to be executed on halt or on earlyStop.

This would typically be used to release any resources acquired by this enumerator.

Note that doOnEarlyStop is subsumed under this operation, the given f being evaluated on both reaching the end or canceling early.

Example:

iterant.doOnEarlyStop(err => Task.eval {
  err match {
    case Some(e) => log.error(e)
    case None =>
      println("Was consumed successfully!")
  }
})

Drops the first n elements (from the start).

Example:

// Yields 4, 5
Iterant[Task].of(1, 2, 3, 4, 5).drop(3)

Drops the longest prefix of elements that satisfy the given predicate and returns a new iterant that emits the rest.

Example:

// Yields 4, 5
Iterant[Task].of(1, 2, 3, 4, 5).dropWhile(_ < 4)

Dumps incoming events to standard output with provided prefix.

Utility that can be used for debugging purposes.

Example:

Iterant[Task].range(0, 4)
  .dump("O")
  .completeL.runAsync

// Results in:

0: O --> 0
1: O --> 1
2: O --> 2
3: O --> 3
4: O completed

Returns true in case the given predicate is satisfied by any of the emitted items, or false in case the end of the stream has been reached with no items satisfying the given predicate.

Example:

val source = Iterant[Coeval].of(1, 2, 3, 4)

// Yields true
source.existsL(_ % 2 == 0)

// Yields false
source.existsL(_ % 7 == 0)

Filters the iterant by the given predicate function, returning only those elements that match.

Example:

// Yields 2, 4, 6
Iterant[Task].of(1, 2, 3, 4, 5, 6).filter(_ % 2 == 0)

Given a predicate, finds the first item that satisfies it, returning Some(a) if available, or None otherwise.

// Yields Some(2)
Iterant[Coeval].of(1, 2, 3, 4).findL(_ % 2 == 0)

// Yields None
Iterant[Coeval].of(1, 2, 3, 4).findL(_ > 10)

The stream is traversed from beginning to end, the process being interrupted as soon as it finds one element matching the predicate, or until the stream ends.

Applies the function to the elements of the source and concatenates the results.

This operation is the monadic "bind", with all laws it entails.

Also note that the implementation can use constant memory depending on usage, thus it can be used in tail recursive loops.

Example:

// Effectively equivalent with .filter
Iterant[Task].of(1, 2, 3, 4, 5, 6).flatMap { elem =>
  if (elem % 2 == 0)
    Iterant[Task].pure(elem)
  else
    Iterant[Task].empty
}

Given an Iterant that generates Iterant elements, concatenates all the generated iterants.

Equivalent with: source.flatMap(x => x)

Given evidence that type A has a cats.Monoid implementation, folds the stream with the provided monoid definition.

For streams emitting numbers, this effectively sums them up. For strings, this concatenates them.

Example:

// Yields 10
Iterant[Task].of(1, 2, 3, 4).foldL

// Yields "1234"
Iterant[Task].of("1", "2", "3", "4").foldL

Note, in case you don't have a Monoid instance in scope, but you feel like you should, try this import:

import cats.instances.all._

Left associative fold using the function op.

On execution the stream will be traversed from left to right, and the given function will be called with the prior result, accumulating state until the end, when the summary is returned.

Example:

// Yields 15 (1 + 2 + 3 + 4 + 5)
Iterant[Task].of(1, 2, 3, 4, 5).foldLeftL(0)(_ + _)

Lazily fold the stream to a single value from the right.

This is the common foldr operation from Haskell's Foldable, or foldRight from Scala's collections, however it has a twist: the user is responsible for invoking early stop in case the processing is short-circuited, hence the signature of function f is different from other implementations, receiving the current earlyStop: F[Unit] as a third parameter.

