tehran_tala/apps/api/scripts/legacy-import/lib/normalize.ts

/**
 * Value conversion from the henza MySQL schema to the Prisma one.
 *
 * Every helper here is pure, so the rules can be read and tested in isolation.
 * Each rule is the one recorded in docs/DATA-MIGRATION-MAP.md.
 */
import { Prisma } from '@prisma/client';

const PERSIAN_DIGITS = '۰۱۲۳۴۵۶۷۸۹';
const ARABIC_DIGITS = '٠١٢٣٤٥٦٧٨٩';

/** Persian/Arabic-Indic digits -> ASCII. Legacy did the same (Utility::faTOen). */
export function latinDigits(s: string): string {
  return s.replace(/[۰-۹٠-٩]/g, (d) => {
    const i = PERSIAN_DIGITS.indexOf(d);
    return String(i >= 0 ? i : ARABIC_DIGITS.indexOf(d));
  });
}

/**
 * Persian text as stored: trimmed, single spaces, and Arabic ي/ك folded to
 * Persian ی/ک. Empty -> null, so '' never reaches a nullable target column.
 */
export function text(v: unknown): string | null {
  if (v === null || v === undefined) return null;
  const s = String(v)
    .replace(/ي/g, 'ی')
    .replace(/ك/g, 'ک')
    .replace(/\s+/g, ' ')
    .trim();
  return s === '' ? null : s;
}

/**
 * Comparison key for matching names across systems. Also drops the zero-width
 * non-joiner, which legacy typed inconsistently ("مدیر اصلی" vs "مدیراصلی").
 * The ي/ك fold alone raised exact city matches from 32 to 72 (§10.4).
 */
export function matchKey(v: unknown): string | null {
  const t = text(v);
  return t === null ? null : t.replace(/‌/g, '').replace(/\s/g, '').toLowerCase();
}

/** Mobile as the login key: ASCII digits only, or null. */
export function mobile(v: unknown): string | null {
  if (v === null || v === undefined) return null;
  const d = latinDigits(String(v)).replace(/\D/g, '');
  return d === '' ? null : d;
}

function num(v: unknown): number {
  if (v === null || v === undefined || v === '') return 0;
  const n = Number(v);
  return Number.isFinite(n) ? n : 0;
}

/**
 * Legacy stores money and weight as `double` (§4). Going through toFixed at the
 * target scale drops the binary noise (12.3 is really 12.30000000000000071…)
 * exactly as Decimal(14,4) would, so a reconciliation can compare like with like.
 */
export const grams = (v: unknown) => new Prisma.Decimal(num(v).toFixed(4));
export const rial = (v: unknown) => new Prisma.Decimal(num(v).toFixed(2));
export const percent = (v: unknown) => new Prisma.Decimal(num(v).toFixed(3));
export const isZero = (v: unknown) => Math.abs(num(v)) < 0.00005;
export const toNumber = num;

/** Integer or null — for tinyint flags and ids that may arrive as strings. */
export function int(v: unknown): number | null {
  if (v === null || v === undefined || v === '') return null;
  const n = Number(v);
  return Number.isInteger(n) ? n : null;
}

/* ───────────── time ───────────── */

/**
 * The Laravel app ran with timezone 'Asia/Tehran' (henza_server config/app.php:70),
 * so every timestamp is Tehran wall-clock time. Read as a UTC instant it would
 * be 3:30 (4:30 under the pre-2022 summer time) off, moving evening documents
 * to the next day on a statement. The ICU zone rules know Iran's historical
 * DST, so the offset is taken from them per instant rather than hard-coded.
 */
const tehranFormat = new Intl.DateTimeFormat('en-US', {
  timeZone: 'Asia/Tehran',
  hourCycle: 'h23',
  year: 'numeric',
  month: '2-digit',
  day: '2-digit',
  hour: '2-digit',
  minute: '2-digit',
  second: '2-digit',
});

function tehranOffsetMs(utcMs: number): number {
  const p: Record<string, string> = {};
  for (const part of tehranFormat.formatToParts(new Date(utcMs))) p[part.type] = part.value;
  const asUtc = Date.UTC(+p.year, +p.month - 1, +p.day, +p.hour, +p.minute, +p.second);
  return asUtc - utcMs;
}

const WALL = /^(\d{4})-(\d{2})-(\d{2})(?:[ T](\d{2}):(\d{2}):(\d{2}))?/;

/** 'YYYY-MM-DD HH:MM:SS' in Tehran wall-clock time -> the UTC instant. */
export function tehran(v: unknown): Date | null {
  if (v === null || v === undefined || v === '') return null;
  const m = WALL.exec(String(v));
  if (!m || m[1] === '0000') return null;
  const wall = Date.UTC(+m[1], +m[2] - 1, +m[3], +(m[4] ?? 0), +(m[5] ?? 0), +(m[6] ?? 0));
  // Two passes: the offset at the guessed instant can differ across a DST edge.
  let utc = wall - tehranOffsetMs(wall);
  utc = wall - tehranOffsetMs(utc);
  return new Date(utc);
}

