locorda_rdf_mapper_generator 0.11.4

locorda_rdf_mapper_generator: ^0.11.4 copied to clipboard

• FIX(locorda_rdf_mapper_generator): preserve imports for generic type arguments in code generation.

0.11.3 #

Removed #

• analyzer v6 support We still had the code for analyzer v6 sort of "commented out", but this is of no use any more. Lets hope that analyzer v8.1 api proves to be somewhat stable

Fixed #

• Code Generation: Fixed bug where generic type arguments in collection types (e.g., List<Chapter>, Map<String, Review>) lost their import information, causing compilation errors in generated code
  • Modified _typeToCode() in analyzer wrapper v8.1 to recursively process type arguments
  • Type arguments now correctly preserve their imports and generate prefixed type references
  • Raw type references (with raw: true) correctly omit type parameters as intended
• Added integration tests for typeToCode() function with TestAnalyzerHelper to verify correct import preservation
  • Tests verify generated code produces correct prefixed types (e.g., List<test.Chapter>)
  • Tests tagged as integration to allow selective execution on CI

• FIX(locorda_rdf_mapper_generator): Fixed analyzer support, min version is 8.1 now.

BREAKING CHANGES #

• File Extension: Generated file extension changed from .locorda_rdf_mapper.g.dart back to .rdf_mapper.g.dart
  • Migration: If you upgraded to 0.11.0, you need to update your imports from .locorda_rdf_mapper.g.dart to .rdf_mapper.g.dart and regenerate your code with dart run build_runner build --delete-conflicting-outputs
  • Context: Version 0.11.0 inadvertently introduced the .locorda_rdf_mapper.g.dart extension during the migration to the Locorda organization. This release reverts to the original naming convention used in all versions prior to 0.11.0
  • Updated build.yaml with correct extensions (.rdf_mapper.cache.json and .rdf_mapper.g.dart)
  • Updated all builders (cache_builder, source_builder, init_file_builder)
  • Updated mapper model builder to generate correct import URIs
  • Updated template renderer to use correct file extensions
  • Regenerated all fixture files with new naming convention
  • Updated all test imports to reference .rdf_mapper.g.dart files

Changed #

• Dart Formatter: Refactored to use dynamic Dart SDK version detection for code formatter, improving compatibility across different Dart SDK versions
• Test Configuration: Added dart_test.yaml with tag definitions for integration and slow tests
• CI Integration: Integration tests that invoke build_runner are now excluded from default CI runs using test tags

Documentation #

• Replaced hardcoded pubspec.yaml dependency examples with dart pub add commands for easier installation
• Updated CLAUDE.md with monorepo-specific testing instructions

• Graduate package to a stable release. See pre-releases prior to this version for changelog entries.

• Version aligned to 0.11.0-dev as part of the Locorda RDF suite. All packages now share synchronized version numbers for simplified dependency management. This release consolidates the monorepo under the locorda/rdf organization in preparation for the upcoming 1.0.0-rc1 release.

Added #

• Directional Mappers: Support for serialize-only and deserialize-only mappers
  • Added serializeOnly and deserializeOnly named constructors to @RdfGlobalResource annotations and a direction parameter to most existing constructors of @RdfGlobalResource, @RdfLocalResource, @RdfIri and @RdfLiteral
  • Generated mappers implement specific interfaces: GlobalResourceSerializer/GlobalResourceDeserializer, LocalResourceSerializer/LocalResourceDeserializer, IriTermSerializer/IriTermDeserializer, LiteralTermSerializer/LiteralTermDeserializer
  • Serialize-only mappers omit deserialization logic (no regex parsing, no fromRdfResource method)
  • Deserialize-only mappers omit serialization logic (no toRdfResource method, IRI strategy optional for global resources)
  • Registry automatically uses correct registration method: registerSerializer(), registerDeserializer(), or registerMapper()

