Vanilla Schema Forms - A JSON schema driven form generator

Vanilla Schema Forms is a web-based application that dynamically generates HTML forms from a given JSON Schema. It is built using Vanilla TypeScript (no heavy frameworks) and Vite for a fast development experience.

Features

• Dynamic Form Generation: Automatically creates HTML forms based on JSON Schema definitions.
• Hyperscript Rendering: Lightweight DOM generation using a hyperscript helper, avoiding innerHTML injection risks.
• Schema Dereferencing: Resolves $ref pointers within your JSON Schema using @apidevtools/json-schema-ref-parser.
• Type-Safe Development: Written in TypeScript for robust and maintainable code.
• Modern Tooling: Utilizes Vite for development and bundling, providing a fast and efficient workflow.
• Customizable UI: Configurable field sorting, visibility, and text overrides via configuration files.
• Unit Testing: Includes tests for core parsing logic using Vitest.

Status

Vanilla Schema Forms is in a very early state. I am not using it yet and it may be changed in the future.

Playground

Check out the playground here

How it Works

The application follows a clear pipeline to transform a JSON Schema into an interactive form:

1. Schema Loading: The index.html file (located in the project root) is the entry point. It loads the main application script (src/index.ts).
2. Schema Fetching & Parsing: The src/index.ts script fetches the schema.json file. It then uses src/parser.ts to parse this schema.
3. Schema Dereferencing and Transformation: src/parser.ts utilizes the @apidevtools/json-schema-ref-parser library to dereference any $ref pointers in the JSON Schema. It then transforms the raw JSON Schema into a simplified, UI-friendly FormNode tree structure.
4. Form Rendering: src/index.ts passes the FormNode tree to src/renderer.ts. This orchestrator delegates the actual DOM creation to src/dom-renderer.ts, which produces the final HTML structure.
5. Live JSON Output: As the user interacts with the generated form, the application dynamically updates a live JSON output, reflecting the current state of the form data.

Validation

The application uses AJV (Another JSON Schema Validator) for robust, standard-compliant validation.

Architecture

1. Initialization: When a schema is parsed in src/parser.ts, a bundled version (resolving external refs but keeping internal refs to avoid recursion issues) is passed to src/validator.ts.
2. Compilation: src/validator.ts initializes an Ajv2020 instance, registers standard formats (via ajv-formats) and custom formats (e.g., uint64), and compiles the schema into a validation function.
3. Mapping: During rendering (src/renderer.ts), the application builds a registry mapping AJV data paths (JSON pointers like /server/port) to HTML Element IDs.
4. Real-time Validation:
   • When the user modifies the form, the current data is extracted from the DOM.
   • The data is passed to the compiled AJV validator.
   • If errors occur, they are mapped back to specific DOM elements using the registry, and error messages are displayed inline.

Supported Formats

In addition to standard JSON Schema formats (email, date-time, etc.), the validator supports the following custom integer formats often found in systems like Protobuf or Go:

• uint, uint8, uint16, uint32, uint64

Project Structure

.
├── index.html            # Main HTML entry point
├── package.json          # Project dependencies and scripts
├── schema.json           # Example JSON Schema
├── tsconfig.json         # TypeScript configuration
├── vite.config.ts        # Vite configuration (including polyfills for Node.js modules)
└── src/
    ├── config.ts         # Configuration for sorting, visibility, and heuristics
    ├── dom-renderer.ts   # DOM generation logic using hyperscript
    ├── i18n.ts           # Text overrides and internationalization mappings
    ├── index.ts          # Main application logic, orchestrates parsing and rendering
    ├── form-data-reader.ts # Logic to read data back from the DOM
    ├── ui-schema-adapter.ts # Adapter for JSON Forms UI Schemas
    ├── parser.ts         # Parses JSON Schema, dereferences, and transforms to FormNode tree
    ├── parser.test.ts    # Unit tests for the parser
    └── renderer.ts       # Renders the FormNode tree into HTML form elements

Setup

1. Install Dependencies:

   npm install

2. Development Server: To start the development server and view the application in your browser:

   npm run dev

   The application will typically be available at http://localhost:5173/ (or a similar port if 5173 is in use).

Troubleshooting

• "Buffer is not defined" or "Module 'path' has been externalized" errors: This project uses Node.js polyfills via vite-plugin-node-polyfills to make @apidevtools/json-schema-ref-parser compatible with browser environments. Ensure this plugin is correctly configured in vite.config.ts and buffer and path are included in its include option.
• 404 errors for index.html: Ensure index.html is located in the root of your project directory.

Customization

Vanilla Schema Forms allows extensive customization through configuration and custom renderers.

