Skip to content

CHANGELOG.md

Latest commit

 

History

History
421 lines (326 loc) · 31.9 KB

CHANGELOG.md

File metadata and controls

421 lines (326 loc) · 31.9 KB

Changelog

All notable changes to this project will be documented in this file.

[Unreleased]

Added

  • enum() on FieldValidator now accepts PHP UnitEnum (non-backed enums): the value must be an instance of the enum or a string equal to one of the case names; BackedEnum behavior is unchanged (int or string backed values via tryFrom())

[0.15.0] - 2026-04-15

Fixed

  • BoolValidator::coerce() no longer string-casts non-scalar values (arrays triggered E_WARNING, objects without __toString threw Error); non-scalars are passed through to type validation, which raises ValidationException as for other invalid types
  • coerceAll() no longer calls coerce() on schema field validators in place (which leaked coercion to every other schema reusing the same FieldValidator instance); coerceAll() now replaces each field with a clone that has coercion enabled
  • coerceAll() is now recursive: nested AssociativeValidator, ObjectValidator, and ArrayValidator (with item validators) all propagate coercion to their children; previously only top-level schema fields were coerced
  • FieldValidator::__clone() skips Closure::bindTo() for static closures (for example from in()), avoiding PHP warnings when cloning such validators; pipeline steps now carry a bindable flag set at insertion time so cloning no longer needs ReflectionFunction
  • satisfies(), satisfiesAll(), satisfiesAny(), satisfiesNone(), and ArrayValidator::contains() now clone any FieldValidator operand at capture time and rebuild those operands when the outer validator is cloned, so later mutations or clone-local pipeline state do not leak across validator instances
  • FieldValidator::tryValidate() now resets transient type-tracking before and after every validation run, so validators that use transform()/pipe() remain repeatable across multiple calls and clones do not inherit stale runtime state
  • default() no longer deep-copies object values; object defaults are returned as-is (shared by handle). Use defaultUsing() when each validation run needs a fresh object instance
  • Base64Variant::UrlSafe now strictly validates URL-safe characters (-_) only; previously it was identical to Base64Variant::Any (accepted both standard and URL-safe)

Added

  • defaultUsing(callable $factory) on FieldValidator: invokes the factory whenever validation resolves to null, providing a fresh default per validation run; intended for mutable domain objects or other defaults that should not be shared across validations or clones
  • passthrough() on AssociativeValidator and ObjectValidator: copy undeclared keys or public properties from the input to the validated output without validating them; declared schema fields are still validated; values already written from the schema (including outputKey targets) are not overwritten by passthrough

[0.14.0] - 2026-03-23

Added

  • uniqueField(string $fieldName, ?string $message = null) on ArrayValidator for cross-item uniqueness validation of a nested field; convenience wrapper around satisfies() that produces field-level error paths by structuring errors as [index => [fieldName => [message]]]; all members of a duplicate group receive errors (e.g., both symlinks.0.destination and symlinks.2.destination); skips items where the field is null or missing; works with both associative arrays and objects; uses strict comparison via serialize() to distinguish types

Fixed

  • CRITICAL BUG: Fixed default() and required() interaction -- required() was rejecting null values before default() had a chance to provide a fallback
    • Issue: Chaining ->default($value)->required() always threw "Value is required" for null inputs because required checked for null before default was applied
    • Root Cause: tryValidate() had a scattered, multi-branch architecture with default and required checks duplicated across 5+ locations (early returns, pre-pipeline, mid-loop, post-loop), making the interaction order unpredictable
    • Fix: Simplified tryValidate() to a clean linear 5-step flow: coerce, type check, pipeline, default, required
    • Impact: default() and required() now work cooperatively -- default() is the last-resort fallback for null, required() only fails if the value is still null after default
    • Real-World Benefit: Enables patterns like ->pipe(fn($x) => null)->default($fallback)->required() where a pipeline step intentionally nullifies a value and default() catches it before required() enforces presence

