/** Read-only Unicode inspection utility for defensive research. */
export type CodePointRecord = {
  position: number;
  codePoint: string;
  printable: string;
  invisibleFormat: boolean;
};

const FORMAT = /\p{Cf}/u;

export function inspectUnicode(input: string): CodePointRecord[] {
  return Array.from(input).map((character, position) => ({
    position,
    codePoint: `U+${character.codePointAt(0)!.toString(16).toUpperCase().padStart(4, '0')}`,
    printable: FORMAT.test(character) ? '[FORMAT]' : character,
    invisibleFormat: FORMAT.test(character),
  }));
}

export function auditText(input: string) {
  const points = inspectUnicode(input);
  return {
    input,
    codePoints: points,
    formatCharacterCount: points.filter((point) => point.invisibleFormat).length,
    normalizedNFC: input.normalize('NFC'),
    normalizedNFKC: input.normalize('NFKC'),
    // Removing format characters is a diagnostic view, not a safe universal sanitizer.
    withoutFormatCharacters: Array.from(input).filter((c) => !FORMAT.test(c)).join(''),
  };
}