class CFPropertyList::Binary
Binary PList parser class
def array_to_binary(val)
def array_to_binary(val) saved_object_count = @written_object_count @written_object_count += 1 #@object_refs += val.value.size values = val.value.map { |v| v.to_binary(self) } bdata = Binary.type_bytes(0b1010, val.value.size) << Binary.pack_int_array_with_size(object_ref_size(@object_refs), values) @object_table[saved_object_count] = bdata saved_object_count end
def bool_to_binary(val)
def bool_to_binary(val) @object_table[@written_object_count] = val ? "\x9" : "\x8" # 0x9 is 1001, type indicator for true; 0x8 is 1000, type indicator for false @written_object_count += 1 @written_object_count - 1 end
def count_object_refs(object)
def count_object_refs(object) case object when CFArray contained_refs = 0 object.value.each do |element| if CFArray === element || CFDictionary === element contained_refs += count_object_refs(element) end end return object.value.size + contained_refs when CFDictionary contained_refs = 0 object.value.each_value do |value| if CFArray === value || CFDictionary === value contained_refs += count_object_refs(value) end end return object.value.keys.size * 2 + contained_refs else return 0 end end
def data_to_binary(val)
def data_to_binary(val) @object_table[@written_object_count] = (Binary.type_bytes(0b0100, val.bytesize) << val) @written_object_count += 1 @written_object_count - 1 end
def date_to_binary(val)
def date_to_binary(val) val = val.getutc.to_f - CFDate::DATE_DIFF_APPLE_UNIX # CFDate is a real, number of seconds since 01/01/2001 00:00:00 GMT @object_table[@written_object_count] = (Binary.type_bytes(0b0011, 3) << [val].pack("E").reverse) @written_object_count += 1 @written_object_count - 1 end
def dict_to_binary(val)
def dict_to_binary(val) saved_object_count = @written_object_count @written_object_count += 1 #@object_refs += val.value.keys.size * 2 keys_and_values = val.value.keys.map { |k| CFString.new(k).to_binary(self) } keys_and_values.concat(val.value.values.map { |v| v.to_binary(self) }) bdata = Binary.type_bytes(0b1101,val.value.size) << Binary.pack_int_array_with_size(object_ref_size(@object_refs), keys_and_values) @object_table[saved_object_count] = bdata return saved_object_count end
def int_to_binary(value)
def int_to_binary(value) nbytes = 0 nbytes = 1 if value > 0xFF # 1 byte integer nbytes += 1 if value > 0xFFFF # 4 byte integer nbytes += 1 if value > 0xFFFFFFFF # 8 byte integer nbytes = 3 if value < 0 # 8 byte integer, since signed Binary.type_bytes(0b0001, nbytes) << if nbytes < 3 [value].pack( if nbytes == 0 then "C" elsif nbytes == 1 then "n" else "N" end ) else # 64 bit signed integer; we need the higher and the lower 32 bit of the value high_word = value >> 32 low_word = value & 0xFFFFFFFF [high_word,low_word].pack("NN") end end
def load(opts)
def load(opts) @unique_table = {} @count_objects = 0 @object_refs = 0 @written_object_count = 0 @object_table = [] @object_ref_size = 0 @offsets = [] fd = nil if(opts.has_key?(:file)) fd = File.open(opts[:file],"rb") file = opts[:file] else fd = StringIO.new(opts[:data],"rb") file = "<string>" end # first, we read the trailer: 32 byte from the end fd.seek(-32,IO::SEEK_END) buff = fd.read(32) offset_size, object_ref_size, number_of_objects, top_object, table_offset = buff.unpack "x6CCx4Nx4Nx4N" # after that, get the offset table fd.seek(table_offset, IO::SEEK_SET) coded_offset_table = fd.read(number_of_objects * offset_size) raise CFFormatError.new("#{file}: Format error!") unless coded_offset_table.bytesize == number_of_objects * offset_size @count_objects = number_of_objects # decode offset table if(offset_size != 3) formats = ["","C*","n*","","N*"] @offsets = coded_offset_table.unpack(formats[offset_size]) else @offsets = coded_offset_table.unpack("C*").each_slice(3).map { |x,y,z| (x << 16) | (y << 8) | z } end @object_ref_size = object_ref_size val = read_binary_object_at(file,fd,top_object) fd.close val end
def num_to_binary(value)
