chrislo-sourceclassifier 0.2.0
Sign up to get free protection for your applications and to get access to all the features.
- data/Manifest +325 -0
- data/README.textile +51 -0
- data/Rakefile +64 -0
- data/examples/example.rb +22 -0
- data/lib/sourceclassifier.rb +22 -0
- data/lib/trainer.rb +51 -0
- data/sourceclassifier.gemspec +35 -0
- data/sources/gcc/ackermann.gcc-2.gcc +93 -0
- data/sources/gcc/ackermann.gcc-3.gcc +20 -0
- data/sources/gcc/ary.gcc +40 -0
- data/sources/gcc/binarytrees.gcc +136 -0
- data/sources/gcc/binarytrees.gcc-2.gcc +140 -0
- data/sources/gcc/binarytrees.gcc-3.gcc +134 -0
- data/sources/gcc/chameneos.gcc +117 -0
- data/sources/gcc/chameneos.gcc-2.gcc +134 -0
- data/sources/gcc/chameneos.gcc-3.gcc +120 -0
- data/sources/gcc/chameneosredux.gcc +306 -0
- data/sources/gcc/echo.gcc +144 -0
- data/sources/gcc/except.gcc +52 -0
- data/sources/gcc/fannkuch.gcc +105 -0
- data/sources/gcc/fannkuch.gcc-2.gcc +94 -0
- data/sources/gcc/fasta.gcc +158 -0
- data/sources/gcc/fasta.gcc-2.gcc +132 -0
- data/sources/gcc/fasta.gcc-3.gcc +145 -0
- data/sources/gcc/fasta.gcc-4.gcc +141 -0
- data/sources/gcc/fibo.gcc +20 -0
- data/sources/gcc/harmonic.gcc +22 -0
- data/sources/gcc/hash.gcc +31 -0
- data/sources/gcc/hash2.gcc +38 -0
- data/sources/gcc/heapsort.gcc +72 -0
- data/sources/gcc/hello.gcc-2.gcc +12 -0
- data/sources/gcc/hello.gcc-3.gcc +20 -0
- data/sources/gcc/knucleotide.gcc +173 -0
- data/sources/gcc/knucleotide.gcc-2.gcc +168 -0
- data/sources/gcc/lists.gcc +226 -0
- data/sources/gcc/magicsquares.gcc +378 -0
- data/sources/gcc/mandelbrot.gcc +84 -0
- data/sources/gcc/mandelbrot.gcc-2.gcc +63 -0
- data/sources/gcc/mandelbrot.gcc-3.gcc +91 -0
- data/sources/gcc/mandelbrot.gcc-4.gcc +78 -0
- data/sources/gcc/matrix.gcc +66 -0
- data/sources/gcc/message.gcc-2.gcc +72 -0
- data/sources/gcc/message.gcc-3.gcc +64 -0
- data/sources/gcc/methcall.gcc +89 -0
- data/sources/gcc/moments.gcc +120 -0
- data/sources/gcc/nbody.gcc +142 -0
- data/sources/gcc/nbody.gcc-2.gcc +143 -0
- data/sources/gcc/nbody.gcc-3.gcc +142 -0
- data/sources/gcc/nestedloop.gcc +25 -0
- data/sources/gcc/nsieve.gcc +35 -0
- data/sources/gcc/nsievebits.gcc-2.gcc +37 -0
- data/sources/gcc/nsievebits.gcc-3.gcc +53 -0
- data/sources/gcc/objinst.gcc +95 -0
- data/sources/gcc/partialsums.gcc +86 -0
- data/sources/gcc/partialsums.gcc-3.gcc +81 -0
- data/sources/gcc/partialsums.gcc-4.gcc +68 -0
- data/sources/gcc/pidigits.gcc +105 -0
- data/sources/gcc/primes.gcc +75 -0
- data/sources/gcc/prodcons.gcc +86 -0
- data/sources/gcc/random.gcc +29 -0
- data/sources/gcc/raytracer.gcc +125 -0
- data/sources/gcc/raytracer.gcc-2.gcc +181 -0
- data/sources/gcc/recursive.gcc +68 -0
- data/sources/gcc/recursive.gcc-2.gcc +55 -0
- data/sources/gcc/regexdna.gcc-2.gcc +126 -0
- data/sources/gcc/regexmatch.gcc +136 -0
- data/sources/gcc/revcomp.gcc +85 -0
- data/sources/gcc/revcomp.gcc-2.gcc +88 -0
- data/sources/gcc/revcomp.gcc-4.gcc +71 -0
- data/sources/gcc/reversefile.gcc +103 -0
- data/sources/gcc/reversefile.gcc-2.gcc +56 -0
- data/sources/gcc/sieve.gcc +34 -0
- data/sources/gcc/spectralnorm.gcc +54 -0
- data/sources/gcc/spellcheck.gcc +72 -0
- data/sources/gcc/spellcheck.gcc-2.gcc +61 -0
- data/sources/gcc/strcat.gcc +38 -0
- data/sources/gcc/sumcol.gcc-2.gcc +98 -0
- data/sources/gcc/sumcol.gcc-3.gcc +22 -0
- data/sources/gcc/sumcol.gcc-4.gcc +18 -0
- data/sources/gcc/sumcol.gcc-5.gcc +32 -0
- data/sources/gcc/takfp.gcc +23 -0
- data/sources/gcc/tcp-stream.gcc +122 -0
- data/sources/gcc/tcpecho.gcc +122 -0
- data/sources/gcc/tcpecho.gcc-2.gcc +136 -0
- data/sources/gcc/tcprequest.gcc +122 -0
- data/sources/gcc/threadring.gcc +73 -0
- data/sources/gcc/wc.gcc +52 -0
