lz77_file_compression.sf

#!/usr/bin/ruby

# Encode and decode small files using LZ77 compression.

define(
    CHUNK_SIZE = (1 << 16),
    SIGNATURE  = ('LZ77' + 2.chr)
)

func lz77_compress (str) {

    var la = 0
    var compressed_data = []

    var prefix = ''
    var chars  = str.chars
    var end    = chars.end

    while (la <= end) {

        var n = 1
        var p = 0

        var token = chars[la]

        while ((n < 255) && (la+n <= end) && ((var tmp = prefix.index(token, p)) >= 0)) {
            p = tmp
            token += chars[la + n]
            ++n
        }

        --n
        compressed_data << (p, n, chars[la+n])
        la += n+1
        prefix += token
    }

    return pack('(SCa)*', compressed_data...)
}

# Compress file
func lz77_compress_file(File input, File output) {

    var in_fh  =  input.open('<:raw') || die "Can't open file <<#{input}>> for reading"
    var out_fh = output.open('>:raw') || die "Can't open file <<#{output}>> for writing"

    var header = SIGNATURE

    # Print the header
    out_fh.print(header)

    # Compress data
    while (in_fh.read(\var chunk, CHUNK_SIZE)) {
        out_fh.print(lz77_compress(chunk))
    }

    # Close the files
    in_fh.close
    out_fh.close
}

# Decompress file
func lz77_decompress_file(File input, File output) {

    var in_fh = input.open('<:raw') || die "Can't open file <<#{input}>> for reading"

    if (SIGNATURE.len.of { in_fh.getc }.join != SIGNATURE) {
        die "Not a LZ77 archive!\n"
    }

    var out_fh = output.open('>:raw') || die "Can't open file <<#{output}>> for writing"

    var chunk = ''
    while (in_fh.read(\var data, 4*CHUNK_SIZE)) {
        ['(SCa)*'.unpack(data)].each_slice(3, {|s,l,c|

            chunk += (chunk.substr(s, l) + c)

            if (chunk.len >= CHUNK_SIZE) {
                out_fh.print(chunk)
                chunk = ''
            }
        })
    }

    out_fh.print(chunk)

    in_fh.close
    out_fh.close
}

ARGV.getopt!('d', \var decode)

var file = File(ARGV.shift) || do {
    say "usage: #{File(__MAIN__).basename} [-d] [input file]"
    Sys.exit(2)
}

if (decode || file.match(/\.lz77\.enc\z/)) {
    lz77_decompress_file(file, File("output.lz77.dec"))
}
else {
    lz77_compress_file(file, File("output.lz77.enc"))
}