Encode a value as a JSON ByteString via its type's codec.

Parse a JSON ByteString using a type's codec.

Implement toJSON via a type's codec.

Implement toJSON via a given codec.

Implement toEncoding via a type's codec.

Implement toEncoding via the given codec.

Implement parseJSON via a type's codec.

Implement parseJSON via a given codec.

A completed autodocodec for parsing and rendering a Value.

You can use a value of this type to get everything else for free:

• Encode values to JSON using toJSONViaCodec or toJSONVia
• Decode values from JSON using parseJSONViaCodec or parseJSONVia
• Produce a JSON Schema using jsonSchemaViaCodec or jsonSchemaVia from autodocodec-schema
• Encode to and decode from Yaml using autodocodec-yaml
• Produce a human-readible YAML schema using renderColouredSchemaViaCodec from autodocodec-yaml
• Produce a Swagger2 schema using autodocodec-swagger2
• Produce a OpenAPI3 schema using autodocodec-openapi3

A completed autodocodec for parsing and rendering a Value.

A class for values which have a canonical codec.

There are no formal laws for this class. If you really want a law, it should be "Whomever uses the codec from your instance should not be surprised."

A class for values which have a canonical object codec.

There are no formal laws for this class. If you really want a law, it should be "Whomever uses the codec from your instance should not be surprised."

An object codec with a given name

Example usage

data Example = Example
  { exampleText :: !Text,
    exampleBool :: !Bool
  }

instance HasCodec Example where
  codec =
    object "Example" $
      Example
        <$> requiredField "text" "a text" .= exampleText
        <*> requiredField "bool" "a bool" .= exampleBool

API Note

This is a forward-compatible version ObjectOfCodec with a name.

object name = ObjectOfCodec (Just name)

Name a codec.

This is used to allow for references to the codec, and that's necessary to produce finite documentation for recursive codecs.

API Note

This is a forward-compatible version of ReferenceCodec.

named = ReferenceCodec

Produce a codec using a type's FromJSON and ToJSON instances.

You will only want to use this if you cannot figure out how to produce a JSONCodec for your type.

Note that this will not have good documentation because, at a codec level, it's just parsing and rendering a Value.

Example usage

>>> toJSONVia (codecViaAeson "Int") (5 :: Int)
Number 5.0
>>> JSON.parseMaybe (parseJSONVia (codecViaAeson "Int")) (Number 5) :: Maybe Int
Just 5

A required field

During decoding, the field must be in the object.

During encoding, the field will always be in the object.

See requiredFieldWith

An optional field

During decoding, the field may be in the object. Nothing will be parsed otherwise.

During encoding, the field will be in the object if it is not Nothing, and omitted otherwise.

See optionalFieldWith

Infix version of lmapCodec

Use this function to supply the rendering side of a codec.

(.=) = flip lmapCodec

Example usage

data Example = Example
  { exampleText :: !Text,
    exampleBool :: !Bool
  }
instance HasCodec Example where
  codec =
    object "Example" $
      Example
        <$> requiredField "text" .= exampleText
        <*> requiredField "bool" .= exampleBool

An optional, or null, field

During decoding, the field may be in the object. Nothing will be parsed if it is not. If the field is null, then it will be parsed as Nothing as well.

During encoding, the field will be in the object if it is not Nothing, and omitted otherwise.

An optional field with a default value

During decoding, the field may be in the object. The default value will be parsed otherwise.

During encoding, the field will be in the object. The default value is ignored.

The shown version of the default value will appear in the documentation.

A required field

During decoding, the field must be in the object.

During encoding, the field will always be in the object.

An optional field

During decoding, the field may be in the object. Nothing will be parsed otherwise.

During encoding, the field will be omitted from the object if it is Nothing.

An optional, or null, field

During decoding, the field may be in the object. Nothing will be parsed if it is not. If the field is null, then it will be parsed as Nothing as well.

