We define the index of our CursorHistory' to be the Count.

Provide some of the type parameters that the underlying DecodeResultT requires. This contains the state and error management as we walk around our zipper and decode our JSON input.

Addtionally we keep our parsing function in a ReaderT such that it's accessible for all of the decoding steps.

Decoder type that is used directly to convert Json structures to other data types.

Wrapper type for the SuccinctCursor

Convenience Error structure for the separate parsing/decoding phases. For when things really aren't that complicated.

Generalise a Decoder that has been specialised to Identity back to some 'Monad f'.

Pretty print the given CursorHistory' to a more useful format compared to a Seq of i.

Function to define a Decoder for a specific data type.

For example, given the following data type:

data Image = Image
  { _imageW        :: Int
  , _imageH        :: Int
  , _imageTitle    :: Text
  , _imageAnimated :: Bool
  , _imageIDs      :: [Int]
  }

We can use withCursor to write a decoder that will be given a cursor that we can use to build the data types that we need.

imageDecoder :: Monad f => Decoder f Image
imageDecoder = withCursor $ \curs -> D.down curs >>= Image
  <$> D.fromKey "Width" D.int curs
  <*> D.fromKey "Height" D.int curs
  <*> D.fromKey "Title" D.text curs
  <*> D.fromKey "Animated" D.bool curs
  <*> D.fromKey "IDs" intArray curs

It's up to you to provide a cursor that is at the correct position for a Decoder to operate, but building decoders in this way simplifies creating decoders for larger structures, as the smaller pieces contain fewer assumptions. This encourages greater reuse of decoders and simplifies the debugging process.

Take a ByteString input and build an index of the JSON structure inside

Lens for accessing the rank of the JsonCursor. The rank forms part of the calculation that is the cursors current position in the index.

Execute the given function n times.

Move the cursor down or into the child of the current cursor position.

The following examples use "*" to represent the cursor position.

Starting position:

 *{"fred": 33, "sally": 44 }

After moving down:

 { *"fred": 33, "sally": 44 }

This function will also move into the elements in an array:

Starting position:

 *[1,2,3]

After moving down:

 [*1,2,3]

This function is essential when dealing with the inner elements of objects or arrays. As you must first move down into the focus. However, you cannot move down into an empty list or empty object. The reason for this is that there will be nothing in the index for the element at the first position. Thus the movement will be considered invalid.

These will fail if you attempt to move down:

 *[]

 *{}

Move the cursor up into the parent of the current cursor position.

The following examples use "*" to represent the cursor position.

Starting position:

 { "fred": 33, *"sally": 44 }

After moving up:

 *{"fred": 33, "sally": 44 }

Attempt a Decoder action that might fail and return a Maybe value instead.

Move the cursor rightwards n times.

Starting position:

 [*1, 2, 3]

After moveRightN 2:

 [1, 2, *3]

Helper function to move right once.

Move the cursor leftwards n times.

Helper function to move left once.

Starting position:

 [1, 2, *3]

Ater moveLeft1:

 [1, *2, 3]

Attempt to move to the value at a given key on the current JSON object. This will only work if you have already moved down into the JSON object, because the cursor allows you to step over an entire object in a single. It has to be told to move into the object first, otherwise it will not look in the correct location for keys.

Cursor position indicated by "*".

Assuming cursor positioned here:

 *{ "foo": 33, "fieldB": "pew pew" }

This won't work, because we're AT the object, not IN the object: moveToKey "foo" cursor

This will work, because we've moved down INTO the object: down cursor >>= moveToKey "foo"

Given a rank value, attempt to move the cursor directly to that position.

Returns a InputOutOfBounds error if that position is invalid.

Using the given parsing function, attempt to decode the value of the ByteString at the current cursor position.

Move to the first occurence of this key, as per moveToKey and then attempt to run the given Decoder on that value, returning the result.

This decoder does not assume you have moved into the object.

...
txtVal <- fromKey "foo" text c
...

A simplified version of fromKey that takes a Text value indicating a key to be moved to and decoded using the given 'Decoder f a'. If you don't need any special cursor movements to reach the list of keys you require, you could use this function to build a trivial Decoder for a record type:

This decoder assumes it is positioned at the top of an object and will move down each time, before attempting to find the given key.

data MyRec = MyRec { fieldA :: Text, fieldB :: Int }

myRecDecoder :: Decoder f MyRec
myRecDecoder = MyRec
  <$> atKey "field_a" text
  <*> atKey "field_b" int

Using the given Decoder, try to decode the current focus.

myIntList <- focus (list int) cursor

A version of fromKey that returns its result in Maybe. If the key is not present in the object, Nothing is returned. If the key is present, decoding will be performed as with fromKey.

