package decoders

The type of values to be decoded (e.g. JSON or Yaml).

The type of decoders.

Use the functions below to construct decoders for your data types.

To run a decoder, pass it to decode_value.

Decode a string.

Decode an int.

Decode a float.

Decode a bool.

Decode a literal value.

Decode a collection into an OCaml list.

Decode a collection into an OCaml list, skipping elements for which the decoder returns None.

Decode a collection with an accumulator.

If we consider that an 'a decoder is basically a type alias for json -> ('a, error) result, the signature of this function is comparable to that of List.fold_left:

val List.fold_left : ('a ->   'b ->                 'a) -> 'a -> 'b list ->                 'a
val list_fold_left : ('a -> json -> ('a, error) result) -> 'a ->    json -> ('a, error) result
val list_fold_left : ('a ->                 'a decoder) -> 'a ->                    'a decoder

Decode a collection, requiring a particular index.

fst |> uncons rest decodes the first element of a list using fst, then decodes the remainder of the list using rest.

For example, to decode this s-expression:

(library
  (name decoders))

we can use this decoder:

string |> uncons (function
  | "library" -> field "name" string
  | _ -> fail "Expected a library stanza")

As another example, say you have a JSON array that starts with a string, then a bool, then a list of integers:

["hello", true, 1, 2, 3, 4]

We could decode it like this:

let (>>=::) fst rest = uncons rest fst

let decoder : (string * bool * int list) decoder =
  string >>=:: fun the_string ->
  bool >>=:: fun the_bool ->
  list int >>= fun the_ints ->
  succeed (the_string, the_bool, the_ints)

(If you squint, the uncons operator >>=:: kind of looks like the cons operator ::.)

Decode an object, requiring a particular field.

Decode an object, where a particular field may or may not be present.

For example, (field_opt "hello" int):

• when run on {"hello": 123}, will succeed with Some 123
• when run on {"hello": null}, will fail
• when run on {"world": 123}, will succeed with None
• when run on ["a", "list", "of", "strings"], will fail

Decode an object, requiring exactly one field.

Decode a nested object, requiring certain fields.

maybe d is a decoder that always succeeds. If d succeeds with x, then maybe d succeeds with Some x, otherwise if d fails, then maybe d succeeds with None.

For example, maybe (field "hello" int):

• when run on {"hello": 123}, will succeed with Some 123
• when run on {"hello": null}, will succeed with None
• when run on {"world": 123}, will succeed with None
• when run on ["a", "list", "of", "strings"], will succeed with None

nullable d will succeed with None if the JSON value is null. If the JSON value is non-null, it wraps the result of running x in a Some.

For example, field "hello" (nullable int):

• when run on {"hello": 123}, will succeed with Some 123
• when run on {"hello": null}, will succeed with None
• when run on {"world": 123}, will fail
• when run on ["a", "list", "of", "strings"], will fail

Try a sequence of different decoders.

Map over the result of a decoder.

Try two decoders and then combine the result. We can use this to decode objects with many fields (but it's preferable to use Infix.(>>=) - see the README).

Decode all of the keys of an object to a list of strings.

Decode an object into a list of key-value pairs.

Decode an object into a list of values, where the value decoder depends on the key.

keys' is for when your keys might not be strings - probably only likely for Yaml.

A decoder that always succeeds with the argument, ignoring the input.

A decoder that always fails with the given message, ignoring the input.

Create decoders that depend on previous results.

Recursive decoders.

let my_decoder = fix (fun my_decoder -> ...) allows you to define my_decoder in terms of itself.

Run a decoder on some input.

Run a decoder on a string.

Run a decoder on a file.