During encoding, the field will be omitted from the object if it is Nothing.

An optional field with default value

During decoding, the field may be in the object. The default value will be parsed otherwise.

During encoding, the field will always be in the object. The default value is ignored.

The shown version of the default value will appear in the documentation.

An optional field with default value that can be omitted when encoding

During decoding, the field may be in the object. The default value will be parsed otherwise.

During encoding, the field will be omitted from the object if it is equal to the default value.

The shown version of the default value will appear in the documentation.

Like optionalFieldWithOmittedDefaultWith, but the value may also be null and that will be interpreted as the default value.

Like requiredField, but without documentation

Like optionalField, but without documentation

Like optionalFieldOrNull, but without documentation

Like optionalFieldWithDefault, but without documentation

Like requiredFieldWith, but without documentation.

Like optionalFieldWith, but without documentation.

Like optionalFieldOrNullWith, but without documentation

Like optionalFieldWithDefaultWith, but without documentation.

Like optionalFieldWithOmittedDefaultWith, but without documentation.

Like optionalFieldWithOmittedDefaultWith', but the value may also be null and that will be interpreted as the default value.

Codec for null

Example usage

>>> toJSONVia nullCodec ()
Null
>>> JSON.parseMaybe (parseJSONVia nullCodec) Null
Just ()
>>> JSON.parseMaybe (parseJSONVia nullCodec) (Number 5)
Nothing

API Note

This is a forward-compatible version of NullCodec.

nullCodec = NullCodec

Codec for boolean values

Example usage

>>> toJSONVia boolCodec True
Bool True

API Note

This is a forward-compatible version of BoolCodec without a name.

boolCodec = BoolCodec Nothing

Codec for text values

Example usage

>>> toJSONVia textCodec "hello"
String "hello"

API Note

This is a forward-compatible version of StringCodec without a name.

textCodec = StringCodec Nothing

Codec for String values

Example usage

>>> toJSONVia stringCodec "hello"
String "hello"

WARNING

This codec uses unpack and pack to dimap a textCodec, so it does not roundtrip.

>>> toJSONVia stringCodec "\55296"
String "\65533"

API Note

This is a String version of textCodec.

Codec for Scientific values

Example usage

>>> toJSONVia scientificCodec 5
Number 5.0
>>> JSON.parseMaybe (parseJSONVia scientificCodec) (Number 3)
Just 3.0

WARNING

Scientific is a type that is only for JSON parsing and rendering. Do not use it for any calculations. Instead, convert to another number type before doing any calculations.

λ> (1 / 3) :: Scientific
*** Exception: fromRational has been applied to a repeating decimal which can't be represented as a Scientific! It's better to avoid performing fractional operations on Scientifics and convert them to other fractional types like Double as early as possible.

API Note

This is a forward-compatible version of NumberCodec without a name.

scientificCodec = NumberCodec Nothing Nothing

Codec for Scientific values with bounds

Example usage

>>> let c = scientificWithBoundsCodec NumberBounds {numberBoundsLower = 2, numberBoundsUpper = 4}
>>> toJSONVia c 3
Number 3.0
>>> toJSONVia c 5
Number 5.0
>>> JSON.parseMaybe (parseJSONVia c) (Number 3)
Just 3.0
>>> JSON.parseMaybe (parseJSONVia c) (Number 5)
Nothing

WARNING

Scientific is a type that is only for JSON parsing and rendering. Do not use it for any calculations. Instead, convert to another number type before doing any calculations.

λ> (1 / 3) :: Scientific
*** Exception: fromRational has been applied to a repeating decimal which can't be represented as a Scientific! It's better to avoid performing fractional operations on Scientifics and convert them to other fractional types like Double as early as possible.

API Note

This is a forward-compatible version of NumberCodec without a name.

scientificWithBoundsCodec bounds = NumberCodec Nothing (Just bounds)

Codec for a Value

This is essentially your escape-hatch for when you would normally need a monad instance for Codec. You can build monad parsing by using valueCodec together with bimapCodec and supplying your own parsing function.