Global Configuration

You can configure global behavior like sorting and visibility using setConfig.

import { setConfig } from "./src/index";

setConfig({
  sorting: {
    defaultPriority: ["name", "id", "type"], // These keys always appear at the top
    defaultRenderLast: ["metadata", "tags"]   // These keys always appear at the bottom
  },
  visibility: {
    hiddenPaths: ["root.internalId"], // Hide specific paths
    hiddenKeys: ["password"]          // Hide keys globally
  }
});

Custom Renderers

You can override the rendering of specific fields based on their key or path.

import { setCustomRenderers, domRenderer } from "./src/index";

setCustomRenderers({
  // Set Custom Renderer: Complete control over the DOM element
  "color": {
    render: (node, path, elementId, dataPath, context) => {
      const input = document.createElement("input");
      input.type = "color";
      input.id = elementId;
      input.value = node.defaultValue || "#000000";
      
      // Use the default wrapper for consistent labeling/layout
      return domRenderer.renderFieldWrapper(node, elementId, input);
    }
  }
});

Customization Guidelines & Best Practices

Customizing the rendering logic provides great flexibility but can lead to maintenance issues if not structured correctly.

1. Preserve Event Bubbling: The core architecture relies on change and input events bubbling up from form elements. If you build a custom widget (e.g., a div-based switch), ensure it dispatches a change event or updates a hidden <input> element. Without this, validation and data retrieval will fail.
2. Use domRenderer Helpers: Avoid re-implementing standard logic. Use domRenderer.renderFieldWrapper to wrap your custom inputs. This ensures consistent rendering of labels, descriptions, and error placeholders.
3. Avoid Monolithic Renderers: Don't create a single "Master Renderer" with huge if/else blocks. Register small, specific renderers for specific keys or types using setCustomRenderers.
4. Security: Avoid using innerHTML to prevent XSS vulnerabilities. Use document.createElement or the internal h() hyperscript helper.

Layout Grouping

You can group fields together (e.g., for horizontal layout) using the layout.groups configuration in src/config.ts.

import { setConfig } from "./config";

setConfig({
  layout: {
    groups: {
      // Key is the ID of the parent object (usually the title)
      "Connection": [
        {
          keys: ["host", "port"], // Fields to group
          className: "d-flex gap-3", // CSS classes for the container
          title: "Endpoint" // Optional title for the group
        }
      ]
    }
  }
});

JSON Forms Adapter

If you prefer using a JSON Forms compatible UI Schema, you can use the built-in adapter.

import { adaptUiSchema } from "./src/index";

const uiSchema = {
  type: "HorizontalLayout",
  elements: [
    { type: "Control", scope: "#/properties/host" },
    { type: "Control", scope: "#/properties/port" }
  ]
};

// Applies the UI Schema to the object with ID "Connection"
adaptUiSchema(uiSchema, "Connection");

Architectural Vision: Supporting UI Frameworks (React, etc.)

While the current implementation is a lightweight, dependency-free DOM renderer, a primary goal is to evolve the architecture to seamlessly support modern UI frameworks like React. The current approach of mounting framework components inside a vanilla renderer can be complex and inefficient for full-form customizations.

To make this easier and more robust, the plan is to refactor the library into a "headless" core engine with separate, optional renderer packages.

The Proposed Structure:

1. @vanilla-schema-forms/core: A framework-agnostic engine that handles:

   • Schema parsing, dereferencing, and transformation into the FormNode tree.
   • AJV-based validation.
   • State management (data and errors).
   • Configuration and UI schema adaptation.
   • It will have zero DOM rendering logic.

2. @vanilla-schema-forms/vanilla-renderer: The default, lightweight renderer (the current implementation), which consumes the core engine to render standard HTML elements.

3. @vanilla-schema-forms/react-renderer: A new package that provides React hooks and components to build forms.

   • A useForm() hook that connects to the core engine.
   • A <Field /> component that can dynamically render the correct input based on the FormNode.
   • A system for providing a custom set of components (e.g., a Material UI renderer set).

What this means for you:

With this architecture, building a fully custom form with React and Material UI would become straightforward:

// Future-state example
import { VanillaSchemaForm } from '@vanilla-schema-forms/react-renderer';
import { materialRenderers } from '@vanilla-schema-forms/material-renderer';

function MyForm({ schema }) {
  return (
    <VanillaSchemaForm
      schema={schema}
      renderers={materialRenderers}
      onChange={(data, errors) => console.log(data, errors)}
    />
  );
}

This approach separates the complex business logic of schema processing from the presentation layer, making the library far more flexible and easier to integrate into any project, while still offering a zero-dependency option for those who need it.

Testing

This project uses Vitest for unit testing.