def num_to_binary(value) @object_table[@written_object_count] = if value.is_a?(CFInteger) int_to_binary(value.value) else real_to_binary(value.value) end @written_object_count += 1 @written_object_count - 1 end
def object_ref_size object_refs
def object_ref_size object_refs Binary.bytes_needed(object_refs) end
def read_binary_array(fname,fd,length)
def read_binary_array(fname,fd,length) ary = [] # first: read object refs if(length != 0) buff = fd.read(length * @object_ref_size) objects = unpack_with_size(@object_ref_size, buff) #buff.unpack(@object_ref_size == 1 ? "C*" : "n*") # now: read objects 0.upto(length-1) do |i| object = read_binary_object_at(fname,fd,objects[i]) ary.push object end end CFArray.new(ary) end
def read_binary_data(fname,fd,length)
def read_binary_data(fname,fd,length) CFData.new(read_fd(fd, length), CFData::DATA_RAW) end
def read_binary_date(fname,fd,length)
def read_binary_date(fname,fd,length) raise CFFormatError.new("Date greater than 8 bytes: #{length}") if length > 3 nbytes = 1 << length buff = fd.read(nbytes) CFDate.new( case length when 0 then # 1 byte CFDate is an error raise CFFormatError.new("#{length+1} byte CFDate, error") when 1 then # 2 byte CFDate is an error raise CFFormatError.new("#{length+1} byte CFDate, error") when 2 then buff.reverse.unpack("e")[0] when 3 then buff.reverse.unpack("E")[0] end, CFDate::TIMESTAMP_APPLE ) end
def read_binary_dict(fname,fd,length)
def read_binary_dict(fname,fd,length) dict = {} # first: read keys if(length != 0) then buff = fd.read(length * @object_ref_size) keys = unpack_with_size(@object_ref_size, buff) # second: read object refs buff = fd.read(length * @object_ref_size) objects = unpack_with_size(@object_ref_size, buff) # read real keys and objects 0.upto(length-1) do |i| key = read_binary_object_at(fname,fd,keys[i]) object = read_binary_object_at(fname,fd,objects[i]) dict[key.value] = object end end CFDictionary.new(dict) end
def read_binary_int(fname,fd,length)
def read_binary_int(fname,fd,length) if length > 3 raise CFFormatError.new("Integer greater than 8 bytes: #{length}") end nbytes = 1 << length buff = fd.read(nbytes) CFInteger.new( case length when 0 then buff.unpack("C")[0] when 1 then buff.unpack("n")[0] when 2 then buff.unpack("N")[0] when 3 hiword,loword = buff.unpack("NN") if (hiword & 0x80000000) != 0 # 8 byte integers are always signed, and are negative when bit 63 is # set. Decoding into either a Fixnum or Bignum is tricky, however, # because the size of a Fixnum varies among systems, and Ruby # doesn't consider the number to be negative, and won't sign extend. -(2**63 - ((hiword & 0x7fffffff) << 32 | loword)) else hiword << 32 | loword end end ) end
def read_binary_null_type(length)
def read_binary_null_type(length) case length when 0 then 0 # null byte when 8 then CFBoolean.new(false) when 9 then CFBoolean.new(true) when 15 then 15 # fill type else raise CFFormatError.new("unknown null type: #{length}") end end
def read_binary_object(fname,fd)
Read an object type byte, decode it and delegate to the correct
def read_binary_object(fname,fd) # first: read the marker byte buff = fd.read(1) object_length = buff.unpack("C*") object_length = object_length[0] & 0xF buff = buff.unpack("H*") object_type = buff[0][0].chr if(object_type != "0" && object_length == 15) then object_length = read_binary_object(fname,fd) object_length = object_length.value end case object_type when '0' # null, false, true, fillbyte read_binary_null_type(object_length) when '1' # integer read_binary_int(fname,fd,object_length) when '2' # real read_binary_real(fname,fd,object_length) when '3' # date read_binary_date(fname,fd,object_length) when '4' # data read_binary_data(fname,fd,object_length) when '5' # byte string, usually utf8 encoded read_binary_string(fname,fd,object_length) when '6' # unicode string (utf16be) read_binary_unicode_string(fname,fd,object_length) when '8' CFUid.new(read_binary_int(fname, fd, object_length).value) when 'a' # array read_binary_array(fname,fd,object_length) when 'd' # dictionary read_binary_dict(fname,fd,object_length) end end
