Method: HTTPI::Dime#initialize

Defined in:

lib/httpi/dime.rb

#initialize(body) ⇒ Dime

# File 'lib/httpi/dime.rb', line 9

def initialize(body)
  bytes = body.unpack('C*')

  while bytes.length > 0
    record = DimeRecord.new

    # Shift out bitfields for the first fields
    byte = bytes.shift
    record.version     = (byte >> 3) & 31         # 5 bits  DIME format version (always 1)
    record.first       = (byte >> 2) & 1          # 1 bit   Set if this is the first part in the message
    record.last        = (byte >> 1) & 1          # 1 bit   Set if this is the last part in the message
    record.chunked     = byte & 1                 # 1 bit   This file is broken into chunked parts
    record.type_format = (bytes.shift >> 4) & 15  # 4 bits  Type of file in the part (1 for binary data, 2 for XML)
                                                  # 4 bits  Reserved (skipped in the above command)

    # Fetch big-endian lengths
    lengths = [] # we can't use a hash since the order will be screwed in Ruby 1.8
    lengths << [:options, (bytes.shift << 8) | bytes.shift]                                             # 2 bytes   Length of the "options" field
    lengths << [:id,      (bytes.shift << 8) | bytes.shift]                                             # 2 bytes   Length of the "ID" or "name" field
    lengths << [:type,    (bytes.shift << 8) | bytes.shift]                                             # 2 bytes   Length of the "type" field
    lengths << [:data,    (bytes.shift << 24) | (bytes.shift << 16) | (bytes.shift << 8) | bytes.shift] # 4 bytes   Size of the included file

    # Read in padded data
    lengths.each do |attribute_set|
      attribute, length = attribute_set
      content = bytes.slice!(0, length).pack('C*')
      if attribute == :data && record.type_format == BINARY
        content = StringIO.new(content)
      end

      record.send "#{attribute.to_s}=", content

      bytes.slice!(0, 4 - (length & 3)) if (length & 3) != 0
    end

    self << record
  end
end