- data/sources/gcc/wc.gcc-2.gcc +40 -0
- data/sources/gcc/wc.gcc-3.gcc +46 -0
- data/sources/gcc/wordfreq.gcc-2.gcc +85 -0
- data/sources/perl/ackermann.perl +28 -0
- data/sources/perl/ackermann.perl-2.perl +25 -0
- data/sources/perl/ackermann.perl-3.perl +20 -0
- data/sources/perl/ackermann.perl-4.perl +18 -0
- data/sources/perl/ary.perl +25 -0
- data/sources/perl/ary.perl-2.perl +23 -0
- data/sources/perl/binarytrees.perl +66 -0
- data/sources/perl/binarytrees.perl-2.perl +71 -0
- data/sources/perl/chameneos.perl +67 -0
- data/sources/perl/echo.perl +81 -0
- data/sources/perl/except.perl +73 -0
- data/sources/perl/fannkuch.perl +44 -0
- data/sources/perl/fannkuch.perl-2.perl +38 -0
- data/sources/perl/fasta.perl +112 -0
- data/sources/perl/fasta.perl-2.perl +135 -0
- data/sources/perl/fasta.perl-4.perl +122 -0
- data/sources/perl/fibo.perl +15 -0
- data/sources/perl/fibo.perl-2.perl +0 -0
- data/sources/perl/fibo.perl-3.perl +0 -0
- data/sources/perl/harmonic.perl +7 -0
- data/sources/perl/hash.perl +23 -0
- data/sources/perl/hash.perl-2.perl +17 -0
- data/sources/perl/hash.perl-3.perl +24 -0
- data/sources/perl/hash2.perl +16 -0
- data/sources/perl/heapsort.perl +65 -0
- data/sources/perl/heapsort.perl-2.perl +0 -0
- data/sources/perl/hello.perl +5 -0
- data/sources/perl/knucleotide.perl-2.perl +30 -0
- data/sources/perl/lists.perl +48 -0
- data/sources/perl/mandelbrot.perl-2.perl +32 -0
- data/sources/perl/matrix.perl +59 -0
- data/sources/perl/matrix.perl-2.perl +0 -0
- data/sources/perl/matrix.perl-3.perl +0 -0
- data/sources/perl/message.perl +27 -0
- data/sources/perl/methcall.perl +66 -0
- data/sources/perl/moments.perl +44 -0
- data/sources/perl/nbody.perl +108 -0
- data/sources/perl/nestedloop.perl +28 -0
- data/sources/perl/nsieve.perl-2.perl +41 -0
- data/sources/perl/nsieve.perl-4.perl +43 -0
- data/sources/perl/nsievebits.perl +37 -0
- data/sources/perl/objinst.perl +73 -0
- data/sources/perl/partialsums.perl-3.perl +31 -0
- data/sources/perl/pidigits.perl +52 -0
- data/sources/perl/pidigits.perl-2.perl +47 -0
- data/sources/perl/process.perl +50 -0
- data/sources/perl/prodcons.perl +47 -0
- data/sources/perl/random.perl-4.perl +17 -0
- data/sources/perl/recursive.perl-2.perl +57 -0
- data/sources/perl/regexdna.perl +48 -0
- data/sources/perl/regexdna.perl-2.perl +43 -0
- data/sources/perl/regexdna.perl-3.perl +50 -0
- data/sources/perl/regexdna.perl-4.perl +49 -0
- data/sources/perl/regexdna.perl-5.perl +42 -0
- data/sources/perl/regexdna.perl-6.perl +43 -0
- data/sources/perl/regexmatch.perl +35 -0
- data/sources/perl/revcomp.perl-2.perl +34 -0
- data/sources/perl/reversefile.perl +8 -0
- data/sources/perl/reversefile.perl-2.perl +0 -0
- data/sources/perl/reversefile.perl-3.perl +0 -0
- data/sources/perl/sieve.perl +23 -0
- data/sources/perl/spectralnorm.perl-2.perl +54 -0
- data/sources/perl/spellcheck.perl +24 -0
- data/sources/perl/strcat.perl +13 -0
- data/sources/perl/strcat.perl-2.perl +0 -0
- data/sources/perl/sumcol.perl +8 -0
- data/sources/perl/takfp.perl +23 -0
- data/sources/perl/takfp.perl-3.perl +20 -0
- data/sources/perl/tcpecho.perl +61 -0
- data/sources/perl/tcprequest.perl +61 -0
- data/sources/perl/tcpstream.perl +61 -0
- data/sources/perl/threadring.perl +55 -0
- data/sources/perl/threadring.perl-2.perl +43 -0
- data/sources/perl/wc.perl +20 -0
- data/sources/perl/wc.perl-2.perl +14 -0
- data/sources/perl/wordfreq.perl +22 -0
- data/sources/perl/wordfreq.perl-3.perl +0 -0
- data/sources/perl/wordfreq.perl3.perl +0 -0
- data/sources/python/ackermann.python +21 -0
- data/sources/python/ary.python +19 -0
- data/sources/python/binarytrees.python +39 -0
- data/sources/python/binarytrees.python-3.python +44 -0