Changed

  • Simplified tryValidate() pipeline architecture -- reduced from 67 lines with 5+ branch points to 33 lines with a linear flow
    • Removed 2 duplicated early returns for null + no-pipeline + not-required
    • Removed mid-loop default() and required() checks (were duplicating post-loop logic)
    • default() now applies in exactly one place: after the pipeline, as the last-resort fallback
    • required() now checks in exactly one place: at the very end, after default has had its chance
    • Pipeline always runs fully without mid-step bailouts
    • Validation flow is now: coerce -> type check -> pipeline -> default -> required

[0.13.0] - 2026-03-06

Added

  • enum(class-string<\BackedEnum> $enumClass, ?string $message = null) on FieldValidator for PHP BackedEnum validation; restricts the value to one of the enum's backed cases (int or string); available on all validator types; use enum(StatusEnum::class) instead of in(array_map(fn($e) => $e->value, StatusEnum::cases()))
  • const(mixed $value, ?string $message = null) on FieldValidator for single-value validation; restricts the value to exactly one allowed constant using strict comparison (===); available on all validator types; use const('active') instead of in(['active']) for clearer intent
  • outputKey(string $key) on FieldValidator for schema fields: output validated values under a different key than the input field (e.g. input service_id -> output service after transform); works with Validator::isAssociative() and Validator::isObject()
  • GitHub Actions CI: lint, platform-check, static analysis, tests on PHP 8.3

Changed

  • Applied Mago formatting and new linter rules across the codebase (readable-literal, no-isset); updated mago.toml (print-width 120, preserve-breaking-* options)
  • Renamed composer fix script to composer format for Mago formatting
  • Mago now a dev dependency (carthage-software/mago) instead of global installation
  • Added config.platform.php: "8.3.0" so dependency resolution targets PHP 8.3; composer update/install will not add packages requiring PHP above 8.3
  • Added composer platform-check script to verify platform requirements

[0.12.0] - 2026-01-24

Changed

  • BREAKING: transform() now skips null values by default (was executing on null)
    • Rationale: Most type conversions don't need null handling, making this safer and more efficient by default
    • Migration: Use transform($fn, skipNull: false) if you need to process null values (e.g., transform(fn($v) => $v ?? 'fallback', skipNull: false))
    • Impact: Existing code that relied on transform() executing on null will need to explicitly set skipNull: false
    • Real-World Benefit: Prevents errors when null values reach type conversions like DateTime::createFromFormat(), explode(), strlen(), etc.

Added

  • notEmpty() convenience method for StringValidator and ArrayValidator to reject empty strings/arrays
  • between() shorthand for string length and numeric range constraints with unified error messages
  • in() alias for oneOf() on primitive validators for clearer allowed-value validation
  • transform() method now accepts optional $skipNull parameter (default: true) to configure null handling

Deprecated

  • oneOf() in favor of in() (alias retained for backward compatibility)

Documentation

  • Clarified fail-fast behavior per field and schema-level error aggregation in guides and examples
  • Updated transformation documentation to reflect new null handling behavior

[0.11.2] - 2026-01-05

Fixed

  • CRITICAL BUG: Fixed array validator error handling to preserve array indices in flattened error messages
    • Issue: Array item validation errors lost array index information, showing paths like symlinks.destination instead of symlinks.0.destination
    • Root Cause: ArrayValidator used validate() which threw immediately on first error, losing index context and preventing error collection
    • Fix: Changed to use tryValidate() for all items, collecting errors with proper indexing (similar to AssociativeValidator behavior)
    • Impact: Flattened error messages now correctly identify which array item has validation errors (e.g., symlinks.0.destination, symlinks.2.destination)
    • Behavior: All array item errors are now collected (not just the first), and indices are preserved in error paths
    • Real-World Benefit: API error responses and form validation can now accurately pinpoint which array element failed validation

