EricksonLopez.Specification.Generators 1.0.0

EricksonLopez.Specification

High-performance, composable, NativeAOT-first Specification Pattern and SQL AST translation ecosystem for modern .NET.

---

EricksonLopez.Specification is an enterprise-grade, AOT-first, provider-agnostic Specification Pattern and SQL AST translation engine targeting .NET 8, .NET 9, and .NET 10. Engineered from the ground up for strict Domain-Driven Design (DDD) layering, it eliminates runtime code generation crashes under NativeAOT via a dedicated in-memory AST interpreter (ExpressionInterpreter), translates LINQ expressions directly into parameterized SQL across 6 database dialects without ORM dependencies for Dapper, and enforces immutability, thread-safety, and domain purity at compile time through 11 custom Roslyn analyzers.

---

Table of Contents

• What Problem It Solves
• Key Features
• Ecosystem
• Documentation
  • Step-by-Step Interactive Showcase (Levels 00 to 10)
  • Technical Reference & Architecture Guides
• Installation
• Quick Start
  • 1. Define a Pure Domain Specification
  • 2. In-Memory Evaluation & Composition (AOT-Safe)
  • 3. Build an Immutable Query Descriptor (Application Layer)
  • 4. Execute via Entity Framework Core (LINQ Adapter)
  • 5. Execute via Dapper (Direct Parameterized SQL Generation)
• Core Use Cases
  • Use Case 1: Clean Architecture / CQRS Query Handler with Repository Pattern
  • Use Case 2: Multi-Step Domain Rule Composition with Short-Circuiting
  • Use Case 3: Keyset / Cursor Pagination on High-Volume Datasets
  • Use Case 4: NativeAOT Microservices with Interpreted Validation
  • Use Case 5: Multi-Dialect SQL Generation for Dapper & Raw ADO.NET
  • Use Case 6: Functional Result Queries with EricksonLopez.Result
• Configuration & Integrations
  • Entity Framework Core & Dependency Injection
  • Dapper & Dialect Configuration
  • MongoDB Driver Integration
  • OpenTelemetry Metrics & Diagnostics
  • Compile-Time Roslyn Analyzers (SPEC001–SPEC011)
• Testing & Quality
  • In-Memory Unit Testing
  • SQL Translation Snapshot Verification
  • Mutation Testing & Quality Score
• Performance Benchmarks
  • Primary Operations & Composition
  • In-Memory Evaluation Benchmark (AOT vs JIT)
  • SQL AST Translation Benchmark
  • LINQ Provider Overhead (QuerySpec.Apply)
  • Span-Based Bulk Predicate Composition (AndAll)
• Compatibility & Technical Matrix
  • Runtime & Target Framework Matrix
  • SQL Dialects Feature Matrix
• Architecture & Design Principles
  • System Flow & Layer Boundaries
  • Expression Lifecycle State Machine
• Best Practices & Anti-Patterns
• Troubleshooting & Common Pitfalls
• Part of the EricksonLopez Ecosystem
• Contributing
• License

---

🎯 What Problem It Solves

The Pain Points of Traditional Specification Libraries

1. NativeAOT Runtime Failures & Dynamic Code Crashes: Traditional libraries rely heavily on Expression.Compile() or runtime IL generation (System.Reflection.Emit) for in-memory predicate evaluation. Under NativeAOT or trimmed deployments, these dynamic paths throw runtime exceptions or fail trimming verification.
2. ORM Coupling & Architectural Leakage in Domain Layers: Legacy implementations bundle ORM concepts (Include, ThenInclude, AsNoTracking, split queries) directly into domain specification base classes. This violates Clean Architecture principles, pollutes domain entities with infrastructure details, and makes domain logic unusable outside Entity Framework.
3. Lack of Provider-Agnostic SQL Generation for Micro-ORMs: Developers wanting high performance with Dapper or raw ADO.NET are forced to manually write string-based SQL queries, discarding domain specifications and introducing SQL injection vulnerabilities and maintainability nightmares.
4. Mutable State & Expression.Invoke Provider Failures: Combining expressions using Expression.Invoke breaks query translation in EF Core, Cosmos DB, and LINQ providers, requiring fragile third-party extensions like LinqKit that fail NativeAOT trimming.

How EricksonLopez.Specification Solves This

• Dual-Engine Architecture (100% NativeAOT Safe): Features a dedicated ExpressionInterpreter that evaluates expression ASTs in memory in just 44.6 nanoseconds without emitting dynamic IL, while preserving an opt-in structural JIT cache (ExpressionCompilationCache) for standard runtimes.
• Strict DDD Layer Separation: Domain specifications (Specification<T>) are strictly pure predicate expressions. Query concerns (sorting, keyset pagination, projection, tracking hints) are isolated in the Application layer via immutable value descriptors (QuerySpec<T>).
• Direct AST-to-SQL Translation: Translates expression trees into parameterized SQL AST models (QueryModel) and renders them across 6 production dialects (PostgreSQL, SQL Server, MySQL, MariaDB, SQLite, Oracle) for zero-overhead Dapper execution.
• Parameter Rebinding without Expression.Invoke: The internal ParameterReplacer and ExpressionComposer rewrite expression parameters in a single pass, guaranteeing 100% compatibility with LINQ providers and SQL translators.
• Compile-Time Architecture Governance: 11 custom Roslyn Analyzers (SPEC001–SPEC011) flag mutable specifications, missing pagination, un-translatable methods, and legacy Ardalis patterns during compilation.

