mcastat

A simple Linux¹ command-line utility that just (at this point) shows .mca region location header information.

It was actually written as a part of research, I wanted to be able to find an average world size depending on the amount on chunks, that is covered in Advanced section.

The program basically reads any 4KiB file and processes it like a .mca region file, because the Anvil format does not define any special format prefix.

Building and Installing

# Generated Makefile for your system
cmake CMakeLists.txt
# builds the project
make
# installs the program into your /usr/local/bin folder, so
# you always have an access to it and can remove the build folder
sudo cp mcastat /usr/local/bin/
# remove build files
make clean

Gathered statistics

Using this tool, I calculated that an average chunk in a nether world (seed 3625364) of 285 700 chunks would weight 6.159660494 KiB.

And an average chunk in an overworld (seed 6416191033863448866) of 1 620 529 chunks weights 10.6779 KiB.

All the data was gathered on a newly generated chunks, the data may and will slightly differ if you have players building lots of stuff.

Usage

The most simple command would be

mcastat world_nether/DIM-1/region/r.0.0.mca

With the following output (data collected using seed 3625364)

File: world_nether/DIM-1/region/r.0.0.mca
Chunks: 1024
Sectors: 1376
Correlation: 1.343750 # turned out to be rather useless if you ask me

Advanced

Getting total number of chunks in regions. I had 285 700 chunks in my nether world.

find region/ -name "*.mca" | xargs -l1 mcastat -qc | awk '{s+=$1} END {print s}'

Getting average correlation using gcalccmd cli calculator. This gave me 1.404533951. So on average 285 700 chunks will have 1.404533951 sectors using seed 3625364 (note that no players ever visited the nether). Knowing that each sector weights 4KiB we can predict that the total size of all compressed sectors would be (285,700 * 1.404533951) * 4096 = 1,568MiB (which is not the world size yet)

# getting correlation for all .mca files
find region/ -name "*.mca" | xargs -l1 mcastat -qp > corr.tmp
# sum up all correlations and divide by their total number
gcalccmd $(awk '{s+=$1} END {print s}' < corr.tmp)/$(wc -l < corr.tmp)
rm corr.tmp

But now let's find region size to chunks number correlation, and then it's average. I got 6.159660494 KiB per chunk on average -> 1,759,816 KiB in total, which really came close to the real size of 1,800,480 KiB.

for file in region/*.mca; do
  size=$(du $file | awk '{print $1}')
  chunks=$(mcastat -qc $file)
  gcalccmd $size/$chunks
done > size_corr.tmp
gcalccmd $(awk '{s+=$1} END {print s}' < size_corr.tmp)/$(wc -l < size_corr.tmp)
rm size_corr.tmp

For an overworld I simply ran the first described command to get the total number of all the chunks within the world and then simply divided the total weight of the world tht I got using du --max-depth=0 -bc world/region/ command by that chunks total number. 17 719 164 928 bytes / 1 620 529 chunks = 10934.18564431738 bytes or 10.6779 KiB.

Footnotes

1. I'm not going to build it on Windows or any other platform, so I cannot confirm that it works on any OS other than Linux. Even though the code does not use any Linux exclusive API, the path to a file depends on the system and, for example, using ~ with mcastat will transform into user's home directory, but it won't on any other OS. (As I'm not going to test and add this feature). Also, I don't think it would have any purpose in Windows :) ↩