Update v2.2.0

• Added the ability to set the root compound's name, both via NBTWriter and NBTOutputStream
  • Thanks @GreenJon902! (#14)

Update v2.1.1

• Fixed NBTList.set() throwing UnsupportedOperationException (#10, thanks @PureGero)
• Throw NullPointerException instead of IllegalArgumentException when argument is null. Applies to:
  • NBTCompound:
    • get() (and all other getters for that class)
    • put()
    • putIfAbsent()
    • containsKey()
    • containsValue()
    • containsTag()

Update v2.1.0

• Partial test coverage
• Fix incorrect int[] decoding (#9)
• Clean-up NBTList and NBTCompound
  • Improved documentation
  • Change superclasses from ArrayList to AbstractList, and HashMap to AbstractMap respectively
• Clean-up TagType
  • Improved documentation
  • Deprecate getClazz() in favor of getRuntimeType(). Functionally, they are exactly the same.

Release v2.0.0

• Introduced the filtering API, giving users more fine-grained control over streamed NBT data. Based on MongoDB projections.
  • Filters are essentially just lists of tag names. The filter's mode (INCLUDE or EXCLUDE) determines how tags with those names are treated in filtered objects.
    • Inclusion filters, when applied to streamed compounds, ensure that those compounds only include tags whose names are explicitly added to the filter.
    • Exclusion filters allow filtered compounds to retain all tags except those whose names are added to the filter.
  • Just like NBTCompound.get(...), tag names can be referenced via dot-notation.
    ⠀
• Anywhere where tags are reference via dot-notation now allows tag names that literally have dots (.) in them to be escaped using backslashes.
  • For string literals, this means adding a double-backslash before any dots that are part of the tag's actual name (\\.).
    ⠀
• Removed dependencies on Lombok and JetBrains annotations.
  ⠀
• Fixed two bugs
  • Root compounds written via NBTOutputStream and NBTWriter didn't have END tags.
  • NBTReader and NBTWriter would automatically close streams that they didn't own.

Update v1.2.0

Minor optimization:

• Added support for interning tag names & values.
  • Methods that now support interning have done so via overloading, and the original methods will have interning disabled by default.
  • When deserializing lots of NBT data that follows the same structure, or uses similar string values, this can help save a significant amount of memory.
  • If deserialization is infrequent and with small data, there should be little-to-no difference in performance or memory consumption.

Update v1.1.1

Two quick changes-

• Wrap NBTOutputStream (and subsequently NBTWriter) output in an unnamed compound
  • This change is to match the new behavior of NBTInputStream
• Add containsTag(String, TagType) to NBTCompound
  • Provides the same functionality as contains(String) while also checking that the matching tag is the desired type

Update v1.1.0

• Added support for reading SNBT (aka "Mojangson") strings
  • This is the format Minecraft uses for NBT data in commands
• NBTReader and NBTInputStream now read from the root tag instead of the implicit one
  • This means most initial paths no longer need to be prefixed with a dot (e.g. .Level.xPos --> Level.xPos)
• Added hashCode() and equals() support for NBTCompound
• Fix extra TAG_End added to data by the NBTOutputStream

Update v1.0.0

• Use only primitive arrays for byte[], int[], and long[]
• Slightly optimize NBTInputStream#readByteArray()

Update v0.1.0

• Changed NBTReader --> NBTInputStream
  • NBTReader is now a utility class for common uses of NBTInputStream
• Added NBT serialization
  • Added NBTOutputStream for directly writing NBT data
  • Added NBTWriter with utilities for common uses of NBTOutputStream
    • Options for writing to base64, file, or OutputStream
• Added fromObject() method to TagType to get the appropriate NBT type for an object

Patch v0.0.2

• Fix NoClassDefFoundError caused by JAXB import
• Use empty quotes ("") in NBTCompound.toString() when tag name is empty