Add preview of 1984-2019
The 2020 winners are hinted it, but the source code is not there,
nor are the hints. Someone could look into the HTML and
see what was commented out, but that is not a problem because
WHO won is known and the name of their AWARD is known already.
Moreover a number of the winners have talked about what their
entirs do, so the commented out spoiler text is not a bit problem.
One will find HTML commented out with:
<!-- XXX remove when 2020 is ready
2020 links here
XXX remove when 2020 is ready -->
The "XXX remove when 2020 is ready" string is found at the
start and end comment blocks. Again, soneone could read
what was commented out, but that is OK since most of that
information is known or can be known. The following files
have such HTML comment blocks in them.
./winners.html
./years-spoiler.html
./years.html
In addition, ./all/summary.txt was modified to not
include the 2020 entries for the momemt.
The 2020 winner code AND write-ups are NOT part of this set.
p.s. The all/all.tar.bz2 does not contain the 2020 source code (yet?).
This commit is contained in:
Landon Curt Noll committed
1 parent
28efc67f5d
commit
c0663537cb
1683 files changed
+220064
No files matched your search
@@ -0,0 +1,117 @@
|
||||
#!/usr/bin/env make
|
||||
#
|
||||
# 2011 makefile
|
||||
#
|
||||
# Copyright (C) 2011, Landon Curt Noll, Simon Cooper, and Leonid A. Broukhis.
|
||||
#
|
||||
# Creative Commons Attribution-ShareAlike 3.0 Unported License.
|
||||
# See: http://creativecommons.org/licenses/by-sa/3.0/
|
||||
|
||||
|
||||
################
|
||||
# tool locations
|
||||
################
|
||||
#
|
||||
SHELL= /bin/bash
|
||||
CP= cp
|
||||
CPP= cpp
|
||||
GUNZIP= gunzip
|
||||
LD= ld
|
||||
MAKE= make
|
||||
RM= rm
|
||||
SED= sed
|
||||
TAR= tar
|
||||
TRUE= true
|
||||
|
||||
# optimization
|
||||
#
|
||||
# Most compiles will safely use -O2. Some can use only -O1 or -O.
|
||||
# A few compilers have broken optimizers or this entry make break
|
||||
# under those buggy optimizers and thus you may not want anything.
|
||||
#
|
||||
OPT= -O2
|
||||
|
||||
# bitness
|
||||
#
|
||||
# Some entries require 32-bitness,
|
||||
# other entries require 64-bitess.
|
||||
# By default we assume nothing.
|
||||
#
|
||||
ARCH=
|
||||
|
||||
# default flags for ANSI C compilation
|
||||
#
|
||||
CFLAGS= -Wall -W -ansi ${ARCH} ${OPT}
|
||||
|
||||
# ANSI compiler
|
||||
#
|
||||
# Set CC to the name of your ANSI compiler.
|
||||
#
|
||||
CC= cc
|
||||
|
||||
|
||||
##############################
|
||||
# Special flags for this entry
|
||||
##############################
|
||||
#
|
||||
ENTRY= fredriksson
|
||||
DATA=
|
||||
|
||||
|
||||
#################
|
||||
# build the entry
|
||||
#################
|
||||
#
|
||||
all: ${ENTRY} ${DATA}
|
||||
@${TRUE}
|
||||
|
||||
${ENTRY}: ${ENTRY}.c
|
||||
${CC} ${CFLAGS} ${ENTRY}.c -o ${ENTRY}
|
||||
|
||||
# alternative executable
|
||||
#
|
||||
alt:
|
||||
@${TRUE}
|
||||
|
||||
|
||||
###############
|
||||
# utility rules
|
||||
###############
|
||||
#
|
||||
everything: all alt
|
||||
|
||||
clean:
|
||||
${RM} -f ${ENTRY}.o ag.c
|
||||
|
||||
clobber: clean
|
||||
${RM} -f ${ENTRY} ag
|
||||
|
||||
nuke: clobber
|
||||
@echo gnab gib!
|
||||
|
||||
install:
|
||||
@echo "Surely you're joking Mr. Feynman!"
|
||||
|
||||
# backwards compatibility
|
||||
#
|
||||
build: all
|
||||
|
||||
|
||||
##################
|
||||
# 133t hacker rulz
|
||||
##################
|
||||
#
|
||||
love:
|
||||
@echo 'not war?'
|
||||
|
||||
haste:
|
||||
$(MAKE) waste
|
||||
|
||||
waste:
|
||||
@echo 'waste'
|
||||
|
||||
easter_egg:
|
||||
@echo you expected to mis-understand this $${RANDOM} magic
|
||||
@echo chongo '<was here>' "/\\oo/\\"
|
||||
@echo Readers shall not be disallowed from being unable to partly misunderstand this final echo.
