Turns a tracery json-string into a generator

import Json.Decode
import Random
import Result.Extra


generate : Int -> String -> String
generate seed json =
    json
        |> Tracery.fromJson
        |> Result.Extra.unpack
            Json.Decode.errorToString
            (\grammar ->
                Random.step (Tracery.run grammar) (Random.initialSeed seed)
                    |> Tuple.first
            )

A tracery json is a object that has a origin field.

The # and \ characters need to be escaped.

"""
{ "origin": "The \\\\# and \\\\\\\\ characters need to be escaped."}
"""
|> generate 42
--> "The # and \\ characters need to be escaped."

If you provide a list, tracer will tick an element at random.

"""
{ "origin": ["I like cats","I like dogs"]}
"""
|> generate 42
--> "I like cats"

You can reference other fields using #..#

"""
{ "origin": ["I have two pets: a #pet# and a #pet#"]
, "pet": ["cat","dog","fish","parrot"]
}
"""
|> generate 42
--> "I have two pets: a dog and a cat"

Definitions may also be recursive.

"""
{ "origin": ["I have #pets#"]
, "pets": ["a #pet#","a #pet# and #pets#"]
, "pet": ["cat","dog","fish","parrot"]
}
"""
|> generate 20
--> "I have a fish and a cat and a dog"

You can define constants by providing a string instead of a list.

"""
{ "origin": ["My #favoritePet# is the best #favoritePet# in the world"]
, "favoritePet" : "#pet#"
, "pet": ["cat","dog","fish","parrot"]
}
"""
|> generate 42
--> "My dog is the best dog in the world"

You may define sub-definitions to organize your definitions.

"""
{ "origin": ["My #cat#","My #dog#"]
, "cat":
  { "origin":"cat is named #name#"
  , "name": ["Cleopatra","Cesar"]
  }
, "dog":
  { "origin":"dog is named #name#"
  , "name": ["Athena","Zeus"]
  }
}
"""
|> generate 42
--> "My cat is named Cleopatra"

Runs a grammar until it ends.

Some recursive definitions might take a long time.

Use runTo if you want to avoid long waiting times.

Runs a grammar until it reaches a key in the list.

compute a single step.

compute a single step.

The algorithm will never choose a recursive option for any key in the given list.

compute a single step.

The algorithm will always choose a recursive option for any key in the given list.

If no such option exists, then the empty list will be used instead.