// Split plain text into alternating text/link segments so a subtask can be
// rendered with real <a> tags via Svelte templating (no `{@html}`, no XSS
// surface — see markdown.ts for the hardened-HTML approach notes take).

export type LinkifyPart = { text: string } | { link: string };

// A run of non-whitespace starting at http(s):// — the trailing-punctuation
// trim below decides how much of the run is really part of the URL.
const URL_RE = /https?:\/\/[^\s<>"']+/g;

// Strip trailing punctuation that's almost certainly sentence punctuation
// rather than part of the URL (e.g. "check https://x.com." or "see
// (https://x.com)."). A trailing ')' is only stripped when it's unbalanced
// within the match — "https://x.com/foo(bar)" keeps its ')'.
function trimTrailingPunct(url: string): string {
  let end = url.length;
  while (end > 0) {
    const ch = url[end - 1];
    if (ch === ")") {
      const slice = url.slice(0, end);
      const opens = (slice.match(/\(/g) ?? []).length;
      const closes = (slice.match(/\)/g) ?? []).length;
      if (closes > opens) {
        end--;
        continue;
      }
      break;
    }
    if (".,;:!?".includes(ch)) {
      end--;
      continue;
    }
    break;
  }
  return url.slice(0, end);
}

/** Break `text` into segments, pulling out http(s) URLs as `{ link }` parts. */
export function linkify(text: string): LinkifyPart[] {
  const parts: LinkifyPart[] = [];
  let lastIndex = 0;
  for (const match of text.matchAll(URL_RE)) {
    const raw = match[0];
    const start = match.index ?? 0;
    const url = trimTrailingPunct(raw);
    if (!url) continue;
    if (start > lastIndex) parts.push({ text: text.slice(lastIndex, start) });
    parts.push({ link: url });
    const consumedEnd = start + url.length;
    if (consumedEnd < start + raw.length) parts.push({ text: raw.slice(url.length) });
    lastIndex = start + raw.length;
  }
  if (lastIndex < text.length) parts.push({ text: text.slice(lastIndex) });
  return parts;
}