Added #

• Cross-File Mapper Dependencies: Mapper generator now processes property field types across file boundaries
  • Automatically detects and generates mappers for classes referenced in properties from other files
  • Supports both @RdfGlobalResource and @RdfLocalResource annotated types used in properties
  • Enables parent-child document patterns where child classes are defined in separate files
  • Internal processing distinguishes between top-level mappers (output to generated file) and nested mappers (used for dependency resolution only)

Added #

• IRI Fragment Template Support: Added fragmentTemplate parameter to IriStrategy for IRI fragment handling

  • Allows separate template for URI fragment component (the part after #)
  • Both base and fragment templates can use the same IRI part variables
  • Variables unused in one template but used in the other won't generate warnings
  • Generated mappers combine base and fragment templates at runtime

• IRI Provider Feature: Added support for providedAs parameter in IriStrategy constructors

  • Parent resources can now provide their IRI to dependent child resources
  • Child resources can reference parent IRIs in their own IRI templates using context variables
  • Generated mappers automatically pass parent IRI via () => subject.value during serialization
  • Enables hierarchical IRI patterns like {+parentIri}/children/{childId}
  • Supports complex parent-child relationships with automatic IRI context propagation

Changed #

• BREAKING: Dropped support for analyzer versions below 7.4.0

  • Updated analyzer constraint from >=6.9.0 <8.0.0 to >=7.4.0 <9.0.0
  • Switched active analyzer wrapper from v6 to v7_4
  • Disabled v6 analyzer wrapper (now provides stub implementations only)
  • Added v8_2 analyzer wrapper skeleton (prepared for future analyzer 8.x/9.x support)

• Updated Dependencies: Upgraded build system and code generation dependencies

  • Updated build constraint from <4.0.0 to <5.0.0
  • Updated analyzer to 8.2.0
  • Updated build to 4.0.1
  • Updated build_runner to 2.9.0
  • Updated dart_style to 3.1.2

Fixed #

• Code Generation Improvements: Enhanced v7_4 analyzer wrapper
  • Filter out synthetic fields, getters, and setters to prevent duplicate property generation
  • Fixed Code.paramsList usage in DartObject-to-Code conversion for better readability
  • Added linter suppressions for deprecated APIs to reduce noise in analyzer output

Changed #

• Updated Dependencies: Updated RDF-related dependencies to latest versions

  • Updated locorda_rdf_mapper from 0.10.1 to 0.10.2 for improved functionality
  • Updated locorda_rdf_core from 0.9.14 to 0.9.15 for bug fixes and enhancements

• Improved IRI Handling: Enhanced IRI creation and processing throughout generated code

  • Changed from direct IriTerm() constructor to context.createIriTerm() for better context awareness
  • Updated generated mappers to use term.value instead of term.iri for consistent IRI access
  • Templates now generate more contextually aware IRI creation and processing

• Enhanced Code Generation: Improved generated code quality and consistency

  • Updated processor utilities to use IriTerm.validated() for safer IRI creation
  • Fixed IRI field access in processor utils to use correct field name (value instead of iri)
  • Improved enum mapper generation with consistent IRI handling patterns

• Documentation Updates: Updated README and API documentation

  • Updated dependency version examples in README to reflect current stable versions
  • Fixed code examples to use proper const IriTerm() constructor calls
  • Updated vocabulary constants to use const constructors for better performance

Fixed #

• IRI Term Processing: Fixed inconsistent IRI field access patterns
  • Corrected processor utilities to access IRI values using proper field names
  • Fixed enum and resource mappers to use consistent IRI value extraction
  • Resolved issues with IRI template processing and regex matching

Fixed #

• Init File Import Resolution: Fixed init file builder to properly apply BroaderImports for import resolution
  • Init files now use clean import aliases like RdfListMapper.new instead of verbose prefixes like rlm.RdfListMapper.new
  • Extended _InitFileTemplateData to include BroaderImports from cache files
  • Improved consistency between individual mapper files and init file generation