---

⚡ Key Features

• 🚀 NativeAOT & Trimming First: Fully verified under .NET 8, 9, and 10 NativeAOT compilers with explicit BCL linker attributes and zero dynamic IL emission on hot paths.
• 🗄️ Multi-Dialect SQL Generation: Native parameterized AST rendering for PostgreSQL, SQL Server (MsSql), MySQL, MariaDB, SQLite, and Oracle Database without ORM dependencies.
• 🧱 Strict DDD Clean Architecture: Pure domain rules in Specification<T>, immutable query descriptors in QuerySpec<T>, and repository adapters in Infrastructure.
• ⚡ Zero Allocations on Hot Paths: Bounded LRU query plan caching (QueryPlanCache), structural expression hashing (ExpressionHasher), and ReadOnlySpan<T> bulk composition (AndAll / OrAny).
• 🔍 Keyset & Offset Pagination: First-class support for both high-throughput deterministic keyset seek (SeekAfter / SeekBefore) and classic offset pagination (Page / Skip / Take).
• 🛡️ Compile-Time Roslyn Governance: 11 analyzers (SPEC001–SPEC011) with automated CodeFix providers to prevent architectural drift and maintain strict domain purity.
• 📊 Enterprise Observability: Integrated OpenTelemetry ActivitySource and Meter instruments tracking specification evaluations, compositions, and SQL translations.

---

📦 Ecosystem

The EricksonLopez.Specification ecosystem is organized into modular, fine-grained packages following strict Clean Architecture boundaries:

Package	Version	Description
EricksonLopez.Specification.Abstractions		Core contracts: ISpecification<T>, QuerySpec<T>, IReadRepository<T>, OrderClause<T>
EricksonLopez.Specification		Core engine: Specification<T>, Spec factory, ExpressionComposer, ExpressionInterpreter (AOT), ExpressionSimplifier
EricksonLopez.Specification.Linq		LINQ provider adapter for IQueryable<T>.Apply(querySpec), Any(spec), and Count(spec)
EricksonLopez.Specification.Sql		AST-based Expression-to-SQL translator (QuerySpecTranslator<T>), QueryModel, ISqlDialect, bounded QueryPlanCache
EricksonLopez.Specification.PostgreSql		PostgreSQL dialect (PostgreSqlDialect: double quotes, $n parameters, LIMIT/OFFSET, collection predicates)
EricksonLopez.Specification.MsSql		Microsoft SQL Server dialect (MsSqlDialect: bracket quoting, @pn parameters, TOP N, OFFSET...FETCH)
EricksonLopez.Specification.MySql		MySQL dialect (MySqlDialect: backtick quoting, @pn parameters, LIMIT/OFFSET, expanded IN)
EricksonLopez.Specification.MariaDb		MariaDB dialect (MariaDbDialect: backtick quoting, @pn parameters, LIMIT/OFFSET, expanded IN)
EricksonLopez.Specification.Sqlite		SQLite dialect (SqliteDialect: double quote quoting, @pn parameters, LIMIT/OFFSET)
EricksonLopez.Specification.Oracle		Oracle Database dialect (OracleDialect: double quotes, :pn positional parameters, OFFSET...FETCH)
EricksonLopez.Specification.Dapper		High-performance Dapper extension methods over IDbConnection (QueryAsync, CountAsync, AnyAsync)
EricksonLopez.Specification.EntityFrameworkCore		Entity Framework Core IReadRepository<T> implementation, evaluators, and DI extensions
EricksonLopez.Specification.MongoDB		MongoDB C# Driver integration (MongoFilterCompiler, MongoSortCompiler, MongoSpecificationExtensions)
EricksonLopez.Specification.DapperExtensions		Execution adapter over internal EricksonLopez.DapperExtensions Unit-of-Work & session tracking
EricksonLopez.Specification.Result		Functional Result<T> query extensions over IReadRepository<T> using EricksonLopez.Result
EricksonLopez.Specification.Analyzers		Roslyn diagnostic analyzers SPEC001–SPEC011 & automated CodeFix providers
EricksonLopez.Specification.Generators		Roslyn source generator for zero-reflection column resolvers ([SpecColumnResolver]) and ordering helpers

---

📚 Documentation

🌐 Official Documentation Hub: https://github.com/ericksonlopezf/dotnet-specification/tree/main/docs

🎓 Step-by-Step Interactive Showcase (Levels 00 to 10)

Explore progressive runnable showcase levels located in the test and sample harness:

Level	Topic	Description
Level 00	Architecture & Conceptual Foundations	Pure DDD specification principles, expression tree encapsulation, and boundary invariants
Level 01	Quick Start & Domain Primitives	Sealed specifications, Spec.For<T>, Spec.True<T>, and in-memory AOT evaluation
Level 02	Configuration & SQL Dialects	Configuring PostgreSQL, SQL Server, MySQL, SQLite, and Oracle AST translators
Level 03	Real-World Enterprise Use Cases	CQRS queries, compound domain rules, multi-condition validation, and business pipelines
Level 04	Advanced ORM & Database Integrations	EF Core IQueryable.Apply, Dapper parameterized execution, and keyset seek pagination
Level 05	Processing & AST Translation Pipeline	Expression AST visitor rewriting, parameter replacement, and boolean constant folding
Level 06	Error Handling & Invariant Enforcement	Null safety, un-translatable expression handling, and Roslyn diagnostic enforcement
Level 07	Scalability & Bounded LRU Caching	High-throughput query plan caching (QueryPlanCache) and structural expression hashing
Level 08	Customization & Custom Column Resolvers	Custom IColumnNameResolver strategies and source-generated [SpecColumnResolver]
Level 09	Ecosystem Extensions (Result & MongoDB)	Railway-oriented Result<T> query integration and MongoDB Filter/Sort compilation
Level 10	Enterprise Architecture & Domain Isolation	Strict Clean Architecture layer isolation, dependency rule enforcement, and microservice patterns

