Compressibility Tool

[thometal]

Hi,

exists a Tool which takes a file as input and prints number as output, which measures the compressibility of the input file?

It needs to working in linux. I thinking at a number between 0- 255 like it is implemented in zpaq.

Thank you in advance.

[Shelwien]

1. Take a codec you like (preferably latest paq8px for more precise results, or zstd for speed)
2. Compress the file
3. Calculate bpc as 8*archive_size/original_size
4. If bpc<8 then file is compressible

So actual "tool" can be implemented as a shell script.
There's really no other way to do it - even for paq8px there'd be some files which it can't compress,
but some other tool can.

At most you can optimize some compression algorithm for this task, by cutting off actual output, etc.
But any fast method would be imprecise - it won't do recompression, for example.

[thometal]

1. Take a codec you like (preferably latest paq8px for more precise results, or zstd for speed)
2. Compress the file
3. Calculate bpc as 8*archive_size/original_size
4. If bpc<8 then file is compressible

So actual "tool" can be implemented as a shell script.
There's really no other way to do it - even for paq8px there'd be some files which it can't compress,
but some other tool can.

At most you can optimize some compression algorithm for this task, by cutting off actual output, etc.
But any fast method would be imprecise - it won't do recompression, for example.

Thank you for proposing this solution, but iam looking in some variant where i dont need to compress the file first, like zpaq does it.

[Shelwien]

1. Any estimation method would be a compression algorithm, just without output
2. zpaq estimation is likely slower than zstd compression... I think it used some simple LZP model