Note that this _does_ mean that the documentation will just say that you are parsing and rendering a value, so you may want to document the extra parsing further using <?>.

API Note

This is a forward-compatible version of Codec.

valueCodec = ValueCodec

A codec for bounded integers like Int, Int8, and Word.

This codec will not have a name, and it will use the boundedNumberBounds to add number bounds.

>>> let c = boundedIntegralCodec :: JSONCodec Int8
>>> toJSONVia c 5
Number 5.0
>>> JSON.parseMaybe (parseJSONVia c) (Number 100)
Just 100
>>> JSON.parseMaybe (parseJSONVia c) (Number 200)
Nothing

NumberBounds for a bounded integral type.

You can call this using TypeApplications: boundedIntegralNumberBounds Word@

A safe codec for Integer.

This codec does a bounds check for the range [-10^1024, 10^1024] it can safely parse very large numbers.

For a codec without this protection, see unsafeUnboundedIntegerCodec.

Example usage

>>> toJSONVia integerCodec 5
Number 5.0
>>> toJSONVia integerCodec (-1000000000000)
Number (-1.0e12)
>>> JSON.parseMaybe (parseJSONVia integerCodec) (Number (-4.0))
Just (-4)
>>> JSON.parseMaybe (parseJSONVia integerCodec) (Number (scientific 1 100000000))
Nothing

This is an unsafe (unchecked) version of integerCodec.

A safe codec for Natural.

This codec does a bounds check for the range [0, 10^1024] it can safely parse very large numbers.

For a codec without this protection, see unsafeUnboundedNaturalCodec.

Example usage

>>> toJSONVia naturalCodec 5
Number 5.0
>>> toJSONVia naturalCodec (1000000000000)
Number 1.0e12
>>> JSON.parseMaybe (parseJSONVia naturalCodec) (Number 4.0)
Just 4
>>> JSON.parseMaybe (parseJSONVia naturalCodec) (Number (scientific 1 100000000))
Nothing

This is an unsafe (unchecked) version of naturalCodec.

A codec for a literal piece of Text.

During parsing, only the given Text is accepted.

During rendering, the given Text is always output.

Example usage

>>> let c = literalTextCodec "hello"
>>> toJSONVia c "hello"
String "hello"
>>> toJSONVia c "world"
String "hello"
>>> JSON.parseMaybe (parseJSONVia c) (String "hello")
Just "hello"
>>> JSON.parseMaybe (parseJSONVia c) (String "world")
Nothing

A codec for a literal value corresponding to a literal piece of Text.

During parsing, only the given Text is accepted.

During rendering, the given value is always output.

Example usage

>>> let c = literalTextValueCodec True "yes"
>>> toJSONVia c True
String "yes"
>>> toJSONVia c False
String "yes"
>>> JSON.parseMaybe (parseJSONVia c) (String "yes") :: Maybe Bool
Just True
>>> JSON.parseMaybe (parseJSONVia c) (String "no") :: Maybe Bool
Nothing

A codec for a Bounded Enum that uses its Show instance to have the values correspond to literal Text values.

Example usage

>>> data Fruit = Apple | Orange deriving (Show, Eq, Enum, Bounded)
>>> let c = shownBoundedEnumCodec
>>> toJSONVia c Apple
String "Apple"
>>> JSON.parseMaybe (parseJSONVia c) (String "Orange") :: Maybe Fruit
Just Orange

A codec for an enum that can be written as constant string values

Example usage

>>> data Fruit = Apple | Orange deriving (Show, Eq)
>>> let c = stringConstCodec [(Apple, "foo"), (Orange, "bar")]
>>> toJSONVia c Orange
String "bar"
>>> JSON.parseMaybe (parseJSONVia c) (String "foo") :: Maybe Fruit
Just Apple

WARNING

If you don't provide a string for one of the type's constructors, the last string in the list will be used instead:

>>> let c = stringConstCodec [(Apple, "foo")]
>>> toJSONVia c Orange
String "foo"

A codec for an enum that can be written each with their own codec.

