llmxml/scripts/build-grammar.js

Convert between markdown and LLM-friendly pseudo-XML

import { readFileSync, writeFileSync, mkdirSync } from 'fs';
import { dirname, join } from 'path';
import { fileURLToPath } from 'url';
import peggy from 'peggy';

const currentDir = dirname(fileURLToPath(import.meta.url));
const rootDir = join(currentDir, '..');
const grammarPath = join(rootDir, 'src', 'grammar', 'llmxml.pegjs');
const outputDir = join(rootDir, 'dist');

// Ensure output directories exist
mkdirSync(join(outputDir, 'cjs', 'grammar'), { recursive: true });
mkdirSync(join(outputDir, 'esm', 'grammar'), { recursive: true });

// Read grammar file
const grammar = readFileSync(grammarPath, 'utf-8');

// Generate parser with createNode function
const parser = peggy.generate(grammar, {
  format: 'bare',
  output: 'source',
  grammarSource: grammarPath,
  allowedStartRules: ['Document'],
  features: {
    // Disable eval usage where possible
    eval: false,
    // Use native functions instead
    native: true
  }
});

// Write ESM version
const esmCode = '// Generated from ' + grammarPath + '\n' +
  '// Note: This is a generated parser that uses eval() for performance.\n' +
  '// The use of eval is safe here as it only processes the grammar rules\n' +
  '// and does not execute any user-provided code.\n' +
  '/* eslint-disable no-eval */\n\n' +
  'import { createNode } from \'../../src/utils/ast.js\';\n\n' +
  'const options = { createNode };\n' +
  'export default ' + parser + ';\n';

writeFileSync(
  join(outputDir, 'esm', 'grammar', 'llmxml.js'),
  esmCode
);

// Write CJS version
const cjsCode = '// Generated from ' + grammarPath + '\n' +
  'const { createNode } = require(\'../../src/utils/ast\');\n\n' +
  'const options = { createNode };\n' +
  'module.exports = ' + parser + ';\n';

writeFileSync(
  join(outputDir, 'cjs', 'grammar', 'llmxml.js'),
  cjsCode
);