Fixed #

• Transitive Import Chain Resolution: Fixed BroaderImports to properly handle transitive import dependencies
  • Added recursive processing of imported libraries to resolve import->import->export chains
  • Fixed issue where libraries imported through intermediate dependencies weren't properly mapped to their broader imports
  • Results in cleaner generated code with simplified import aliases (e.g., RdfListMapper.new instead of rlm.RdfListMapper.new)
  • Added comprehensive test coverage for complex transitive import scenarios including deep dependency chains

Added #

• Annotation Subclassing Support: Added comprehensive support for inheritance of RDF annotations

  • Classes can now inherit RDF mappings from their parent classes
  • Proper handling of annotation inheritance chains for complex class hierarchies
  • Validation and error handling for annotation inheritance conflicts
  • Enhanced mapper generation to support inherited annotation patterns

• Named Factory Constructor Support: Added full support for named factory constructors in RDF mapping

  • Enhanced analyzer wrapper models (both v6 and v7.4) to detect and handle named constructors
  • Updated mapper model builder to properly configure named factory constructors
  • Generated mappers now support classes using named factory constructors for instantiation
  • Improved constructor resolution and parameter handling in code generation

Changed #

• Enhanced Mapper Model Building: Improved internal mapper model building infrastructure

  • Updated IriModelBuilderSupport and MapperModel classes for better constructor handling
  • Enhanced parameter resolution and type inference for complex constructor scenarios
  • Better integration between analyzer wrapper and mapper generation systems

• Updated Dependencies: Updated to latest versions of RDF-related dependencies

  • Updated locorda_rdf_mapper_annotations to version 0.10.2 for enhanced annotation features
  • Refreshed other RDF ecosystem dependencies for improved compatibility and features

Fixed #

• Constructor Resolution: Fixed issues with constructor detection and parameter mapping

  • Resolved problems with named factory constructor parameter resolution
  • Improved handling of complex constructor scenarios in inheritance hierarchies
  • Fixed edge cases in constructor validation and error reporting

• Code Generation Consistency: Improved consistency in generated mapper code

  • Fixed formatting and structure issues in generated test files
  • Enhanced code generation for complex inheritance and constructor scenarios
  • Better handling of import dependencies in generated code

Added #

• Contextual Property Mapping Improvements: Enhanced support for context-dependent property serialization/deserialization

  • Refactored ContextualMapping to use proper SerializationProvider dependencies instead of named providers
  • Improved dependency injection for contextual mappers in generated code
  • Added support for @RdfUnmappedTriples(globalUnmapped: true) to capture all unmapped triples globally

• Raw Type Support: Added raw parameter to type code generation for better mapper instantiation

  • New raw parameter in DartType.toCode() method allows getting raw class names without generic parameters
  • Enables proper mapper constructor calls with raw class types while preserving generic type information
  • Implemented across all analyzer wrapper versions (v6 and v7.4)

Changed #

• Code Generation Improvements: Enhanced template system and dependency management

  • Removed deprecated Code.constructor() factory method in favor of more flexible Code.type() and Code.literal() combination
  • Improved mapper dependency resolution with better parameter naming (configurable suffix support)
  • Enhanced contextual mapping code generation with proper SerializationProvider integration
  • Simplified constructor code generation using Code.paramsList() for better readability

• Mapper Reference Handling: Improved mapper reference tracking and instantiation

  • Enhanced MapperRefInfo to track both full type and raw type separately
  • Better handling of generic type parameters in mapper references
  • Improved mapper dependency default value generation

• Test Infrastructure Updates: Enhanced testing capabilities and examples

  • Updated document example to demonstrate contextual mapping with IriRelativeSerializationProvider
  • Improved test fixtures with better generic type handling
  • Enhanced validation test infrastructure with proper project setup

Fixed #

• Constructor Code Generation: Fixed issues with const constructor and parameter handling

  • Corrected const constructor generation to use proper Code.literal() and Code.type() combination
  • Fixed parameter list generation for cleaner, more readable generated code
  • Resolved issues with import alias handling in constructor calls

• Dependency Management: Improved dependency resolution and parameter handling

  • Fixed contextual mapping dependency generation with proper SerializationProvider types
  • Corrected mapper dependency suffix handling for better naming consistency
  • Improved error handling for unresolved mapper dependencies

