// IRANYekan *fanum* maps Latin digit codepoints (0-9) to Persian glyphs.
// Montserrat (`.en`) keeps real Latin digits — so any Latin-letter run that
// also carries model numbers (HP R5, 8GB, EF2126WM…) must render under `.en`.
// Pure digit runs (e.g. "13" inside «آیفون 13 پرو») stay on the FA font and
// correctly show as Persian digits.
const EN_RUN = /[A-Za-z0-9][A-Za-z0-9\s\-_|./+#%]*/g;

export type TextSegment = { type: "fa" | "en"; text: string };

export function splitMixedText(text: string): TextSegment[] {
  const segments: TextSegment[] = [];
  let cursor = 0;
  let match: RegExpExecArray | null;
  const re = new RegExp(EN_RUN.source, "g");

  while ((match = re.exec(text)) !== null) {
    let start = match.index;
    let end = start + match[0].length;
    while (end > start && /\s/.test(text[end - 1]!)) end--;
    const chunk = text.slice(start, end);
    if (!/[A-Za-z]/.test(chunk)) continue;

    if (start > cursor) segments.push({ type: "fa", text: text.slice(cursor, start) });
    segments.push({ type: "en", text: chunk });
    cursor = end;
  }

  if (cursor < text.length) segments.push({ type: "fa", text: text.slice(cursor) });
  return segments;
}

/** Wrap Latin runs in a plain string as `.en` spans (text is assumed already HTML-safe). */
export function wrapLatinRuns(text: string): string {
  return splitMixedText(text)
    .map((seg) =>
      seg.type === "en" ? `<span class="en" dir="ltr" lang="en">${seg.text}</span>` : seg.text,
    )
    .join("");
}

/** Walk HTML text nodes (content between tags) and wrap Latin runs. */
export function wrapLatinInHtml(html: string): string {
  return html.replace(/(^|>)([^<]+)/g, (_, prefix: string, text: string) => prefix + wrapLatinRuns(text));
}