- data/sources/python/chameneos.python-6.python +73 -0
- data/sources/python/chameneosredux.python +126 -0
- data/sources/python/chameneosredux.python-2.python +122 -0
- data/sources/python/dispatch.python +176 -0
- data/sources/python/dispatch.python-2.python +136 -0
- data/sources/python/echo.python +64 -0
- data/sources/python/except.python +62 -0
- data/sources/python/fannkuch.python +50 -0
- data/sources/python/fannkuch.python-2.python +54 -0
- data/sources/python/fasta.python-2.python +79 -0
- data/sources/python/fibo.python +17 -0
- data/sources/python/fibo.python-2.python +0 -0
- data/sources/python/fibo.python-3.python +0 -0
- data/sources/python/harmonic.python-2.python +9 -0
- data/sources/python/hash.python +21 -0
- data/sources/python/hash.python-2.python +0 -0
- data/sources/python/hash2.python +30 -0
- data/sources/python/heapsort.python-3.python +66 -0
- data/sources/python/hello.python +5 -0
- data/sources/python/implicitode.python +231 -0
- data/sources/python/knucleotide.python +55 -0
- data/sources/python/lists.python +44 -0
- data/sources/python/magicsquares.python +145 -0
- data/sources/python/mandelbrot.python +44 -0
- data/sources/python/mandelbrot.python-2.python +35 -0
- data/sources/python/mandelbrot.python-3.python +46 -0
- data/sources/python/matrix.python +34 -0
- data/sources/python/matrix.python-2.python +23 -0
- data/sources/python/message.python +24 -0
- data/sources/python/message.python-2.python +20 -0
- data/sources/python/message.python-3.python +19 -0
- data/sources/python/meteor.python +210 -0
- data/sources/python/meteor.python-2.python +192 -0
- data/sources/python/methcall.python +51 -0
- data/sources/python/moments.python +65 -0
- data/sources/python/nbody.python +123 -0
- data/sources/python/nbody.python-2.python +120 -0
- data/sources/python/nestedloop.python +24 -0
- data/sources/python/nsieve.python +27 -0
- data/sources/python/nsieve.python-2.python +23 -0
- data/sources/python/nsieve.python-4.python +25 -0
- data/sources/python/nsievebits.python +27 -0
- data/sources/python/nsievebits.python-2.python +43 -0
- data/sources/python/objinst.python +53 -0
- data/sources/python/partialsums.python +37 -0
- data/sources/python/partialsums.python-2.python +35 -0
- data/sources/python/partialsums.python-3.python +48 -0
- data/sources/python/pidigits.python +38 -0
- data/sources/python/pidigits.python-3.python +63 -0
- data/sources/python/pidigits.python-4.python +24 -0
- data/sources/python/process.python +51 -0
- data/sources/python/process.python-2.python +133 -0
- data/sources/python/prodcons.python +51 -0
- data/sources/python/prodcons.python-2.python +0 -0
- data/sources/python/random.python +27 -0
- data/sources/python/raytracer.python +203 -0
- data/sources/python/recursive.python +35 -0
- data/sources/python/regexdna.python +39 -0
- data/sources/python/regexdna.python-2.python +34 -0
- data/sources/python/regexmatch.python +36 -0
- data/sources/python/revcomp.python-3.python +31 -0
- data/sources/python/reversefile.python +13 -0
- data/sources/python/reversefile.python-2.python +0 -0
- data/sources/python/reversefile.python-3.python +0 -0
- data/sources/python/sieve.python +50 -0
- data/sources/python/spectralnorm.python-2.python +36 -0
- data/sources/python/spellcheck.python +17 -0
- data/sources/python/strcat.python +35 -0
- data/sources/python/strcat.python-2.python +0 -0
- data/sources/python/sumcol.python-2.python +0 -0
- data/sources/python/sumcol.python-3.python +0 -0
- data/sources/python/takfp.python +19 -0
- data/sources/python/tcpecho.python +67 -0
- data/sources/python/tcprequest.python +67 -0
- data/sources/python/tcpstream.python +67 -0