📖 Technical Reference & Architecture Guides

• System Architecture Overview — Comprehensive architectural blueprint, memory layouts, and clean layering boundaries.
• Architectural Decision Records (ADRs) — 28 official ADRs documenting design rationale, structural tradeoffs, and rejected proposals.
• Features & Technical Matrix — Detailed feature classification, tier boundaries, and verified competitor comparisons.
• NativeAOT & Trimming Guide — Linker attributes, AST interpreter node support, and zero-dynamic-code deployment rules.
• Verified Performance Benchmarks — BenchmarkDotNet suite results across expression composition, in-memory validation, and SQL translation.
• Enterprise Cookbook & Recipes — 27 production-ready copy-paste recipes for DDD, EF Core, Dapper, NativeAOT, and OpenTelemetry.
• Best Practices & Anti-Patterns — Architectural rules, coding guidelines, and analyzer diagnostic compliance.
• Competitive Audit & Matrix — In-depth technical comparison against Ardalis.Specification, LinqKit, and native EF Core.
• Migration from Ardalis.Specification — Step-by-step guide and Roslyn automated code fixes for legacy migrations.
• Final Quality Audit — Comprehensive technical audit, Stryker.NET mutation score verification, and quality gates.

---

📥 Installation

Install the required packages using the .NET CLI or Visual Studio Package Manager:

1. Core Packages (Domain & Application Layers)

# Core contracts and immutable query descriptors (zero dependencies)
dotnet add package EricksonLopez.Specification.Abstractions

# Core specification engine, composite combinators, and AOT interpreter
dotnet add package EricksonLopez.Specification

2. Database Dialects & Providers (Infrastructure Layer — Dapper / SQL)

# AST-to-SQL translation engine and Dapper execution extensions
dotnet add package EricksonLopez.Specification.Sql
dotnet add package EricksonLopez.Specification.Dapper

# Install your target database dialect:
dotnet add package EricksonLopez.Specification.PostgreSql  # PostgreSQL ($1, $2, LIMIT/OFFSET)
dotnet add package EricksonLopez.Specification.MsSql       # Microsoft SQL Server (TOP, OFFSET FETCH)
dotnet add package EricksonLopez.Specification.MySql       # MySQL (`col`, @p1, LIMIT/OFFSET)
dotnet add package EricksonLopez.Specification.MariaDb     # MariaDB (`col`, @p1, LIMIT/OFFSET)
dotnet add package EricksonLopez.Specification.Sqlite      # SQLite ("col", @p1, LIMIT/OFFSET)
dotnet add package EricksonLopez.Specification.Oracle      # Oracle Database (:p1, OFFSET FETCH)

3. ORM & Document Database Adapters (Infrastructure Layer)

# Entity Framework Core & generic LINQ provider adapter
dotnet add package EricksonLopez.Specification.Linq
dotnet add package EricksonLopez.Specification.EntityFrameworkCore

# MongoDB C# Driver FilterDefinition & SortDefinition compiler
dotnet add package EricksonLopez.Specification.MongoDB

4. Compile-Time Governance & Source Generators (Development Only)

# Roslyn analyzers (SPEC001–SPEC011) and automated CodeFix providers
dotnet add package EricksonLopez.Specification.Analyzers

# Incremental Source Generator for zero-reflection column resolvers
dotnet add package EricksonLopez.Specification.Generators

---

🚀 Quick Start

1. Define a Pure Domain Specification

Domain specifications represent pure business rules. They are immutable, stateless, sealed, and have zero dependencies on ORMs or database drivers.

using System.Linq.Expressions;
using EricksonLopez.Specification;

namespace MyProject.Domain.Specifications;

// Domain layer — pure, stateless, sealed (enforced by SPEC001 analyzer)
public sealed class ActivePremiumCustomerSpec : Specification<Customer>
{
    private readonly decimal _minimumPurchases;

    public ActivePremiumCustomerSpec(decimal minimumPurchases = 1_000m)
        => _minimumPurchases = minimumPurchases;

    protected override Expression<Func<Customer, bool>> BuildExpression()
        => c => c.IsActive && c.TotalPurchases >= _minimumPurchases && c.DeletedAt == null;
}

2. In-Memory Evaluation & Composition (AOT-Safe)

Evaluate candidates directly in memory without emitting dynamic IL code, or compose multiple business rules using logical combinators:

using EricksonLopez.Specification;

var activeSpec = new ActivePremiumCustomerSpec(minimumPurchases: 500m);
var candidate = new Customer { IsActive = true, TotalPurchases = 750m, DeletedAt = null };

// 100% NativeAOT-safe in-memory evaluation (evaluated via ExpressionInterpreter in ~44.6 ns)
bool isEligible = activeSpec.IsSatisfiedBy(candidate);

// Fluent predicate composition without Expression.Invoke (inlinable by SQL translators)
var verifiedEmailSpec = Spec.For<Customer>(c => c.IsEmailVerified);
var highRiskSpec = Spec.For<Customer>(c => c.RiskScore > 80);

var qualifiedPromoSpec = activeSpec
    .And(verifiedEmailSpec)
    .And(highRiskSpec.Not());

// Bulk composition using static combinators
Specification<Customer> allRules = Spec.All(
    new ActivePremiumCustomerSpec(),
    Spec.For<Customer>(c => c.CountryCode == "US"),
    Spec.For<Customer>(c => c.CreditLimit > 0m)
);

