Skip to main content

🛡️ Test Structure & Application Quality

Quality in InventoryAlert is enforced through a comprehensive, multi-layered approach. We have consolidated our testing infrastructure into a single, high-performance integration project that covers everything from static architecture analysis to full system orchestration.

🏗️ Test Solution Structure​

The solution is divided into three core testing projects:

1. InventoryAlert.UnitTests (Isolated Logic)​

Goal: Verify pure business logic, calculations, and state changes with 100% mocked dependencies.

  • Scope: Handles edge cases, null inputs, and finance math accuracy (e.g., PercentDropFromCost).
  • Speed: Instant feedback (< 2 seconds for full suite).

2. InventoryAlert.IntegrationTests (Unified System Tests)​

Goal: Verify that the application communicates correctly with its persistence (Postgres/Redis) and messaging (SQS/SNS) layers.

  • Architecture: Organized into Three Logical Tiers (see Integration Testing Tiers):
    • Tier 1 (White-box): Direct logic testing using production DI.
    • Tier 2 (Black-box): API surface testing with log-correlation.
    • Tier 3 (Orchestration): End-to-end "Whole Flow" validation.
  • Isolation: Uses Respawn for DB reset and WireMock for external API simulation.

3. InventoryAlert.ArchitectureTests (Static Enforcement)​

Goal: Prevent architectural drift and tech debt using NetArchTest.

  • Enforcement: Ensures Domain never references Infrastructure and all controllers are secured with [Authorize].

🛡️ Comprehensive Quality Coverage Matrix​

How we guarantee quality across different operational domains:

Domain/ComponentHow Quality is Ensured
Business LogicTier 1 Integration tests resolve real services from the production DI graph to verify logic against a real DB.
Database IntegrityRespawn ensures every test starts with a clean DB. Tier 1/2 tests verify EF Core constraints and complex LINQ queries.
External APIsWireMock simulates Finnhub. We test success paths, 429 rate limits, and 500 server errors without hitting real endpoints.
System OrchestrationTier 3 tests verify the Whole Flow: API Call → SQS → Worker Job → DB Update → API Notification Feed.
ObservabilityDockerLogReader allows tests to "scrape" container logs, proving that internal request traces (CorrelationId) are correctly logged.

📈 Measuring Quality: Code Coverage​

We use Coverlet and ReportGenerator to visualize coverage.

  • Script: Run ./code-coverage.bat from the root.
  • Target: 80%+ coverage in the Domain and Application layers.

⚙️ Automated Quality Gates (CI/CD)​

Every Pull Request is blocked until the following passes:

  1. Build & Lint: Next.js and .NET 10 compilation.
  2. Architecture Check: Runs NetArchTest rules.
  3. Unit Tests: Executes isolated logic tests.
  4. Integration Suite: Spins up the full Docker stack and executes all 3 Tiers of integration tests to prove "Whole-Flow" stability.

🩺 Runtime Quality (Observability)​

  • Structured Logging (Seq): Trace single user actions across projects via CorrelationId.
  • Health Checks (/health): Active probes for Postgres, Redis, and SQS connectivity.
  • DLQ isolation: Poison messages are moved to inventory-event-dlq to prevent worker crash loops.