- data/sources/python/threadring.python +47 -0
- data/sources/python/threadring.python-2.python +40 -0
- data/sources/python/threadring.python-3.python +34 -0
- data/sources/python/wc.python-2.python +19 -0
- data/sources/python/wordfreq.python +43 -0
- data/sources/python/wordfreq.python-2.python +0 -0
- data/sources/python/wordfreq.python-3.python +28 -0
- data/sources/python/wordfreq.python-4.python +38 -0
- data/sources/python/wordfreq.python-5.python +39 -0
- data/sources/ruby/ackermann.ruby +17 -0
- data/sources/ruby/ackermann.ruby-5.ruby +153 -0
- data/sources/ruby/ary.ruby +22 -0
- data/sources/ruby/binarytrees.ruby-2.ruby +55 -0
- data/sources/ruby/chameneos.ruby-2.ruby +71 -0
- data/sources/ruby/dispatch.ruby +114 -0
- data/sources/ruby/echo.ruby +41 -0
- data/sources/ruby/except.ruby +61 -0
- data/sources/ruby/except.ruby-2.ruby +61 -0
- data/sources/ruby/fannkuch.ruby +42 -0
- data/sources/ruby/fasta.ruby +81 -0
- data/sources/ruby/fibo.ruby +15 -0
- data/sources/ruby/harmonic.ruby-2.ruby +15 -0
- data/sources/ruby/hash.ruby +19 -0
- data/sources/ruby/hash2.ruby +23 -0
- data/sources/ruby/heapsort.ruby +55 -0
- data/sources/ruby/hello.ruby +6 -0
- data/sources/ruby/knucleotide.ruby-2.ruby +44 -0
- data/sources/ruby/lists.ruby +46 -0
- data/sources/ruby/mandelbrot.ruby-3.ruby +63 -0
- data/sources/ruby/matrix.ruby +40 -0
- data/sources/ruby/matrix.ruby-2.ruby +30 -0
- data/sources/ruby/message.ruby +29 -0
- data/sources/ruby/message.ruby-2.ruby +24 -0
- data/sources/ruby/meteor.ruby +386 -0
- data/sources/ruby/meteor.ruby-2.ruby +561 -0
- data/sources/ruby/methcall.ruby +58 -0
- data/sources/ruby/methcall.ruby-2.ruby +54 -0
- data/sources/ruby/moments.ruby +64 -0
- data/sources/ruby/nbody.ruby-2.ruby +145 -0
- data/sources/ruby/nestedloop.ruby +22 -0
- data/sources/ruby/nsieve.ruby +36 -0
- data/sources/ruby/nsieve.ruby-2.ruby +25 -0
- data/sources/ruby/nsievebits.ruby-2.ruby +42 -0
- data/sources/ruby/objinst.ruby +58 -0
- data/sources/ruby/partialsums.ruby +39 -0
- data/sources/ruby/pidigits.ruby +92 -0
- data/sources/ruby/pidigits.ruby-2.ruby +109 -0
- data/sources/ruby/prodcons.ruby +41 -0
- data/sources/ruby/random.ruby +17 -0
- data/sources/ruby/recursive.ruby-2.ruby +53 -0
- data/sources/ruby/regexdna.ruby +32 -0
- data/sources/ruby/regexdna.ruby-2.ruby +38 -0
- data/sources/ruby/regexmatch.ruby +33 -0
- data/sources/ruby/revcomp.ruby +28 -0
- data/sources/ruby/reversefile.ruby +7 -0
- data/sources/ruby/sieve.ruby +30 -0
- data/sources/ruby/spectralnorm.ruby +48 -0
- data/sources/ruby/spellcheck.ruby +18 -0
- data/sources/ruby/spellcheck.ruby-2.ruby +0 -0
- data/sources/ruby/strcat.ruby +12 -0
- data/sources/ruby/strcat.ruby-2.ruby +12 -0
- data/sources/ruby/sumcol.ruby +12 -0
- data/sources/ruby/sumcol.ruby-2.ruby +5 -0
- data/sources/ruby/takfp.ruby +15 -0
- data/sources/ruby/tcpecho.ruby +45 -0
- data/sources/ruby/tcprequest.ruby +45 -0
- data/sources/ruby/tcpstream.ruby +45 -0
- data/sources/ruby/threadring.ruby +61 -0
- data/sources/ruby/threadring.ruby-2.ruby +33 -0
- data/sources/ruby/wc.ruby +15 -0
- data/sources/ruby/wordfreq.ruby +17 -0
- data/sources/ruby/wordfreq.ruby2.ruby +0 -0
- data/test/fixtures/sources/gcc/ackermann.gcc-2.gcc +93 -0
- data/test/fixtures/sources/python/ackermann.python +21 -0
- data/test/fixtures/sources/ruby/ackermann.ruby +17 -0
- data/test/test_source_classifier.rb +40 -0
- data/test/test_trainer.rb +34 -0
- data/trainer.bin +1193 -0
- metadata +393 -0
@@ -0,0 +1,12 @@
|
|
1
|
+
/* The Computer Language Benchmarks Game
|
2
|
+
* http://shootout.alioth.debian.org/
|
3
|
+
* contributed by Joe Tucek 2008-03-29
|
4
|
+
* Clearly printf is worse than fputs, which is worse than write...