WARNING

If you don't provide a string for one of the type's constructors, the last codec in the list will be used instead.

Either codec

During encoding, parse a value according to either codec. During encoding, use the corresponding codec to encode either value.

HasCodec instance for sum types

To write a HasCodec instance for sum types, you will need to decide whether encoding is disjoint or not. The default, so also the implementation of this function, is possiblyJointEitherCodec, but you may want to use disjointEitherCodec instead.

Ask yourself: Can the encoding of a Left value be decoded as Right value (or vice versa)?

Yes -> use possiblyJointEitherCodec.

No -> use disjointEitherCodec.

Example usage

>>> let c = eitherCodec codec codec :: JSONCodec (Either Int String)
>>> toJSONVia c (Left 5)
Number 5.0
>>> toJSONVia c (Right "hello")
String "hello"
>>> JSON.parseMaybe (parseJSONVia c) (String "world") :: Maybe (Either Int String)
Just (Right "world")

API Note

This is a forward-compatible version of possiblyJointEitherCodec.

eitherCodec = possiblyJointEitherCodec

Possibly joint either codec

During encoding, parse a value according to either codec. During encoding, use the corresponding codec to encode either value.

This codec is for the case in which parsing must be disjoint.

HasCodec instance for sum types with an encoding that is definitely disjoint.

The eitherCodec can be used to implement HasCodec instances for sum types for which the encoding is definitely disjoint.

>>> data War = WorldWar Word8 | OtherWar Text deriving (Show, Eq)
>>> :{
  instance HasCodec War where
   codec =
     dimapCodec f g $
       disjointEitherCodec
         (codec :: JSONCodec Word8)
         (codec :: JSONCodec Text)
     where
       f = \case
         Left w -> WorldWar w
         Right t -> OtherWar t
       g = \case
         WorldWar w -> Left w
         OtherWar t -> Right t
:}

Note that this incoding is indeed disjoint because an encoded String can never be parsed as an Word8 and vice versa.

>>> toJSONViaCodec (WorldWar 2)
Number 2.0
>>> toJSONViaCodec (OtherWar "OnDrugs")
String "OnDrugs"
>>> JSON.parseMaybe parseJSONViaCodec (String "of the roses") :: Maybe War
Just (OtherWar "of the roses")

WARNING

If it turns out that the encoding of a value is not disjoint, decoding may fail and documentation may be wrong.

>>> let c = disjointEitherCodec (codec :: JSONCodec Int) (codec :: JSONCodec Int)
>>> JSON.parseMaybe (parseJSONVia c) (Number 5) :: Maybe (Either Int Int)
Nothing

Encoding still works as expected, however:

>>> toJSONVia c (Left 5)
Number 5.0
>>> toJSONVia c (Right 6)
Number 6.0

Example usage

>>> toJSONVia (disjointEitherCodec (codec :: JSONCodec Int) (codec :: JSONCodec String)) (Left 5)
Number 5.0
>>> toJSONVia (disjointEitherCodec (codec :: JSONCodec Int) (codec :: JSONCodec String)) (Right "hello")
String "hello"

API Note

This is a forward-compatible version of 'EitherCodec DisjointUnion'.

disjointEitherCodec = EitherCodec DisjointUnion

Possibly joint either codec

During encoding, parse a value according to either codec. During encoding, use the corresponding codec to encode either value.

This codec is for the case in which parsing may not be disjoint.

HasCodec instance for sum types with an encoding that is not disjoint.