3. Build an Immutable Query Descriptor (Application Layer)

Use QuerySpec<T> in Application layer handlers to encapsulate filtering, sorting, pagination, and projection as an immutable value record:

using EricksonLopez.Specification;

// Application layer — immutable query descriptor
var querySpec = QuerySpec<Customer>.Empty
    .Where(new ActivePremiumCustomerSpec(1_000m)) // Accepts IExpressionSpecification<T>
    .Where(c => c.CountryCode == "US")           // Or inline lambda criteria
    .OrderByDescending(c => c.TotalPurchases)
    .ThenBy(c => c.Name)
    .Page(page: 1, pageSize: 25)
    .TagWith("Handler:GetTopCustomersQuery");

4. Execute via Entity Framework Core (LINQ Adapter)

Apply query specifications directly to any EF Core IQueryable<T> without leaking ORM concerns into your domain layer:

using Microsoft.EntityFrameworkCore;
using EricksonLopez.Specification.Linq;

// Seamlessly applies WHERE predicates, ORDER BY, SKIP, TAKE, and query tags
List<Customer> topCustomers = await dbContext.Customers
    .Apply(querySpec)
    .ToListAsync(cancellationToken);

5. Execute via Dapper (Direct Parameterized SQL Generation)

Translate domain specifications into high-performance, injection-safe parameterized SQL without requiring an ORM:

using EricksonLopez.Specification.Sql;
using EricksonLopez.Specification.PostgreSql;
using EricksonLopez.Specification.Dapper;

// 1. Initialize translator with table name and column naming strategy
var translator = new QuerySpecTranslator<Customer>("customers", new SnakeCaseColumnNameResolver());

// 2. Select target dialect (PostgreSQL, SQL Server, MySQL, SQLite, Oracle)
var dialect = PostgreSqlDialect.Default;

// 3. Execute directly over any standard IDbConnection via Dapper extension
IEnumerable<Customer> results = await dbConnection.QueryAsync(
    querySpec, 
    translator, 
    dialect, 
    cancellationToken: cancellationToken
);

// Generated SQL:
// SELECT * FROM "customers" 
// WHERE "is_active" = $1 AND "total_purchases" >= $2 AND "deleted_at" IS NULL AND "country_code" = $3 
// ORDER BY "total_purchases" DESC, "name" ASC 
// LIMIT $4 OFFSET $5

---

💡 Core Use Cases

Use Case 1: Clean Architecture / CQRS Query Handler with Repository Pattern

Encapsulate data retrieval in CQRS query handlers using IReadRepository<T> without coupling application logic to specific database technologies:

using EricksonLopez.Specification;

public sealed record GetActiveVipCustomersQuery(decimal MinPurchases, int Page, int PageSize) 
    : IRequest<IReadOnlyList<CustomerDto>>;

public sealed class GetActiveVipCustomersQueryHandler 
    : IRequestHandler<GetActiveVipCustomersQuery, IReadOnlyList<CustomerDto>>
{
    private readonly IReadRepository<Customer> _repository;

    public GetActiveVipCustomersQueryHandler(IReadRepository<Customer> repository)
        => _repository = repository;

    public async Task<IReadOnlyList<CustomerDto>> Handle(
        GetActiveVipCustomersQuery request, 
        CancellationToken cancellationToken)
    {
        var spec = QuerySpec<Customer, CustomerDto>.Empty
            .Where(new ActivePremiumCustomerSpec(request.MinPurchases))
            .OrderByDescending(c => c.TotalPurchases)
            .Page(request.Page, request.PageSize)
            .Select(c => new CustomerDto(c.Id, c.Name, c.TotalPurchases));

        return await _repository.ListAsync(spec, cancellationToken);
    }
}

Use Case 2: Multi-Step Domain Rule Composition with Short-Circuiting

Dynamically assemble complex business rules across domain services with automatic boolean constant folding:

public sealed class LoanEligibilityService
{
    public bool EvaluateEligibility(Customer customer, LoanApplication application)
    {
        Specification<Customer> rules = new MinimumAgeSpec(minAge: 21)
            .And(new VerifiedIdentitySpec())
            .And(new DebtToIncomeRatioSpec(maxRatio: 0.43m));

        if (application.RequiresCollateral)
        {
            rules = rules.And(new VerifiedCollateralSpec(application.RequestedAmount));
        }

        // Evaluates in-memory with automatic short-circuit semantics
        return rules.IsSatisfiedBy(customer);
    }
}

Use Case 3: Keyset / Cursor Pagination on High-Volume Datasets

Prevent slow SQL OFFSET table scans on massive datasets using deterministic keyset seek pagination:

// Forward pagination (Next Page after specific cursor ID)
var nextBatchQuery = QuerySpec<Order>.Empty
    .Where(o => o.Status == OrderStatus.Completed)
    .OrderByDescending(o => o.Id)
    .SeekAfter(o => o.Id, cursorValue: 1_084_250L, take: 50);

// Generates: WHERE status = @p1 AND id < @cursor ORDER BY id DESC LIMIT 50
var orders = await _readRepository.ListAsync(nextBatchQuery, cancellationToken);

Use Case 4: NativeAOT Microservices with Interpreted Validation

Deploy lightweight containerized microservices and serverless functions under NativeAOT with zero reflection warnings:

[DapperAot]
public partial class MicroserviceDapperContext : IDapperContext;