Here's for example how existsL, forallL and ++ could be expressed in terms of foldRightL:

def exists[F[_], A](fa: Iterant[F, A], p: A => Boolean)
  (implicit F: Sync[F]): F[Boolean] = {

  fa.foldRightL(F.pure(false)) { (a, next, stop) =>
    if (p(a)) stop.map(_ => true) else next
  }
}

def forall[F[_], A](fa: Iterant[F, A], p: A => Boolean)
  (implicit F: Sync[F]): F[Boolean] = {

  fa.foldRightL(F.pure(true)) { (a, next, stop) =>
    if (!p(a)) stop.map(_ => false) else next
  }
}

def concat[F[_], A](lh: Iterant[F, A], rh: Iterant[F, A])
  (implicit F: Sync[F]): Iterant[F, A] = {

  Iterant.suspend[F, A] {
    lh.foldRightL(F.pure(rh)) { (a, rest, stop) =>
      F.pure(Iterant.nextS(a, rest, stop))
    }
  }
}

In this example we are short-circuiting the processing in case we find the one element that we are looking for, otherwise we keep traversing the stream until the end, finally returning the default value in case we haven't found what we were looking for.

WARNING

The implementation cannot ensure resource safety automatically, therefore it falls on the user to chain the stop reference in the processing, in case the right parameter isn't factored in.

In other words:

• in case the processing fails in any way with exceptions, it is the user's responsibility to chain stop
• in case the processing is short-circuited by not using the F[B] right param, it is the user responsibility to chain stop

This is in contrast with all operators (unless explicitly mentioned otherwise).

See the examples provided above, as they are correct in their handling of stop.

Left associative fold using the function op that can be short-circuited.

On execution the stream will be traversed from left to right, and the given function will be called with the prior result, accumulating state either until the end, or until op returns a Right result, when the summary is returned.

The results are returned in the F[_] functor context, meaning that we can have lazy or asynchronous processing and we can suspend side effects, depending on the F data type being used.

Example using cats.effect.IO:

// Sums first 10 items
Iterant[IO].range(0, 1000).foldWhileLeftEvalL(IO((0, 0))) {
  case ((sum, count), e) =>
    IO {
      val next = (sum + e, count + 1)
      if (count + 1 < 10) Left(next) else Right(next)
    }
}

// Implements exists(predicate)
Iterant[IO].of(1, 2, 3, 4, 5).foldWhileLeftEvalL(IO(false)) {
  (default, e) =>
    IO { if (e == 3) Right(true) else Left(default) }
}

// Implements forall(predicate)
Iterant[IO].of(1, 2, 3, 4, 5).foldWhileLeftEvalL(IO(true)) {
  (default, e) =>
    IO { if (e != 3) Right(false) else Left(default) }
}

Left associative fold using the function op that can be short-circuited.

On execution the stream will be traversed from left to right, and the given function will be called with the prior result, accumulating state either until the end, or until op returns a Right result, when the summary is returned.

Example:

// Sums first 10 items
Iterant[Task].range(0, 1000).foldWhileLeftL((0, 0)) {
  case ((sum, count), e) =>
    val next = (sum + e, count + 1)
    if (count + 1 < 10) Left(next) else Right(next)
}

// Implements exists(predicate)
Iterant[Task].of(1, 2, 3, 4, 5).foldWhileLeftL(false) {
  (default, e) =>
    if (e == 3) Right(true) else Left(default)
}

// Implements forall(predicate)
Iterant[Task].of(1, 2, 3, 4, 5).foldWhileLeftL(true) {
  (default, e) =>
    if (e != 3) Right(false) else Left(default)
}

Returns true in case the given predicate is satisfied by all of the emitted items, or false in case the given predicate fails for any of those items.

Example:

val source = Iterant[Coeval].of(1, 2, 3, 4)

// Yields false
source.forallL(_ % 2 == 0)

// Yields true
source.existsL(_ < 10)

Consumes the source iterable, executing the given callback for each element.

Example:

// Prints all elements, each one on a different line
Iterant[Task].of(1, 2, 3).foreachL { elem =>
  println("Elem: " + elem.toString)
}

Optionally selects the first element.

// Yields Some(1)
Iterant[Task].of(1, 2, 3, 4).headOptionL

// Yields None
Iterant[Task].empty[Int].headOptionL

Given mapping functions from F to G, lifts the source into an iterant that is going to use the resulting G for evaluation.

