0.4.3

This version of Lorenz introduces a number of improvements to the Enigma IO classes, namely
fixing support that has been broken under 0.4.x.

0.4.2

This version of Lorenz addresses a critical bug in the Enigma reader, that prevents inner
classes from preceding the other members of a class.

0.4.1

This version of Lorenz addresses a critical bug in the Enigma writer, that causes invalid
output (totally broken).

Broken

CLASS none/aet

METHOD c getDeathMessage (Lnone/afq;)Lnone/iw;


ARG 0 entity

Fixed

CLASS none/aet
	METHOD c getDeathMessage (Lnone/afq;)Lnone/iw;
		ARG 0 entity

0.4.0

This version of Lorenz consolidates the work towards supporting additional formats, even
going as far to support those formats. A number of breaking changes have been made in order
to support these formats, in addition to new constructs to make using Lorenz easier.

The following formats are now supported:

• Enigma
• JAM
• Kin

Big thanks to @Minecrell, for his help towards this release!

• Contributions to Bombe
• Contributions to Lorenz

Changes

Type Model

Lorenz 0.4.0 has received a major overhaul of the type model, migrating that section of
the codebase over to its own project (Bombe),
and introducing a great deal of safety to types and their uses.

• PrimitiveType has been split up to BaseType and VoidType, becoming more in line
  with the JVM specification.
  • However, there now exists a PrimtiveType interface that both BaseType and
    VoidType inherit.
• FieldType has been introduced, providing type safety for field types (void is not a
  valid field or parameter descriptor as per the JVM specification).
• The #getObfuscated() and #getDeobfuscated(MappingSet) methods no longer exist in
  either Type or MethodDescriptor.
  • Calls to #getObfuscated() should be replaced with calls to #toString().
  • Calls to #getDeobfuscated(MappingSet) should be replaced with calls to
    MappingSet#deobfuscate for the appropriate type.
• MethodDescriptor#compile() has been renamed to MethodDescriptor#of().
• The 'convenience' signature constructors have been replaced with #of(String, String)
  methods.

I/O

Previously Lorenz had only supported text-based mapping formats and gave them a rather
strict means of being implemented. Lorenz 0.4.0 has generified the I/O classes to be more
adaptable to a greater number of situations, including binary-based formats.

The provided implementations of mapping format I/O classes have also been moved to their
own packages in preparation of the modularisation that will happen in Lorenz 0.5.0 (more
on that later).

• MappingsReader and MappingsWriter have been generified and are no longer dependent
  on text-based formats.
  • There is abstract classes for both text-based formats and binary-based formats.
• The Processors of text-based formats are now contained as inner classes of their
  parent reader.
  • The processors no longer use Guava's LineProcessor, instead using Java's Consumer.
• MappingsReader#parse() has been refactored to MappingsReader#read() becoming more
  consistent with the reader naming.
• The mapping readers will now throw exceptions when given bad input, as opposed to ignoring
  it as it did in the past.
  • Do note that SRG package mappings are'nt considered bad input, and will be silently
    ignored.
  • This remains to be something that needs to be looked into further, and expect changes
    to be made in 0.5.0/1.0.0.
• Introduced a MappingFormat interface, with convenience methods that will be useful to
  consumers of Lorenz.
  • Notably there are convenience methods for operating with Paths directly.
    • MappingFormat#read(Path)
    • MappingFormat#write(MappingSet, Path)
  • There is a psuedo-enumeration of MappingFormats implemented within Lorenz available
    through the MappingFormats class.

Other

These changes, though notable, are not large enough to warrant their own section. Most of
the changes listed here are features / changes to better support the new formats that
Lorenz supports.

• Fields are now stored using their signature as a key, not their obfuscated name.
• FieldTypeProviders no longer guarantee the availability of field type information.
• MappingSet makes use of a CascadingFieldTypeProvider allowing multiple providers to be used.
• Method Parameter Mappings are now supported, using the index of the parameter as the
  obfuscated name and key.
  • To make this change, MemberMapping was altered to become generic - allowing for any
    parent mapping of type Mapping.
• The dependency on Guava has been dropped.

Modularisation

As previously mentioned, Lorenz 0.5.0 will have complete modularisation. This will go
further than the "primitive" modularisation present in 0.4.0. There will exist individual
modules for each format supported in Lorenz, as opposed to being included in the base
module.

Complete modularisation was deferred until 0.5.0 to not hold back any further on a 0.4.0
release, and allow for adequate time to consider a few technical points that would be nice
to have solved.

Lorenz-ASM

The first module has appeared! Lorenz-ASM provides a number of utilities that may be useful
when using Lorenz with ASM.

• An implementation of ASM's Remapper class, backed by a MappingSet and an
  InheritanceProvider (from Bombe).
  • This class was previously available through
    Survey and notably the Lorenz version
    lacks the SignatureAdapter that can handle broken signatures produced by obfuscators.
    This omission was made simply as fixing issues with obfuscators does not fall within
    Lorenz's remit.
  • Bombe-ASM (which Lorenz-ASM depends on) has 2 implementations of InheritanceProvider.
• An implementation of Lorenz's FieldTypeProvider, backed by a SourceSet (from
  Bombe-ASM).

0.3.2

This version of Lorenz addresses a critical bug in the TSRG writer, that causes invalid output (missing class names).

Lorenz is available through Maven Central.

0.3.1

The revision of Lorenz brings about no new features or bug-fixes, rather completes the effort begun in 0.3.0 to better document Lorenz.

Lorenz is available through Maven Central.

0.3.0

This version of Lorenz largely builds on the efforts of the previous release, in addition to some preparation for supporting the Enigma and JAM mapping formats.

Changes

• In accordance with the Java specification, I have corrected the naming of MethodDescriptor / MethodSignature (I had them the reverse way round).
• Model construction is now handled through a MappingSetModelFactory - allowing models to be used together
• Introduced an (unused) FieldSignature and FieldTypeProvider

Migrating from 0.2.0

Migration should be a fairly easy task, unless MethodDescriptor/MethodSignature was used. There have been no other breaking changes.

Lorenz is available through Maven Central.

0.2.0

This version of Lorenz begins to address some of the issues that present using it within Nocturne - which is largely driving the functionality I desire for 1.0.0.

Changes

• The API has been split from the implementation
• The parser classes have been renamed: *Parser -> *Processor, .parser. -> .reader.
• The two variants of SRG are now supported: CSRG and TSRG

Migrating from 0.1.0

Despite the far-reaching internal changes, migrating from 0.1.0 is a rather simple procedure.

• Model classes can no longer be constructed manually, instead the createMapping methods should be used. To create a MappingSet, MappingSet.create() should be used
• The reader class imports will need to be changes, to use the new package name and class names mentioned above

Lorenz is available through Maven Central.

0.1.0

This is the first production-ready release of Lorenz, despite it being a 'pre-release'. The API is not stable, and will be subject to lots of changes between now and 1.0.0.

Lorenz releases are available through Maven Central.