Changed #

• Breaking Change: Updated rdf_vocabularies dependency to use the new multipackage structure:
  • Replaced rdf_vocabularies: ^0.3.0 with locorda_rdf_terms_core: ^0.4.1 and locorda_rdf_terms_schema: ^0.4.1
  • Updated all import statements throughout the codebase to use the new package structure
  • import 'package:rdf_vocabularies/rdf.dart' → import 'package:locorda_rdf_terms_core/rdf.dart'
  • import 'package:rdf_vocabularies/xsd.dart' → import 'package:locorda_rdf_terms_core/xsd.dart'
  • import 'package:rdf_vocabularies/schema.dart' → import 'package:locorda_rdf_terms_schema/schema.dart'
  • And similar updates for other vocabulary imports (foaf, vcard, etc.)

Added #

• Automated dependency management with Dependabot configuration for weekly updates

Changed #

• BREAKING: Updated build dependency from 2.5.4 to 3.0.0 for improved build performance and compatibility, but continue to support older build versions
• Updated dependency versions:
  • dart_style from 3.1.0 to 3.1.1
  • analyzer from 7.7.0 to 7.7.1
  • locorda_rdf_core from 0.9.7 to 0.9.11
  • locorda_rdf_mapper from 0.9.2 to 0.9.3
  • locorda_rdf_mapper_annotations from 0.3.1 to 0.3.2
  • Various test dependencies updated to latest versions

Fixed #

• Compatibility with build package 3.0.0 by adding explicit type cast in analyzer wrapper
• Code cleanup in analyzer v7.4 wrapper by removing unused exports

Added #

• Added hasInitializer and isSettable properties to field analysis for better code generation control
• Added comprehensive collection mapping test coverage including all collection types and edge cases

Changed #

• Refactored internal model class names for improved clarity:
  • FieldInfo → PropertyInfo
  • PropertyInfo → RdfPropertyInfo
  • RdfPropertyInfo → RdfPropertyAnnotationInfo
• Renamed collection type detection methods for consistency:
  • isList → isCoreList
  • isMap → isCoreMap
  • isSet → isCoreSet
  • isCollection → isCoreCollection
• Updated internal property structures to use properties instead of fields for consistency with RDF terminology

Fixed #

• Improved analyzer wrapper to properly detect field initializers and settability
• Removed warning about missing constructor parameter during build_runner that happened for initialized final fields and getters without accompanying setters
• Detection if something is a collection sometimes did not work, itemType override thus failed for custom collection types
• important constants like rdfList, rdfSeq etc. were not documented at all

Added #

• Added additional test cases for collection property annotation mapper generating.

Fixed #

• Fixed collection item mapper type generation where item mappers were incorrectly typed for the collection type instead of the individual item type (e.g., IriTermMapper<List<String>> now correctly generates as IriTermMapper<String>)
• Fixed parameter naming inconsistency in collection mappers where custom serializer/deserializer parameters are now correctly named itemSerializer/itemDeserializer instead of serializer/deserializer when dealing with collection item mappers
• Improved documentation formatting in generated IRI mappers by properly wrapping class names in backticks

Changed #

• Support powerfull collection mapping
• BREAKING: Updated toRdfResource return type from (RdfSubject, List<Triple>) to (RdfSubject, Iterable<Triple>) for improved performance and flexibility
• BREAKING: Standardized custom mapper parameter names in generated code:
  • iriTermDeserializer/iriTermSerializer → deserializer/serializer
  • literalTermDeserializer/literalTermSerializer → deserializer/serializer
  • globalResourceDeserializer/resourceSerializer → deserializer/serializer
  • localResourceDeserializer/resourceSerializer → deserializer/serializer
• BREAKING: Updated @RdfLiteral.custom method signatures to use LiteralContent instead of LiteralTerm:
  • toLiteralTermMethod and fromLiteralTermMethod now work with LiteralContent objects
  • Methods are automatically wrapped with proper datatype handling in generated code
• Added datatype field to generated LiteralTermMapper and IriTermMapper implementations for better type safety