This can be used for replacing the underlying F type into something else. For example say we have an iterant that uses Coeval, but we want to convert it into one that uses Task for evaluation:

// Source is using Coeval for evaluation
val source = Iterant[Coeval].of(1, 2, 3, 4)

// Transformation to an iterant based on Task
source.liftMap(_.toTask, _.toTask)

Given a functor transformation from F to G, lifts the source into an iterant that is going to use the resulting G for evaluation.

This can be used for replacing the underlying F type into something else. For example say we have an iterant that uses Coeval, but we want to convert it into one that uses Task for evaluation:

import cats.~>

// Source is using Coeval for evaluation
val source = Iterant[Coeval].of(1, 2, 3, 4)

// Transformation to an iterant based on Task
source.liftMapK(new (Coeval ~> Task) {
  def apply[A](fa: Coeval[A]): Task[A] =
    fa.task
})

This operator can be used for more than transforming the F type into something else.

Returns a new stream by mapping the supplied function over the elements of the source.

// Yields 2, 4, 6
Iterant[Task].of(1, 2, 3).map(_ * 2)

Given a mapping function that returns a possibly lazy or asynchronous result, applies it over the elements emitted by the stream.

Iterant[Task].of(1, 2, 3, 4).mapEval { elem =>
  Task.eval {
    println("Received: " + elem.toString)
    elem * 2
  }
}

Takes the elements of the source iterant and emits the element that has the maximum key value, where the key is generated by the given function.

Example:

case class Person(name: String, age: Int)

// Yields Some(Person("Peter", 23))
Iterant[Coeval].of(Person("Peter", 23), Person("May", 21))
  .maxByL(_.age)

// Yields None
Iterant[Coeval].empty[Int].maxByL(_.age)

Given a cats.Order over the stream's elements, returns the maximum element in the stream.

Example:

// Yields Some(20)
Iterant[Coeval].of(1, 10, 7, 6, 8, 20, 3, 5).maxL

// Yields None
Iterant[Coeval].empty[Int].maxL

Takes the elements of the source iterant and emits the element that has the minimum key value, where the key is generated by the given function.

Example:

case class Person(name: String, age: Int)

// Yields Some(Person("May", 21))
Iterant[Coeval].of(Person("Peter", 23), Person("May", 21))
  .minByL(_.age)

// Yields None
Iterant[Coeval].empty[Int].minByL(_.age)

Given a cats.Order over the stream's elements, returns the minimum element in the stream.

Example:

// Yields Some(3)
Iterant[Coeval].of(10, 7, 6, 8, 20, 3, 5).minL

// Yields None
Iterant[Coeval].empty[Int].minL

Returns an Iterant that mirrors the behavior of the source, unless the source is terminated with an error, in which case the streaming of events fallbacks to an iterant emitting a single element generated by the backup function.

Example:

val prefix = Iterant[Task].of(1, 2, 3, 4)
val suffix = Iterant[Task].raiseError(DummyException("dummy"))
val fa = prefix ++ suffix

fa.onErrorHandle { _ => 5 }

See onErrorRecover for the version that takes a partial function as a parameter.

Returns an Iterant that mirrors the behavior of the source, unless the source is terminated with an error, in which case the streaming of events continues with the specified backup sequence generated by the given function.

Example:

val prefix = Iterant[Task].of(1, 2, 3, 4)
val suffix = Iterant[Task].raiseError(DummyException("dummy"))
val fa = prefix ++ suffix

fa.onErrorHandleWith {
  case _: DummyException =>
    Iterant[Task].pure(5)
  case other =>
    Iterant[Task].raiseError(other)
}

See onErrorRecoverWith for the version that takes a partial function as a parameter.

Returns an Iterant that mirrors the behavior of the source, unless the source is terminated with an error, in which case the streaming of events fallbacks to an iterant emitting a single element generated by the backup function.

