parity(api,apps): security hotfix stage 2 - real customer sessions, close IDOR

Stage 1 narrowed GET /customers but did not close the leak: the response still
carried every customer id, and /customers/:id, /:id/orders and
/:id/ledger-entries were still public, so the full customer book was still two
requests away. The user caught this. Stage 2 deletes the id-bearing surface.

Customer JWT, separate from the admin one (T1):
  - new src/modules/customer-auth/ with its own passport strategy,
    CustomerJwtGuard, @CustomerAuth() and @CurrentCustomer()
  - signed with JWT_CUSTOMER_SECRET, and every token carries typ:'customer'
    which the strategy verifies; both checks are present deliberately
  - customer state re-read per request, so deactivating takes effect at once
  - mirrors the reference's own split: auth:api vs auth('api-customer')

Real OTP (D6), porting ApiMobileController::userAuth/userVerfiy (:39-64):
  - new CustomerOtp table, additive migration, nothing existing touched
  - parity behaviour: rand(1000,9999), active-customer filter, faTOen digit
    normalisation on both number and code, rejection on mismatch
  - approved improvements over the reference: bcrypt-hashed code, 2-minute
    expiry, 5 attempts then lockout, single use, send throttle (60s per
    number, 5/number and 20/IP per 10 min), and no number enumeration

Routes removed, replaced by token-derived equivalents:
  GET customers, customers/:id, /:id/orders, /:id/ledger-entries
    -> GET customer/me, customer/me/orders, customer/me/ledger-entries
No route accepts a customer id any more, so the IDOR is closed structurally
rather than by a check. The admin panel was unaffected (admin/customers-omde/*).

Both apps migrated in the same commit per the v1/v2 parity rule: token-based
session store, Bearer header, 401 -> clear session and route to login, real
two-step OTP login. Each app keeps its own markup and styling; only logic moved.

Tests: auth-smoke.sh extended with the stage-2 assertions, 46 passed 0 failed
against a live server. jest installed dev-only (user-approved) so pnpm test
works for the first time: 7 passed covering faToEn, which moved to its own
digits.ts to be testable without booting Nest. Typecheck clean on api,
user-app and user-app-v2; both apps build.

[ref: henza_server ApiMobileController.php:39-78]