|
||||
@${TRUE}
|
||||
@@ -0,0 +1,21 @@
|
||||
#include<stdio.h>
|
||||
void exit(int);
|
||||
int(*Q)(int,char**),b['['+']']={'-'-'-'},i,j,s,t,D['C'*' '-'-'],z,system(char*);
|
||||
char O['C'*' '-'-'],p[]="@cs \"$ | ,& |& (&)x ) &,( ,(| ,( ,{ |,},; ,( ;,> ,( ,( ,;, );g cgt3odt - <~+<+- + ~ <&~<~+ <&+~+ ~ <=- = ( - !() c !(&( ! \n$P & m& )P )o )e')\")\"' ) w_ 'w)e ')x') ')q 'a_ )x'\"up)q &]v\") &t &t-k[]t\\.nivc -:l O CI:OnO lCl ={ -: =(- +[=+ =C =]\\.[- & - ~ -(|,g& -&- <- ( ~ -g()= Q| b) *,t, Q- ( - (\n& ~ & Q \n &\n,&g r& \n& \n ,&~\nt &=\n &n) rt)+=\n& (~ &k= &g(-&= (Q -( P ) \n )\n ) \"+\"' )z\"ujlpx)\"*bgQ Px )(\")x)\" P \"P\"'i x )\")m') axP \")\" Pv pP)e )O\"_\"apvP(t(c )d) \")\"pqpmpleP \"*te\" t\n \"] \"O_\" g c *P )\" =+ * r\n&r\nr \n! * (* (* + *. )x ')+) +)* ,\n )&* + ma +) 'j )d)' m)O )e)nc )m )n ) ')\"'jl _) _)_ )nps' )cix'iml )tit_eqp)m )rcnMm+wnhdl)esakn )\")ryPiam)O)v} no)zdOui =( \" '+'= ~(=(= (\n= ()=(@_& ~@' ( = '%; go. PO 'x )p)g <h )kh ) \"k\" M ai) <eu + =z q =x =eowe =v+kO= c= ! =dMe ij lmp c p _) c_O |tp=^ c sz^o ^sibnj^jis <^q^e aQ ^ thug^eiawyq=!=a<]= vhyh > '@ =\n '= &=,-\n=,=\n (& [ ()* , * ,*= )& (\n, \n , \n ,& ,*('l'c ) l \\ 's 'l' ci lcad eilm nq_dkv<=-^ (@, \n(& !\n=, ( , (/\n @ m h =*(.c. (&' ( =,(=(| (|(|(|\n&(,& [g(, (,\n tni^-,=(=(u t&(=)(~,=d(,( = ,sep wpg\n@\nan,\n&,* n a,$=,=|\n.( c'=~,~,=(,\n=(,[(o ad#( \n (d x=~,\n,='e, ( = '~),~=(, O=\n ( oc -~= ,\n =<, <=,(@='\\i|~ i~, =\n =i o i=-\n,i(=v t l t l * ,+ *+ *, p u=\n!=,\n(=,( e a*=, \n=\n[f(< n~('n i n,n \n,eu s( ,|&^,(=l op='&|,\n ,+\n^\n !,(= o, =/-=(/=%+ , e, (= ,()= | ] |OtvO)\n ) \n)k ')n\n' - (=( | - ( > - =(";
|
||||
FILE*f;
|
||||
int main(int _,char**v){
|
||||
_>(_^_)&&(t=_,i='C'*' '-'-'+~!!"",Q=main,Q(-!!"",v),i^=i,Q(~!!"",v),i=s^=s,Q(~-~!!"",v),i='C'*' '-'-',j=-~('('*'('),Q((-~!!""*~!!""),v),
|
||||
f=fopen(O,"w"),fprintf(f,O+~-(~-~!!""*~!!""),'C'*' '-'-'-((~!!""*~!!"")*''+-~-~!!""),O+((~!!""*~!!"")*''+-~-~!!"")),fclose(f),
|
||||
sprintf(p,O+(~-(~-~!!""*~!!"")<<!!"")),i=!!"",Q(~-(-~!!""*~!!""),v),exit(system(p)),'r')
|
||||
||
|
||||
_==~!_&&i>=(i^i)&&
|
||||
((p[i]==' '&&(p[i]=(-~i)[p])||p[i]=='@'&&(p[i]=' ')||p[i]=='$'&&(p[i]^=p[i])),i=~-i,Q(_,v))
|
||||
||
|
||||
_==~!!""&&i<'C'*' '-'-'&&(D[i]=~-(i[p][b]=-~i[p][b]),i=-~i,Q(_,v))
|
||||
||
|
||||
_==~-~!!""&&i<(!!""<<-~(~-~!!""*~!!""))&&(z=b[i],i[b]=s,s+=z,i=-~i,Q(_,v))
|
||||
||
|
||||
_==(-~!!""*~!!"")&&-~(i=~-i)&&(i[O]=p[j],j=D[j]+j[p][b],Q(_,v))
|
||||
||
|
||||
_==~-(-~!!""*~!!"")&&i<t&&(sprintf(p,"%s %s",p,i[v]),i=-~i,Q(_,v));
|
||||
}
|
||||
@@ -0,0 +1,315 @@
|
||||
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
|
||||
<html>
|
||||
<head>
|
||||
<meta name="GENERATOR" content="mkd2html 2.1.3 DL=DISCOUNT">
|
||||
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=utf-8">
|
||||
<link rel="stylesheet"
|
||||
type="text/css"
|
||||
href="../hint.css">
|
||||
<title>Most useful</title>
|
||||
</head>
|
||||
<body>
|
||||
<h1>Most useful</h1>
|
||||
|
||||
<p>Kimmo Fredriksson<br>
|
||||
<a href="mailto:kimmo.k.k.fredriksson@gmail.com">kimmo.k.k.fredriksson@gmail.com</a></p>
|
||||
|
||||
<h2>Judges' comments:</h2>
|
||||
|
||||
<h3>To build:</h3>
|
||||
|
||||
<pre><code>make fredriksson
|
||||
</code></pre>
|
||||
|
||||
<h3>To run:</h3>
|
||||
|
||||
<pre><code>./fredriksson [-icvtnk#] regexp < file
|
||||
</code></pre>
|
||||
|
||||
<h3>Try:</h3>
|
||||
|
||||
<pre><code>./fredriksson -k1 -i POOT < /etc/group
|
||||
ulimit -s unlimited
|
||||
./fredriksson -k2 -t -n bofuskate < /usr/share/dict/words
|
||||
</code></pre>
|
||||
|
||||
<h3>Selected Judges Remarks:</h3>
|
||||
|
||||
<p>The author of this entry was making sure that it does not win the
|
||||
“Best Short Program” award. On the outside, the program behaves
|
||||
as an approximate grep. On the inside, it is a Burrows-Wheeler
|
||||
transform decompressor that produces the source code of the actual
|
||||