The eitherCodec can be used to implement HasCodec instances for sum types. If you just have two codecs that you want to try in order, while parsing, you can do this:

>>> :{
  data Ainur
    = Valar Text Text
    | Maiar Text
    deriving (Show, Eq)
:}

>>> :{
  instance HasCodec Ainur where
    codec =
      dimapCodec f g $
        possiblyJointEitherCodec
          (object "Valar" $
            (,)
             <$> requiredField "domain" "Domain which the Valar rules over" .= fst
             <*> requiredField "name" "Name of the Valar" .= snd)
          (object "Maiar" $ requiredField "name" "Name of the Maiar")
      where
        f = \case
          Left (domain, name) -> Valar domain name
          Right name -> Maiar name
        g = \case
          Valar domain name -> Left (domain, name)
          Maiar name -> Right name
:}

Note that this encoding is indeed not disjoint, because a Valar object can parse as a Maiar value.

>>> toJSONViaCodec (Valar "Stars" "Varda")
Object (fromList [("domain",String "Stars"),("name",String "Varda")])
>>> toJSONViaCodec (Maiar "Sauron")
Object (fromList [("name",String "Sauron")])
>>> JSON.parseMaybe parseJSONViaCodec (Object (Compat.fromList [("name",String "Olorin")])) :: Maybe Ainur
Just (Maiar "Olorin")

WARNING

The order of the codecs in a possiblyJointEitherCodec matters.

In the above example, decoding works as expected because the Valar case is parsed first. If the Maiar case were first in the possiblyJointEitherCodec, then Valar could never be parsed.

API Note

This is a forward-compatible version of 'EitherCodec PossiblyJointUnion'.

possiblyJointEitherCodec = EitherCodec PossiblyJointUnion

Wrap up a value of type b with its codec to produce and encoder for as that ignores its input and instead encodes the value b. This is useful for building discriminatedUnionCodecs.

Map a codec for decoding bs into a decoder for as. This is useful for building discriminatedUnionCodecs.

Encode/decode a discriminated union of objects

The type of object being encoded/decoded is discriminated by a designated "discriminator" property on the object which takes a string value.

When encoding, the provided function is applied to the input to obtain a new encoder for the input. The function mapToEncoder is provided to assist with building these encoders. See examples in hs.

When decoding, the value of the discriminator property is looked up in the HashMap to obtain a decoder for the output. The function mapToDecoder is provided to assist with building these decoders. See examples in hs.

The HashMap is also used to generate schemas for the type. In particular, for OpenAPI 3, it will generate a schema with a discriminator, as defined by https://swagger.io/docs/specification/data-models/inheritance-and-polymorphism/

API Note

This is a forward-compatible version of DiscriminatedUnionCodec.

discriminatedUnionCodec = 'DiscriminatedUnionCodec'

Map both directions of a codec

You can use this function to change the type of a codec as long as the two functions are inverses.

HasCodec instance for newtypes

A good use-case is implementing HasCodec for newtypes:

newtype MyInt = MyInt { unMyInt :: Int }
instance HasCodec MyInt where
  codec = dimapCodec MyInt unMyInt codec

Map a codec's input and output types.

This function allows you to have the parsing fail in a new way.

If you use this function, then you will most likely want to add documentation about how not every value that the schema specifies will be accepted.

This function is like BimapCodec except it also combines one level of a nested BimapCodecs.

Example usage

logLevelCodec :: JSONCodec LogLevel logLevelCodec = bimapCodec parseLogLevel renderLogLevel codec ? "Valid values include DEBUG, INFO, WARNING, ERROR."

Map the output part of a codec

You can use this function if you only need to map the parsing-side of a codec. This function is probably only useful if the function you map does not change the codec type.

WARNING: This can be used to produce a codec that does not roundtrip.

>>> JSON.parseMaybe (parseJSONVia (rmapCodec (*2) codec)) (Number 5) :: Maybe Int
Just 10

Map the input part of a codec

You can use this function if you only need to map the rendering-side of a codec. This function is probably only useful if the function you map does not change the codec type.

WARNING: This can be used to produce a codec that does not roundtrip.