Fixed #

• Improved datatype handling in custom literal mappers with explicit datatype support
• Enhanced method call generation for custom literal conversion methods

Added #

• Lossless Mapping Support: Added complete support for @RdfUnmappedTriples annotation to enable lossless round-trip RDF mapping
  • Fields annotated with @RdfUnmappedTriples capture all triples not explicitly mapped to other class properties
  • Generated code includes reader.getUnmapped<T>() calls for deserialization and builder.addUnmapped(field) calls for serialization
  • Supports custom types with registered UnmappedTriplesMapper<T> implementations
  • RdfGraph type has built-in support and is recommended for most use cases
• Enhanced Validation: Added comprehensive validation for @RdfUnmappedTriples usage
  • Error when multiple fields per class are annotated with @RdfUnmappedTriples
  • Warning when non-RdfGraph types are used with clear guidance about custom mapper registration
• Improved Developer Experience: Enhanced warning and error message formatting
  • Professional, structured warning messages with bullet points for better readability
  • Clear, actionable guidance for resolving validation issues
  • Improved build-time feedback for annotation usage problems

Fixed #

• Fixed logging configuration for proper warning display during build_runner execution
• Ensured getUnmapped() calls are always executed last during deserialization for correct lossless mapping behavior

Fixed #

• Corrected example code in README.md

Added #

• Professional homepage at doc/index.html with modern design and feature highlights
• ignore_for_file: unused_field directive to generated files to suppress analyzer warnings

Changed #

• Major rewrite and modernization of README.md with clearer feature descriptions
• Improved documentation accuracy and removed overstated claims
• Enhanced onboarding experience with better "Try It" section
• Restructured documentation to be more compelling and professional for new users
• Cleaned up test fixtures for better maintainability

Fixed #

• Fixed Stack Overflow error caused by infinite recursion in BroaderImports when processing circular library exports
• Added cycle detection to prevent infinite loops in library import/export resolution
• Fixed regex pattern rendering bug where regex patterns were not properly escaped as raw strings in generated code

Changed #

• Expanded analyzer package version support to include older versions (>6.9.0 <8.0.0)
• Added analyzer API wrapper layer to ensure compatibility across analyzer versions
• Updated dart_style dependency constraints to support broader version range

Fixed #

• Compatibility issues with analyzer package versions 6.x
• Build compatibility with projects using older analyzer versions

Added #

• Full support for all annotations from locorda_rdf_mapper_annotations 0.2.2:
  • Resource annotations (@RdfGlobalResource, @RdfLocalResource)
  • IRI processing (@RdfIri, @RdfIriPart) with template support
  • Literal processing (@RdfLiteral, @RdfValue)
  • Property annotations (@RdfProperty) with various mapping options
  • Collection support (Lists, Sets, Maps)
  • Map entry processing (@RdfMapEntry, @RdfMapKey, @RdfMapValue)
  • Enumeration support with @RdfEnumValue
  • Language tag support via @RdfLanguageTag
• Dynamic IRI resolution with template variables and context variables
• Provider-based IRI strategies for flexible URI creation
• Type-safe mapper generation for global and local resources
• Automatic registration of generated mappers
• Custom mapper integration (named, by type, and by instance)
• Smart type inference for RDF annotations with registerGlobally: false option
• Validation and comprehensive error messages
• Complete serialization and deserialization between Dart objects and RDF triples
• Support for complex object graphs
• Mustache templates for code generation
• Regular expression-based IRI parsing and creation
• Auto-generated mapper documentation
• Nested resource serialization and deserialization

Engineering #

• Multi-layered architecture with model and resolved model layers
• Structured code generation system with separate processors for different annotation types
• Clean separation between parsing, analysis, and code generation
• Comprehensive test suite covering all generated mapper scenarios
• Advanced type inference system for property resolution
• Modular template system for code generation
• Performance optimizations for working with complex object graphs
• Code abstraction for better maintainability