obfuscated approximate grep program and calls it. As an exercise,
|
||||
try writing the compressor.</p>
|
||||
|
||||
<h2>Author’s comments:</h2>
|
||||
|
||||
<h1>Approximate grep</h1>
|
||||
|
||||
<p>Implements a variant of grep.</p>
|
||||
|
||||
<h2>Usage</h2>
|
||||
|
||||
<p>To search regexp from file, do:</p>
|
||||
|
||||
<pre><code>./fredriksson [-icvtnk#] regexp < file
|
||||
</code></pre>
|
||||
|
||||
<p>Pipes may not work, see the section Limitations.
|
||||
The options <code>-i -c -v</code> and <code>-n</code> are the classic ones:</p>
|
||||
|
||||
<ul>
|
||||
<li><p><code>-i</code> Ignore case.</p></li>
|
||||
<li><p><code>-c</code> Count only, prints the number of matches, but not the matching lines.</p></li>
|
||||
<li><p><code>-v</code> Invert match.</p></li>
|
||||
<li><p><code>-n</code> Prefixes each printed line by its line number (only if
|
||||
<code>-c</code> was not given).</p></li>
|
||||
</ul>
|
||||
|
||||
|
||||
<p>This version of grep recognizes two more options, not present in normal
|
||||
grep implementations:</p>
|
||||
|
||||
<ul>
|
||||
<li><p><code>-kN</code> Permit <code>N</code> insertions, deletions or mismatches of characters in the
|
||||
matches (aka edit-distance). E.g. “obfuscated” matches “obfuscation”
|
||||
with one deletion and two mismatches. Likewise, “obfuscate” and
|
||||
“oversimplify” match with 7 edit operations. The default is <code>-k0</code>
|
||||
(i.e. exact match).</p></li>
|
||||
<li><p><code>-t</code> Add local transpositions to the set of allowed edit-operations.
|
||||
That is, “ab” matches “ba” with one transposition (swap), and
|
||||
“obfuscated” matches “bofsuact” with three swaps and two insertions,
|
||||
i.e. use <code>-tk5</code> to find this.</p></li>
|
||||
</ul>
|
||||
|
||||
|
||||
<p>Of course, one could construct a standard regexp that matches the same
|
||||
patterns, but the problem is that such a regexp grows exponentially in
|
||||
size when the number of allowed edit operations is increased.</p>
|
||||
|
||||
<p>This version does not recognize all regular expressions, but the
|
||||
following are allowed:</p>
|
||||
|
||||
<ul>
|
||||
<li><p>Wild card: <code>.</code> (dot) matches any character.</p></li>
|
||||
<li><p>Bracket expression: list of charactres enclosed by <code>[</code> and <code>]</code>,
|
||||
matches any single character in the list.</p></li>
|
||||
<li><p>Range expression: two characters separated by hyphen and enclosed in
|
||||
brackets, matches any character that is lexicographically between the
|
||||
two characters (inclusive). E.g. <code>[a-d]</code> is the same as <code>[abcd]</code>.</p></li>
|
||||
</ul>
|
||||
|
||||
|
||||
<p>The special characters must be escaped, if they are to be taken
|
||||
literally (i.e. use e.g. <code>\.</code> to match period). You might want to protect
|
||||
these with <code>'</code>, i.e. use <code>'foo\[bar'</code> instead of <code>foo\[bar</code>. If you want to
|
||||
use the literal <code>-</code> in bracket expression, protect that too (<code>\-</code>),
|
||||
otherwise this is not needed. Note that the syntax differs from standard
|
||||
(?) grep a bit, i.e. in plain grep you can use <code>]</code> and <code>-</code> in bracket
|
||||
expressions by putting <code>]</code> as the first character, and putting <code>-</code> last.</p>
|
||||
|
||||
<h3>Other uses</h3>
|
||||
|
||||
<ul>
|
||||
<li><p>You can use this in place of cat. Just say</p>
|
||||
|
||||
<pre><code> ./fredriksson -k3 cat < file
|
||||
</code></pre></li>
|
||||
<li><p>Like with cat, you can append line numbers too:</p>
|
||||
|
||||
<pre><code> ./fredriksson -nk3 cat < file
|
||||
</code></pre></li>
|
||||
</ul>
|
||||
|
||||
|
||||
<h3>Other features</h3>
|
||||
|
||||
<ul>
|
||||
<li><p>It is assumed that the user knows how to feed correct input to the
|
||||
program. If undefined options are given, the result is also undefined.