[0.11.1] - 2025-12-28

Fixed

  • CRITICAL BUG: Fixed null handling for transformation methods to match documented behavior
    • Issue: pipe() transformations (like trim()) were executing on null values, causing TypeError exceptions
    • Root Cause: All transformations were executing on null, but pipe() transformations are type-preserving and expect specific types
    • Fix: Implemented skipNull flag in pipeline structure - pipe() now skips null (type-preserving, expects specific type), transform() executes on null (can handle null and change types)
    • Impact: pipe() transformations correctly skip null values, preventing errors with functions like trim() that don't accept null
    • Behavior: transform() continues to execute on null as designed (can transform null to other values/types)
    • Clone Support: Updated __clone() method to preserve skipNull flag when cloning validators
    • Real-World Benefit: Form validation with optional fields using pipe(trim(...)) now works correctly without errors

Documentation

  • Clarified null handling differences between pipe() and transform() in transformation guidance.

[0.11.0] - 2025-12-11

Added

  • getFlattenedErrors() method on ValidationException returning a flattened list of errors suitable for API consumption. Each error entry contains path (field path with '_root' for root-level errors) and message (error message). Supports nested structures, array items, and maintains consistent path notation.
  • ValidationException::flattenErrors() static method for flattening errors from tryValidate() results without needing to catch exceptions. Returns empty array when passed null.

[0.10.0] - 2025-12-10

Added

  • Numeric comparison helpers (gt, gte, lt, lte), non-negative/non-positive validators, and clampToRange(min, max) transformation for numeric validators.
  • String format validators: hostname(), time(), base64(), hex(), domain(), and regex() alias method for pattern() on StringValidator.
  • port() method for IntValidator to validate port numbers (1-65535) as numeric constraints.
  • Enum-based variant flags: IpVersion enum for IP address validation, Base64Variant enum for Base64 validation, and UuidVariant enum for UUID validation.
  • UUID version 7 support in UuidVariant enum (Unix timestamp-based, sortable).
  • Array length constraints: minItems() and maxItems() methods for ArrayValidator to validate array size.
  • Array contains validation: contains() method for ArrayValidator supporting both specific value matching and validator-based item matching.
  • Comprehensive documentation clarifying library philosophy: focus on core validation principles over exhaustive validator coverage, with external libraries encouraged for advanced/specialized validators.

Changed

  • BREAKING: Refactored ip() method to use IpVersion enum flag instead of separate ipv4()/ipv6() methods. Use ip(IpVersion::IPv4) or ip(IpVersion::IPv6) for version-specific validation. Enum flag comes first, message parameter last.
  • BREAKING: Refactored base64() method to use Base64Variant enum flag. Use base64(Base64Variant::UrlSafe) for URL-safe Base64 validation or base64(Base64Variant::Any) to accept both variants. Enum flag comes first, message parameter last.
  • BREAKING: Refactored uuid() method to use UuidVariant enum flag. Use uuid(UuidVariant::V1) through uuid(UuidVariant::V7) for version-specific validation, or uuid(UuidVariant::Any) (default) to accept all versions (1-7). Enum flag comes first, message parameter last.
  • Enhanced documentation to clarify library philosophy: primary focus on core validation principles rather than implementing every possible validator. External libraries are strongly encouraged for advanced, specialized, or frequently-evolving validators.
  • Added declare(strict_types=1); across the codebase for stricter type enforcement and clearer errors.
  • Replaced php-cs-fixer with Mago for linting/formatting; scripts now map composer lint/composer fix to Mago.

[0.9.0] - 2025-11-18

Added

  • Validator cloning: Added clone() support to all validators for safe reuse of prebuilt pipelines without shared state. Pipelines are rebound to the cloned instance and nested schemas/item validators deep-clone to avoid cross-contamination between validators.

[0.8.0] - 2025-11-06

Changed

  • Migrated all runtime classes to the Lemmon\Validator namespace and aligned documentation/examples with the new structure.