>>> toJSONVia (lmapCodec (*2) (codec :: JSONCodec Int)) 5
Number 10.0

Maybe codec

This can be used to also allow null during decoding of a Maybe value.

During decoding, also accept a null value as Nothing. During encoding, encode as usual.

Example usage

>>> toJSONVia (maybeCodec codec) (Just 'a')
String "a"
>>> toJSONVia (maybeCodec codec) (Nothing :: Maybe Char)
Null

List codec

Build a codec for lists of values from a codec for a single value.

Example usage

>>> toJSONVia (listCodec codec) ['a','b']
Array [String "a",String "b"]

API Note

This is the list version of vectorCodec.

Build a codec for nonempty lists of values from a codec for a single value.

Example usage

>>> toJSONVia (nonEmptyCodec codec) ('a' :| ['b'])
Array [String "a",String "b"]

API Note

This is the non-empty list version of vectorCodec.

Single or list codec

This codec behaves like listCodec, except the values may also be simplified as a single value.

During parsing, a single element may be parsed as the list of just that element. During rendering, a list with only one element will be rendered as just that element.

Example usage

>>> let c = singleOrListCodec codec :: JSONCodec [Int]
>>> toJSONVia c [5]
Number 5.0
>>> toJSONVia c [5,6]
Array [Number 5.0,Number 6.0]
>>> JSON.parseMaybe (parseJSONVia c) (Number 5) :: Maybe [Int]
Just [5]
>>> JSON.parseMaybe (parseJSONVia c) (Array [Number 5, Number 6]) :: Maybe [Int]
Just [5,6]

WARNING

If you use nested lists, for example when the given value codec is also a listCodec, you may get in trouble with ambiguities during parsing.

Single or nonempty list codec

This codec behaves like nonEmptyCodec, except the values may also be simplified as a single value.

During parsing, a single element may be parsed as the list of just that element. During rendering, a list with only one element will be rendered as just that element.

Example usage

>>> let c = singleOrNonEmptyCodec codec :: JSONCodec (NonEmpty Int)
>>> toJSONVia c (5 :| [])
Number 5.0
>>> toJSONVia c (5 :| [6])
Array [Number 5.0,Number 6.0]
>>> JSON.parseMaybe (parseJSONVia c) (Number 5) :: Maybe (NonEmpty Int)
Just (5 :| [])
>>> JSON.parseMaybe (parseJSONVia c) (Array [Number 5, Number 6]) :: Maybe (NonEmpty Int)
Just (5 :| [6])

WARNING

If you use nested lists, for example when the given value codec is also a nonEmptyCodec, you may get in trouble with ambiguities during parsing.

API Note

This is a nonempty version of singleOrListCodec.

Vector codec

Build a codec for vectors of values from a codec for a single value.

Example usage

>>> toJSONVia (vectorCodec codec) (Vector.fromList ['a','b'])
Array [String "a",String "b"]

API Note

This is a forward-compatible version of ArrayOfCodec without a name.

vectorCodec = ArrayOfCodec Nothing

Like parseAlternatives, but with only one alternative codec

Example usage

Values

>>> data Fruit = Apple | Orange deriving (Show, Eq, Bounded, Enum)
>>> let c = parseAlternative shownBoundedEnumCodec (stringConstCodec [(Apple, "foo"), (Orange, "bar")])
>>> toJSONVia c Apple
String "Apple"
>>> JSON.parseMaybe (parseJSONVia c) (String "foo") :: Maybe Fruit
Just Apple
>>> JSON.parseMaybe (parseJSONVia c) (String "Apple") :: Maybe Fruit
Just Apple

