All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog and this project adheres to Semantic Versioning.
- Issue #188 Made BeanProperty-annotation repeatable. Added targets-property to BeanProperty-annotation, allowing the user to specify which mappings a BeanProperty should apply to.
- Added diagnostics, allowing users to check what mappings and conversion are performed as part of a given mapping.
Diagnostics are only collected when enabled. Diagnostics can be enabled by using the BeanMapper.wrap(DiagnosticsDetailLevel)-method.
Each DiagnosticsDetailLevel comes with a different logging strategy.
DiagnosticsDetailLevel.COUNT_TOTAL: Logs the total number of mappings and conversion performed as part of the top-level mapping, in the format:[Mapping ] SourceClass -> TargetClass {total mappings: 38, total conversions: 70, max depth: 7}DiagnosticsDetailLevel.COUNT_PER_PAIR: Logs the total number of mappings and conversion per type pair.DiagnosticsDetailLevel.TREE_COMPLETE: Logs the entire mapping and conversion tree.
- Issue #203 Expand caching to include other expensive, unchanging operations; Added the BeanConverterStore and the CanonicalClassNameStore, to reduce reflective calls.
- Caching for BeanConverter-selection.
- Caching for canonical class names.
- Issue #196 Calls to the logger should not contain a call to String.formatted(Object...); Fixed by removing any formatting from performance logging. Trace-logging should keep the level of detail provided by the formatting.
- Issue #197 Implement caching for Unproxy results.; Created UnproxyResultStore, allowing for thread-safe caching and retrieval of unproxied classes.
- Issue #199 Use of OverrideField causes excessive resource consumption.; Use of OverrideField was phased out. OverrideField itself is now deprecated for removal.
- Issue #201 Calls from OverrideConfiguration to parentConfiguration cause unnecessary overhead.; Introduced fields in the OverrideConfiguration, to limit the overhead of calls to the parent configuration.
- Issue #190 Enum with abstract overrides cannot be mapped.; Due to a change in the way Class.getCanonicalName() works in newer versions of Java, this would return null for Enums with an abstract method.
- PR #193 Create Logger, replace existing logging.; Added performance logging and replaced regular logging with a trace-logger.
- Issue #181 RecordToAnyConverter#copyFieldsToIntermediary(Object, Map) throws a NPE if Map contains an Entry where the value is null.; Added null-check to RecordToAnyConverter#copyFieldsToIntermediary(Object, Map), preventing a NPE whenever a field in the source-object is null.
- Issue #185 OverrideConfiguration#strictMappingProperties cause of severe hit to performance.; Replaced usage of OverrideField, with direct call to underlying configuration.
- Upgraded JUnit 5.8.2 to 5.9.0
- Upgraded SLF4J 2.0.0 to 2.0.3
- Upgraded Javassist 3.29.1-GA to 3.29.2-GA
- Upgraded jackson-databind 2.13.4 to 2.13.4.2
- Issue #130 @BeanConstruct does not map collection constructor arguments; Added DefaultBeanInitializer#mapParameterizedArguments(Type[], Object[]), allowing BeanMapper to properly map constructor parameters with a type-parameter, when using the @BeanConstruct-annotation.
- Issue #132 Mapping the same field twice ignores one of the mappings.; Whenever a field is mapped, a check will be performed to detect field shadowing in this way. If field shadowing is detected, a FieldShadowingException is thrown. However, a BeanProperty-annotation may take on the name of a field that is non-public and for which an accessor is not exposed. Fields annotated with BeanIgnore are not considered when checking for shadowing.
- Issue #149 Support mapping of JDK 16+ records to classes.; Added support for mapping record through the usual BeanMapper#map(Object, Class<?>)-method.
- Issue #152 Methods that return a Collection should never return null.; All methods that return a Collection, will return an empty Collection of the target type, rather than returning null.
- Issue #153 #153; Rather than using the Boolean-wrapper, all occurrences of Boolean that are not absolutely necessary due to generics, have been replaced with the primitive boolean, or the Trinary-enum.
- Issue #166 Source with BeanAlias-annotated fields cannot be downsized.; Modified ClassGenerator#createClass, to check whether a generated field is annotated with BeanAlias. If so, the name of the generated field will be set to the value on the BeanAlias-annotation, and the annotation will be removed from the generated field.
