Function telegramHtml

packages/bot-telegram/src/telegram-html.ts:41–124

(input: string)

}

export function telegramHtml(input: string): string {
  if (!input) return input;

  // ── 1. Pull code regions out so we don't touch them. ──
  const codeRegions: string[] = [];

  // Fenced code blocks first (```…```)
  let body = input.replace(/```([\s\S]*?)```/g, (_match, inner: string) => {
    const escaped = escapeHtml(inner.replace(/^\n/, "").replace(/\n$/, ""));
    codeRegions.push(`<pre>${escaped}</pre>`);
    return codePlaceholder(codeRegions.length - 1);
  });

  // Inline code `…`
  body = body.replace(/`([^`\n]*)`/g, (_match, inner: string) => {
    const escaped = escapeHtml(inner);
    codeRegions.push(`<code>${escaped}</code>`);
    return codePlaceholder(codeRegions.length - 1);
  });

  // ── 1b. Pull markdown links out before bold/italic transforms mangle URLs. ──
  // The link text and URL will be HTML-escaped in step 2 below via the
  // global escapeHtml pass on the placeholder-free body; the final <a> tag
  // is assembled after escaping so the URL keeps its escaped form (single-escape,
  // consistent with the existing `&`-in-URL behaviour).
  const linkRegions: Array<{ text: string; url: string }> = [];

  body = body.replace(
    /\[([^\]\n]+)\]\(([^)\s]+)\)/g,
    (_m, text: string, url: string) => {
      linkRegions.push({ text, url });
      return linkPlaceholder(linkRegions.length - 1);
    },
  );

  // ── 2. Escape HTML-special chars in the remaining (non-code) text. ──
  body = escapeHtml(body);

  // ── 3. Bold first, into a sentinel; then italic won't eat its output. ──
  const BOLD_OPEN = "\x01B\x01";
  const BOLD_CLOSE = "\x02B\x02";

  body = body.replace(/\*\*([^\n*]+?)\*\*/g, `${BOLD_OPEN}$1${BOLD_CLOSE}`);
  body = body.replace(/__([^\n_]+?)__/g, `${BOLD_OPEN}$1${BOLD_CLOSE}`);

  // Headings (#…) → bold
  body = body.replace(
    /^\s{0,3}#{1,6}\s+(.*)$/gm,
    (_m, text: string) => `${BOLD_OPEN}${text.trim()}${BOLD_CLOSE}`,
  );

  // Strikethrough ~~text~~ → <s>text</s>
  body = body.replace(/~~([^\n~]+?)~~/g, "<s>$1</s>");

  // Italic *text* or _text_ → <i>text</i> (skip bold sentinels)
  body = body.replace(/(^|[^*\w])\*(\S(?:[^*\n]*\S)?)\*(?!\w)/g, "$1<i>$2</i>");
  body = body.replace(/(^|[^_\w])_(\S(?:[^_\n]*\S)?)_(?!\w)/g, "$1<i>$2</i>");