doc_ai_frontend/src/lib/ommlConverter.ts

/**
 * MathML to OMML Converter
 * 
 * Uses 'mathml2omml' library to convert MathML to Office Math Markup Language (OMML).
 * This is a pure JavaScript implementation and does not require external XSLT files.
 * 
 * @see https://github.com/fiduswriter/mathml2omml
 */

import { mml2omml } from 'mathml2omml';

/**
 * Cleans up OMML output from mathml2omml to fix known library issues.
 * These are workarounds for bugs in mathml2omml that produce invalid OMML.
 * 
 * @param omml Raw OMML string from mathml2omml
 * @returns Cleaned OMML string compatible with Microsoft Word
 */
function cleanOmml(omml: string): string {
  return omml
    // Fix: m:sty m:val="undefined" is invalid (library bug with mathvariant="normal")
    .replace(/<m:sty m:val="undefined"\/>/g, '')
    // Remove empty control properties that cause parsing issues in some Word versions
    .replace(/<m:ctrlPr\/>/g, '')
    .replace(/<w:rPr\/>/g, '')
    // Remove inline namespace declarations (will use wrapper's namespaces)
    .replace(/ xmlns:m="[^"]*"/g, '')
    .replace(/ xmlns:w="[^"]*"/g, '');
}

/**
 * Converts MathML string to OMML string using mathml2omml library.
 * 
 * @param mathml The MathML content string
 * @returns Promise resolving to cleaned OMML string
 * @see https://github.com/fiduswriter/mathml2omml
 */
export async function convertMathmlToOmml(mathml: string): Promise<string> {
  try {
    // Convert using mml2omml function
    const omml = mml2omml(mathml);
    
    // Apply fixes for known library issues
    return cleanOmml(omml);
  } catch (error) {
    console.error('MathML to OMML conversion failed:', error);
    return '';
  }
}

/**
 * Wraps OMML for clipboard/paste into Microsoft Word.
 * Uses Office Open XML namespaces (Office 2007+).
 * 
 * @param omml Cleaned OMML string from convertMathmlToOmml
 * @returns XML string ready for clipboard
 */
export function wrapOmmlForClipboard(omml: string): string {
  // Extract inner content from <m:oMath> tags to avoid nesting
  const innerContent = omml
    .replace(/<m:oMath[^>]*>/g, '')
    .replace(/<\/m:oMath>/g, '');

  // Return with proper Office Open XML structure
  return `<?xml version="1.0" encoding="UTF-8"?>
<m:oMath xmlns:m="http://schemas.openxmlformats.org/officeDocument/2006/math" xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
${innerContent}
</m:oMath>`;
}