// 100% NativeAOT: In-memory evaluation uses interpreted AST walk
var fraudCheck = new NonFlaggedAccountSpec().And(new IpReputationSpec("US"));
if (!fraudCheck.IsSatisfiedBy(account))
{
    return Results.Forbid();
}

// Database query rendered via compile-time source generated column resolvers
var results = await connection.QueryAsync(querySpec, aotTranslator, PostgreSqlDialect.Default);

Use Case 5: Multi-Dialect SQL Generation for Dapper & Raw ADO.NET

Execute identical business specifications across heterogeneous database engines:

var spec = new ActivePremiumCustomerSpec().ToQuerySpec().Page(1, 10);
var translator = new QuerySpecTranslator<Customer>("Customers");

// PostgreSQL: SELECT * FROM "Customers" WHERE ... LIMIT 10 OFFSET 0
var pgQuery = PostgreSqlDialect.Default.Render(translator.Translate(spec));

// SQL Server: SELECT * FROM [Customers] WHERE ... OFFSET 0 ROWS FETCH NEXT 10 ROWS ONLY
var msSqlQuery = MsSqlDialect.Default.Render(translator.Translate(spec));

// MySQL: SELECT * FROM `Customers` WHERE ... LIMIT 10 OFFSET 0
var mySqlQuery = MySqlDialect.Default.Render(translator.Translate(spec));

// Oracle: SELECT * FROM "Customers" WHERE ... OFFSET 0 ROWS FETCH NEXT 10 ROWS ONLY
var oracleQuery = OracleDialect.Default.Render(translator.Translate(spec));

Use Case 6: Functional Result Queries with EricksonLopez.Result

Integrate Railway-Oriented Programming for resilient, exception-free repository queries:

using EricksonLopez.Result;
using EricksonLopez.Specification.Result;

// Returns Result<Customer> instead of throwing exceptions on missing entities
Result<Customer> customerResult = await repository.FirstOrDefaultResultAsync(
    QuerySpec<Customer>.Empty.Where(c => c.Email == requestEmail),
    cancellationToken
);

return customerResult.Match(
    onSuccess: customer => Results.Ok(customer),
    onFailure: error => Results.NotFound(error.Message)
);

---

🔌 Configuration & Integrations

Entity Framework Core & Dependency Injection

Register the EF Core specification evaluator and generic read repositories in your IServiceCollection:

using Microsoft.Extensions.DependencyInjection;
using EricksonLopez.Specification.EntityFrameworkCore;

public static void ConfigureServices(IServiceCollection services, IConfiguration configuration)
{
    services.AddDbContext<AppDbContext>(options =>
        options.UseNpgsql(configuration.GetConnectionString("PostgresDb")));

    // Registers IReadRepository<T> and ISpecificationEvaluator backed by EF Core
    services.AddSpecificationEntityFramework<AppDbContext>();

    // Or register explicit typed read repository:
    services.AddEfReadRepository<AppDbContext, Customer>();
}

Dapper & Dialect Configuration

Optimize SQL generation with compile-time source-generated column resolvers:

using EricksonLopez.Specification.Generators;
using EricksonLopez.Specification.PostgreSql;
using EricksonLopez.Specification.Sql;

// 1. Declare source-generated column resolver (eliminates runtime reflection)
[SpecColumnResolver(typeof(Customer), Convention = NamingConvention.SnakeCase)]
public sealed partial class CustomerColumnResolver;

// 2. Register translator in DI as singleton
services.AddSingleton(new QuerySpecTranslator<Customer>("customers", new CustomerColumnResolver()));
services.AddSingleton<ISqlDialect>(PostgreSqlDialect.Default);

MongoDB Driver Integration

Compile specification expressions directly into native MongoDB FilterDefinition<T> and SortDefinition<T>:

using EricksonLopez.Specification.MongoDB;
using MongoDB.Driver;

var querySpec = QuerySpec<CustomerDoc>.Empty
    .Where(c => c.IsActive && c.Score >= 100)
    .OrderByDescending(c => c.CreatedAt)
    .Page(1, 20);

// Compile to MongoDB native filter and sort definitions
FilterDefinition<CustomerDoc> filter = MongoFilterCompiler.Compile(querySpec);
SortDefinition<CustomerDoc> sort = MongoSortCompiler.Compile(querySpec);

var results = await mongoCollection
    .Find(filter)
    .Sort(sort)
    .Skip(querySpec.Pagination?.Skip)
    .Limit(querySpec.Pagination?.Take)
    .ToListAsync(cancellationToken);

OpenTelemetry Metrics & Diagnostics

Monitor specification evaluations, composition overhead, and SQL translation counts in production:

using OpenTelemetry.Metrics;

var meterProvider = Sdk.CreateMeterProviderBuilder()
    .AddMeter("EricksonLopez.Specification")
    .AddPrometheusExporter()
    .Build();

// Automatically recorded metrics:
// - specification.evaluations_total (Counter)
// - specification.compositions_total (Counter)
// - specification.sql_translations_total (Counter)

Compile-Time Roslyn Analyzers (SPEC001–SPEC011)

The library includes 11 automated analyzers to enforce architectural purity and prevent misuse during compilation:

Diagnostic ID	Severity	Category	Description	CodeFix Available
SPEC001	Warning	Architecture	Specification classes must be declared sealed or abstract.	✅ Yes
SPEC002	Warning	Immutability	Specifications must not contain mutable state, fields, or properties.	❌ No
SPEC003	Error	Correctness	Expression.Invoke is prohibited in specification expression trees.	❌ No
SPEC004	Info	Performance	Unbounded query detected; recommend adding .Take() or .Page().	❌ No
SPEC005	Info	Correctness	Ordering clause applied without pagination limits.	❌ No
SPEC006	Info	Layering	Domain specification declared outside Domain layer boundary.	❌ No
SPEC007	Warning	SQL Translation	Non-translatable method invocation inside BuildExpression.	❌ No
SPEC008	Error	Purity	Prohibits infrastructure dependencies (DbContext, IServiceProvider) in constructors.	❌ No
SPEC009	Error	Correctness	Disallows async lambdas inside BuildExpression.	❌ No
SPEC010	Error	Correctness	Disallows calling IsSatisfiedBy inside BuildExpression.	❌ No
SPEC011	Warning	Migration	Flags inheritance from legacy Ardalis.Specification base class.	✅ Yes

---

🧪 Testing & Quality

In-Memory Unit Testing

Test pure domain specifications in unit test projects without databases or mocks:

using Xunit;
using EricksonLopez.Specification;

public sealed class ActivePremiumCustomerSpecTests
{
    [Fact]
    public void IsSatisfiedBy_WhenCustomerIsActiveAndMeetsPurchases_ReturnsTrue()
    {
        // Arrange
        var spec = new ActivePremiumCustomerSpec(minimumPurchases: 1_000m);
        var customer = new Customer { IsActive = true, TotalPurchases = 1_500m, DeletedAt = null };

        // Act
        bool result = spec.IsSatisfiedBy(customer);

        // Assert
        Assert.True(result);
    }

    [Fact]
    public void IsSatisfiedBy_WhenCustomerIsDeleted_ReturnsFalse()
    {
        // Arrange
        var spec = new ActivePremiumCustomerSpec(minimumPurchases: 1_000m);
        var customer = new Customer { IsActive = true, TotalPurchases = 2_000m, DeletedAt = DateTime.UtcNow };

        // Act
        bool result = spec.IsSatisfiedBy(customer);

        // Assert
        Assert.False(result);
    }
}

SQL Translation Snapshot Verification

Verify AST translation output deterministically:

[Fact]
public void Translate_WithFilterAndPaging_GeneratesExpectedPostgreSql()
{
    var spec = QuerySpec<Customer>.Empty
        .Where(c => c.IsActive)
        .OrderByDescending(c => c.CreatedAt)
        .Page(page: 2, pageSize: 10);

    var translator = new QuerySpecTranslator<Customer>("customers", new SnakeCaseColumnNameResolver());
    var model = translator.Translate(spec);
    var sqlQuery = PostgreSqlDialect.Default.Render(model);

    Assert.Equal(
        "SELECT * FROM \"customers\" WHERE \"is_active\" = $1 ORDER BY \"created_at\" DESC LIMIT $2 OFFSET $3",
        sqlQuery.Sql
    );
    Assert.Equal(true, sqlQuery.Parameters["$1"]);
    Assert.Equal(10, sqlQuery.Parameters["$2"]);
    Assert.Equal(10, sqlQuery.Parameters["$3"]);
}

Mutation Testing & Quality Score

Every release undergoes exhaustive mutation testing via Stryker.NET to ensure test assertions catch all logic regressions:

• Mutation Score: ≥ 99%
• Test Invariants: 100% coverage of boolean folding, parameter rewriting, and dialect rendering nodes.
• Linker Verification: Zero trimming warnings under .NET SDK 10.0 NativeAOT compiler.

---

⚡ Performance Benchmarks

Environment: .NET 10.0.10 (10.0.1026.32716), X64 RyuJIT AVX-512, Windows 11 Pro, BenchmarkDotNet v0.14.0

Primary Operations & Composition

Compairing combining predicates (c => c.IsActive and c => !c.IsDeleted) via ExpressionComposer.And versus manual dynamic lambda construction:

Method	Mean	Ratio	Gen0	Allocated	Alloc Ratio
Manual: x => left && right	184.71 ns	1.00	0.0110	560 B	1.00
EricksonLopez: ExpressionComposer.And	93.51 ns	0.51	0.0088	448 B	0.80
EricksonLopez: 5-way AND Composition	343.08 ns	1.86	0.0343	1,736 B	3.10

Key Takeaway: ExpressionComposer.And outperforms manual lambda creation by 49% in execution time and 20% in memory allocation via single-pass parameter rebinding.

---

In-Memory Evaluation Benchmark (AOT vs JIT)

Evaluates a composite specification (ActiveCustomerSpec.And(NotDeletedSpec)) against a candidate instance:

Method	Mean	Median	Gen0	Allocated	Execution Mode
Manual Direct Delegate (JIT Inlined)	0.0017 ns	0.0004 ns	-	-	Direct compiled C# delegate
EricksonLopez: IsSatisfiedBy (Interpreted)	44.64 ns	44.63 ns	0.0019	96 B	100% NativeAOT Safe (Zero Dynamic IL)
EricksonLopez: IsSatisfiedBy (Compiled Cache)	64.60 ns	64.49 ns	0.0010	48 B	JIT Cached Structural Delegate

Key Takeaway: In-memory interpreted evaluation executes in just 44.6 nanoseconds, enabling sub-microsecond validation on NativeAOT without dynamic code generation.