[0.7.0] - 2025-10-19

Added

  • Type-Safe Pipeline Architecture: Enhanced internal architecture with PipelineType enum for improved developer experience
    • IDE Autocomplete Support: PipelineType::VALIDATION and PipelineType::TRANSFORMATION with full IDE integration
    • Refactoring Safety: No more magic strings - IDE handles all references automatically during refactoring
    • Self-Documenting Code: Clear intent with documented enum cases explaining validation vs transformation behavior
    • Future Extensibility: Easy to add new pipeline types (e.g., CONDITIONAL, ASYNC) with type safety
    • Zero Runtime Cost: Enums compile to same string values with no performance impact
    • Perfect Static Analysis: PHPStan level max compliance with proper type annotations
  • Smart Null Handling: Revolutionary null handling system that makes validation intuitive and order-independent
    • Intelligent Null Processing: Validations automatically skip null values unless field is marked as required()
    • Order Independence: ->email()->required() and ->required()->email() both work identically with null inputs
    • Transformation Consistency: Transformations (pipe(), transform(), nullifyEmpty()) always execute, even on null values
    • Global Required Flag: required() works as a global flag (like coerce()), not a pipeline step
    • Smart Default Application: Default values applied after pipeline execution if final result is null
    • Real-World Benefit: Eliminates confusing execution order dependencies - write validation chains naturally
    • Backward Compatible: All existing code works unchanged with the full suite
  • Unified Pipeline Architecture: Revolutionary single pipeline design that maintains conceptual clarity while optimizing execution
    • Hybrid Execution Model: Pure validations (like email(), minLength()) collect all errors for better UX, while transformations (like required(), pipe()) fail fast for correct behavior
    • Execution Order Guarantee: All methods execute in the exact order written in the fluent chain - ->email()->required()->pipe('trim') works exactly as expected
    • Conceptual Simplicity: From developer perspective, there's one pipeline where callbacks do whatever they need (validate, transform, or both)
    • Performance Optimized: Error collection only where beneficial, fail-fast where appropriate
    • Backward Compatible: All existing code works unchanged across the full suite
    • Internal Architecture: Maintains separate $validations (error collection) and $pipeline (transformations) arrays for optimal execution
  • New satisfies* API: Enhanced instance logical combinators for improved API consistency and flexibility
    • satisfiesAny(array $validations, ?string $message = null) - Validates that value passes ANY of the provided validators or callables
    • satisfiesAll(array $validations, ?string $message = null) - Validates that value passes ALL of the provided validators or callables
    • satisfiesNone(array $validations, ?string $message = null) - Validates that value satisfies NONE of the provided validators or callables
    • Enhanced satisfies() Method: Now accepts callable|FieldValidator instances for maximum flexibility
    • Backward Compatibility: Deprecated anyOf(), allOf(), not() instance methods maintained as aliases
    • API Clarity: Eliminates confusion between static Validator::anyOf() and instance $validator->anyOf() methods
    • Updated test metrics to reflect expanded coverage
  • Refactored oneOf() to OneOfTrait: Improved type safety and execution order consistency
    • Type Safety: oneOf() now only available on primitive validators (StringValidator, IntValidator, FloatValidator, BoolValidator)
    • Execution Order: oneOf() now implemented as transformation, respecting fluent API chain position
    • Logical Consistency: Removed from complex types (ArrayValidator, ObjectValidator) where it doesn't make semantic sense
    • Breaking Change: ArrayValidator and ObjectValidator no longer have oneOf() method
    • Added comprehensive test coverage and removed invalid test cases
  • Form-Safe Empty String Coercion: Enhanced ObjectValidator and AssociativeValidator coercion for better form handling
    • ObjectValidator::coerce(): Empty strings ('') now coerce to empty stdClass objects
    • AssociativeValidator::coerce(): Empty strings ('') now coerce to empty arrays []
    • Real-World Benefit: Form parameters like ?settings= now create empty structures instead of failing validation
    • Type Safety Maintained: Non-empty strings still fail validation as expected
    • Added comprehensive test coverage
  • Enhanced required() Method: Added optional custom error message parameter for consistency with other validation methods
    • required(?string $message = null) - Custom error messages for required field validation
    • API Consistency: All validation methods now support optional custom error messages
    • Better UX: Context-specific error messages for different forms and use cases
    • Maintains backward compatibility with existing code
  • Internal API Modernization: Complete migration from deprecated addValidation() to satisfies() throughout codebase
    • StringValidator: All 10 validation methods (email(), url(), uuid(), ip(), minLength(), maxLength(), length(), pattern(), datetime(), date()) now use satisfies() internally
    • NumericConstraintsTrait: All 5 constraint methods (min(), max(), multipleOf(), positive(), negative()) now use satisfies() internally
    • Static Methods: Validator::anyOf(), Validator::allOf(), Validator::not() now use new satisfies* API internally
    • Future-Proofing: All internal code ready for addValidation() removal in v1.0.0
    • Consistency: Single source of truth for custom validation logic across entire codebase
    • Maintains perfect backward compatibility for users still using deprecated methods