def read_binary_object_at(fname,fd,pos)
def read_binary_object_at(fname,fd,pos) position = @offsets[pos] fd.seek(position,IO::SEEK_SET) read_binary_object(fname,fd) end
def read_binary_real(fname,fd,length)
def read_binary_real(fname,fd,length) raise CFFormatError.new("Real greater than 8 bytes: #{length}") if length > 3 nbytes = 1 << length buff = fd.read(nbytes) CFReal.new( case length when 0 # 1 byte float? must be an error raise CFFormatError.new("got #{length+1} byte float, must be an error!") when 1 # 2 byte float? must be an error raise CFFormatError.new("got #{length+1} byte float, must be an error!") when 2 then buff.reverse.unpack("e")[0] when 3 then buff.reverse.unpack("E")[0] else fail "unexpected length: #{length}" end ) end
def read_binary_string(fname,fd,length)
def read_binary_string(fname,fd,length) buff = read_fd fd, length @unique_table[buff] = true unless @unique_table.has_key?(buff) CFString.new(buff) end
def read_binary_unicode_string(fname,fd,length)
def read_binary_unicode_string(fname,fd,length) # The problem is: we get the length of the string IN CHARACTERS; # since a char in UTF-16 can be 16 or 32 bit long, we don't really know # how long the string is in bytes buff = fd.read(2*length) @unique_table[buff] = true unless @unique_table.has_key?(buff) CFString.new(Binary.charset_convert(buff,"UTF-16BE","UTF-8")) end
def read_fd fd, length
def read_fd fd, length length > 0 ? fd.read(length) : "" end
def real_to_binary(val)
def real_to_binary(val) Binary.type_bytes(0b0010,3) << [val].pack("E").reverse end
def string_to_binary(val)
def string_to_binary(val) val = val.to_s @unique_table[val] ||= begin if !Binary.ascii_string?(val) val = Binary.charset_convert(val,"UTF-8","UTF-16BE") bdata = Binary.type_bytes(0b0110, Binary.charset_strlen(val,"UTF-16BE")) val.force_encoding("ASCII-8BIT") if val.respond_to?("encode") @object_table[@written_object_count] = bdata << val else bdata = Binary.type_bytes(0b0101,val.bytesize) @object_table[@written_object_count] = bdata << val end @written_object_count += 1 @written_object_count - 1 end end
def to_str(opts={})
def to_str(opts={}) @unique_table = {} @count_objects = 0 @object_refs = 0 @written_object_count = 0 @object_table = [] @offsets = [] binary_str = "bplist00" @object_refs = count_object_refs(opts[:root]) opts[:root].to_binary(self) next_offset = 8 offsets = @object_table.map do |object| offset = next_offset next_offset += object.bytesize offset end binary_str << @object_table.join table_offset = next_offset offset_size = Binary.bytes_needed(table_offset) if offset_size < 8 # Fast path: encode the entire offset array at once. binary_str << offsets.pack((%w(C n N N)[offset_size - 1]) + '*') else # Slow path: host may be little or big endian, must pack each offset # separately. offsets.each do |offset| binary_str << "#{Binary.pack_it_with_size(offset_size,offset)}" end end binary_str << [offset_size, object_ref_size(@object_refs)].pack("x6CC") binary_str << [@object_table.size].pack("x4N") binary_str << [0].pack("x4N") binary_str << [table_offset].pack("x4N") binary_str end
def uid_to_binary(value)
def uid_to_binary(value) nbytes = 0 nbytes = 1 if value > 0xFF # 1 byte integer nbytes += 1 if value > 0xFFFF # 4 byte integer nbytes += 1 if value > 0xFFFFFFFF # 8 byte integer nbytes = 3 if value < 0 # 8 byte integer, since signed @object_table[@written_object_count] = Binary.type_bytes(0b1000, nbytes) << if nbytes < 3 [value].pack( if nbytes == 0 then "C" elsif nbytes == 1 then "n" else "N" end ) else # 64 bit signed integer; we need the higher and the lower 32 bit of the value high_word = value >> 32 low_word = value & 0xFFFFFFFF [high_word,low_word].pack("NN") end @written_object_count += 1 @written_object_count - 1 end
def unpack_with_size(nbytes, buff)
def unpack_with_size(nbytes, buff) format = ["C*", "n*", "N*", "N*"][nbytes - 1]; if nbytes == 3 buff = "\0" + buff.scan(/.{1,3}/).join("\0") end return buff.unpack(format) end