sidef-scripts/Encoding/variable_length_run_encoding_2.sf

#!/usr/bin/ruby

# Implementation of the Variable Length Run Encoding.

# Reference:
#   Data Compression (Summer 2023) - Lecture 5 - Basic Techniques
#   https://youtube.com/watch?v=TdFWb8mL5Gk

func VLR_encoding (String str) {

    var bitstream = []

    for c,v in (str.chars.run_length) {
        bitstream << c.ascii2bin
        if (v == 1) {
            bitstream << '0'
        }
        else {
            var t = v.as_bin
            bitstream << ('1'*(t.len-1) + '0' + t.substr(1))
        }
    }

    bitstream.join
}

func VLR_decoding (String bitstring) {

    var decoded = []
    var bits_fh = bitstring.open_r(:raw)

    while (!bits_fh.eof) {

        var s = 8.of { bits_fh.getc }.join
        var c = pack('B*', s)
        var bl = (^Inf -> first { bits_fh.getc != '1' })

        decoded << c

        if (bl > 0) {
            decoded << c*(Num('1' + bl.of { bits_fh.getc }.join, 2)-1)
        }
    }

    decoded.join
}

var str = join('', 'a'*13, 'b'*14, 'c'*10, 'd'*3, 'e'*1, 'f'*1, 'g'*4)

var enc = VLR_encoding(str)
var dec = VLR_decoding(enc)

say enc
say dec

assert_eq(dec, str)

with (File(__FILE__).read(':raw')) {|str|
    var enc = VLR_encoding(str)
    assert(enc.len < 8*str.len)
    assert_eq(VLR_decoding(enc), str)
}

__END__
011000011110101011000101110110011000111110010011001001010110010100110011000110011111000
aaaaaaaaaaaaabbbbbbbbbbbbbbccccccccccdddefgggg