ioccc/2012/omoikane

To build:

    make

There is an alternate version that will read from the compiled file itself, in case you don't have a /dev/urandom file.

To use:

    ./nyaruko [seed.txt] < original.bin > output.c
    bash output.c > key.c
    perl output.c > data.c

    cat key.c data.c > output.c

    cc output.c -o output
    ./output > regenerated.bin

NOTE: if your OS does not have /dev/urandom then you should specify a seed file or use the alternate version which reads in the compiled binary itself.

Try:

    ./try.sh

The script try.sh will check if you have perl before trying to use the perl parts of the script.

Alternate code:

The alternate version will, if no file is specified, read in the compiled binary itself rather than trying to open /dev/urandom, which is the default for the original entry. This is useful if your system does not have a /dev/urandom file and you do not want to specify an extra file.

Alternate build:

    make alt

Alternate use:

Use nyaruko.alt as you would nyaruko above.

Alternate try:

For the first alternate version:

    ./try.alt.sh

Judges' remarks:

The judges have nothing to add that has not already been written about in the obfuscation.html (requires JavaScript) file! :-)

Author's remarks:

Usage

Nyaruko is a binary to text filter. Given some input on stdin, Nyaruko will produce C code that reproduces this input on stdout:

    ./nyaruko < original.bin > output.c
    gcc output.c -o output
    ./output > regenerated.bin

Output is encrypted, but both key and data are included in the output. To separate the key from the data, run these commands:

    bash output.c > key.c
    perl output.c > data.c

The key-less data.c still compiles, but produces a different message on stdout instead of the original input. This message is a hint to why the code is formatted the way it is.

To combine the key and data, concatenate them together in either order:

    cat key.c data.c > output.c
    cat data.c key.c > output.c

By default, Nyaruko generates a unique random key for every message, using /dev/urandom as the seed. If given an extra command line argument, Nyaruko will seed using that file instead of /dev/urandom:

    ./nyaruko seed.txt < input.bin > output.c

This makes the output key deterministic, allowing the same key to be shared across different files. On operating systems that do not have /dev/urandom, users should always specify this extra seed argument to avoid indeterministic keys, or else use the alternate version which will by default read the compiled alternate version itself, which might or might not be deterministic, depending on where and how it's used.

Features

Implementation details that make Nyaruko more obfuscated than usual programs:

• Nyaruko recycles variables and buffers to reduce internal state. The variable names are also carefully chosen for mixed-case madness.
• Nyaruko employs some preprocessor magic to share much of the same code and state between the encoder and decoder, and to increase occurrences of smileys ;)

Output code has these features:

• Fits nicely under 80 columns, and does not contain any trigraphs.
• Uses a fairly efficient encoding scheme, better than uuencode for files larger than ~13K, and better than base64 for files larger than 20K.
• Encryption with ISAAC, a cryptographically secure pseudorandom number generator.
• Trivial 3 language polyglot.

Code layout is meant to resemble Nyaruko, also known as Nyarlathotep, the Crawling Chaos. The most obvious thing to do with chaos is to make a random number generator, and the most obvious thing to do with a random number generator is to make one-time-pads for encryption.

Compatibility

Nyaruko has these environment dependencies:

• Requires ASCII character set.
• Assumes sizeof(unsigned int) == 4.
• Best viewed with tab stops set to 8 spaces.

Output code has the same dependencies, with the additional requirement that the compiler must support arbitrarily long string literals. Maximum input size that can be encoded while still producing standard-compliant output is ~276 bytes for C89, and ~3168 bytes for C99.

Nyaruko has been to verified to work with these compiler/OS combinations:

• gcc 4.4.5 on Linux (32bit and 64bit)
• clang 3.1 on Windows (Cygwin)
• gcc 4.5.3 on Windows (Cygwin)
• gcc 4.5.3 on Windows (MingW)
• gcc 4.3.5 on JSLinux
• tcc 0.9.25 on JSLinux

Note that on MingW, stdin and stdout are not opened in binary mode by default, this means Nyaruko may not faithfully reproduce files on MingW. As a workaround, a binmode_main.h is included, so it's possible to produce an encoder that's binary-clean like this:

    x86_64-w64-mingw32-gcc -include binmode_main.h nyaruko.c -o nyaruko.exe

The C code produced by the encoder will need to be compiled with the same -include binmode_main.h hack if you want its output to faithfully reproduce the original binary input.

NOTICE to those who wish for a greater challenge:

If you want a greater challenge, don't read any further: just try to understand the program via the source.

If you get stuck, come back and read below for additional hints and information.

Extra files

obfuscation.html - Contains full recording of how the code went from blank state to an obfuscated program, gzipped to fit under 1MB. Requires JavaScript.