The created Iterant mirrors the behavior of the source in case the source does not end with an error or if the thrown Throwable is not matched.

Example:

val prefix = Iterant[Task].of(1, 2, 3, 4)
val suffix = Iterant[Task].raiseError(DummyException("dummy"))
val fa = prefix ++ suffix

fa.onErrorRecover {
  case _: DummyException => 5
}

See onErrorHandle for the version that takes a total function as a parameter.

Returns an Iterant that mirrors the behavior of the source, unless the source is terminated with an error, in which case the streaming of events continues with the specified backup sequence generated by the given partial function.

The created Iterant mirrors the behavior of the source in case the source does not end with an error or if the thrown Throwable is not matched.

Example:

val prefix = Iterant[Task].of(1, 2, 3, 4)
val suffix = Iterant[Task].raiseError(DummyException("dummy"))
val fa = prefix ++ suffix

fa.onErrorRecoverWith {
  case _: DummyException =>
    Iterant[Task].pure(5)
}

See onErrorHandleWith for the version that takes a total function as a parameter.

Lazily zip two iterants together, the elements of the emitted tuples being fetched in parallel.

This is the parallel version of zip, the results are still ordered, but it can yield non-deterministic ordering of effects when fetching the elements of an emitted tuple.

Lazily zip two iterants together, in parallel, using the given function f to produce output values.

This is like zipMap, except that the element pairs are processed in parallel (ordered results, but non-deterministic ordering of effects).

Reduces the elements of the source using the specified associative binary operator, going from left to right, start to finish.

Example:

// Yields Some(10)
Iterant[Coeval].of(1, 2, 3, 4).reduceL(_ + _)

// Yields None
Iterant[Coeval].empty[Int].reduceL(_ + _)

Applies a binary operator to a start value and all elements of this Iterant, going left to right and returns a new Iterant that emits on each step the result of the applied function.

Similar to foldLeftL, but emits the state on each step. Useful for modeling finite state machines.

Example showing how state can be evolved and acted upon:

sealed trait State[+A] { def count: Int }
case object Init extends State[Nothing] { def count = 0 }
case class Current[A](current: A, count: Int) extends State[A]

val scanned = source.scan(Init : State[A]) { (acc, a) =>
  acc match {
    case Init => Current(a, 1)
    case Current(_, count) => Current(a, count + 1)
  }
}

scanned
  .takeWhile(_.count < 10)
  .collect { case Current(a, _) => a }

Applies a binary operator to a start value and all elements of this Iterant, going left to right and returns a new Iterant that emits on each step the result of the applied function.

Similar with scan, but this can suspend and evaluate side effects in the F[_] context, thus allowing for asynchronous data processing.

Similar to foldLeftL and foldWhileLeftEvalL, but emits the state on each step. Useful for modeling finite state machines.

Example showing how state can be evolved and acted upon:

sealed trait State[+A] { def count: Int }
case object Init extends State[Nothing] { def count = 0 }
case class Current[A](current: Option[A], count: Int)
  extends State[A]

case class Person(id: Int, name: String)

// Initial state
val seed = Task.now(Init : State[Person])

val scanned = source.scanEval(seed) { (state, id) =>
  requestPersonDetails(id).map { person =>
    state match {
      case Init =>
        Current(person, 1)
      case Current(_, count) =>
        Current(person, count + 1)
    }
  }
}

scanned
  .takeWhile(_.count < 10)
  .collect { case Current(a, _) => a }

Skips over Iterant.Suspend states, along with Iterant.NextCursor and Iterant.NextBatch states that signal empty collections.

Will mirror the source, except that the emitted internal states might be different. Can be used as an optimization if necessary.

Drops the first element of the source iterant, emitting the rest.

Example:

// Yields 2, 3, 4
Iterant[Task].of(1, 2, 3, 4).tail

Creates a new iterant that upon evaluation will select the first n elements from the source and then stop, in the order they are emitted by the source.

Example:

// Yields 1, 2, 3
Iterant[Task].of(1, 2, 3, 4, 5, 6).take(3)

Creates a new iterable that only emits the last n elements emitted by the source.

