nx-json-parser

nx-json-parser

A robust Markdown-to-JSON transformer SDK built on top of unified, remark, and remark-gfm.

Designed for structured data extraction from LLM outputs and other Markdown-formatted strings.

Features

• AST-Based Parsing: Uses remark for reliable, production-grade Markdown parsing (no regex hell).
• Table Support: Automatically converts Markdown tables into arrays of JavaScript objects using remark-gfm.
• List Support: Smartly handles lists, converting them to arrays.
• Bullet-Style Sections: Includes a custom plugin to detect and transform bulleted "sections" (e.g., - Title: Content) into structured JSON keys.
• CamelCase Normalization: Automatically transforms headings and table headers into camelCase keys.
• Extensible: Built on the unified ecosystem, allowing for additional plugins (math, frontmatter, etc.).

Installation

npm install nx-json-parser

Usage

Basic Usage

import { markdownToJson } from 'nx-json-parser';

const md = `
# Summary
The product is a high-performance database.

## Key Features
- Scalable
- Secure
- Fast
`;

const result = markdownToJson(md);
/*
{
  summary: "The product is a high-performance database.",
  keyFeatures: ["Scalable", "Secure", "Fast"]
}
*/

Table Support

const md = `
### Database Performance
| Metric | Value | Status |
|--------|-------|--------|
| Latency| 5ms   | OK     |
| Throughput | 10k ops/s | High |
`;

const result = markdownToJson(md);
/*
{
  databasePerformance: [
    { metric: "Latency", value: "5ms", status: "OK" },
    { metric: "Throughput", value: "10k ops/s", status: "High" }
  ]
}
*/

Bullet-Style Sections

LLMs often output sections using bullet points. nx-json-parser detects this pattern and treats them as structured keys:

const md = `
- Short Answer
The sky appears blue due to Rayleigh scattering.

- Technical Detail
Short-wavelength light is scattered more strongly by atmospheric gases.
`;

const result = markdownToJson(md);
/*
{
  shortAnswer: "The sky appears blue due to Rayleigh scattering.",
  technicalDetail: "Short-wavelength light is scattered more strongly by atmospheric gases."
}
*/

API

markdownToJson(markdown: string): any

Converts a Markdown string to a JSON object.

JSONTransformer

The main class for transformation. You can provide a custom RemarkParser if needed.

import { JSONTransformer, RemarkParser } from 'nx-json-parser';

const transformer = new JSONTransformer({
  parser: new RemarkParser()
});
const result = transformer.transform(md);

License

ISC