- Issue #168 ClassGeneratorTest#shouldNotFailConcurrently always succeeds.; By catching Throwable, and saving Throwable in the List, rather than exception, the test should work correctly, and fail when it should.
- BeanMapper#map(Collection, Class) allowing users to map from a Collection to the type actual type of the
- Support for mapping Queue. A Queue will be mapped to an ArrayDeque by default. The order of elements is guaranteed to be preserved, except when the Queue is mapped to a Queue that inherently modifies the order of elements (e.g. PriorityQueue).
- BeanMapper#map(Queue, Class) allowing users to map a Queue to a new Queue, mapping the elements of the source to the target class.
- BeanMapper#map(Object[], Class) allowing users to map an array to an array with elements of the type of the target class. Also works for primitive arrays.
- Support for mapping to and from JDK16 record-classes.
- @BeanRecordConstruct-annotation, used to give BeanMapper instructions on how to map a record.
- BeanRecordConstructMode-enum, which allows the user to exclude certain constructors from being used to map a record, or force one to be used.
- Configuration#addCustomDefaultValueForClass(Class, Object value) and Configuration#getDefaultValueForClass(Class), which can be used to define default values for classes during runtime.
- RecordToAnyConverter, which creates a dynamic class based off of the given Record, which contains only public fields, corresponding to the RecordComponents. The dynamic class is then filled with values from the source Record, and used as an intermediary between the source and the target.
- MappingException, to serve as a common super-class for Exceptions thrown while mapping classes, and records.
- Default values for Optional, List, Set and Map in the DefaultValues.
- BeanMapper#map(Object, ParameterizedType), allowing for smarter mapping of collections and optionals.
- BeanMatchStore#detectBeanPropertyFieldShadowing(PropertyAccessor, io.beanmapper.annotations.BeanProperty), which will throw a FieldShadowingException when appropriate.
- Mapping of Optional-objects to a target class, using BeanMapper#map(Optional, Class).
- Methods in OverrideConfiguration that used to perform NOP, now throw a BeanConfigurationOperationNotAllowedException.
- Removed deprecated method BeanMapper#config(), BeanMapper#wrapConfig().
- Removed deprecated class BeanMapperAware.
- @BeanAlias now applicable to RecordComponent.
- Expanded the use of type parameters to various methods throughout the library, including the BeanMapper-class.
- BeanConverters may be offered null-values now. Will default to returning the default value for the relevant type.
- slf4j-api dependency, as it is transitive through jcl-over-slf4j.
- Issue #121 Mapping an Enum field to an Enum field of the same type fails when a custom toString method is present.; When an Enum with a custom toString-method was mapped to an Enum, the mapping would fail. Fixed by adding an instanceof check in the AnyToEnumConverter, making it use Enum#name() to get the name of an Enum, rather than toString.
- Issue #137 #137; Mapping a class with a getter that returns an Optional, would fail, as an Optional can typically not be mapped to the target class. Fixed implementing an OptionalToObjectConverter, which handles unpacking an Optional, and additionally delegates further conversion back to the BeanMapper.
- DynamicClassGeneratorHelper was removed, due to returning null, and replaced with passing the baseclass for a generated class immediately to the constructor of GeneratedClass.
- Upgraded JUnit 4 to JUnit 5
- Replaced travis build with Github actions workflow
- Upgraded jackson-databind dependency which had an owasp error
- Upgraded to Java 17, from Java 8
- Issue #141, Dynamic Mapping throws InaccessibleObjectException in Java 16+; when a form is dynamically
mapped to a class, an InaccessibleObjectException would be thrown, due to updated Reflection requirements. Fixed by updating
GeneratedClassto make use ofCtClass#toClass(Class), rather thanCtClass#toClass().
