An efficient serialization and deserialization library for Motoko.

The library contains four modules:

• Candid

  • fromText() - Converts Candid text to its serialized form.
  • toText() - Converts serialized candid to its textual representation.
  • encode() - Converts the Candid variant to a blob.
  • decode() - Converts a blob to the Candid variant.

• CBOR

  • encode() - Converts serialized candid to CBOR.
  • decode() - Converts CBOR to a serialized candid.

• JSON

  • fromText() - Converts JSON text to serialized candid.
  • toText() - Converts serialized candid to JSON text.

• URL-Encoded Pairs

  • fromText() - Converts URL-encoded text to serialized candid.
  • toText() - Converts serialized candid to URL-encoded text.

1. Install mops.
2. Inside your project directory, run:

mops install serde

To start, import the necessary modules:

import { JSON; Candid; CBOR; UrlEncoded } from "mo:serde";

The following code can be used for converting data between the other modules (Candid and URL-Encoded Pairs).

Example: JSON to Motoko

1. Defining Data Type: This critical step informs the conversion functions (from_candid and to_candid) about how to handle the data.

   Consider the following JSON data:

   [
       {
           "name": "John",
           "id": 123
       },
       {
           "name": "Jane",
           "id": 456,
           "email": "[email protected]"
       }
   ]

   The optional email field translates to:

   type User = {
       name: Text;
       id: Nat;
       email: ?Text;
   };

2. Conversion: a. Parse JSON text into a candid blob using JSON.fromText. b. Convert the blob to a Motoko data type with from_candid.

   let jsonText = "[{\"name\": \"John\", \"id\": 123}, {\"name\": \"Jane\", \"id\": 456, \"email\": \"[email protected]\"}]";
   
   let #ok(blob) = JSON.fromText(jsonText, null); // you probably want to handle the error case here :)
   let users : ?[User] = from_candid(blob);
   
   assert users == ?[
       {
           name = "John";
           id = 123;
           email = null;
       },
       {
           name = "Jane";
           id = 456;
           email = ?"[email protected]";
       },
   ];

Example: Motoko to JSON

1. Record Keys: Collect all unique record keys from your data type into an array. This helps the module convert the record keys correctly instead of returning its hash.

   let UserKeys = ["name", "id", "email"];

2. Conversion:

   let users: [User] = [
       {
           name = "John";
           id = 123;
           email = null;
       },
       {
           name = "Jane";
           id = 456;
           email = ?"[email protected]";
       },
   ];
   
   let blob = to_candid(users);
   let json_result = JSON.toText(blob, UserKeys, null);
   
   assert json_result == #ok(
        "[{\"name\": \"John\",\"id\": 123},{\"name\": \"Jane\",\"id\":456,\"email\":\"[email protected]\"}]"
    );

Example: Renaming Fields

• Useful way to rename fields with reserved keywords in Motoko.

import Serde from "mo:serde";

    // type JsonSchemaWithReservedKeys = {
    //     type: Text; // reserved
    //     label: Text;  // reserved
    //     id: Nat;
    // };

type Item = {
    item_type: Text;
    item_label: Text;
    id: Nat
};

let jsonText = "{\"type\": \"bar\", \"label\": \"foo\", \"id\": 112}";
let options: Serde.Options = { 
    renameKeys = [("type", "item_type"), ("label", "item_label")] 
};

let #ok(blob) = Serde.JSON.fromText(jsonText, ?options);
let renamedKeys: ?Item = from_candid(blob);

assert renamedKeys == ?{ item_type = "bar"; item_label = "foo"; id = 112 };

Checkout the usage guide for additional examples:

• Candid Text
• URL-Encoded Pairs

• Users must provide a list of record keys and variant names during conversions from Motoko to other data formats due to constraints in the candid format.
• Lack of specific syntax for conversion between Blob, Principal, and bounded Nat/Int types.

1. Install dependencies:

   • mops
   • mocv
   • wasmtime

2. Inside the project directory, run:

mops test

Happy coding with serde! 🚀