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

apps/api/src/modules/customer-auth/customer-auth.service.ts

@@ -0,0 +1,248 @@
+import { HttpException, HttpStatus, Injectable, Logger, UnauthorizedException } from '@nestjs/common';
+import { JwtService } from '@nestjs/jwt';
+import { MessageAudience } from '@prisma/client';
+import * as bcrypt from 'bcryptjs';
+import { PrismaService } from '../../prisma/prisma.service';
+import { CUSTOMER_TOKEN_TYPE, type CustomerJwtPayload } from './customer-jwt.strategy';
+import { faToEn } from './digits';
+
+/** Nest 10 has no TooManyRequestsException; 429 with a Persian message. */
+class TooManyRequestsException extends HttpException {
+  constructor(message: string) {
+    super(message, HttpStatus.TOO_MANY_REQUESTS);
+  }
+}
+
+/** How long a code stays usable. The reference had no expiry at all (D6). */
+const OTP_TTL_MS = 2 * 60 * 1000;
+/** Wrong guesses allowed against one challenge before it is burned. */
+const MAX_ATTEMPTS = 5;
+/** Sends allowed per mobile number inside the window. */
+const MAX_SENDS_PER_MOBILE = 5;
+/** Sends allowed per source IP inside the window. */
+const MAX_SENDS_PER_IP = 20;
+const SEND_WINDOW_MS = 10 * 60 * 1000;
+/** Minimum gap between two sends to the same number. */
+const RESEND_COOLDOWN_MS = 60 * 1000;
+
+@Injectable()
+export class CustomerAuthService {
+  private readonly logger = new Logger(CustomerAuthService.name);
+
+  constructor(
+    private readonly prisma: PrismaService,
+    private readonly jwt: JwtService,
+  ) {}
+
+  /**
+   * Step 1 — port of ApiMobileController::userAuth (henza_server:39-49).
+   *
+   * Reference: validate `username`, find an ACTIVE customer by mobile, generate
+   * rand(1000,9999), store it, send it by SMS, return `true`.
+   *
+   * Kept: the active-customer filter, the 4-digit range, the opaque `true`
+   * response (it never reveals whether the number exists).
+   * Changed: the code is hashed, it expires, and sends are rate limited.
+   */
+  async requestOtp(rawMobile: string, ip: string | undefined) {
+    const mobile = faToEn(rawMobile);
+
+    await this.assertSendAllowed(mobile, ip);
+
+    // Legacy used Customer::active(), and the archived filter matches the
+    // customer-list lookup the app login previously relied on.
+    const customer = await this.prisma.customer.findFirst({
+      where: { mobile, isActive: true, isArchived: false },
+      select: { id: true },
+    });
+
+    // Deliberately opaque: the reference used firstOrFail() and leaked a 404
+    // for unknown numbers, which let anyone enumerate registered customers.
+    // Recorded as a deviation in docs/parity/DECISIONS.md.
+    if (!customer) {
+      this.logger.warn(`OTP requested for unknown or inactive mobile ${mobile}`);
+      return { sent: true };
+    }
+
+    const code = String(Math.floor(1000 + Math.random() * 9000));
+
+    await this.prisma.customerOtp.create({
+      data: {
+        mobile,
+        customerId: customer.id,
+        codeHash: await bcrypt.hash(code, 10),
+        expiresAt: new Date(Date.now() + OTP_TTL_MS),
+        requestIp: ip ?? null,
+      },
+    });
+
+    await this.deliver(mobile, code);
+    return { sent: true };
+  }
+
+  /**
+   * Step 2 — port of ApiMobileController::userVerfiy (henza_server:51-64).
+   *
+   * Reference: `if (faTOen($request->password) != $customer->send_password) abort(422);`
+   * then issue a token and return THAT ONE customer.
+   *
+   * The parity behaviour is the comparison and the rejection. The attempt
+   * counter, the expiry check and the single-use consumption are the D6
+   * additions.
+   */
+  async verifyOtp(rawMobile: string, rawCode: string, ip: string | undefined) {
+    const mobile = faToEn(rawMobile);
+    const code = faToEn(rawCode);
+
+    const challenge = await this.prisma.customerOtp.findFirst({
+      where: { mobile, consumedAt: null },
+      orderBy: { createdAt: 'desc' },
+      include: {
+        customer: {
+          select: { id: true, mobile: true, isActive: true, isArchived: true },
+        },
+      },
+    });
+
+    // One message for every failure mode below, so a caller cannot tell "no
+    // code was sent" from "wrong code" from "expired".
+    const reject = () => new UnauthorizedException('کد وارد شده صحیح نیست');
+
+    if (!challenge) throw reject();
+    if (challenge.expiresAt.getTime() < Date.now()) throw reject();
+
+    if (challenge.attempts >= MAX_ATTEMPTS) {
+      throw new TooManyRequestsException(
