/**
 * Pure function(s) for extracting canonical tournament identity
 * (name + year) from a `scoring_events` row.
 *
 * Used by the Phase 2 backfill to group per-window events into
 * canonical `tournaments` rows.
 */

export interface ScoringEventInput {
  name: string;
  eventDate: string | null;
  eventType: string;
}

export interface TournamentIdentity {
  /** Tournament name with any trailing 4-digit year stripped. */
  name: string;
  year: number;
}

const TRAILING_YEAR_RE = / (\d{4})$/;

/**
 * Extracts the canonical `(name, year)` identity for a tournament
 * from a scoring event.
 *
 * Resolution order for year:
 *   1. A trailing 4-digit year on the event name (e.g. "Wimbledon 2026").
 *      The year is stripped from the returned name.
 *   2. The first 4 characters of `eventDate` (format `YYYY-MM-DD`).
 *
 * Throws if neither source supplies a year.
 */
export function extractTournamentIdentity(
  ev: ScoringEventInput,
): TournamentIdentity {
  const trimmedName = ev.name.trim();
  const match = trimmedName.match(TRAILING_YEAR_RE);

  if (match) {
    const yearFromName = Number(match[1]);
    const nameWithoutYear = trimmedName.slice(0, match.index).trim();
    return { name: nameWithoutYear, year: yearFromName };
  }

  if (ev.eventDate) {
    const yearStr = ev.eventDate.slice(0, 4);
    const yearFromDate = Number(yearStr);
    if (Number.isFinite(yearFromDate) && yearStr.length === 4) {
      return { name: trimmedName, year: yearFromDate };
    }
  }

  throw new Error(
    `cannot determine year for scoring event "${ev.name}" (eventDate=${ev.eventDate ?? "null"})`,
  );
}