/**
 * Birthdays (owner ruling 2026-10-07, §10.7): of 915 set values only 541 are
 * real. Dropped: the sign-up default (birth date == sign-up date, 190 rows) and
 * the impossible ones (a time of day, or born after 2008; 184 rows). A
 * birthday is a calendar date, so it is kept as UTC midnight of that date.
 */
export function birthday(birth: unknown, signup: unknown): Date | null {
  const b = WALL.exec(String(birth ?? ''));
  if (!b) return null;
  if ((b[4] ?? '00') !== '00' || (b[5] ?? '00') !== '00' || (b[6] ?? '00') !== '00') return null;
  if (+b[1] > 2008 || +b[1] < 1900) return null;
  const s = WALL.exec(String(signup ?? ''));
  if (s && s[1] === b[1] && s[2] === b[2] && s[3] === b[3]) return null;
  return new Date(Date.UTC(+b[1], +b[2] - 1, +b[3]));
}

/* ───────────── Jalali dates ───────────── */

/**
 * `cashes.recive_date` is a varchar holding a Jalali date ("1403/05/12"; 88 of
 * 89 values start with 14). This is the standard jalaali-js arithmetic
 * (Borkowski), inlined so the tool needs no extra dependency.
 *
 * Its integer division TRUNCATES toward zero. Math.floor is not a substitute:
 * g2d's (gm - 8) / 6 is negative for March, and flooring it shifted every
 * converted date by a year.
 */
const div = (a: number, b: number) => Math.trunc(a / b);
const mod = (a: number, b: number) => a - Math.trunc(a / b) * b;

function jalCal(jy: number) {
  const breaks = [-61, 9, 38, 199, 426, 686, 756, 818, 1111, 1181, 1210, 1635, 2060, 2097, 2192, 2262, 2324, 2394, 2456, 3178];
  const gy = jy + 621;
  let leapJ = -14;
  let jp = breaks[0];
  let jump = 0;
  for (let i = 1; i < breaks.length; i++) {
    const jm = breaks[i];
    jump = jm - jp;
    if (jy < jm) break;
    leapJ += div(jump, 33) * 8 + div(mod(jump, 33), 4);
    jp = jm;
  }
  const n = jy - jp;
  leapJ += div(n, 33) * 8 + div(mod(n, 33) + 3, 4);
  if (mod(jump, 33) === 4 && jump - n === 4) leapJ += 1;
  const leapG = div(gy, 4) - div((div(gy, 100) + 1) * 3, 4) - 150;
  return { gy, march: 20 + leapJ - leapG };
}

function g2d(gy: number, gm: number, gd: number) {
  const d =
    div((gy + div(gm - 8, 6) + 100100) * 1461, 4) +
    div(153 * mod(gm + 9, 12) + 2, 5) +
    gd - 34840408;
  return d - div(div(gy + 100100 + div(gm - 8, 6), 100) * 3, 4) + 752;
}

function d2g(jdn: number) {
  let j = 4 * jdn + 139361631;
  j = j + div(div(4 * jdn + 183187720, 146097) * 3, 4) * 4 - 3908;
  const i = div(mod(j, 1461), 4) * 5 + 308;
  const gd = div(mod(i, 153), 5) + 1;
  const gm = mod(div(i, 153), 12) + 1;
  const gy = div(j, 1461) - 100100 + div(8 - gm, 6);
  return { gy, gm, gd };
}

/** Jalali y/m/d -> Gregorian y/m/d. */
export function jalaliToGregorian(jy: number, jm: number, jd: number) {
  const r = jalCal(jy);
  const jdn = g2d(r.gy, 3, r.march) + (jm - 1) * 31 - div(jm, 7) * (jm - 7) + jd - 1;
  return d2g(jdn);
}

/**
 * A free-text Jalali date ("1403/5/12", "۱۴۰۳-۰۵-۱۲") -> UTC midnight of the
 * same day, or null when it does not parse. Callers keep the raw text too.
 */
export function jalaliDate(v: unknown): Date | null {
  const s = latinDigits(String(v ?? '')).trim();
  const m = /^(1[34]\d{2})\D(\d{1,2})\D(\d{1,2})$/.exec(s);
  if (!m) return null;
  const jm = +m[2];
  const jd = +m[3];
  if (jm < 1 || jm > 12 || jd < 1 || jd > 31) return null;
  const g = jalaliToGregorian(+m[1], jm, jd);
  return new Date(Date.UTC(g.gy, g.gm - 1, g.gd));
}