+        'تعداد تلاش‌های ناموفق بیش از حد مجاز است. لطفاً کد جدیدی درخواست کنید.',
+      );
+    }
+
+    const ok = await bcrypt.compare(code, challenge.codeHash);
+    if (!ok) {
+      await this.prisma.customerOtp.update({
+        where: { id: challenge.id },
+        data: { attempts: { increment: 1 } },
+      });
+      throw reject();
+    }
+
+    if (!challenge.customer.isActive || challenge.customer.isArchived) {
+      throw reject();
+    }
+
+    // Single use: burn the challenge before issuing the token.
+    await this.prisma.customerOtp.update({
+      where: { id: challenge.id },
+      data: { consumedAt: new Date() },
+    });
+
+    const payload: CustomerJwtPayload = {
+      sub: challenge.customer.id,
+      mobile: challenge.customer.mobile,
+      typ: CUSTOMER_TOKEN_TYPE,
+    };
+
+    // Reference reused api_token forever once set (:60-62); a fresh token is
+    // issued on every verify here (T1).
+    const token = await this.jwt.signAsync(payload, {
+      secret: customerJwtSecret(),
+      // jsonwebtoken types expiresIn as a template-literal union, which a plain
+      // string from process.env does not satisfy — same cast AuthModule uses.
+      expiresIn: (process.env.JWT_CUSTOMER_EXPIRES_IN ?? '30d') as `${number}d`,
+    });
+
+    return {
+      token,
+      customer: { id: challenge.customer.id, mobile: challenge.customer.mobile },
+    };
+  }
+
+  /**
+   * Per-mobile and per-IP send throttle.
+   *
+   * Counted off the CustomerOtp rows rather than an in-memory map, so the limit
+   * survives a restart and holds across replicas. The reference had no limit:
+   * userAuth could be called in a loop, each call sending a real (billed) SMS.
+   */
+  private async assertSendAllowed(mobile: string, ip: string | undefined) {
+    const since = new Date(Date.now() - SEND_WINDOW_MS);
+
+    const lastToMobile = await this.prisma.customerOtp.findFirst({
+      where: { mobile },
+      orderBy: { createdAt: 'desc' },
+      select: { createdAt: true },
+    });
+    if (lastToMobile && Date.now() - lastToMobile.createdAt.getTime() < RESEND_COOLDOWN_MS) {
+      const wait = Math.ceil(
+        (RESEND_COOLDOWN_MS - (Date.now() - lastToMobile.createdAt.getTime())) / 1000,
+      );
+      throw new TooManyRequestsException(`لطفاً ${wait} ثانیه دیگر تلاش کنید`);
+    }
+
+    const sentToMobile = await this.prisma.customerOtp.count({
+      where: { mobile, createdAt: { gte: since } },
+    });
+    if (sentToMobile >= MAX_SENDS_PER_MOBILE) {
+      throw new TooManyRequestsException(
+        'تعداد درخواست‌های کد بیش از حد مجاز است. لطفاً بعداً تلاش کنید.',
+      );
+    }
+
+    if (ip) {
+      const sentFromIp = await this.prisma.customerOtp.count({
+        where: { requestIp: ip, createdAt: { gte: since } },
+      });
+      if (sentFromIp >= MAX_SENDS_PER_IP) {
+        throw new TooManyRequestsException(
+          'تعداد درخواست‌های کد بیش از حد مجاز است. لطفاً بعداً تلاش کنید.',
+        );
+      }
+    }
+  }
+
+  /**
+   * Hands the code to the SMS gateway.
+   *
+   * No gateway is wired up in this project yet — SmsMessage is persistence only
+   * (logged as STUB-02 in docs/parity/ISSUES.md), exactly as the reference's
+   * Sms::Send would have been the integration point. Until one is connected the
+   * code is logged, and only outside production, so a deployment cannot print
+   * live codes into its logs.
+   *
+   * The reference's message text is kept verbatim for when the gateway lands
+   * (ApiMobileController.php:47).
+   */
+  private async deliver(mobile: string, code: string) {
+    const body =
+      `<#> code is ${code}\n` +
+      `رمز عبور یکبار مصرف شما جهت ورود به اپلیکشن هنزا گلد ${code} می باشد\n` +
+      `aMo6iFyUQaY`;
+
+    if (process.env.NODE_ENV !== 'production') {
+      this.logger.log(`[dev] OTP for ${mobile}: ${code}`);
+    }
+
+    // Recorded the same way the admin panel's own SMS sends are.
+    await this.prisma.smsMessage.create({
+      data: {
+        audience: MessageAudience.SINGLE_NUMBER,
+        recipients: mobile,
+        body,
+        sentCount: 1,
+      },
+    });
+  }
+}
+
+/**
+ * Customer tokens are signed with their own secret so an admin token can never
+ * validate as a customer token, or the reverse, even if the `typ` check were
+ * ever removed (T1).
+ */
+export function customerJwtSecret(): string {
+  return (
+    process.env.JWT_CUSTOMER_SECRET ??
+    process.env.JWT_SECRET ??
+    'dev-only-insecure-customer-secret'
+  );
+}