Fixed

  • CRITICAL BUG: Fixed execution order for required() and nullifyEmpty() methods to respect fluent API contract
    • Issue: Validator::isString()->pipe('trim')->nullifyEmpty()->required()->validate(' ') incorrectly passed validation
    • Root Cause: required() and nullifyEmpty() executed before transformations, violating principle of least surprise
    • Fix: Moved both methods to transformation pipeline to respect their position in the fluent chain
    • Impact: Fluent API now works exactly as written - transformations execute in the order they appear
    • Breaking Change: Execution order now matches developer expectations (trim → nullifyEmpty → required)
    • Real-World Benefit: Form validation with trimming and empty string handling now works correctly
  • CRITICAL BUG: Fixed floating-point precision issue in multipleOf() validation
    • Issue: Validator::isFloat()->multipleOf(0.01)->validate(500.01) failed due to floating-point arithmetic precision
    • Root Cause: fmod(500.01, 0.01) returned 0.009999999999980497 instead of 0.0
    • Fix: Implemented epsilon comparison (1e-9 tolerance) for floating-point remainder validation
    • Impact: Decimal multiples (currency, measurements) now validate correctly
    • Added comprehensive test coverage for floating-point precision edge cases
  • CRITICAL BUG: Fixed schema validation field inclusion behavior in ObjectValidator and AssociativeValidator
    • Issue: Validators were incorrectly including ALL schema fields in results, even when not provided in input
    • Fix: Now only includes fields that were actually provided in input OR have default values applied
    • Impact: Results now accurately reflect validated data without unexpected properties
    • Behavior: Required field validation still works correctly (missing required fields still fail)
    • Added comprehensive test coverage to prevent regression
    • Documentation: Added "Field Inclusion Behavior" section to Object Validation guide with clear examples

Documentation

  • Complete type-aware transformation system documentation - Added comprehensive coverage for transform() and pipe() methods
    • Added "Data Transformations" section to Core Concepts guide with detailed type context switching explanation
    • Added transformation examples to Basic Usage guide with clear usage guidelines
    • Documented filterEmpty() method for ArrayValidator with practical examples and real-world use cases
    • Added "Type-Specific Methods" section to API Reference with complete array method documentation
    • Fixed API reference to match actual implementation (removed unimplemented boolean parameter methods)
    • Updated ROADMAP to remove unimplemented enhanced methods and focus on existing features
  • Complete nullifyEmpty() documentation - Added comprehensive coverage across all guides and API reference
    • Added "Universal Methods" section to API Reference with detailed nullifyEmpty() documentation
    • Added "Form-Safe String Handling" section to String Validation Guide with practical examples
    • Added "Explicit Empty String Nullification" section to Numeric Validation Guide
    • Added "Form-Safe Validation with nullifyEmpty()" section to Form Validation Examples
    • Added "Empty String Nullification" section to Basic Usage Guide with fundamental patterns
    • Updated README.md Quick Start example to showcase nullifyEmpty() usage
  • Fixed all dead links in documentation - Replaced 15 dead links with existing comprehensive guides
    • Fixed missing API reference links across all documentation files
    • Ensured all internal navigation works properly for seamless user experience
  • Added development notice - Clear indication that the library is in active development with potential API changes

