Home > Net >  Python -> Haskell FFI: The best way how to represent Haskell recursive datastructure in Python
Python -> Haskell FFI: The best way how to represent Haskell recursive datastructure in Python

Time:02-11

Let's have some finite recursive datastructure in Haskell. Eg.

data Tree = Node Tree Tree | Nil

I need to be able to load such datastructure from Haskell to Python, change it and return it back to Haskell.

Is there some standard/elegant way how to do it without much pain? Eg. using some directory like objects?

Thanks.

CodePudding user response:

The easiest option is probably to go via JSON, because Haskell has easy support for saving data as JSON and Python can directly load it as dicts.

{-# LANGUAGE DeriveGeneric, DeriveAnyClass #-}

import GHC.Generics

import Data.Aeson
import Data.Aeson.TH

data Tree = Node Tree Tree | Nil
 deriving (Generic, FromJSON, ToJSON)

This generates rather awkward JSON though, like, Node (Node Nil Nil) Nil becomes

        "tag": "Node",
        "contents": [
            {
                "tag": "Node",
                "contents": [
                    {
                        "tag": "Nil"
                    },
                    {
                        "tag": "Nil"
                    }
                ]
            },
            {
                "tag": "Nil"
            }
        ]

It gets much more compact with

data TreeNode = Node { lSubtree, rSubtree :: Tree }
 deriving (Generic, FromJSON, ToJSON)

type Tree = Maybe TreeNode

where the equivalent Node (Just (Node Nothing Nothing)) Nothing is now saved as

        {
            "rSubtree": null,
            "lSubtree": {
                "rSubtree": null,
                "lSubtree": null
            }
        }
  • Related