- Added value alias for
@BeanPropertyannotation, so usingname =is no longer needed. It is still supported to maintain backwards compatibility
- Issue #118, Nested ParameterizedTypes throws a ClassCastException; when a type is a nested generic (eg,
List<List>) it threw a ClassCastException, even if no mapping took place. Fixed by checking the Type in
AbstractBeanPropertyClassand calling for rawType if it concerns a ParameterizedType. - Issue #116, Mappable nested classes analyzed for wrong direction; when dealing with a mappable nested class, the applied direction is used as-is. This resulted in errors when the field could only be accessed through a method accessor and that the complementing method accessor was not available (ie, get and no set, or set and no get). The problem has been fixed by inverting the method direction on dealing with a mappable nested class.
- BREAKING CHANGE: BeanConverter implementations no longer work with BeanField, but with BeanProperty. The name BeanField is no longer correct. A BeanProperty is the combination of BeanField and BeanMethod accessors (get/set).
- Issue #114, Bean property Class types are determined depending on the accessor that will be used; previously the Class was determined on the basis of the bean field. Now, it will look at the way a property will be accessed (getter/setter/field) and use the proper Class, including generics (ie, Type).
- Issue #109, Specify the return type for AbstractBeanConverter.doConvert; on extending
AbstractBeanConverter, it is beneficial for the developer to immediately see the expected return type for the
doConvertmethod. - Bug #111, BeanCollection null list overrides the existing list; when a source and target have been assigned to a collection handler, it will now treat a null value for the source as special, subscribing it to the BeanCollectionUsage rules (default: CLEAR). That is, it will REUSE the target, CLEAR it, or CONSTRUCT a new one. This is the most logical behaviour with Hibernate on the other side.
- Bug #112, Strict mapping messes up the build sequence; The validation took place before all configuration
was done, resulting in collection handlers not being available in some cases. The validation of strict classes is done as part of the last step of
the
BeanMapperBuilder.build()method, after all required steps have been taken.
- Issue #107, Test for access by running against LogicSecuredCheck instance; ability to add LogicSecuredCheck classes to BeanMapper's configuration. These classes can be called upon using the @BeanLogicSecured annotation. It allows for more complex interaction with the enveloping security system, such as not only checking against roles, but also comparing fields in the source or target against information known about the Principal.
- Issue #105, Ability to deal with @BeanRoleSecured by delegating to a RoleSecuredCheck; when a field is tagged as @BeanRoleSecured, BeanMapper will query its attached SecuredPropertyHandler. The handler will most likely be associated with a security implementation, such as Spring Security (not handled here). If no handler is present, access is granted by default.
- Issue #106, When a @BeanRoleSecured is found without a RoleSecuredCheck being set, throw an exception; the absence of a RoleSecuredCheck is by default a reason to throw an exception when @BeanRoleSecured is used anywhere within the application.
- Issue #104, IllegalAccessExceptions that occurred after JMV garbage collection; since all PropertyDescriptor instances are kept in a static cache and PD relies on Java's SoftReference, the reference can be discarded. This means that if the accessible boolean was set to true, it will be discarded. The error can manifest itself after a GC run and if the application relies on package-protected classes. The fix: the setAccessible method should be called very soon prior to calling the invoke method rather than only once at the creation of the PropertyDescriptorPropertyAccessor.
- Issue #99, Collections in superclasses did not get their generic types read; one of the attributes of getDeclaredField(field) is that it only works on the active class. The fix means that the superclasses will be checked for presence of the field. When found, it will call getDeclaredField on that class to get its generic type. Also, collection mapping instructions are not used when no collection element type can be determined; that is the one crucial element required for mapping collections.
- Issue #100, Collections.EmptySet can not have add() called on; BeanCollectionUsage will check the canonical classname of the collection. If it starts with "java.util.Collections.", it will be tagged as reconstructable.
- Issue #92, BeanCollection no longer required for mapping collections; BeanMapper is now capable of determining the collection element type of the target collection by examining the generic type. It will use this type as input for the mapping process. BeanCollection.elementType is no longer a required value. BeanMapper will merge collection information from target and source, giving preference to target information.
- Issue #97, StringToEnumConverter replaced with AnyToEnumConverter; this makes it possible to convert enums to enums as well. Functionality is required because lists are no longer copied by reference by default.
