declepticon

declarative validation and transformation of inconsistently structured data

(the name is a truly terrible pun, because I had to come up with something)

Usage

$ npm install declepticon

let { transformation, optional, eager, validators } = require("declepticon");

let { nonBlankString, integer } = validators;
let transform = transformation({
    name: "Item",
    fields: {
        id: integer,
        type: value => value === "item" || value === "entity",
        name: nonBlankString
    },
    slots: {
        id: eager,
        type: true,
        designation: eager("name")
    },
    stringify: ({ id }) => `#${id}`
});

let data = require("/path/to/data.json");
let record = transform(data);

This will turn the incoming JSON data into a record <Item #123> with the properties described by slots, reporting discrepancies with regard to the expectations expressed within fields.

Note that transformed records can be turned back into JSON via JSON.stringify(record), provided slots' values can be serialized.

See tests for more elaborate examples.

API

Anything described here can be imported directly from the declepticon package (e.g. let { transformation, eager } = require("declepticon")).

• transformation(descriptor) is the primary entry point, returning a function transform(data, context) to validate and transform data objects (optionally supplying an arbitrary context object for slot transformations) as described by descriptor properties (see below).

• struct is used to describe nested structures within descriptor fields.

• optional is a function used to indicate non-essential fields within descriptors. It accepts one more validators.

• eager is used to indicate that a slot should be populated before validation, for identification within error reporting (i.e. slots used within stringify). eager can be used by iteself or as a function with the name of the respective original field.

• skipSlot, if returned from a slot transformation, indicates that the respective slot should be omitted from an individual record.

• validators is a set of common type validators provided for convenience; see validators.js for details.

• repr(value, jsonify) is a utility function to return a value's technical representation (i.e. wrapping it in backticks), optionally converting it to JSON.

Descriptors

Any descriptor object has the following properties:

• name is a string identifier used for error reporting:

  name: "Item"

• stringify is an optional function which is passed the respective record and returns a string for use in error reporting:

  stringify: ({ id, name }) => `${repr(id)} "${name}"`

  This will automatically be combined with name to result in <Item `123` "foo">.

• onError is an optional function which is invoked with the respective message if a validation error occurs.

  onError(message) {
      throw new Error(message);
  }

• fields is an object describing the expected shape of data, both with regard to its properties and their respective values:

  fields: {
      id: value => Number.isInteger(value),
      type: "item",
      label: optional(value => value === true || value === false),
      details: struct("ItemDetails", {
          url: value => true
      })
  }

  The right-hand side consists of a validator or an array thereof (OR conjunction). A validator is typically a function which is passed the original value and returns false if that value is considered invalid. As a shortcut, the expected value may be used as a validator instead (strict equality comparison).

  Non-essential fields are flagged by passing the respective validator(s) to the optional function (see above).

  Nested structures are described by passing a name and fields object to the struct function (see above).

• slots is an object describing the shape of the transformed record:

  slots: {
      id: eager,
      name: eager("name"),
      caption: "label",
      url: ({ details }, context) => details ? details.url + context : skipSlot
  }

  The right-hand side consists of a transformer. This is typically a function which is passed the original data object and the context object passed to the transform function (if any). It returns either the desired value (e.g. converting the original value into a number) or skipSlot to conditionally omit that slot.

  As a shortcut, true can be used to copy over the corresponding original value. If a string is used, the respective field's value is copied over instead.

Contributing

• ensure Node is installed
• npm install downloads dependencies
• npm test checks code for stylistic consistency

Alternatives

• joi
• clojure.spec