0pt/src/document-workbench/source-values.ts

/**
 * Deterministic, locale-aware value parsing used for evidence checks.
 *
 * The model sees text; code decides whether a source span really supports a
 * value. These helpers never guess when a format is ambiguous: a span that can
 * be read two ways yields no match, and the field goes to review.
 */

function isCalendarDate(year: number, month: number, day: number): boolean {
  if (year < 1 || month < 1 || month > 12 || day < 1) return false;
  const leap = year % 4 === 0 && (year % 100 !== 0 || year % 400 === 0);
  const days = [31, leap ? 29 : 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31];
  return day <= days[month - 1]!;
}

const pad = (n: number): string => String(n).padStart(2, '0');

/** Returns every unambiguous ISO rendering of the span (empty when ambiguous). */
export function dateCandidates(text: string): string[] {
  const t = text.trim();
  const iso = /^(\d{4})-(\d{1,2})-(\d{1,2})$/.exec(t);
  if (iso) {
    const [y, m, d] = [Number(iso[1]), Number(iso[2]), Number(iso[3])];
    return isCalendarDate(y, m, d) ? [`${y}-${pad(m)}-${pad(d)}`] : [];
  }
  const compact = /^(\d{4})(\d{2})(\d{2})$/.exec(t);
  if (compact) {
    const [y, m, d] = [Number(compact[1]), Number(compact[2]), Number(compact[3])];
    return isCalendarDate(y, m, d) ? [`${y}-${pad(m)}-${pad(d)}`] : [];
  }
  const numeric = /^(\d{1,2})[./-](\d{1,2})[./-](\d{4})$/.exec(t);
  if (numeric) {
    const a = Number(numeric[1]);
    const b = Number(numeric[2]);
    const y = Number(numeric[3]);
    // Both day-first and month-first are possible for <=12/<=12: no match.
    if (a <= 12 && b <= 12) return [];
    const [day, month] = a > 12 ? [a, b] : [b, a];
    return isCalendarDate(y, month, day) ? [`${y}-${pad(month)}-${pad(day)}`] : [];
  }
  return [];
}

/** Returns every unambiguous decimal rendering of the span. */
export function decimalCandidates(text: string): string[] {
  const t = text.trim().replace(/[\s\u00a0\u202f]/g, '');
  if (!/^[+-]?[\d.,]+$/.test(t) || !/\d/.test(t)) return [];
  const sign = t.startsWith('-') ? '-' : '';
  const body = t.replace(/^[+-]/, '');
  const hasComma = body.includes(',');
  const hasDot = body.includes('.');
  const out = new Set<string>();
  const add = (value: string) => {
    if (/^(0|[1-9]\d*)(\.\d+)?$/.test(value)) out.add(`${sign}${value}`);
  };
  if (hasComma && hasDot) {
    const lastComma = body.lastIndexOf(',');
    const lastDot = body.lastIndexOf('.');
    if (lastComma > lastDot) add(body.replace(/\./g, '').replace(',', '.'));
    else add(body.replace(/,/g, ''));
  } else if (hasComma) {
    const parts = body.split(',');
    if (parts.length === 2 && parts[1]!.length <= 2) add(`${parts[0]}.${parts[1]}`);
    add(body.replace(/,/g, ''));
  } else if (hasDot) {
    const parts = body.split('.');
    if (parts.length === 2 && parts[1]!.length <= 2) add(body);
    add(body.replace(/\./g, ''));
  } else {
    add(body);
  }
  return [...out];
}

/**
 * Canonical decimal string: no exponent, no leading '+', no trailing zeros in
 * the fraction. Used only for equality checks; the delivered value keeps its
 * original precision.
 */
export function canonicalDecimal(value: string): string | undefined {
  const match = /^([+-]?)(0|[1-9]\d*)(?:\.(\d+))?$/.exec(value.trim());
  if (!match) return undefined;
  const sign = match[1] === '-' ? '-' : '';
  const fraction = (match[3] ?? '').replace(/0+$/, '');
  return `${sign}${match[2]}${fraction ? `.${fraction}` : ''}`;
}