It sorts files on deflate streams and entropy(compression) and writes filenames on logs.
Code:
Log_deflate.log : files containing deflate streams
Log_c80.log : files having projected compression ratio above 80
Log_c50a.log : files having projected compression ratio above 50 below 80
Log_c50b.log : files having projected compression ratio below 50
you can then use arc.exe on the file. f.e
arc a -di -i2 -s; -ep1 -ed
-n@Log_deflate.log -wtemp\ -mpzlib+srep+lzma "%output%" "%Input%"
arc a -di -i2 -s; -ep1 -ed
-n@Log_c80.log -wtemp\ -mlz4 "%output%" "%Input%"
arc a -di -i2 -s; -ep1 -ed
-n@Log_c50a.log -wtemp\ -msrep+4x4:lzma "%output%" "%Input%"
arc a -di -i2 -s; -ep1 -ed
-n@Log_c50b.log -wtemp\ -msrep+lzma "%output%" "%Input%"
scanning is also very fast, but it only scans the first 64mb of file
uses Bulat Ziganshin's Data Smoke