|
||||
This is a feature.</p></li>
|
||||
<li><p>It may consume huge amounts of memory, and a stack overflow may occur.
|
||||
This results in Segmentation fault (or the equivalent of your favorite
|
||||
system, such as Bus error). This is a feature, that can be used to detect
|
||||
overly long lines in the input file, such as checking the lenght of one
|
||||
liners in IOCCC entries, just adjust your stack size to a suitable
|
||||
threshold, and search for “ioccc”, with option <code>-vk5</code>.</p></li>
|
||||
</ul>
|
||||
|
||||
|
||||
<h2>Limitations and remarks</h2>
|
||||
|
||||
<ul>
|
||||
<li><p>Using warning options (such as -Wall) when building gives a lot of warning,
|
||||
such as <code>suggest parentheses around '&&' within '||'</code>, <code>value computed is
|
||||
not used</code>, <code>implicit declaration of function 'putchar'</code>, <code>implicit
|
||||
declaration of function 'getchar'</code>, <code>control reaches end of non-void
|
||||
function</code> and <code>format not a string literal and no format arguments</code>.
|
||||
Please ignore them all, since they just try to tell you that the source
|
||||
is obfuscated.</p></li>
|
||||
<li><p>Only standard input is handled. And <em>only</em> if stdin comes from a file,
|
||||
i.e. pipes may or may not work. If you want to count only (<code>-c</code> option),
|
||||
then this shouldn’t be an issue.</p></li>
|
||||
<li><p>The total lenght of the (preprocessed) pattern is at most
|
||||
<code>sizeof(long)*CHAR_BIT</code>. That does not count things like bracket expressions
|
||||
or escape chars, that is, the pattern <code>'a[bcd]e\.f'</code> is 5 chars long, not
|
||||
10 chars. That does not mean that the matched substring in the standard
|
||||
input has this limit, as using the edit operations can make the string
|
||||
longer. However, no assumptions are made for <code>sizeof(long)</code>. In fact,
|
||||
you can easily use <code>long long</code> instead,</p></li>
|
||||
<li><p>Depending on the file, may require a lot of stack space. If the
|
||||
program segfaults (because of this), say</p>
|
||||
|
||||
<pre><code> ulimit -s unlimited
|
||||
</code></pre>
|
||||
|
||||
<p>cross your fingers, hope it is enough, and try again.</p></li>
|
||||
<li><p>Assumes C99 standard, because of the one very long line. Most C compilers
|
||||
handle this just fine even in C89 mode, though.</p></li>
|
||||
<li><p>The code should be reasonably portable, it assumes standard ASCII codes
|
||||
for each char, and that a C compiler is on $PATH; the latter is hard coded
|
||||
to be ‘gcc’, but any compiler would work, provided that it supports
|
||||
C99, or long lines as an extension. This code has been tested on:</p>
|
||||
|
||||
<ul>
|
||||
<li>GNU/Linux FC4, i386, gcc</li>
|
||||
<li>GNU/Linux FC7, x86_64, gcc 4.1.2, tcc 0.9.24 and icc 10.0</li>
|
||||
<li>GNU/Linux FC9, x86_64, gcc 4.?.?</li>
|
||||
<li>GNU/Linux CentOS 5, gcc 4.1.2</li>
|
||||
<li>GNU/Linux Ubuntu 11.04, gcc 4.5.2</li>
|
||||
<li>SunOS 5.8, UltraSPARC IIIi, gcc 4.1.2</li>
|
||||
<li>SunOS 5.9, UltraSPARC IIIi, gcc 4.1.1 and Sun ONE Studio 8 CC</li>
|
||||
<li>AIX 5.3, IBM eServer p5-550, gcc 3.3.3 and AIX CC Version 6</li>
|
||||
</ul>
|
||||
|
||||
|
||||
<p>(Btw, tcc has its roots in IOCCC…)</p></li>
|
||||
</ul>
|
||||
|
||||
|
||||
<h2>Obfuscations (a.k.a. spoilers…)</h2>
|
||||
|
||||
<ul>
|
||||
<li><p>The initialized char array includes the source code of the real program,
|
||||
as well as its name and commands to build and run it, all Burrows-Wheeler
|
||||
-transformed, then run-length-encoded with unary coding. The main program
|
||||
extracts and builds that, and then calls it. More precicely, the command
|
||||
to build is</p>
|
||||
|
||||
<pre><code> gcc -O3 ag.c -o ag
|
||||
</code></pre>
|
||||
|
||||
<p>To change that you need to edit the char array <code>p[]</code> and (possibly) two
|
||||
specially encoded constants, <code>-~('('*'(')</code> and <code>'C'*' '-'-'</code>.