In case the source triggers an error, then the underlying buffer gets dropped and the error gets emitted immediately.

Example:

// Yields 1, 2, 3
Iterant[Task].of(1, 2, 3, 4, 5, 6).take(3)

Takes longest prefix of elements that satisfy the given predicate and returns a new iterant that emits those elements.

Example:

// Yields 1, 2, 3
Iterant[Task].of(1, 2, 3, 4, 5, 6).takeWhile(_ < 4)

Aggregates all elements in a List and preserves order.

Example:

// Yields List(1, 2, 3, 4)
Iterant[Task].of(1, 2, 3, 4).toListL

Note that this operation is dangerous, since if the iterant is infinite then this operation is non-terminating, the process probably blowing up with an out of memory error sooner or later.

Converts this Iterant into an org.reactivestreams.Publisher.

Meant for interoperability with other Reactive Streams implementations. Also useful because it turns the Iterant into another data type with a push-based communication protocol with back-pressure.

Usage sample:

import monix.eval.Task
import monix.execution.rstreams.SingleAssignmentSubscription
import org.reactivestreams.{Publisher, Subscriber, Subscription}

def sum(source: Publisher[Int], requestSize: Int): Task[Long] =
  Task.create { (_, cb) =>
    val sub = SingleAssignmentSubscription()

    source.subscribe(new Subscriber[Int] {
      private[this] var requested = 0L
      private[this] var sum = 0L

      def onSubscribe(s: Subscription): Unit = {
        sub := s
        requested = requestSize
        s.request(requestSize)
      }

      def onNext(t: Int): Unit = {
        sum += t
        if (requestSize != Long.MaxValue) requested -= 1

        if (requested <= 0) {
          requested = requestSize
          sub.request(request)
        }
      }

      def onError(t: Throwable): Unit =
        cb.onError(t)
      def onComplete(): Unit =
        cb.onSuccess(sum)
    })

    // Cancelable that can be used by Task
    sub
  }

val pub = Iterant[Task].of(1, 2, 3, 4).toReactivePublisher

// Yields 10
sum(pub, requestSize = 128)

See the Reactive Streams for details.

Applies the function to the elements of the source and concatenates the results.

This variant of flatMap is not referentially transparent, because it tries to apply function f immediately, in case the Iterant is in a NextCursor or NextBatch state.

To be used for optimizations, but keep in mind it's unsafe, as its application isn't referentially transparent.

Explicit covariance operator.

The Iterant type isn't covariant in type param A, because covariance doesn't play well with a higher-kinded type like F[_]. So in case you have an Iterant[F, A], but need an Iterant[F, B], knowing that A extends B, then you can do an upcast.

Example:

val source: Iterant[Task, List[Int]] = ???

// This will trigger an error because of the invariance:
val sequences: Iterant[Task, Seq[Int]] = source

// But this will work just fine:
val sequences: Iterant[Task, Seq[Int]] = source.upcast[Seq[Int]]

Lazily zip two iterants together.

The length of the result will be the shorter of the two arguments.

Example:

val lh = Iterant[Task].of(11, 12, 13, 14)
val rh = Iterant[Task].of(21, 22, 23, 24, 25)

// Yields (11, 21), (12, 22), (13, 23), (14, 24)
lh.zip(rh)

Lazily zip two iterants together, using the given function f to produce output values.

The length of the result will be the shorter of the two arguments.

Example:

val lh = Iterant[Task].of(11, 12, 13, 14)
val rh = Iterant[Task].of(21, 22, 23, 24, 25)

// Yields 32, 34, 36, 38
lh.zipMap(rh) { (a, b) => a + b }

Zips the emitted elements of the source with their indices.

The length of the result will be the same as the source.

Example:

val source = Iterant[Task].of("Sunday", "Monday", "Tuesday", "Wednesday")

// Yields ("Sunday", 0), ("Monday", 1), ("Tuesday", 2), ("Wednesday", 3)
source.zipWithIndex