Jolt.Json.Newtonsoft 1.0.1

Jolt

Welcome! This is a JSON transformation language inspired by XSLT and the wonderful .Net JSON adaptation of it over at JUST.Net. This project provides an expression-based interpreter for the language and a highly extensible way of approaching the same problem, namely How To Transform JSON Into Different JSON.

There are two .Net Standard 2.1 packages available: Jolt.Json.Newtonsoft and Jolt.Json.DotNet. These use Newtonsoft or System.Text.Json functionality, respectively.

Build/Release Status

Package Name	NuGet Version
Jolt.Json.DotNet
Jolt.Json.Newtonsoft

Syntax

The transformation language itself is pretty straightforward and primarily relies on defining and using methods to handle the transformation work needed, both as provided by the library within the package or external methods that you will create and register for use.

Methods

All method calls begin with a # symbol and are case-sensitive. Open and close parentheses surround the method arguments, which are comma-separated. Let's take a look at a an example from the library.

{
    "value": "#valueOf($.some.json.path)"
}

You'll notice the use of a JSON Path query as a parameter. This allows you to query the source document for the JSON value that exists at that path and use it in your transformation. Most of the library methods can take either a path, like above, or the value from the source document itself.

{
    "valueExists": "#exists($.some.json.path)",
    "otherValueExists": "#exists(#valueOf($.some.other.json.path))"
}

You can also optionally use the piped method syntax with an arrow in order to use the result of one method as the first parameter of another method. In that case, that parameter is removed from the second call.

{
    "valueExists": "#valueOf($.some.json.path)->#exists()"
}

If you are familiar with C# extension methods, this is the same sort of concept. We can interchangeably use any method (including your own external methods) as either standalone or piped and pretend that a method call is an instance method for a little while to help with readability. We'll go into the available library methods further down, but let's talk first about how you'd possibly go about implementing your own method to use in a transformer.

External Methods

These are methods that you have created in your code. In order to use them, you will need to register them prior to using the transformer. Let's assume that you have the following C# method that you would like to be able to use. The easiest way to approach this is to use the JoltExternalMethod attribute from the Jolt.Library namespace. If the method name needs to be different when used from a transformer, specify an alias in the attribute constructor, such as [JoltExternalMethod("differentName")].

using Jolt.Library;

namespace JoltExample;

public class TransformerMethods
{
    [JoltExternalMethod]
    public static bool IsNotNull(string value) => value != null;
}

You will need to create a JSON transformer and register your methods with it prior to the actual transformation. There are two .Net Standard 2.1 packages available by default: Jolt.Json.Newtonsoft and Jolt.Json.DotNet, which use either Newtonsoft or System.Text.Json as their particular flavor of implementation. Also, note that the Jolt.Json.DotNet package has an additional dependency, namely on JsonPath.Net, due to there being no default JSON Path support out of the box in System.Text.Json as of yet.

For our demonstration purposes here, we'll assume Newtonsoft.

using Jolt.Json.Newtonsoft;

// Omitting boilerplate class definition.

public string TransformationExample(string transformerJson, string sourceJsonDocument)
{
    var transformer = JoltJsonTransformer.DefaultWith<TransformerMethods>(transformerJson);
    return transformer.Transform(sourceJsonDocument);
}

Now that we've got the transformer set up in code and the method registered, let's create a JSON transformer that makes use of the new method.

{
    "result": "#IsNotNull(#valueOf($.stringValue))"
}

You'll notice that the JSON path above is to a property called stringValue. This transformer assumes that you know that a string property with that name exists on the source document so, for example, this document here could be your source.

{
    "stringValue": "not null"
}

That will execute your method with the value "not null" and create the following output.

{
    "result": true
}

Your external methods that provide their output at the root level, as in the example above, must return a value that is either an acceptable JSON value (e.g. string, boolean), an IJsonToken instance, an EvaluationResult instance, or an instance that will be serialized to JSON. Your other non-root external methods may return and consume whatever instance types they want.

The external methods you create may be either static or instance methods. We took a look at how a static method would be used above, and you can follow the same path for instance methods with one extra step. Let's assume we added the following instance method to the TransformerMethods class above.

[JoltExternalMethod]
public string ReverseString(string value) => value.Reverse();

In order to use the instance method, we need to pass an instance of TransformerMethods to the transformer creation call.

var transformer = JoltJsonTransformer.DefaultWith<TransformerMethods>(transformerJson, new TransformerMethods());

This instance will be used for all instance method calls during the transformation step, so you could easily modify your JSON transformer to include this in the same way as before. Keep in mind that your own methods can also take advantage of the piped method syntax, so let's demonstrate that here.

