Back to top

ReQL command: fold

Command syntax

sequence.fold(base, function) → value

sequence.fold(base, function).optArg("emit", function)[.optArg("final_emit", function)] → sequence


Apply a function to a sequence in order, maintaining state via an accumulator. The fold command returns either a single value or a new sequence.

In its first form, fold operates like reduce, returning a value by applying a combining function to each element in a sequence. The combining function takes two parameters: the previous reduction result (the accumulator) and the current element. However, fold has the following differences from reduce:

  • it is guaranteed to proceed through the sequence from first element to last.
  • it passes an initial base value to the function with the first element in place of the previous reduction result.

combiningFunction(accumulator | base, element) → newAccumulator

In its second form, fold operates like concat_map, returning a new sequence rather than a single value. When an emit function is provided, fold will:

  • proceed through the sequence in order and take an initial base value, as above.
  • for each element in the sequence, call both the combining function and a separate emitting function. The emitting function takes three parameters: the previous reduction result (the accumulator), the current element, and the output from the combining function (the new value of the accumulator).

If provided, the emitting function must return a list.

emit(previousAccumulator, element, accumulator) → array

A finalEmit function may also be provided, which will be called at the end of the sequence. It takes a single parameter: the result of the last reduction through the iteration (the accumulator), or the original base value if the input sequence was empty. This function must return a list, which will be appended to fold’s output stream.

finalEmit(accumulator | base) → array

Example: Concatenate words from a list.

    (acc, word) -> acc.add(r.branch(r.eq(acc, ""), "", ", ")).add(word)

(This example could be implemented with reduce, but fold will preserve the order when words is a RethinkDB table or other stream, which is not guaranteed with reduce.)

Example: Return every other row in a table.

    (acc, row) -> r.add(acc, 1)
    (acc, row, new_acc) -> r.branch(r.mod(new_acc, 2).eq(0),
                                    r.array(row), r.array())

The first function increments the accumulator each time it’s called, starting at 0; the second function, the emitting function, alternates between returning a single-item list containing the current row or an empty list. The fold command will return a concatenated list of each emitted value.

Example: Compute a five-day running average for a weight tracker.

    r.hashMap("name", "bob")
    (acc, row) -> r.add(r.array(row), acc).limit(5)
    (acc, row, new_acc) -> r.branch(new_acc.size().eq(5),

Get more help

Couldn't find what you were looking for?