Cody Boone Ferguson
1ac055fa85
In most cases the word was changed to 'obfuscation' as it was about
obfuscation, not deobfuscation. In some cases like 1995/leo since the
file was (in 1995/leo) about a secret switch it was changed to
'secret.md' (and 'secret.html') but there were others like that as well.
One entry had the words 'spoiling the size rule' but this had been
changed to 'deobfuscating the size rule' which was obviously wrong.
In the latter YYYY/README.md files the judges had written after some
entries (brief summaries) '...
...We'll stop spouting spoilers now. Have fun exploring all the
entries!
but that was changed to:
...We'll stop spouting deobfuscation information now. Have fun
exploring all the entries!
which wasn't really about deobfuscation but just a brief summary of what
the entries did.
And in the most problematic case (at least for me :-), 2018/ferguson) I
had put a joke in:
... but it's the only Easter egg I won't spoil
but it was changed to 'won't deobfuscate' which not only is not what it
is but it spoilt the joke.
In some cases the word was changed back to 'SPOILER' (or similar) but in
most cases not. I believe it should be better now. There are some
caveats, however, as below.
Although make find_missing_links is okay I do not remember if this will
catch links that refer to '%%REPO_URL%%' or other pandoc tokens. This
means I might have to go back to some earlier years to verify links are
okay. That should not take that much time I hope.
It might also be that some wording can still be improved some but it is
hoped this will be discovered when checking links and looking over the
manifest (which obviously changed).
4.7 KiB
To build:
make
To use:
./prog [PIN] < input.txt > output.c
Try:
./try.sh
The author provided the file cppp.cc which can preprocess the generated output to make them more acceptable to typical compilers. To build:
make cppp
See the author's remarks for more details.
Judges' remarks:
You might think of this entry as code obfuscater with a code. If you use the wrong code, you're gonna have a bad time. What is compiling this entry about? Having a good .. time!
This winning entry also comes with a JavaScript time-lapse obfuscation that shows how this entry was written.
Author's remarks:
Synopsis
echo "Hello world" | ./prog 12345 > encoded.c
gcc -DPIN=12345 encoded.c -o ./decoded && ./decoded
Details
Mile is a C code generator. Run without arguments for the help message below:
usage: [PIN] < input.txt > output.c
where [PIN] is an integer in the range of 1..65535
When run with an encoding PIN (first command line argument), Mile will
read bytes from stdin and generate C code on stdout. Example:
echo "Hello, world!" | ./prog 56789 > encoded.c
This generated code will reproduce the same input bytes on stdout when
executed, if it was compiled with the same encoding PIN.
gcc -DPIN=56789 encoded.c -o decoded
The generated code reconstructs all data at compile time by seeding a
random number generator with the given PIN to reproduce the input bytes,
and simply does one fwrite(3) call at run time to print those bytes. Not
specifying a PIN is a compile time error. Specify the wrong PIN and
output code will still compile, but will likely produce garbage instead of
the input bytes. If the original input was empty then any PIN works, but
you might get a compile time warning complaining about empty arrays.
Because all data is reconstructed at compile time, expect the compile time to be excruciatingly long if you try to encode too much data. Code generated by encoding a haiku will probably compile in seconds, a limerick may require tens of seconds, a sonnet will take minutes. Give Mile an epic poem and you can probably write a faster preprocessor before GCC is done (Clang probably already failed long before then).
I see that there is a lack of preprocessor abuse in recent years, probably because programming in preprocessor is too tedious. Mile is meant to help with that by producing C code filled with at least 98% preprocessor lines for any input. Now you too can enjoy long compile times with minimal effort.
Features
-
Output is randomized using system clock, so if you wait a second between runs, you should get a different generated code even for the same encoding PIN.
-
Customized bytecode encoding of preprocessor boilerplates, mostly for compression but should also make reverse engineering more fun.
-
CRC32 of the source code is embedded in the source code itself.
-
Code layout shaped to resemble "Mile" from "Didn't I Say to Make My Abilities Average in the Next Life?!". Mile is known to be able to talk to nanomachines directly, which is exactly what programming in preprocessor feels like.
Compatibility
Mile has been verified to work in these environments (also, compiles without warnings on all of these):
- gcc 9.2.0 on Cygwin
- gcc 8.3.0 on Linux
- gcc 7.4.0 on Cygwin
- gcc 6.1.0 on JSLinux
- gcc 5.4.0 on Linux
- clang 8.0.1 on Linux
- clang 8.0.1 on Cygwin
- tcc 0.9.27 on Linux
- tcc 0.9.25 on JSLinux
Output of Mile is a bit more taxing, and only works with a subset of the above:
- gcc: works just fine when using 8.3.0 and below, but may trigger warning for large files. gcc 9.2.0 and above may crash with large files.
- clang: works fine for small files, fails on large files due to preprocessor bug.
- tcc: works just fine on Linux. Fails silently on JSLinux.
For people eager to encode large files, an extra utility has been packaged with this entry to preprocess the generated output to make them more acceptable to typical compilers. View the cppp.cc source and then compile it with a c++ compiler of your choice.