[0.6.0] - 2025-10-09

Fixed

  • CRITICAL BUG: Fixed ObjectValidator null property handling - isset() was excluding null properties from result objects
    • ObjectValidator now correctly includes all validated properties in the result, even when they are null
    • Maintains consistency with AssociativeValidator behavior
    • Added comprehensive test coverage to prevent regression

Changed

  • BREAKING CHANGE: Enhanced form safety for empty string coercion across all validators
  • IntValidator::coerce(): Empty strings ('') now convert to null instead of 0
  • FloatValidator::coerce(): Empty strings ('') now convert to null instead of 0.0
  • BoolValidator::coerce(): Empty strings ('') now convert to null instead of validation failure
  • Rationale: Prevents dangerous zero defaults in form fields (e.g., bank balances, quantities)
  • Migration: Use explicit ->default(0) if you need zero defaults for empty form fields
  • Added comprehensive test coverage for new behavior

Added

  • New satisfies() Method: Intuitive custom validation with optional error messages
    • Replaces verbose addValidation() with natural language: ->satisfies(fn($v) => $v > 0)
    • Optional error message parameter with sensible default: "Custom validation failed"
    • Maintains backward compatibility with deprecated addValidation() method
    • Added comprehensive test coverage
    • Updated all documentation (65+ references) to use new method
  • BoolValidator Test Suite: Complete test coverage for boolean validation and coercion
    • Tests for string boolean coercion ('true', 'false', 'on', 'off', '1', '0')
    • Case-insensitive coercion support
    • Empty string safety validation

Improved

  • Developer Experience: Natural language validation with satisfies() method reads like English
  • API Consistency: Optional error messages eliminate forced message requirements
  • Documentation Quality: Comprehensive updates across all guides with practical examples
  • Form Safety: Real-world protection against dangerous zero defaults in financial and quantity fields

[0.5.0] - 2025-10-08

Added

  • Enhanced Numeric Coercion: IntValidator and FloatValidator coercion improvements for form handling
    • ⚠️ SUPERSEDED: This behavior was changed in the next release for safety reasons
    • See "Unreleased" section for current empty string coercion behavior
  • Array Filtering with Auto-Reindexing: New filterEmpty() method for ArrayValidator
    • Removes empty strings ('') and null values while preserving valid falsy values (0, false, [])
    • Automatically reindexes arrays to maintain indexed array structure (0, 1, 2, 3...)
    • Works seamlessly with item validators for comprehensive array cleanup
  • Type-Aware Transformation System: Revolutionary transformation infrastructure with intelligent type handling
    • transform(callable $transformer) - Type-flexible transformations that can change data types
    • pipe(callable ...$transformers) - Type-preserving transformations that maintain current type context
    • Dynamic Type Context: Automatically tracks and switches type context during transformation chains
    • Smart Array Coercion: pipe() operations on arrays automatically reindex to maintain structure
    • Associative Key Preservation: AssociativeValidator preserves keys during pipe() operations
    • Enables complex multi-type transformation chains (Array → String → Int) with intuitive syntax

Improved

  • Form Data Handling: Enhanced coercion makes form validation significantly more practical and intuitive
  • Array Data Integrity: filterEmpty() maintains ArrayValidator's contract of returning properly indexed arrays
  • Extensibility: Universal transformation methods enable unlimited flexibility with external library integration
  • Test Coverage: Expanded coverage across numeric, array, and type-aware transformation validators