{
    "reversed": "#valueOf($.stringValue)->#ReverseString()"
}

This will create the following output when used to transform the same source JSON we used earlier.

{
    "reversed": "llun ton"
}

Math And Equality

The usual basic math operators are implemented, namely addition, subtraction, multiplication, and division, using +, -, *, and / respectively, along with comparison and equality as =, !=, >, >=, <, and <=. Operations can be parenthesized as well. These are all used much like you're used to and can take the results of methods as operands. For example, assuming the following source JSON:

{
    "first": 5,
    "second": 4,
    "third": 3,
    "fourth": 2,
    "fifth": 1
}

You could create a transformer that would look something like this:

{
    "result": "#valueOf($.first) * #valueOf($.second) - #valueOf($.third) + #valueOf($.fourth) / #valueOf($.fifth) = 19"
}

And after transformation it would look like this:

{
    "result": true
}

It's also worth noting that the addition operator + is overloaded and can serve as a shortcut to string concatenation, either with a source document value or a literal string. For example, in the following source JSON:

{
    "value": "example"
}

And assuming that you have a transformer that looks like this:

{
    "result": "#valueOf($.value) + #valueOf($.value) + 'hello'"
}

The output would look like this:

{
    "result": "exampleexamplehello"
}

Range Variables

There are times that you may want to perform a particular transform and then do something additional to it, whether that's simple math or looping over it. Most of the time, you'd also rather that the intermediary transform not stick around in the final JSON output, it's just there to do some preprocessing. This is an area where you may find that range variables work for your needs.

You can declare and set a variable in the property name one of two ways, either as the direct result of an expression or as the output of a loop. Let's take a look at both.

{
    "@firstVar": "#valueOf($.someInteger)",
    "@secondVar": "#valueOf($.someOtherInteger)",
    "sum": "@firstVar + @secondVar"
}

This is one way to directly declare and use variables in your transformations. All variables are prefixed with the @ symbol and specifying it in the property name indicates that you would like to set it and also remove that JSON node from the transformed result. Let's see how you would be able to do some preprocessing on an array now.

{
    "#loop($.someArray)->@tempResult": [
        {
            "intermediateValue": "#loopValueOf($.someInteger) * 2"
        }
    ],
    "#loop(@tempResult)->'actualResult'": [
        {
            "finalValue": "#valueOf($.intermediateValue) + 5"
        }
    ]
}

The first loop will set the variable @tempResult to the value of the transformed array, which will use the provided content template to transform each array element it finds, then removes the node from the output. The second loop takes the variable results as input and loops over that intermediary array, applying additional transformation to the results, naming the final array property as actualResult.

You can also declare variables within the content template of a loop, but they only exist within it and on a per-iteration basis (meaning that the variable is redeclared, set, and thrown away every time through the loop). Let's see an example of that here.

{
    "#loop($.someArray)->'finalResult'": [
        {
            "@scopedVar": "#loopValueOf($.someInteger) / 2",
            "multipliedValue": "@scopedVar - 3"
        }
    ],
    "invalidValue": "@scopedVar"
}

In the resulting JSON output, the finalResult property will be populated by the array loop but the invalidValue property will be null due to accessing a variable which has already been destroyed.

Library Methods

This package comes with quite a few methods built into it to get you started, all of which are documented here and represent the most common things that you may want to do when transforming a JSON file. If you find that an opportunity for a new library method exists, please raise an issue and it will be considered. For more detail on the individual library methods, please see the Jolt wiki over here.

Alternate External Method Registrations

There are additional ways available to register your external methods, let's take a look at another one.

Method Registration

The slight bit of added complexity that the previous way (using the [JoltExternalMethod] attribute) hid from you is the registration process. If you'd like to do this process manually, you will need a few things.

The first thing is the MethodRegistration class. This wraps up all of the relevant information about your external method in a way that the transformer can understand in order to call it. You can either choose static or instance methods. Let's register the ones we defined earlier.

var staticRegistration = MethodRegistration.FromStaticMethod(typeof(TransformerMethods), nameof(TransformerMethods.IsNotNull));
var instanceRegistration = MethodRegistration.FromInstanceMethod(nameof(TransformerMethods.ReverseString));

You'll notice that you don't have to provide an instance type for the instance method, just the name of the method itself. This is because the transformer will assume that you sent in an appropriate method context instance when it was created and will just use whatever you provide during method resolution. Include the registrations and an instance for your methods (if needed) when you create the JoltJsonTransformer instance.

var transformer = JoltJsonTransformer.DefaultWith(transformerJson, new[] { staticRegistration, instanceRegistration }, new TransformerMethods());

That's it! You can pass in your source JSON document to the Transform method call just like before and collect the transformed result.