|
5
|
+
*/
|
6
|
+
|
7
|
+
#include <unistd.h>
|
8
|
+
|
9
|
+
int main() {
|
10
|
+
write(1, "hello world\n", 12);
|
11
|
+
return(0);
|
12
|
+
}
|
@@ -0,0 +1,20 @@
|
|
1
|
+
/* The Computer Language Benchmarks Game
|
2
|
+
* http://shootout.alioth.debian.org/
|
3
|
+
* contributed by Joe Tucek 2008-03-31
|
4
|
+
*
|
5
|
+
* Tell GCC that we don't want atexit, we don't want to use the heap,
|
6
|
+
* and we really don't want anything. Can't even call write the "normal"
|
7
|
+
* way, because write() isn't linked in....
|
8
|
+
*
|
9
|
+
* Compile flags are picky for this. I used:
|
10
|
+
* gcc -pipe -Wall -O3 -fomit-frame-pointer -march=pentium4 -ffreestanding -nostartfiles -s -static -o start3 start3.c
|
11
|
+
*/
|
12
|
+
|
13
|
+
#include <sys/syscall.h>
|
14
|
+
#include <unistd.h>
|
15
|
+
|
16
|
+
int _start() {
|
17
|
+
syscall(__NR_write, 1, "hello world\n", 12);
|
18
|
+
syscall(__NR_exit, 0);
|
19
|
+
return(0);
|
20
|
+
}
|
@@ -0,0 +1,173 @@
|
|
1
|
+
/* The Computer Language Benchmarks Game
|
2
|
+
http://shootout.alioth.debian.org/
|
3
|
+
|
4
|
+
Contributed by Josh Goldfoot
|
5
|
+
to compile, use gcc -O3
|
6
|
+
|
7
|
+
This revision uses "simple_hash.h," available from
|
8
|
+
http://alioth.debian.org/plugins/scmcvs/cvsweb.php/shootout/bench/Include/?cvsroot=shootout
|
9
|
+
|
10
|
+
*/
|
11
|
+
#include <stdio.h>
|
12
|
+
#include <string.h>
|
13
|
+
#include <ctype.h>
|
14
|
+
#include <stdlib.h>
|
15
|
+
#include "../../Include/simple_hash.h"
|
16
|
+
|
17
|
+
long
|
18
|
+
hash_table_size (int fl, long buflen)
|
19
|
+
{
|
20
|
+
long maxsize1, maxsize2;
|
21
|
+
|
22
|
+
maxsize1 = buflen - fl;
|
23
|
+
maxsize2 = 4;
|
24
|
+
while (--fl > 0 && maxsize2 < maxsize1)
|
25
|
+
maxsize2 = maxsize2 * 4;
|
26
|
+
if (maxsize1 < maxsize2)
|
27
|
+
return maxsize1;
|
28
|
+
return maxsize2;
|
29
|
+
}
|
30
|
+
|
31
|
+
struct ht_ht *
|
32
|
+
generate_frequencies (int fl, char *buffer, long buflen)
|
33
|
+
{
|
34
|
+
struct ht_ht *ht;
|
35
|
+
char *reader;
|
36
|
+
long i;
|
37
|
+
char nulled;
|
38
|
+
|
39
|
+
if (fl > buflen)
|
40
|
+
return NULL;
|
41
|
+
|
42
|
+
ht = ht_create (hash_table_size (fl, buflen));
|
43
|
+
for (i = 0; i < buflen - fl + 1; i++)
|
44
|
+
{
|
45
|
+
reader = &(buffer[i]);
|
46
|
+
nulled = reader[fl];
|
47
|
+
reader[fl] = 0x00;
|
48
|
+
ht_find_new (ht, reader)->val++;
|
49
|
+
reader[fl] = nulled;
|
50
|
+
}
|
51
|
+
return ht;
|
52
|
+
}
|
53
|
+
|
54
|
+
typedef struct ssorter
|
55
|
+
{
|
56
|
+
char *string;
|
57
|
+
int num;
|
58
|
+
} sorter;
|
59
|
+
|
60
|
+
void
|
61
|
+
write_frequencies (int fl, char *buffer, long buflen)
|
62
|
+
{
|
63
|
+
struct ht_ht *ht;
|
64
|
+
long total, i, j, size;
|
65
|
+
struct ht_node *nd;
|
66
|
+
sorter *s;
|
67
|
+
sorter tmp;
|
68
|
+
|
69
|
+
ht = generate_frequencies (fl, buffer, buflen);
|
70
|
+
total = 0;
|
71
|
+
size = 0;
|
72
|
+
for (nd = ht_first (ht); nd != NULL; nd = ht_next (ht))
|
73
|
+
{
|
74
|
+
total = total + nd->val;
|
75
|
+
size++;
|
76
|
+
}
|
77
|
+
s = calloc (size, sizeof (sorter));
|
78
|
+
i = 0;
|
79
|
+
for (nd = ht_first (ht); nd != NULL; nd = ht_next (ht))
|
80
|
+
{
|
81
|
+
s[i].string = nd->key;
|
82
|
+
s[i++].num = nd->val;
