Function decodeHtmlEntities

sdk/src/tools/read-url.ts:74–100

(text: string)

}

function decodeHtmlEntities(text: string): string {
  const namedEntities: Record<string, string> = {
    amp: '&',
    apos: "'",
    copy: '(c)',
    hellip: '...',
    gt: '>',
    lt: '<',
    mdash: '-',
    middot: '*',
    nbsp: ' ',
    ndash: '-',
    quot: '"',
    rsquo: "'",
  }

  return text.replace(/&(#x?[0-9a-fA-F]+|[a-zA-Z]+);/g, (entity, body) => {
    if (body[0] === '#') {
      const isHex = body[1]?.toLowerCase() === 'x'
      const value = Number.parseInt(body.slice(isHex ? 2 : 1), isHex ? 16 : 10)
      return Number.isFinite(value) && value >= 0 && value <= 0x10ffff
        ? String.fromCodePoint(value)
        : entity
    }
    return namedEntities[body] ?? entity
  })
}

function normalizeText(text: string): string {
  return text