19

and vice versa.

I am running a RedHat if relevant.

Fermat's Little Student
  • 555
  • 1
  • 3
  • 12
  • 6
    What type of binary file? – Stéphane Chazelas Oct 29 '15 at 16:47
  • 2
    In any case, for any file format, I'm sure you could write a one-liner to convert its endianness using Perl and [pack](http://perldoc.perl.org/functions/pack.html) / [unpack](http://perldoc.perl.org/functions/unpack.html). For some formats, it'll just be a longer line than for others. ;-) – Ilmari Karonen Oct 29 '15 at 21:13

4 Answers4

38

You can byteswap with dd. Is that sufficent? If not, please update your question to give an example of an input file and the expected outfile.

echo hello >infile
dd conv=swab <infile >outfile

hex infile
   0000 68 65 6c 6c 6f 0a                                 hello.
hex outfile
   0000 65 68 6c 6c 0a 6f                                 ehll.o
mikeserv
  • 57,448
  • 9
  • 113
  • 229
roaima
  • 107,089
  • 14
  • 139
  • 261
35

You cannot do this because for such a conversion, you need to know the meaning of the binary content.

If e.g. there is a string inside a binary file it must not be converted and a 4 byte integer may need different treatment than a two byte integer.

In other words, for a byte order conversion, you need a data type description.

schily
  • 18,806
  • 5
  • 38
  • 60
  • 7
    I was just to say that. it is not as simple as just swapping every pair of bytes. for example a 4 byte integer 01 02 03 04 in bigendian could look like this in little endian 04 02 03 01 NOT 02 01 04 03 https://www.cs.umd.edu/class/sum2003/cmsc311/Notes/Data/endian.html – Rob Oct 29 '15 at 16:36
  • 1
    Agreed, endianness makes no sense on a raw, typeless byte stream... this should be the accepted answer if OP does not edit his question – Thomas Oct 30 '15 at 11:02
  • [citation needed] - AFAIK, unless an application explicitly preserves some byte order and not others, then it is usual for bytes or words to be consistently ordered throughout. – Dennis Williamson Oct 30 '15 at 15:52
  • 1
    @DennisWilliamson: A UTF-8 or ASCII string does not have endianness. UTF-8 specifies byte order while ASCII is 7 bit and doesn't have multi-byte tokens. A UTF-16 string does have endianness and may or may not have a BOM to mark that endianness. A Unix binary which is intended to interact with Windows or Java might have both types of string, which means you have to distinguish them when converting. – Kevin Oct 30 '15 at 18:42
  • I came here looking for a program where you give it the data description and the data and it converts it. Thus far I have not found such a program but that does not mean it cannot be done. – Levi Morrison Aug 10 '17 at 16:37
6

In order to change file endianess, assuming word (32-bit) size, this 1 liner should work for you:

hexdump -v -e '1/4 "%08x"' -e '"\n"' input_file | xxd -r -p > output_file
hesham_EE
  • 196
  • 1
  • 2
2

If you don't care about file contents and just want to swap bytes, then try endconv. It is just a wrapper around standard byte conversion functions, so it supports conversion by 2, 4 and 8 byte long integers. It's not one liner though because it is separate program.