/** What a piece of JSON text is, for highlighting. */
export type JsonTokenKind = 'key' | 'string' | 'number' | 'literal' | 'plain';

/** A run of JSON text of one kind. */
export interface JsonToken {
  kind: JsonTokenKind;
  text: string;
}

// A string (a key when a colon follows), a number, a literal, or anything
// else (punctuation and spaces), which is plain.
const TOKEN =
  /("(?:\\.|[^"\\])*")(\s*:)?|(-?\d+(?:\.\d+)?(?:[eE][+-]?\d+)?)|\b(true|false|null)\b/g;

/**
 * The tokens of JSON text, in order, so it can be highlighted as text rather
 * than as HTML. Neighbouring tokens of the same kind are joined, so a
 * document gives as few as possible. Text that isn't valid JSON still comes
 * back whole, with what doesn't match as plain.
 */
export function tokeniseJson(text: string): JsonToken[] {
  const tokens: JsonToken[] = [];
  const push = (kind: JsonTokenKind, part: string): void => {
    if (part === '') return;
    const last = tokens.at(-1);
    if (last?.kind === kind) last.text += part;
    else tokens.push({ kind, text: part });
  };

  let end = 0;
  for (const match of text.matchAll(TOKEN)) {
    push('plain', text.slice(end, match.index));
    const [whole, string, colon, number, literal] = match;
    if (string !== undefined) {
      push(colon === undefined ? 'string' : 'key', string);
      push('plain', colon ?? '');
    } else if (number !== undefined) {
      push('number', number);
    } else if (literal !== undefined) {
      push('literal', literal);
    }
    end = match.index + whole.length;
  }
  push('plain', text.slice(end));
  return tokens;
}