Skip to main content

​
Overview

The Legacy JSON generator produces the legacy version of the API documentation in JSON format. This generator maintains backward compatibility with existing tools while newer JSON schemas are being developed.

​
Metadata

​
name
string
default:"legacy-json"
Generator identifier
​
version
string
default:"1.0.0"
Generator version
​
dependsOn
string
default:"metadata"
Requires metadata generator output as input
​
hasParallelProcessor
boolean
default:"true"
Supports parallel processing for improved performance

​
Configuration

​
ref
string
default:"main"
Git reference (branch/tag) for source links
​
minify
boolean
default:"false"
Enable JSON minification (removes whitespace)
​
output
string
Output directory path for generated JSON files

​
Output Format

The generator produces:
  • JSON files: One per API module (e.g., http.json, fs.json)
  • Format: Legacy schema structure for backward compatibility
  • Indentation: 2 spaces (unless minified)

​
Usage

doc-kit -t legacy-json
With minification: 
doc-kit -t legacy-json --config.legacy-json.minify=true

​
Dependency Chain

ast → metadata → legacy-json
The legacy JSON generator depends on:
  1. ast - Parses Markdown to AST
  2. metadata - Extracts API metadata
  3. legacy-json - Converts to legacy JSON format

​
Implementation Details

The generator:
  • Groups nodes by module using groupNodesByModule
  • Filters for depth-1 heading nodes (top-level modules)
  • Processes chunks in parallel using worker threads
  • Builds JSON sections using buildSection utility
  • Converts to legacy JSON format via legacyToJSON
  • Streams results as they complete

​
Parallel Processing

The processChunk function runs in worker threads:
  • Takes pre-grouped module entries to avoid redundant processing
  • Builds sections for assigned modules
  • Returns JSON-ready objects

​
Performance Optimization

The generator optimizes performance by:
  • Pre-grouping entries to avoid sending all ~4900+ entries to each worker
  • Streaming chunks as they complete for immediate file writing
  • Using worker threads for CPU-intensive JSON building
Source: src/generators/legacy-json/

Build docs developers (and LLMs) love

Get started for freeTalk to us