|
||||
The extracted source (<code>ag.c</code>) is obfuscated as well.</p>
|
||||
|
||||
<p>(The original version also deleted the generated source and the
|
||||
corresponding binary after running it, but this feature was removed
|
||||
from the submitted entry.)</p></li>
|
||||
<li><p>The code demonstrates several useful programming paradigms, such as</p>
|
||||
|
||||
<ul>
|
||||
<li>recursion to remove all loops</li>
|
||||
<li>sub routines (<code>main()</code> contains several logical sub routines,
|
||||
selected by the first argument)</li>
|
||||
<li>function pointers</li>
|
||||
</ul>
|
||||
</li>
|
||||
<li><p>Algorithmically, it incorporates the following</p>
|
||||
|
||||
<ul>
|
||||
<li>(inverse) Burrows-Wheeler transform (used also e.g. in bzip2)</li>
|
||||
<li>data compression (really expansion :-), run-lenght-encoding,
|
||||
unary coding (this is why it is expansion rather than compression)</li>
|
||||
<li>approximate string matching</li>
|
||||
<li>dynamic programming to compute the edit distance</li>
|
||||
<li>parallel computation in sequential computers (by packing
|
||||
several objects into a single long, to speed up the dynamic
|
||||
programming)</li>
|
||||
<li>self extracting, compiling and running code</li>
|
||||
</ul>
|
||||
</li>
|
||||
<li><p>As a C program, it demonstrates that</p>
|
||||
|
||||
<ul>
|
||||
<li>a C subset without reserved words is complete enough,
|
||||
(excluding data type specific things (<code>int</code>, <code>char</code>, <code>void</code> …))</li>
|
||||
<li>only letters are needed (i.e. the digits 0-9 are not)</li>
|
||||
<li>only one statement is enough to code any program (plus
|
||||
variable definitions)</li>
|
||||
<li>white space has other uses than just indenting the code</li>
|
||||
</ul>
|
||||
</li>
|
||||
<li><p>It also shows the power of boolean logic and bitwise arithmetic.</p></li>
|
||||
<li><p>This is the first (?) entry in the history of IOCCC, that cannot be
|
||||
totally un-obfuscated by the author. This is due to the clever choice of
|
||||
the actual algorithm, that is obfucated by nature, and it is an open
|
||||
research problem whether a simpler algorithm exists, with the same time
|
||||
complexity.</p></li>
|
||||
<li><p>The main obfuscation comes from using only arithmetic and boolean
|
||||
(short-circuit) logic combined with recursion to remove all loops and
|
||||
if-then-else constructs (also, no ternary <code>?:</code> constructs are present).
|
||||
Two’s complement representation of integers is also abused. E.g. instead
|
||||
of writing <code>x++</code> or <code>x-1</code>, it is much better to write <code>x=-~x</code> and <code>~-x</code>.</p></li>
|
||||
<li><p>No numerical values, other than in the initialized char array, try</p>
|
||||
|
||||
<pre><code> ./fredriksson -ck0 [0-9] < fredriksson.c
|
||||
</code></pre>
|
||||
|
||||
<p>This should print 1, as there is a numer ‘3’ in the char array (only).
|
||||
You can also try</p>
|
||||
|
||||
<pre><code> ./fredriksson -ck0 [0-9] < ag.c
|
||||
</code></pre>
|
||||
|
||||
<p>after running fredriksson at least once (since this generates that
|
||||
<code>ag.c</code> file), and this should print 0.</p></li>
|
||||
<li><p>The source code is best viewed with cat (or using the program itself),
|
||||
instead of some editor. (This is a somewhat system dependent thing.)</p>
|
||||
|
||||
<p>One reason for this is that the code builds (some of the) constant
|
||||
values from string literal <code>"\b"</code>, except that the backspace <code>\b</code> is not
|
||||
escaped like that, the source uses the raw ascii value (010 octal) instead.
|
||||
Thus when viewing with cat or something, that backspace erases the leading
|
||||
quotation mark. Makes it appear uncompilable, and it is: if you copy-paste
|
||||
the cat output from terminal to some editor, save it, the result would not
|
||||
compile.</p>
|
||||
|
||||
<p>Some other constants are built from ASCII values of chars.</p>
|
||||
|
||||
<p>Besides of making the code hard to read, all this makes it also quite large.</p></li>
|
||||
<li><p>Running it through GNU indent or bcpp (C beautifiers) does not beautify
|
||||
it a bit. In fact, indent makes it only worse (at least with the default
|
||||
options). Running it through the preprocessor does basically nothing,
|
||||
since no <code>#define</code>s are used.</p></li>
|
||||
<li><p>Some <em>abuse of the rules</em>. The number of non-whitespace and <code>;</code>, <code>{</code>
|
||||
and <code>}</code> chars is below the limit, but most of the whitespace chars are
|
||||
in an <em>initialized character array</em>. That is, the number of chars in the
|
||||
program exceeds the 2048 limit. However, whitespace is whitespace and
|
||||
the rules did not specify that these must be in between the statements.