|
83
|
+
}
|
84
|
+
for (i = 0; i < size - 1; i++)
|
85
|
+
for (j = i + 1; j < size; j++)
|
86
|
+
if (s[i].num < s[j].num)
|
87
|
+
{
|
88
|
+
memcpy (&tmp, &(s[i]), sizeof (sorter));
|
89
|
+
memcpy (&(s[i]), &(s[j]), sizeof (sorter));
|
90
|
+
memcpy (&(s[j]), &tmp, sizeof (sorter));
|
91
|
+
}
|
92
|
+
for (i = 0; i < size; i++)
|
93
|
+
printf ("%s %.3f\n", s[i].string, 100 * (float) s[i].num / total);
|
94
|
+
printf ("\n");
|
95
|
+
ht_destroy (ht);
|
96
|
+
free (s);
|
97
|
+
}
|
98
|
+
|
99
|
+
void
|
100
|
+
write_count (char *searchFor, char *buffer, long buflen)
|
101
|
+
{
|
102
|
+
struct ht_ht *ht;
|
103
|
+
|
104
|
+
ht = generate_frequencies (strlen (searchFor), buffer, buflen);
|
105
|
+
printf ("%d\t%s\n", ht_find_new (ht, searchFor)->val, searchFor);
|
106
|
+
ht_destroy (ht);
|
107
|
+
}
|
108
|
+
|
109
|
+
int
|
110
|
+
main ()
|
111
|
+
{
|
112
|
+
char c;
|
113
|
+
char *line, *buffer, *tmp, *x;
|
114
|
+
int i, linelen, nothree;
|
115
|
+
long buflen, seqlen;
|
116
|
+
|
117
|
+
line = malloc (256);
|
118
|
+
if (!line)
|
119
|
+
return -1;
|
120
|
+
seqlen = 0;
|
121
|
+
nothree = 1;
|
122
|
+
|
123
|
+
while (nothree && fgets (line, 255, stdin))
|
124
|
+
if (line[0] == '>' && line[1] == 'T' && line[2] == 'H')
|
125
|
+
nothree = 0;
|
126
|
+
free (line);
|
127
|
+
|
128
|
+
buflen = 10240;
|
129
|
+
buffer = malloc (buflen + 1);
|
130
|
+
if (!buffer)
|
131
|
+
return -1;
|
132
|
+
x = buffer;
|
133
|
+
|
134
|
+
while (fgets (x, 255, stdin))
|
135
|
+
{
|
136
|
+
linelen = strlen (x);
|
137
|
+
if (linelen)
|
138
|
+
{
|
139
|
+
if (x[linelen - 1] == '\n')
|
140
|
+
linelen--;
|
141
|
+
c = x[0];
|
142
|
+
if (c == '>')
|
143
|
+
break;
|
144
|
+
else if (c != ';')
|
145
|
+
{
|
146
|
+
seqlen = seqlen + linelen;
|
147
|
+
if (seqlen + 512 >= buflen)
|
148
|
+
{
|
149
|
+
buflen = buflen + 10240;
|
150
|
+
tmp = realloc (buffer, buflen + 1);
|
151
|
+
if (tmp == NULL)
|
152
|
+
return -1;
|
153
|
+
buffer = tmp;
|
154
|
+
x = &(buffer[seqlen]);
|
155
|
+
}
|
156
|
+
else
|
157
|
+
x = &(x[linelen]);
|
158
|
+
x[0] = 0;
|
159
|
+
}
|
160
|
+
}
|
161
|
+
}
|
162
|
+
for (i = 0; i < seqlen; i++)
|
163
|
+
buffer[i] = toupper (buffer[i]);
|
164
|
+
write_frequencies (1, buffer, seqlen);
|
165
|
+
write_frequencies (2, buffer, seqlen);
|
166
|
+
write_count ("GGT", buffer, seqlen);
|
167
|
+
write_count ("GGTA", buffer, seqlen);
|
168
|
+
write_count ("GGTATT", buffer, seqlen);
|
169
|
+
write_count ("GGTATTTTAATT", buffer, seqlen);
|
170
|
+
write_count ("GGTATTTTAATTTATAGT", buffer, seqlen);
|
171
|
+
free (buffer);
|
172
|
+
return 0;
|
173
|
+
}
|
@@ -0,0 +1,168 @@
|
|
1
|
+
/* The Computer Language Benchmarks Game
|
2
|
+
http://shootout.alioth.debian.org/
|
3
|
+
|
4
|
+
Contributed by Josh Goldfoot
|
5
|
+
*/
|
6
|
+
|
7
|
+
#include <stdio.h>
|
8
|
+
#include <string.h>
|
9
|
+
#include <ctype.h>
|
10
|
+
#include <stdlib.h>
|
11
|
+
|
12
|
+
long
|
13
|
+
hash_table_size (int fl, long buflen)
|
14
|
+
{
|
15
|
+
long maxsize1, maxsize2;
|
16
|
+
|
17
|
+
maxsize1 = buflen - fl;
|
18
|
+
maxsize2 = 4;
|
19
|
+
while (--fl > 0 && maxsize2 < maxsize1)
|
20
|
+
maxsize2 = maxsize2 * 4;
|
21
|
+
if (maxsize1 < maxsize2)