A version of atKey that returns its result in Maybe. If the key is not present in the object, Nothing is returned. If the key is present, decoding will be performed as with atKey.

For example, if a key could be absent and could be null if present, it could be decoded as follows:

join <$> atKeyOptional "key" (maybeOrNull text)

Attempt to work with a JCurs provided the type of Json at the current position matches your expectations.

Such as:

withType JsonTypeArray d

d will only be entered if the cursor at the current position is a JSON array: '[]'.

From the current cursor position, move leftwards one position at a time and push each a onto the front of some Cons structure.

From the current cursor position, move rightwards one position at a time, and append the a to some Snoc structure.

Higher order function for combining a folding function with repeated cursor movements. This lets you combine arbitrary cursor movements with an accumulating function.

The functions leftwardCons and rightwardSnoc are both implemented using this function.

leftwardCons = foldCursor (flip cons) moveLeft1

rightwardSnoc = foldCursor snoc moveRight1

At the given cursor position, return the Count or rank of that position. Useful if you want to build a map of a complicated structure such that you're able to optimise your Decoder by using moveToRankN instead of individual cursor movements.

Create a Decoder from a Prism'.

As per prismD but fail the Decoder if unsuccessful.

Like prismDOrFail, but lets you use the a to construct the error.

Decode the Json structure at the cursor. Useful if you don't have a need to convert the Json and only want to make changes before sending it on its way.

Decode an Int.

Decode a Scientific number value.

Decoder for some Integral type. This conversion is walked through Mayan, I mean, Scientific to try to avoid numeric explosion issues.

Decode a String value.

Decode a strict ByteString value.

Decode a lazy ByteString value.

Decode a Char value that is equivalent to a Haskell Char value, as Haskell Char supports a wider range than JSON.

Decode a Char that will fail if the Char is outside of the range U+D800 to U+DFFF.

Decode Text

Decode a Bool value.

Decode an explicit null value at the current cursor position.

Given a Decoder for a, attempt to decode a NonEmpty list of a at the current cursor position.

Helper to create a 'NonEmpty a' Decoder.

Like nonemptyAt, this takes a Decoder of a and at the given cursor will try to decode a '[a]'.

Helper function to simplify writing a '[]' decoder.

Try to decode an object using the given key and value Decoders at the given cursor.

Helper function to simplify writing a '{}' decoder.

Try to decode an optional value, returning the given default value if Nothing is returned.

Named to match it's Encoder counterpart, this function will decode an optional value.

Decode either an a or a b, failing if neither Decoder succeeds. The Right decoder is attempted first.

Helper function for "pattern matching" on a decoded value to some Haskell value. The Text argument is used in the error message should this decoder fail. Normally it would simply be the name of the type you are writing the decoder for.

This is useful for decoding sum types, such as:

data MyEnum
  = A
  | B
  | C

decodeMyEnum :: Monad f => Decoder f MyEnum
decodeMyEnum = D.oneOf D.text "MyEnum"
  [ ("a", A)
  , ("b", B)
  , ("c", C)
  ]

decodeMyEnumFromInt :: Monad f => Decoder f MyEnum
decodeMyEnumFromInt = D.oneOf D.int "MyEnum"
  [ (1, A)
  , (2, B)
  , (3, C)
  ]

A specialised decoder for moving over a JSON object where the keys are values that you would like to have as part of the value at the different keys.

An example of such an input is:

{ "Collection" : {
  "BobsInput_ce43dff22": {
    "someValue": "Some data"
  },
  "FredsInput_a4b32def": {
    "someValue": "Some different data"
  }
}

Where those key values like "XInput_YYYY" are to be included in the object.

Given a type like this:

data ContainsItsKey = ContainsItsKey
  { _containsItsKey_KeyValue :: Text
  , _containsItsKey_SomeValue :: Text
  }

To decode the above you would use this function like so:

takesKeyDecoder :: Monad f => Text -> Decoder f ContainsItsKey
takesKeyDecoder k = ContainsItsKey k <$> D.atKey "someValue" D.text

collectionDecoder :: Monad f => Decoder f [ContainsItsKey]
collectionDecoder = D.atKey "Collection" $ D.passKeysToValues [] D.text takesKeyDecoder