import pdf from 'pdf-parse';

/**
 * Extracts plain text from a CV PDF file buffer.
 * @param {Buffer} fileBuffer - The PDF file content as a buffer.
 * @returns {Promise<string>} - The extracted plain text.
 */
export const parsePdfCv = async (fileBuffer) => {
  try {
    const data = await pdf(fileBuffer);
    
    // Clean up the text: remove double newlines and trim whitespace
    let text = data.text;
    text = text.replace(/\n\s*\n/g, '\n');
    text = text.trim();
    
    return text;
  } catch (error) {
    console.error('Error parsing PDF CV:', error);
    throw new Error('Failed to parse PDF: ' + error.message);
  }
};