|
||||
A character is a whitespace if <code>isspace()</code> returns non-zero value. Even
|
||||
if deleting even one of those chars breaks the program! (This does not
|
||||
apply to the generated source <code>ag.c</code>.)</p></li>
|
||||
</ul>
|
||||
|
||||
|
||||
<hr />
|
||||
|
||||
<!--
|
||||
(c) Copyright 1984-2012, [Leo Broukhis, Simon Cooper, Landon Curt Noll][judges] - All rights reserved
|
||||
This work is licensed under a [Creative Commons Attribution-ShareAlike 3.0 Unported License][cc].
|
||||
|
||||
[judges]: http://www.ioccc.org/judges.html
|
||||
[cc]: http://creativecommons.org/licenses/by-sa/3.0/
|
||||
-->
|
||||
|
||||
<TABLE><TR>
|
||||
<TD><a rel="license" href="http://creativecommons.org/licenses/by-sa/3.0/"><img alt="Creative Commons License" style="border-width:0" src="http://i.creativecommons.org/l/by-sa/3.0/88x31.png" /></a></TD>
|
||||
<TD><P>© Copyright 1984-2012,
|
||||
<A HREF="/judges.html">Leo Broukhis, Simon Cooper, Landon Curt Noll</A>
|
||||
- All rights reserved<BR>
|
||||
This work is licensed under a <a rel="license" href="http://creativecommons.org/licenses/by-sa/3.0/">Creative Commons Attribution-ShareAlike 3.0 Unported License</a>.</P></TD>
|
||||
<TD> <!--<a href="http://validator.w3.org/check?uri=referer"><img src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01 Transitional" height="31" width="88"></a>--></TD>
|
||||
</TR></TABLE>
|
||||
</body>
|
||||
</html>
|
||||
@@ -0,0 +1,274 @@
|
||||
# Most useful
|
||||
|
||||
Kimmo Fredriksson
|
||||
<kimmo.k.k.fredriksson@gmail.com>
|
||||
|
||||
|
||||
## Judges' comments:
|
||||
### To build:
|
||||
|
||||
make fredriksson
|
||||
|
||||
### To run:
|
||||
|
||||
./fredriksson [-icvtnk#] regexp < file
|
||||
|
||||
### Try:
|
||||
|
||||
./fredriksson -k1 -i POOT < /etc/group
|
||||
ulimit -s unlimited
|
||||
./fredriksson -k2 -t -n bofuskate < /usr/share/dict/words
|
||||
|
||||
### Selected Judges Remarks:
|
||||
|
||||
The author of this entry was making sure that it does not win the
|
||||
"Best Short Program" award. On the outside, the program behaves
|
||||
as an approximate grep. On the inside, it is a Burrows-Wheeler
|
||||
transform decompressor that produces the source code of the actual
|
||||
obfuscated approximate grep program and calls it. As an exercise,
|
||||
try writing the compressor.
|
||||
|
||||
|
||||
|
||||
## Author's comments:
|
||||
# Approximate grep
|
||||
|
||||
Implements a variant of grep.
|
||||
|
||||
## Usage
|
||||
|
||||
To search regexp from file, do:
|
||||
|
||||
./fredriksson [-icvtnk#] regexp < file
|
||||
|
||||
Pipes may not work, see the section Limitations.
|
||||
The options `-i -c -v` and `-n` are the classic ones:
|
||||
|
||||
* `-i` Ignore case.
|
||||
|
||||
* `-c` Count only, prints the number of matches, but not the matching lines.
|
||||
|
||||
* `-v` Invert match.
|
||||
|
||||
* `-n` Prefixes each printed line by its line number (only if
|
||||
`-c` was not given).
|
||||
|
||||
This version of grep recognizes two more options, not present in normal
|
||||
grep implementations:
|
||||
|
||||
* `-kN` Permit `N` insertions, deletions or mismatches of characters in the
|
||||
matches (aka edit-distance). E.g. "obfuscated" matches "obfuscation"
|
||||
with one deletion and two mismatches. Likewise, "obfuscate" and
|
||||
"oversimplify" match with 7 edit operations. The default is `-k0`
|
||||
(i.e. exact match).
|
||||
|
||||
* `-t` Add local transpositions to the set of allowed edit-operations.
|
||||
That is, "ab" matches "ba" with one transposition (swap), and
|
||||
"obfuscated" matches "bofsuact" with three swaps and two insertions,
|
||||
i.e. use `-tk5` to find this.
|
||||
|
||||
Of course, one could construct a standard regexp that matches the same
|
||||
patterns, but the problem is that such a regexp grows exponentially in
|
||||
size when the number of allowed edit operations is increased.
|
||||
|
||||
This version does not recognize all regular expressions, but the
|
||||
following are allowed:
|
||||
|
||||
* Wild card: `.` (dot) matches any character.