- BREAKING CHANGE Issue #93, config() not threadsafe; it is possible for override configurations to be reused between threads, theoretically allowing fields to be changed before the map is called. This is not threadsafe. Now, nowhere is override configuration reused; it will always create a new override configuration. Both config() and wrapConfig() have been replaced with wrap(), which does the same as wrapConfig() . Internally, some OverrideConfiguration properties have been delegated to an OverrideField which takes care of returning the right value for a property. The clear() method has been removed; calling wrap automatically resets these properties (expect for downsize source/target, which are primarily used internally).
- Issue #60, Unmatched BeanProperty did not throw an exception; properties annotated with BeanProperty must match. If they do not, an exception must be thrown. Due to a bug, this did not always occur (only with BeanProperty on the target side). The current mechanism keep tabs on matched properties and does a final verification. If unmatched properties remain that should have been matched, an exception is thrown.
- Issue #89, Use sensible implementation for Set; when a set is created an no preferredCollectionClass is passed, the handler will look at the collection element type. If the type is Comparable, it will return a TreeSet. If not, it will return a HashSet.
- Issue #90, Introduce a global flushEnabled, BeanCollection.flushAfterClear default true; a global flushEnabled setting has been introduced, which is false by default. The BeanCollection.flushAfterClear has changed from a default false to true. BeanMapper asserts that both settings must be true before flushing. BeanMapper Spring #28 sets flushEnabled=true when Lazy is used, because this offers the best chance of the EntityManager running in a transaction context.
- Issue #83, The name field from an enum is not mapped to target field; in the resolution of issue #78 the definition of getter fields has been tightened, because previously all private fields were tagged as available as well. One project made use of this loophole by reading the name field of an enumeration class to a String field. With the new fix this is no longer possible, since the name field is private. This fix makes an exception for the name field of an enum class. It will be considered available for reading.
- Issue #84, BeanMapper executes the list of AfterClearFlusher instances; when a clear method on a collection is called, BeanMapper makes sure to run down the list of registered AfterClearFlusher instances. On every instance, the flush method will be called. By default no AfterClearFlusher instances are added. BeanMapper has no notion of ORMs; this is left to beanmapper-spring and beanmapper-spring-boot-starter . Note that the flusher chain is only called when it has been set to do so in the BeanCollection annotation or the override configuration (flushAfterClear).
- POSSIBLE BREAKING CHANGE! Issue #59, @BeanCollection usage should be CLEAR not REUSE; the current collection usage is REUSE, which means the collection is used as is, keeping the target elements in the collection. In Hibernate, this will not trigger the deletion of orphans. The dominant option is to use CLEAR, which means that any target collection instance will be used, but its contents removed by calling clear(). By calling this method, a managed list (such as Hibernate does), will keep track of the removed elements (orphan deletion, if enabled).
- Issue #76, Support anonymous classes; Instances that were created anonymously were not supported by the Unproxy. Now, it will check whether the class is anonymous. If this is the case, it will refer to its superclass and re-enter the unproxy process.
- POSSIBLE BREAKING CHANGE! Issue #68, Refactored collection mapping; collection mapping happened in two different styles. One was handled by MapCollectionStrategy, the other by CollectionConverter. CollectionConverter now defers to MapCollectionStrategy. The latter has been extended to be able to deal with more precise mappings and clearing/reusing collections. BeanCollection no longer supports the targetCollectionType(), which is now inferred and handled by the appropriate CollectionHandler. Instead, use preferredCollectionType if you care for a specific collection class instead of the one provided by the CollectionHandler.
- Issue #78, BeanMapper contained an error that considered all fields (even private fields) as readable. This dormant error, previously seems not to have manifested itself, but with the strict handling of mappings it did. The fix has been to make sure to check the field modifiers for public access.
- Issue #75, Optional strict handling of BeanMapper mappings; two layers of protection have been introduced. The first is the possibility to register a class pair with one side being strict. The strict side must have matching properties for all its valid properties. If properties are not matched, an exception will be thrown detailing the mismatches. The second layer of protection works on the Form/Result convention. It checks whether the source is consider a form Ie, classname has the suffix 'Form') or a target is a result (ie, classname has the suffix 'Result'). If this is the case, the other side must have matching properties as well. This second layer of defense works runtime right before the mapping takes place. Note that the suffix can be changed and the convention for strict mapping can be disabled.