---

SQL AST Translation Benchmark

Measures translating a QuerySpec<T> into a parameterized SQL string and parameter dictionary for PostgreSQL:

Method	Mean	Gen0	Allocated	Description
EricksonLopez: Simple Spec → SQL	108.63 ns	0.0225	1.11 KB	Single filter (WHERE is_active = $1)
EricksonLopez: Complex Spec → SQL	412.62 ns	0.0634	3.13 KB	3 filters + ORDER BY + LIMIT/OFFSET

---

LINQ Provider Overhead (QuerySpec.Apply)

Measures applying a QuerySpec<T> with filtering, sorting, and paging against an IQueryable<T> data source of 1,000 entities:

Method	Mean	Ratio	Gen0	Allocated	Alloc Ratio
Manual LINQ Query	688.7 μs	1.00	0.9766	67.96 KB	1.00
EricksonLopez: QuerySpec.Apply	702.2 μs	1.02	0.9766	67.12 KB	0.99

---

Span-Based Bulk Predicate Composition (AndAll)

Bulk composition of 5 predicates using ReadOnlySpan<T> versus chained .And() invocations:

Method	Mean	Ratio	Gen0	Allocated	Alloc Ratio
Chained .And() × 4	520.0 ns	1.00	0.0391	1.94 KB	1.00
AndAll(ReadOnlySpan) × 5	477.6 ns	0.92	0.0391	1.94 KB	1.00

---

🌐 Compatibility & Technical Matrix

Runtime & Target Framework Matrix

Package	.NET 8.0 LTS	.NET 9.0 STS	.NET 10.0	NativeAOT	Trimmable	Linker Notes
EricksonLopez.Specification.Abstractions	✅ Full	✅ Full	✅ Full	✅ Full	✅ Full	BCL-only pure contracts
EricksonLopez.Specification	✅ Full	✅ Full	✅ Full	✅ Full	✅ Full	Core AOT safe; JIT cache marked [RequiresDynamicCode]
EricksonLopez.Specification.Linq	✅ Full	✅ Full	✅ Full	✅ Full	✅ Full	Passes expressions to LINQ provider
EricksonLopez.Specification.Sql	✅ Full	✅ Full	✅ Full	⚠️ Annotated	✅ Full	Reflection closures annotated [RequiresUnreferencedCode]
EricksonLopez.Specification.PostgreSql	✅ Full	✅ Full	✅ Full	✅ Full	✅ Full	Pure AST string renderer
EricksonLopez.Specification.MsSql	✅ Full	✅ Full	✅ Full	✅ Full	✅ Full	Pure AST string renderer
EricksonLopez.Specification.MySql	✅ Full	✅ Full	✅ Full	✅ Full	✅ Full	Pure AST string renderer
EricksonLopez.Specification.MariaDb	✅ Full	✅ Full	✅ Full	✅ Full	✅ Full	Pure AST string renderer
EricksonLopez.Specification.Sqlite	✅ Full	✅ Full	✅ Full	✅ Full	✅ Full	Pure AST string renderer
EricksonLopez.Specification.Oracle	✅ Full	✅ Full	✅ Full	✅ Full	✅ Full	Pure AST string renderer
EricksonLopez.Specification.Dapper	✅ Full	✅ Full	✅ Full	✅ Full	✅ Full	Fully compatible with Dapper.AOT
EricksonLopez.Specification.EntityFrameworkCore	✅ Full	✅ Full	✅ Full	✅ Full	✅ Full	Compatible with EF Core compiled models
EricksonLopez.Specification.MongoDB	✅ Full	✅ Full	✅ Full	✅ Full	✅ Full	Compatible with MongoDB Driver v3+
EricksonLopez.Specification.Analyzers	N/A	N/A	N/A	N/A	N/A	Roslyn compile-time only

SQL Dialects Feature Matrix

Dialect	Identifier Quote	Parameter Format	Paging Clause	String Matching	Collection Predicates
PostgreSQL	"column"	$1, $2, ...	LIMIT n OFFSET m	LIKE, ILIKE	= ANY(@p)
SQL Server	[column]	@p1, @p2, ...	OFFSET m ROWS FETCH NEXT n ROWS ONLY	LIKE	IN (@p1, @p2)
MySQL	`column`	@p1, @p2, ...	LIMIT n OFFSET m	LIKE	IN (@p1, @p2)
MariaDB	`column`	@p1, @p2, ...	LIMIT n OFFSET m	LIKE	IN (@p1, @p2)
SQLite	"column"	@p1, @p2, ...	LIMIT n OFFSET m	LIKE	IN (@p1, @p2)
Oracle	"COLUMN"	:p1, :p2, ...	OFFSET m ROWS FETCH NEXT n ROWS ONLY	LIKE	IN (:p1, :p2)

---

🏛️ Architecture & Design Principles

System Flow & Layer Boundaries

The following diagram illustrates how domain business rules travel from the Domain layer into application query descriptors and execute through LINQ or native SQL infrastructure:

flowchart TD
    %% Architecture Layers
    subgraph Domain["Domain Layer"]
        Spec["Specification<T>"]
        SpecDesc["Encapsulates Pure Predicates as Expressions"]
        Spec --- SpecDesc
    end

    subgraph Application["Application Layer"]
        QuerySpec["QuerySpec<T> / QuerySpec<T, TResult>"]
        QueryDesc["Adds Sorting, Keyset/Offset Paging, & Projection"]
        QuerySpec --- QueryDesc
    end

    subgraph CoreEngine["Specification Engine"]
        Composer["ExpressionComposer (Parameter Replacer)"]
        Simplifier["ExpressionSimplifier (Constant Folding)"]
        Hasher["ExpressionHasher (Structural Equality)"]
    end

    subgraph Infrastructure["Infrastructure Layer"]
        direction LR
        subgraph LinqProvider["LINQ Provider (EF Core)"]
            ApplyExt["QuerySpecLinqExtensions.Apply"]
            IQueryable["IQueryable<T>"]
        end
        subgraph SqlProvider["SQL Provider (Dapper)"]
            Translator["QuerySpecTranslator<T>"]
            QueryModel["QueryModel AST"]
            Dialect["ISqlDialect (PostgreSQL, SQL Server, MySQL, SQLite, Oracle)"]
            SqlQuery["SqlQuery (Parameterized SQL + Parameters)"]
        end
    end