|
||||
|
||||
* Bracket expression: list of charactres enclosed by `[` and `]`,
|
||||
matches any single character in the list.
|
||||
|
||||
* Range expression: two characters separated by hyphen and enclosed in
|
||||
brackets, matches any character that is lexicographically between the
|
||||
two characters (inclusive). E.g. `[a-d]` is the same as `[abcd]`.
|
||||
|
||||
The special characters must be escaped, if they are to be taken
|
||||
literally (i.e. use e.g. `\.` to match period). You might want to protect
|
||||
these with `'`, i.e. use `'foo\[bar'` instead of `foo\[bar`. If you want to
|
||||
use the literal `-` in bracket expression, protect that too (`\-`),
|
||||
otherwise this is not needed. Note that the syntax differs from standard
|
||||
(?) grep a bit, i.e. in plain grep you can use `]` and `-` in bracket
|
||||
expressions by putting `]` as the first character, and putting `-` last.
|
||||
|
||||
### Other uses
|
||||
|
||||
* You can use this in place of cat. Just say
|
||||
|
||||
./fredriksson -k3 cat < file
|
||||
|
||||
* Like with cat, you can append line numbers too:
|
||||
|
||||
./fredriksson -nk3 cat < file
|
||||
|
||||
### Other features
|
||||
|
||||
* It is assumed that the user knows how to feed correct input to the
|
||||
program. If undefined options are given, the result is also undefined.
|
||||
This is a feature.
|
||||
|
||||
* It may consume huge amounts of memory, and a stack overflow may occur.
|
||||
This results in Segmentation fault (or the equivalent of your favorite
|
||||
system, such as Bus error). This is a feature, that can be used to detect
|
||||
overly long lines in the input file, such as checking the lenght of one
|
||||
liners in IOCCC entries, just adjust your stack size to a suitable
|
||||
threshold, and search for "ioccc", with option `-vk5`.
|
||||
|
||||
## Limitations and remarks
|
||||
|
||||
* Using warning options (such as -Wall) when building gives a lot of warning,
|
||||
such as `suggest parentheses around '&&' within '||'`, `value computed is
|
||||
not used`, `implicit declaration of function 'putchar'`, `implicit
|
||||
declaration of function 'getchar'`, `control reaches end of non-void
|
||||
function` and `format not a string literal and no format arguments`.
|
||||
Please ignore them all, since they just try to tell you that the source
|
||||
is obfuscated.
|
||||
|
||||
* Only standard input is handled. And _only_ if stdin comes from a file,
|
||||
i.e. pipes may or may not work. If you want to count only (`-c` option),
|
||||
then this shouldn't be an issue.
|
||||
|
||||
* The total lenght of the (preprocessed) pattern is at most
|
||||
`sizeof(long)*CHAR_BIT`. That does not count things like bracket expressions
|
||||
or escape chars, that is, the pattern `'a[bcd]e\.f'` is 5 chars long, not
|
||||
10 chars. That does not mean that the matched substring in the standard
|
||||
input has this limit, as using the edit operations can make the string
|
||||
longer. However, no assumptions are made for `sizeof(long)`. In fact,
|
||||
you can easily use `long long` instead,
|
||||
|
||||
* Depending on the file, may require a lot of stack space. If the
|
||||
program segfaults (because of this), say
|
||||
|
||||
ulimit -s unlimited
|
||||
|
||||
cross your fingers, hope it is enough, and try again.
|
||||
|
||||
* Assumes C99 standard, because of the one very long line. Most C compilers
|
||||
handle this just fine even in C89 mode, though.
|
||||
|
||||
* The code should be reasonably portable, it assumes standard ASCII codes
|
||||
for each char, and that a C compiler is on $PATH; the latter is hard coded
|
||||
to be 'gcc', but any compiler would work, provided that it supports
|
||||
C99, or long lines as an extension. This code has been tested on:
|
||||
|
||||
- GNU/Linux FC4, i386, gcc
|
||||
- GNU/Linux FC7, x86_64, gcc 4.1.2, tcc 0.9.24 and icc 10.0
|
||||
- GNU/Linux FC9, x86_64, gcc 4.?.?
|
||||
- GNU/Linux CentOS 5, gcc 4.1.2
|
||||
- GNU/Linux Ubuntu 11.04, gcc 4.5.2
|
||||
- SunOS 5.8, UltraSPARC IIIi, gcc 4.1.2
|
||||
- SunOS 5.9, UltraSPARC IIIi, gcc 4.1.1 and Sun ONE Studio 8 CC
|
||||
- AIX 5.3, IBM eServer p5-550, gcc 3.3.3 and AIX CC Version 6
|
||||
|
||||
(Btw, tcc has its roots in IOCCC...)
|
||||
|
||||
## Obfuscations (a.k.a. spoilers...)
|
||||
|
||||
* The initialized char array includes the source code of the real program,
|
||||
as well as its name and commands to build and run it, all Burrows-Wheeler
|
||||
-transformed, then run-length-encoded with unary coding. The main program
|
||||
extracts and builds that, and then calls it. More precicely, the command
|
||||
to build is
|
||||
|
||||
gcc -O3 ag.c -o ag
|
||||
|
||||
To change that you need to edit the char array `p[]` and (possibly) two
|
||||
specially encoded constants, `-~('('*'(')` and `'C'*' '-'-'`.
