Amulet can't decompress LZ4-compressed chunks (Minecraft Java Edition 1.20.5)
msbmteam opened this issue · comments
Bug Report
In Minecraft Java Edition 1.20.5, some additional compression schemes are being added for dedicated servers, including LZ4 (compression type 4). Amulet cannot currently decompress LZ4-compressed chunks.
Current Behaviour:
ERROR - Error loading chunk 181 186 minecraft:overworld
Traceback (most recent call last):
File "amulet\api\wrapper\format_wrapper.py", line 442, in _safe_load
File "amulet\api\wrapper\format_wrapper.py", line 483, in _load_chunk
File "amulet\level\formats\anvil_world\format.py", line 666, in _get_raw_chunk_data
File "amulet\level\formats\anvil_world\dimension.py", line 66, in get_chunk_data_layers
File "amulet\level\formats\anvil_world\dimension.py", line 159, in get_chunk_data
File "amulet\level\formats\anvil_world\region.py", line 236, in get_data
File "amulet\level\formats\anvil_world\region.py", line 59, in _decompress
amulet.api.errors.ChunkLoadError: Invalid compression type 4
Expected behavior:
Amulet should be able to decompress chunks that were compressed with LZ4 block format.
Steps To Reproduce:
The LZ4 compression scheme is only accessible to dedicated Minecraft servers, so to reproduce this issue, you have to host a Minecraft server, not a single player world
- Prepare to run a Minecraft server with the latest 1.20.5 snapshot (24w07a)
- Before running the server, modify
server.propertiesto have the following line:region-file-compression=lz4 - Start the server
- Make sure a few chunks are generated before closing the server
- Access the .mca files in the server's world directory
Environment:
- OS: Windows 11 Home
- Minecraft Platform: Java
- Minecraft Version: 24w07a
- Amulet Version: V0.10.27
Attachments
Here are two versions of the same MCA file, but with different compression schemes:
The example you have sent starts with LZ4Block so I assume it is using block compression.
I think it is using this Java library
https://github.com/lz4/lz4-java/blob/master/src/java/net/jpountz/lz4/LZ4BlockInputStream.java#L129
Looks like this bit here is what we need.
https://github.com/lz4/lz4-java/blob/7c931bef32d179ec3d3286ee71638b23ebde3459/src/java/net/jpountz/lz4/LZ4BlockInputStream.java#L200
The java library adds header data to each block so we need to parse that before decompressing.
That works. I will hold off on releasing a new version because there is probably other things that need updating.