|
22
|
+
return maxsize1;
|
23
|
+
return maxsize2;
|
24
|
+
}
|
25
|
+
|
26
|
+
struct ht_ht *
|
27
|
+
generate_frequencies (int fl, char *buffer, long buflen)
|
28
|
+
{
|
29
|
+
struct ht_ht *ht;
|
30
|
+
char *reader;
|
31
|
+
long i;
|
32
|
+
char nulled;
|
33
|
+
|
34
|
+
if (fl > buflen)
|
35
|
+
return NULL;
|
36
|
+
|
37
|
+
ht = ht_create (hash_table_size (fl, buflen));
|
38
|
+
for (i = 0; i < buflen - fl + 1; i++)
|
39
|
+
{
|
40
|
+
reader = &(buffer[i]);
|
41
|
+
nulled = reader[fl];
|
42
|
+
reader[fl] = 0x00;
|
43
|
+
ht_find_new (ht, reader)->val++;
|
44
|
+
reader[fl] = nulled;
|
45
|
+
}
|
46
|
+
return ht;
|
47
|
+
}
|
48
|
+
|
49
|
+
typedef struct ssorter
|
50
|
+
{
|
51
|
+
char *string;
|
52
|
+
int num;
|
53
|
+
} sorter;
|
54
|
+
|
55
|
+
void
|
56
|
+
write_frequencies (int fl, char *buffer, long buflen)
|
57
|
+
{
|
58
|
+
struct ht_ht *ht;
|
59
|
+
long total, i, j, size;
|
60
|
+
struct ht_node *nd;
|
61
|
+
sorter *s;
|
62
|
+
sorter tmp;
|
63
|
+
|
64
|
+
ht = generate_frequencies (fl, buffer, buflen);
|
65
|
+
total = 0;
|
66
|
+
size = 0;
|
67
|
+
for (nd = ht_first (ht); nd != NULL; nd = ht_next (ht))
|
68
|
+
{
|
69
|
+
total = total + nd->val;
|
70
|
+
size++;
|
71
|
+
}
|
72
|
+
s = calloc (size, sizeof (sorter));
|
73
|
+
i = 0;
|
74
|
+
for (nd = ht_first (ht); nd != NULL; nd = ht_next (ht))
|
75
|
+
{
|
76
|
+
s[i].string = nd->key;
|
77
|
+
s[i++].num = nd->val;
|
78
|
+
}
|
79
|
+
for (i = 0; i < size - 1; i++)
|
80
|
+
for (j = i + 1; j < size; j++)
|
81
|
+
if (s[i].num < s[j].num)
|
82
|
+
{
|
83
|
+
memcpy (&tmp, &(s[i]), sizeof (sorter));
|
84
|
+
memcpy (&(s[i]), &(s[j]), sizeof (sorter));
|
85
|
+
memcpy (&(s[j]), &tmp, sizeof (sorter));
|
86
|
+
}
|
87
|
+
for (i = 0; i < size; i++)
|
88
|
+
printf ("%s %.3f\n", s[i].string, 100 * (float) s[i].num / total);
|
89
|
+
printf ("\n");
|
90
|
+
ht_destroy (ht);
|
91
|
+
free (s);
|
92
|
+
}
|
93
|
+
|
94
|
+
void
|
95
|
+
write_count (char *searchFor, char *buffer, long buflen)
|
96
|
+
{
|
97
|
+
struct ht_ht *ht;
|
98
|
+
|
99
|
+
ht = generate_frequencies (strlen (searchFor), buffer, buflen);
|
100
|
+
printf ("%d\t%s\n", ht_find_new (ht, searchFor)->val, searchFor);
|
101
|
+
ht_destroy (ht);
|
102
|
+
}
|
103
|
+
|
104
|
+
int
|
105
|
+
main ()
|
106
|
+
{
|
107
|
+
char c;
|
108
|
+
char *line, *buffer, *tmp, *x;
|
109
|
+
int i, linelen, nothree;
|
110
|
+
long buflen, seqlen;
|
111
|
+
|
112
|
+
line = malloc (256);
|
113
|
+
if (!line)
|
114
|
+
return -1;
|
115
|
+
seqlen = 0;
|
116
|
+
nothree = 1;
|
117
|
+
|
118
|
+
while (nothree && fgets (line, 255, stdin))
|
119
|
+
if (line[0] == '>' && line[1] == 'T' && line[2] == 'H')
|
120
|
+
nothree = 0;
|
121
|
+
free (line);
|
122
|
+
|
123
|
+
buflen = 10240;
|
124
|
+
buffer = malloc (buflen + 1);
|
125
|
+
if (!buffer)
|
126
|
+
return -1;
|
127
|
+
x = buffer;
|
128
|
+
|
129
|
+
while (fgets (x, 255, stdin))
|
130
|
+
{
|
131
|
+
linelen = strlen (x);
|
132
|
+
if (linelen)
|
133
|
+
{
|
134
|
+
if (x[linelen - 1] == '\n')
|
135
|
+
linelen--;
|
136
|
+
c = x[0];
|
137
|
+
if (c == '>')
|
138
|
+
break;
|
139
|
+
else if (c != ';')
|
140
|
+
{
|
141
|
+
seqlen = seqlen + linelen;
|
142
|
+