|
||||
The extracted source (`ag.c`) is obfuscated as well.
|
||||
|
||||
(The original version also deleted the generated source and the
|
||||
corresponding binary after running it, but this feature was removed
|
||||
from the submitted entry.)
|
||||
|
||||
* The code demonstrates several useful programming paradigms, such as
|
||||
|
||||
- recursion to remove all loops
|
||||
- sub routines (`main()` contains several logical sub routines,
|
||||
selected by the first argument)
|
||||
- function pointers
|
||||
|
||||
* Algorithmically, it incorporates the following
|
||||
|
||||
- (inverse) Burrows-Wheeler transform (used also e.g. in bzip2)
|
||||
- data compression (really expansion :-), run-lenght-encoding,
|
||||
unary coding (this is why it is expansion rather than compression)
|
||||
- approximate string matching
|
||||
- dynamic programming to compute the edit distance
|
||||
- parallel computation in sequential computers (by packing
|
||||
several objects into a single long, to speed up the dynamic
|
||||
programming)
|
||||
- self extracting, compiling and running code
|
||||
|
||||
* As a C program, it demonstrates that
|
||||
|
||||
- a C subset without reserved words is complete enough,
|
||||
(excluding data type specific things (`int`, `char`, `void` ...))
|
||||
- only letters are needed (i.e. the digits 0-9 are not)
|
||||
- only one statement is enough to code any program (plus
|
||||
variable definitions)
|
||||
- white space has other uses than just indenting the code
|
||||
|
||||
* It also shows the power of boolean logic and bitwise arithmetic.
|
||||
|
||||
* This is the first (?) entry in the history of IOCCC, that cannot be
|
||||
totally un-obfuscated by the author. This is due to the clever choice of
|
||||
the actual algorithm, that is obfucated by nature, and it is an open
|
||||
research problem whether a simpler algorithm exists, with the same time
|
||||
complexity.
|
||||
|
||||
* The main obfuscation comes from using only arithmetic and boolean
|
||||
(short-circuit) logic combined with recursion to remove all loops and
|
||||
if-then-else constructs (also, no ternary `?:` constructs are present).
|
||||
Two's complement representation of integers is also abused. E.g. instead
|
||||
of writing `x++` or `x-1`, it is much better to write `x=-~x` and `~-x`.
|
||||
|
||||
* No numerical values, other than in the initialized char array, try
|
||||
|
||||
./fredriksson -ck0 [0-9] < fredriksson.c
|
||||
|
||||
This should print 1, as there is a numer '3' in the char array (only).
|
||||
You can also try
|
||||
|
||||
./fredriksson -ck0 [0-9] < ag.c
|
||||
|
||||
after running fredriksson at least once (since this generates that
|
||||
`ag.c` file), and this should print 0.
|
||||
|
||||
* The source code is best viewed with cat (or using the program itself),
|
||||
instead of some editor. (This is a somewhat system dependent thing.)
|
||||
|
||||
One reason for this is that the code builds (some of the) constant
|
||||
values from string literal `"\b"`, except that the backspace `\b` is not
|
||||
escaped like that, the source uses the raw ascii value (010 octal) instead.
|
||||
Thus when viewing with cat or something, that backspace erases the leading
|
||||
quotation mark. Makes it appear uncompilable, and it is: if you copy-paste
|
||||
the cat output from terminal to some editor, save it, the result would not
|
||||
compile.
|
||||
|
||||
Some other constants are built from ASCII values of chars.
|
||||
|
||||
Besides of making the code hard to read, all this makes it also quite large.
|
||||
|
||||
* Running it through GNU indent or bcpp (C beautifiers) does not beautify
|
||||
it a bit. In fact, indent makes it only worse (at least with the default
|
||||
options). Running it through the preprocessor does basically nothing,
|
||||
since no `#define`s are used.
|
||||
|
||||
* Some _abuse of the rules_. The number of non-whitespace and `;`, `{`
|
||||
and `}` chars is below the limit, but most of the whitespace chars are
|
||||
in an _initialized character array_. That is, the number of chars in the
|
||||
program exceeds the 2048 limit. However, whitespace is whitespace and
|
||||
the rules did not specify that these must be in between the statements.
|
||||
A character is a whitespace if `isspace()` returns non-zero value. Even
|
||||
if deleting even one of those chars breaks the program! (This does not
|
||||
apply to the generated source `ag.c`.)
|
||||
|
||||
|
||||
--------------------------------------------------------------------------------
|
||||
<!--
|
||||
(c) Copyright 1984-2012, [Leo Broukhis, Simon Cooper, Landon Curt Noll][judges] - All rights reserved
|
||||
This work is licensed under a [Creative Commons Attribution-ShareAlike 3.0 Unported License][cc].
|
||||
|
||||
[judges]: http://www.ioccc.org/judges.html
|
||||
[cc]: http://creativecommons.org/licenses/by-sa/3.0/
|
||||
-->
|
||||
Reference in new issue
Block a user