    %% Relationships
    Spec -->|.And() / .Or() / .Not()| Composer
    Composer --> Simplifier
    Simplifier --> QuerySpec
    Spec -->|Direct Where| QuerySpec
    
    QuerySpec -->|LINQ Path| ApplyExt
    ApplyExt --> IQueryable
    IQueryable -->|EF Core Execution| Database[(Database)]

    QuerySpec -->|Native SQL Path| Translator
    Translator -->|Translates to AST| QueryModel
    QueryModel -->|Renders Dialect| Dialect
    Dialect --> SqlQuery
    SqlQuery -->|Executes via Dapper| Database

---

Expression Lifecycle State Machine

The state machine below depicts the transformation of specification expressions through simplification, caching, and multi-path provider execution:

stateDiagram-v2
    [*] --> ExpressionTree: Instantiate Specification<T>
    
    state ExpressionTree {
        [*] --> Unsimplified
        Unsimplified --> Simplified: ExpressionSimplifier.Simplify()
        Simplified --> [*]
    }
    
    ExpressionTree --> InterpretedAOT: spec.IsSatisfiedBy(candidate)
    InterpretedAOT --> Evaluated: ExpressionInterpreter AST Walk (Zero IL)
    
    ExpressionTree --> JITCompiled: spec.ToCompiledPredicate()
    JITCompiled --> Evaluated: ExpressionCompilationCache (JIT Only)
    
    ExpressionTree --> IQueryable: QuerySpecLinqExtensions.Apply()
    IQueryable --> EFCoreSQL: EF Core Query Translation
    
    ExpressionTree --> SqlAST: QuerySpecTranslator.Translate()
    SqlAST --> ParameterizedSQL: ISqlDialect.Render()
    
    EFCoreSQL --> Executed: Provider ExecuteAsync()
    ParameterizedSQL --> Executed: Dapper connection.QueryAsync()
    
    Evaluated --> [*]
    Executed --> [*]

---

🛡️ Best Practices & Anti-Patterns

Scenario	❌ Avoid	✅ Recommended
Class Declaration	Declaring unsealed specification classes	Sealing all concrete specifications (SPEC001)
Specification State	Adding mutable fields/properties to specifications	Using readonly constructor parameters captured in closures (SPEC002)
Expression Composition	Using Expression.Invoke to combine lambdas	Using .And(), .Or(), and ExpressionComposer (SPEC003)
Query Limits	Executing unbounded queries without pagination	Always applying .Page() or .Take() on database queries (SPEC004)
Ordering & Paging	Paginating without a deterministic ordering	Specifying .OrderBy() before applying pagination clauses (SPEC005)
Clean Architecture	Referencing DbContext or IDbConnection in Domain specs	Keeping specifications pure and injecting IReadRepository<T> in Application (SPEC008)
Async Operations	Using async / await inside BuildExpression	Keeping expression trees pure and executing async in repositories (SPEC009)
In-Memory Validation in AOT	Calling spec.ToCompiledPredicate() under NativeAOT	Using spec.IsSatisfiedBy() (evaluates via ExpressionInterpreter)
High-Volume Pagination	Using large OFFSET values on big tables	Using keyset pagination via .SeekAfter() / .SeekBefore()
SQL Column Mapping	Relying on runtime reflection in AOT paths	Using source-generated column resolvers ([SpecColumnResolver])

---

⚠️ Troubleshooting & Common Pitfalls

---

🌐 Part of the EricksonLopez Ecosystem

EricksonLopez.Specification is part of the standardized, high-performance .NET enterprise ecosystem:

• 🧱 EricksonLopez.SharedKernel — Foundational domain primitives, value objects, entity identifiers, and domain event dispatching.
• ⚡ EricksonLopez.Result — High-performance, zero-allocation struct-based Result Pattern & Railway-Oriented Programming.
• 🔍 EricksonLopez.Specification — Composable, AOT-first Specification Pattern & multi-dialect SQL AST translation.
• 📬 EricksonLopez.Mediator — Zero-allocation, source-generated mediator and CQRS pipeline engine.
• 🔄 EricksonLopez.Concurrency — Resilient asynchronous concurrency primitives, rate limiting, and execution gates.
• 🏢 EricksonLopez.MultiTenancy — High-isolation multitenancy infrastructure with PostgreSQL Row-Level Security (RLS).

---

🤝 Contributing

Contributions, bug reports, and pull requests are welcome!

Local Development Setup

1. Prerequisites:

   • .NET SDK 10.0 (or .NET SDK 8.0/9.0)
   • Docker (optional, for running integration test database containers)

2. Clone & Build:

   git clone https://github.com/ericksonlopezf/dotnet-specification.git
   cd dotnet-specification
   dotnet restore EricksonLopez.Specifications.slnx
   dotnet build EricksonLopez.Specifications.slnx -c Release
   

3. Run Unit Tests:

   dotnet test EricksonLopez.Specifications.slnx -c Release --no-build
   

4. Run Mutation Testing (Stryker.NET):

   dotnet tool restore
   dotnet stryker --config-file-path stryker-config.json
   

Please read our Contributing Guide, Code of Conduct, and Security Policy before submitting pull requests.

---

📄 License

Distributed under the MIT License. Copyright © 2026 Erickson Lopez.