gistfile1.rb

HASH_CHUNK_SIZE = 65536
UINT64_MAX = 2**64 - 1

require 'xmlrpc/client'
require 'time'
require 'pp'

def withOST
  server = XMLRPC::Client.new_from_uri 'http://api.opensubtitles.org/xml-rpc'
  login = server.call 'LogIn', '', '', 'en', 'OS Test User Agent'
  status = login['status']
  raise status unless status == '200 OK'
  token = login['token']
  begin
    yield server, token
  ensure
    server.call 'LogOut', token
  end
end

def log msg
  STDERR.puts "#{DateTime.now.rfc3339}: #{msg}"
end

def hashFile(file)
  # beginning of file
  file.seek(0, IO::SEEK_SET)
  buffer = file.sysread(HASH_CHUNK_SIZE)
  file.seek(-HASH_CHUNK_SIZE, IO::SEEK_END)
  buffer << file.sysread(HASH_CHUNK_SIZE)

  bufsize = buffer.size
  if bufsize != 2 * HASH_CHUNK_SIZE
    raise "Only read #{bufsize} bytes"
  end

  buffhash = buffer.unpack('Q*').reduce do |acc, v|
    (acc + v) & UINT64_MAX
  end

  return (file.pos + buffhash) & UINT64_MAX
end

def scrobble(files)
  md = Hash.new do
    {:filenames => []}
  end

  files.each do |name|
    hash = File.open(name, 'rb') { |f| '%08x' % hashFile(f) }
    md[hash][:filenames] << name
    # Added for #ruby question
    STDERR.puts "#{hash}: #{name}"
  end
  # Added for #ruby question
  STDERR.puts "md: #{md}"

  ## Output:
  # a2ce46cd62b81668: 1312.7128v1.pdf
  # fa3ed90f2ffcedba: Berkeley-Latency-Mar2012.pdf
  # bb960143a7334619: MIT-CSAIL-TR-2009-002.pdf
  # 311cdf88ad358bef: S0002-9947-1965-0170805-7.pdf
  # e39b083c3485094: acm-float-10.1.1.102.244.pdf
  # e925f184a114cecf: bitcoin.pdf
  # ca50f53cac1082d8: edmonds.pdf
  # 9a38c83de294a47f: gidra13asplos-naps.pdf
  # f6d2f7e5b3e8e844: karp.pdf
  # cd3200dd8af596ac: p761-thompson.pdf
  # f5cbf07550c62ab5: paper-reading.pdf
  # f798d8a6d059e58b: rcix-oopsla-2013.pdf
  # 593caea3394f5e61: re-writing.pdf
  # de53d1e1dfa988fd: shannon1948.pdf
  # ad78d8fa3dd06b69: thenightwatch.pdf
  # f358aa1753759610: turing-1936.pdf
  # md: {}

  res = withOST do |server, token|
    server.call 'CheckMovieHash2', token, md.keys
  end

  status = res['status']
  raise status unless status == '200 OK'
  
  res['data'].each do |hash, infos|
    md[hash][:infos] = infos
  end

  require 'awesome_print'
  ap md
end

scrobble(ARGV) if __FILE__ == $0