// Test decoding
const fs = require('fs');

// Read actual file content
const content = fs.readFileSync('data/processed/pages.json', 'utf8');
const idx = content.indexOf('bg_image=');
const snippet = content.substring(idx, idx + 30);

console.log('File snippet:', snippet);
console.log('Bytes:', Buffer.from(snippet).toString('hex'));

// The file has: bg_image=”10432″
// Which is: bg_image= + ” + 10432 + ″

// Test with the actual entity string
const test = 'bg_image=”10432″';
console.log('\nTest string:', test);

// Decode decimal entities
const decoded = test
  .replace(/”/g, '"')  // ” -> "
  .replace(/″/g, '"'); // ″ -> "

console.log('After replace:', decoded);

// The issue is that the file has literal ” not the entity
// Let me check what the actual characters are
console.log('\nCharacter analysis:');
for (let i = 0; i < test.length; i++) {
  const char = test[i];
  const code = test.charCodeAt(i);
  if (code > 127 || char === '&' || char === '#' || char === ';') {
    console.log(i, char, code, '0x' + code.toString(16));
  }
}