[0.4.0] - 2025-10-06

Added

  • Static Logical Combinators: New static factory methods for advanced validation logic
    • Validator::anyOf(array $validators) - Creates validator that passes if ANY validator passes (perfect for mixed-type validation)
    • Validator::allOf(array $validators) - Creates validator that passes if ALL validators pass (ideal for combining constraints)
    • Validator::not(FieldValidator $validator) - Creates validator that passes if the provided validator does NOT pass (exclusion logic)
  • Enhanced Mixed-Type Support: Clean syntax for validating arrays with mixed item types
  • Comprehensive Documentation: Complete documentation restructure with focused guides, API reference, and practical examples
    • Getting Started guides (Installation, Basic Usage, Core Concepts)
    • Validation Guides (String, Numeric, Array, Object, Custom, Error Handling)
    • API Reference with complete method documentation including new static combinators
    • Real-world Examples (Form validation, API validation, E-commerce)
    • Enhanced README: Concise overview with quick start examples and organized navigation
    • Comprehensive Test Suite: Expanded coverage for static logical combinators

Improved

  • API Consistency: Static logical combinators provide cleaner syntax than instance-only methods
  • Mixed-Type Validation: Simplified array validation with Validator::anyOf() for mixed types
  • Documentation Coverage: All new methods fully documented with practical examples
  • Type Safety: Static combinators work with any data type while maintaining type safety

[0.3.0] - 2025-09-25

Added

  • Complete String Validation Suite: Added comprehensive string format validators:
    • url() - Validates URLs with support for various protocols
    • uuid() - Validates UUID strings (all versions)
    • ip() - Validates IPv4 and IPv6 addresses
    • minLength(), maxLength(), length() - String length constraints
    • pattern() - Regular expression pattern matching
    • datetime(), date() - Date and datetime format validation
  • FloatValidator: New dedicated validator for floating-point numbers with full constraint support
  • NumericConstraintsTrait: Shared trait for numeric validations eliminating code duplication between IntValidator and FloatValidator
  • Enhanced IntValidator: Added numeric constraints (min(), max(), multipleOf(), positive(), negative())
  • Logical Combinators: Advanced validation logic with allOf(), anyOf(), and not() methods
  • Comprehensive Error Collection: Validators now collect all validation errors instead of stopping at the first failure
  • Enhanced Validation Context: Custom validators receive (value, key, input) parameters for context-aware validation

Changed

  • BREAKING: Renamed NumberValidator to FloatValidator for clearer semantics
  • BREAKING: Renamed Validator::isNumber() to Validator::isFloat() for consistency with PHP types
  • Enhanced addValidation() method to support context-aware custom validation functions
  • Improved error messages and validation feedback throughout the library
  • Refactored numeric constraint methods to use shared NumericConstraintsTrait

Improved

  • Test Organization: Split monolithic test suite into focused test files:
    • AssociativeValidatorTest.php - Associative array validation
    • ObjectValidatorTest.php - stdClass object validation
    • ArrayValidatorTest.php - Plain array validation
    • StringValidatorTest.php - String validation and formats
    • IntValidatorTest.php - Integer validation
    • FloatValidatorTest.php - Float validation
    • FieldValidatorTest.php - Base validator functionality
    • NumericConstraintsTraitTest.php - Trait-specific tests
  • Code Quality: Achieved 100% PHPStan compliance and PHP-CS-Fixer standards
  • Performance: Optimized validation logic and eliminated code duplication

[0.2.0] - 2025-09-07

Added

  • ObjectValidator for validating stdClass objects, created with Validator::isObject().
  • Coercion support for ObjectValidator to convert associative arrays into stdClass objects.
  • Coercion support for AssociativeValidator to convert stdClass objects into associative arrays.

Changed

  • BREAKING: Renamed SchemaValidator to AssociativeValidator for better API consistency.
  • The factory method Validator::isAssociative() now returns an AssociativeValidator instance.
  • Improved error message for associative array type validation to be more specific.

Fixed

  • Changed the type hint on FieldValidator::tryValidate() to mixed to correctly accept both array and object payloads as context.

[0.1.0] - 2025-09-05

Added

  • Initial release of the lemmon/validator package.