bingrep

#!/usr/bin/env ruby
require 'optparse'

# @param bytes [String] Byte sequence as needle.
# @param io [IO] IO object as haystack.
# @param max [Integer, nil] Stop search when `max` number of `bytes` are found.
# @return [Array<Integer>]
def search(bytes, io, max = nil)
  buf_size = bytes.size * 10
  buf = ''
  buf_tail = ''
  results = []
  loop do
    pos = io.pos
    buf = io.read(buf_size)
    break if buf.nil? || buf.size < bytes.size
    buf = buf_tail + buf
    pos -= buf_tail.size
    idx = 0
    loop do
      idx = buf.index(bytes, idx + 1)
      break if idx.nil?
      results << pos + idx
      return results if results.size == max
    end
    buf_tail = buf.byteslice(-bytes.size, bytes.size)
  end
  results
end

def optparse!(argv)
  r = {
    data_io: nil,
    key: nil,
    max_num: nil
  }

  key_as_file = false

  opt = OptionParser.new
  opt.banner = "Usage: #{opt.program_name} [options] <search-key> [<file>]"
  opt.on('-f', '--key-as-file') { key_as_file = true }
  opt.on('-n', '--max-num=0') { |v| r[:max_num] = v.to_i > 0 ? v.to_i : nil }
  opt.parse!(argv)

  if argv.size < 1
    $stderr.puts 'ERROR: Invalid length of arguments.'
    $stderr.puts
    $stderr.puts opt.help
    exit 1
  end

  r[:key] = key_as_file ? IO.binread(argv[0]) : argv[0].dup.force_encoding('ASCII-8BIT')
  r[:data_io] = argv.size == 2 ? open(argv[1], 'rb') : $stdin

  r
end

def main
  opts = optparse! ARGV.dup
  search(opts[:key], opts[:data_io], opts[:max_num]).each { |idx| puts idx }
end

main