Required object fields

>>> data Fruit = Apple | Orange deriving (Show, Eq, Bounded, Enum)
>>> let c = shownBoundedEnumCodec
>>> let o = parseAlternative (requiredFieldWith "current" c "current key for this field") (requiredFieldWith "legacy" c "legacy key for this field")
>>> toJSONObjectVia o Apple
fromList [("current",String "Apple")]
>>> JSON.parseMaybe (parseJSONObjectVia o) (KM.fromList [("current",String "Apple")]) :: Maybe Fruit
Just Apple
>>> JSON.parseMaybe (parseJSONObjectVia o) (KM.fromList [("legacy",String "Apple")]) :: Maybe Fruit
Just Apple
>>> JSON.parseMaybe (parseJSONObjectVia o) (KM.fromList [("current",String "Tomato")]) :: Maybe Fruit
Nothing

Required object fields

While parseAlternative works exactly like you would expect it would with requiredField, using parseAlterternative with optional fields has some pitfalls.

>>> data Fruit = Apple | Orange deriving (Show, Eq, Bounded, Enum)
>>> let c = shownBoundedEnumCodec
>>> let o = parseAlternative (optionalFieldWith "current" c "current key for this field") (optionalFieldWith "legacy" c "legacy key for this field")
>>> toJSONObjectVia o (Just Apple)
fromList [("current",String "Apple")]
>>> toJSONObjectVia o Nothing
fromList []
>>> JSON.parseMaybe (parseJSONObjectVia o) (KM.fromList [("current",String "Apple")]) :: Maybe (Maybe Fruit)
Just (Just Apple)

! This is the important result ! The second optionalFieldWith is not tried because the first one _succeeds_ in parsing Nothing

>>> JSON.parseMaybe (parseJSONObjectVia o) (KM.fromList [("legacy",String "Apple")]) :: Maybe (Maybe Fruit)
Just Nothing

Here the parser succeeds as well, because it fails to parse the current field, so it tries to parse the legacy field, which is missing.

>>> JSON.parseMaybe (parseJSONObjectVia o) (KM.fromList [("current",String "Tomato")]) :: Maybe (Maybe Fruit)
Just Nothing

Use one codec for the default way of parsing and rendering, but then also use a list of other codecs for potentially different parsing.

You can use this for keeping old ways of parsing intact while already rendering in the new way.

Example usage

>>> data Fruit = Apple | Orange deriving (Show, Eq, Bounded, Enum)
>>> let c = parseAlternatives shownBoundedEnumCodec [stringConstCodec [(Apple, "foo"), (Orange, "bar")]]
>>> toJSONVia c Apple
String "Apple"
>>> JSON.parseMaybe (parseJSONVia c) (String "foo") :: Maybe Fruit
Just Apple
>>> JSON.parseMaybe (parseJSONVia c) (String "Apple") :: Maybe Fruit
Just Apple
>>> JSON.parseMaybe (parseJSONVia c) (String "Tomato") :: Maybe Fruit
Nothing

A choice codec, but unlike eitherCodec, it's for the same output type instead of different ones.

While parsing, this codec will first try the left codec, then the right if that fails.

While rendering, the provided function is used to decide which codec to use for rendering.

Note: The reason this is less primitive than the eitherCodec is that Either makes it clear which codec you want to use for rendering. In this case, we need to provide our own function for choosing which codec we want to use for rendering.

Example usage

>>> :{
  let c =
       matchChoiceCodec
        (literalTextCodec "even")
        (literalTextCodec "odd")
        (\s -> if s == "even" then Left s else Right s)
:}

>>> toJSONVia c "even"
String "even"
>>> toJSONVia c "odd"
String "odd"
>>> JSON.parseMaybe (parseJSONVia c) (String "even") :: Maybe Text
Just "even"
>>> JSON.parseMaybe (parseJSONVia c) (String "odd") :: Maybe Text
Just "odd"

API Note

This is a forward-compatible version of 'matchChoiceCodecAs PossiblyJointUnion':

disjointMatchChoiceCodec = matchChoiceCodecAs PossiblyJointUnion

Disjoint version of matchChoiceCodec

API Note

This is a forward-compatible version of 'matchChoiceCodecAs DisjointUnion':

disjointMatchChoiceCodec = matchChoiceCodecAs DisjointUnion

An even more general version of matchChoiceCodec and disjointMatchChoiceCodec.

