Unsigned LEB128 Converter

leb128.rb

#!/usr/bin/env ruby

# Unsigned LEB128 Converter
# https://en.wikipedia.org/wiki/LEB128

def encode( num, with_hex=false)
  bs_n = num.to_s(2)
  len = bs_n.length
  # 0 padding for 7-bit
  padding_len = (len / 7 + ( len % 7 > 0 ? 1 : 0 )) * 7
  zero_padded = bs_n.rjust(padding_len,"0")
  # split into 7-bit groups
  groups = []
  (zero_padded.length/7).times {|i|
    high1bit = i==0 ? "0" : "1"
    bits = high1bit + zero_padded.slice!(0,7)
    groups.push(bits.to_i(2).to_s(16).rjust(2,'0'))
  }
  hex = groups.reverse.join
  # hex to bin
  if with_hex
    hex
  else
    [hex].pack("H*")
  end
end

def decode( b128, with_hex=false)
  # hex to bin
  if with_hex
    b128 = [b128].pack("H*")
  end
  # set BINARY
  b128.force_encoding('BINARY')
  # decode
  groups = []
  b128.each_char {|c|
    groups.push(c.ord.to_s(2).rjust(8,'0').slice(1,8))
    if 0 == c.ord & 0x80
      break
    end
  }
  groups.reverse.join.to_i(2)
end

def parse(bytes, offset)
  # set BINARY
  bytes = bytes.force_encoding('BINARY')
  bytes = bytes.slice( offset, bytes.length )
  groups = []
  bytes.each_char {|c|
    groups.push(c)
    if 0 == c.ord & 0x80
      break
    end
  }
  groups.join.force_encoding('BINARY')
end

# Examples
# encode: Integer to LEB128 Hex String
encode(0, true) # 00
encode(127, true) # 7f
encode(543210, true) # ea9321
encode(543210, true).encoding # <Encoding:US-ASCII>

# encode: Integer to LEB128 Binary
encode(0) # "\x00"
encode(127) # "\x7F"
encode(543210) # "\xEA\x93!"
encode(543210).encoding # <Encoding:ASCII-8BIT>

# decode: LEB128 Hex String to Integer
decode("00", true) # 0
decode("7f", true) # 127
decode("ea9321", true) # 543210

# decode: LEB128 Binary to Integer
decode("\x00") # 0
decode("\x7F") # 127
decode("\xEA\x93!") # 543210

# parse LEB128 by offset
parse("C\x00",1) # "\x00"
parse("C\x7F",1) # "\x7F"
parse("C\xEA\x93!",1) # "\xEA\x93!"
parse("C\xEA\x93!",1).encoding # <Encoding:ASCII-8BIT>