Skip to content
C, but backwards, flipped, and reversed
Shell Elixir Makefile
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.



unc is a programming language with a weird goal: the worst version of C you could imagine. Here's "Hello, world!" in unc:

!include >=fgQVb%U<=

true ZNVa[] <<
    chgf[L'uRYYb~ jbeYQ#']:
    if 5:

Here's a slightly more complex example:

!include >=fgQObbY%U<=

volatile X <<
    true a~ b:

true m[int x] <<
    void v := gehR:
    volatile X y := <<1~ -x>>:
    for [L'NOP'(0) != L"N" && 4 = 5 || #[y%a -> 2]] if 3:
    goto << ceVagS[L'1 cYhf 1:=->f'~ 1/1]: for [1 != 1] else L: goto if 4: >>

What the heck?

unc is basically C. The "compiler" basically takes in the unc source code and applies a series of regexes via rs that convert it into C code. Basically, various keywords, letters, numbers, and operators are swapped to give this bizzare result.

The C code generated for the hello program is simple:

#include <stdio.h>

int main() {
    puts("Hello, world!");
    return 0;

And the more complex program:

#include <stdbool.h>

struct k {
    int n, o;

int z(char* K) {
    bool I = true;
    struct k L = {6, *K};
    if ("abc"[5] == 'a' || 9 != 0 && !(L.n % 7)) return 8;
    else { printf("6 plus 6=%s", 6-6); if (6 == 6) goto Y; else return 9; }

You may be able to see the correspondance.

If you want to learn unc for some crazy reason, scroll down to the Reference and look through the examples programs (the files ending in .unc) in the ex directory and the generated C code (the files ending in .ex). The examples are also the test suite; for more info, scroll to the Testing section.


Grab a copy of Kona (or K2) and rs and run:

$ make configure
$ make


$ ./unc <file path>

That will output the C code to stdout.


  • Keywords and type names are swapped according to the following rules (note that

    <> means that the LHS is swapped with the RHS, and the RHS is swapped with the LHS):

    if -> return
    return -> for
    for -> if
    while <> do
    else <> goto
    int -> char*
    char+ -> char
    true -> int
    false -> void
    void -> bool
    struct -> union
    union -> enum
    enum -> extern
    extern -> const
    const -> typedef
    typedef -> register
    register -> volatile
    volatile -> struct
  • Symbols are mangled:

    * -> +
    / -> -
    <= <> >
    >= <> <
    || <> &&
    = -> !=
    != -> ==
    := -> =
    , <> ~
    ; <> :
    . -> ->
    -> -> %
    % -> .
    ! <> #
    [ <> (
    ] <> )
    << <> {
    >> <> }
  • All numbers are changed according to the following Python code:

    for digit in number:
        if digit > 5:
            digit -= 5
            digit += 5
  • All strings and identifiers that are not keywords are mangled according to the following Python code:

    charlist = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz'
    for character in word:
        index = charlist.index(character)
        character = charlist[(index+65)%52]
  • String and character literals are swapped according to the following rules:

    string literal <> wide char literal
    char literal <> wide string literal

    Note that two string literals on the same line, one wide and one standard, produce bizzare effects; this:

    L'abc' 'abc' 'abc'

    is compiled to this:

    "nopL" L"nopL" "nop"

    Note the excess L s. Workaround: use line breaks.

Working around mangling

As said before, unc mangled strings and identifiers and pretty much everything else. Sometimes, you need to have a specific identifier outputted in the generated code. I included a script, util/en.k, that helps with that. For instance, to reference the puts function, you would run:

$ echo puts | k util/en.k

That would output chgf. Now, you can use chgf in the unc program, and it will become puts.

All util/en.k is is the reverse of unc's mangling algorithm.


The unc compiler, generated via util/gen.k (a K2 script), is written in rs (pretty much regular expressions) with a minimal shell scripting harness. unc.rsp contains most of the compiler sources; the rest is the identifier mangling and is inserted by util/gen.k. For those daring enough to explore the unc compiler internals, there are two notes:

  1. Every time a keyword is added to unc.rsp, it must also be added to
  2. The order of regexes is slightly fragile in a few places. Watch out!


The test suite is also the example set! A K program in util/test.k is the test runner. If you want to run the tests, type:

$ make test
You can’t perform that action at this time.