A choice codec for a list of options, each with their own rendering matcher.

During parsing, each of the codecs are tried from first to last until one succeeds.

During rendering, each matching function is tried until either one succeeds and the corresponding codec is used, or none succeed and the fallback codec is used.

Example usage

>>> :{
  let c =
       matchChoicesCodec
         [ (\s -> if s == "even" then Just s else Nothing, literalTextCodec "even")
         , (\s -> if s == "odd" then Just s else Nothing, literalTextCodec "odd")
         ] (literalTextCodec "fallback")
:}

>>> toJSONVia c "even"
String "even"
>>> toJSONVia c "odd"
String "odd"
>>> toJSONVia c "foobar"
String "fallback"
>>> JSON.parseMaybe (parseJSONVia c) (String "even") :: Maybe Text
Just "even"
>>> JSON.parseMaybe (parseJSONVia c) (String "odd") :: Maybe Text
Just "odd"
>>> JSON.parseMaybe (parseJSONVia c) (String "foobar") :: Maybe Text
Nothing
>>> JSON.parseMaybe (parseJSONVia c) (String "fallback") :: Maybe Text
Just "fallback"

API Note

This is a forward-compatible version of 'matchChoicesCodecAs DisjointUnion'.

disjointMatchChoiceCodec = matchChoicesCodecAs DisjointUnion

Disjoint version of matchChoicesCodec

API Note

This is a forward-compatible version of 'matchChoicesCodecAs DisjointUnion'.

disjointMatchChoiceCodec = matchChoicesCodecAs DisjointUnion

An even more general version of matchChoicesCodec and disjointMatchChoicesCodec

Add a comment to a codec

This is an infix version of CommentCodec > (?) = flip CommentCodec

A version of <?> that lets you supply a list of lines of text instead of a single text.

This helps when you use an automated formatter that deals with lists more nicely than with multi-line strings.

A Self-documenting encoder and decoder,

also called an Autodocodec.

In an ideal situation, this type would have only one type parameter: 'Codec value'. This does not work very well because we want to be able to implement Functor and Applicative, which each require a kind '* -> *'. So instead we use two type parameters.

The two type parameters correspond to the phase in which they are used:

• The input parameter is used for the type that is used during encoding of a value, so it's the input to the codec.
• The output parameter is used for the type that is used during decoding of a value, so it's the output of the codec.
• Both parameters are unused during documentation.

A codec within the Value context.

An Codec can be used to turn a Haskell value into a Value or to parse a Value into a haskell value.

This cannot be used in certain places where ObjectCodec could be used, and vice versa.

A codec within the Value context.

An Object can be used to turn a Haskell value into a Value or to parse a Value into a haskell value.

This cannot be used in certain places where Codec could be used, and vice versa.

Produce a value without parsing any part of an Object.

This function exists to implement Applicative (ObjectCodec input).

API Note

This is a forward-compatible version of PureCodec.

pureCodec = PureCodec

Sequentially apply two codecs that parse part of an Object.

This function exists to implement Applicative (ObjectCodec input).

API Note

This is a forward-compatible version of ApCodec.

apCodec = ApCodec

Autodocodec is a wrapper to provide codec-based deriving strategies.

Example usage

data Via = Via {viaOne :: !Text, viaTwo :: !Text}
  deriving stock (Show, Eq, Generic)
  deriving (FromJSON, ToJSON) via (Autodocodec Via)

instance HasCodec Via where
  codec =
    object "Via" $
      Via
        <$> requiredField "one" "first field" .= viaOne
        <*> requiredField "two" "second field" .= viaTwo

Show a codec to a human.

This function exists for codec debugging. It omits any unshowable information from the output.