if (seqlen + 512 >= buflen)
|
143
|
+
{
|
144
|
+
buflen = buflen + 10240;
|
145
|
+
tmp = realloc (buffer, buflen + 1);
|
146
|
+
if (tmp == NULL)
|
147
|
+
return -1;
|
148
|
+
buffer = tmp;
|
149
|
+
x = &(buffer[seqlen]);
|
150
|
+
}
|
151
|
+
else
|
152
|
+
x = &(x[linelen]);
|
153
|
+
x[0] = 0;
|
154
|
+
}
|
155
|
+
}
|
156
|
+
}
|
157
|
+
for (i = 0; i < seqlen; i++)
|
158
|
+
buffer[i] = toupper (buffer[i]);
|
159
|
+
write_frequencies (1, buffer, seqlen);
|
160
|
+
write_frequencies (2, buffer, seqlen);
|
161
|
+
write_count ("GGT", buffer, seqlen);
|
162
|
+
write_count ("GGTA", buffer, seqlen);
|
163
|
+
write_count ("GGTATT", buffer, seqlen);
|
164
|
+
write_count ("GGTATTTTAATT", buffer, seqlen);
|
165
|
+
write_count ("GGTATTTTAATTTATAGT", buffer, seqlen);
|
166
|
+
free (buffer);
|
167
|
+
return 0;
|
168
|
+
}
|
@@ -0,0 +1,226 @@
|
|
1
|
+
/* -*- mode: c -*-
|
2
|
+
* $Id: lists.gcc,v 1.1.1.1 2004-05-19 18:10:23 bfulgham Exp $
|
3
|
+
* http://www.bagley.org/~doug/shootout/
|
4
|
+
*/
|
5
|
+
|
6
|
+
#include <stdio.h>
|
7
|
+
#include <stdlib.h>
|
8
|
+
#include <string.h>
|
9
|
+
#include <unistd.h>
|
10
|
+
|
11
|
+
#define SIZE 10000
|
12
|
+
|
13
|
+
// a simple Double Linked List
|
14
|
+
// the head node is special, it's val is length of list
|
15
|
+
typedef struct DLL {
|
16
|
+
int val;
|
17
|
+
struct DLL *next; /* points to next or head (if at tail) */
|
18
|
+
struct DLL *prev; /* points to prev or tail (if at head) */
|
19
|
+
} DLL;
|
20
|
+
|
21
|
+
inline int list_length(DLL *head) { return(head->val); }
|
22
|
+
inline int list_empty(DLL *head) { return(list_length(head) == 0); }
|
23
|
+
inline DLL *list_first(DLL *head) { return(head->next); }
|
24
|
+
inline DLL *list_last(DLL *head) { return(head->prev); }
|
25
|
+
|
26
|
+
void list_push_tail(DLL *head, DLL *item) {
|
27
|
+
DLL *tail = head->prev;
|
28
|
+
tail->next = item;
|
29
|
+
item->next = head;
|
30
|
+
head->prev = item;
|
31
|
+
item->prev = tail;
|
32
|
+
head->val++;
|
33
|
+
}
|
34
|
+
|
35
|
+
DLL *list_pop_tail(DLL *head) {
|
36
|
+
DLL *prev, *tail;
|
37
|
+
if (list_empty(head)) return(NULL);
|
38
|
+
tail = head->prev;
|
39
|
+
prev = tail->prev;
|
40
|
+
prev->next = head;
|
41
|
+
head->prev = prev;
|
42
|
+
head->val--;
|
43
|
+
return(tail);
|
44
|
+
}
|
45
|
+
|
46
|
+
void list_push_head(DLL *head, DLL *item) {
|
47
|
+
DLL *next = head->next;
|
48
|
+
head->next = item;
|
49
|
+
next->prev = item;
|
50
|
+
item->next = next;
|
51
|
+
item->prev = head;
|
52
|
+
head->val++;
|
53
|
+
}
|
54
|
+
|
55
|
+
DLL *list_pop_head(DLL *head) {
|
56
|
+
DLL *next;
|
57
|
+
if (list_empty(head)) return(NULL);
|
58
|
+
next = head->next;
|
59
|
+
head->next = next->next;
|
60
|
+
next->next->prev = head;
|
61
|
+
head->val--;
|
62
|
+
return(next);
|
63
|
+
}
|
64
|
+
|
65
|
+
int list_equal(DLL *x, DLL *y) {
|
66
|
+
DLL *xp, *yp;
|
67
|
+
// first val's checked will be list lengths
|
68
|
+
for (xp=x, yp=y; xp->next != x; xp=xp->next, yp=yp->next) {
|
69
|
+
if (xp->val != yp->val) return(0);
|
70
|
+
}
|
71
|
+
if (xp->val != yp->val) return(0);
|
72
|
+
return(yp->next == y);
|
73
|
+
}
|
74
|
+
|
75
|
+
void list_print(char *msg, DLL *x) {
|
76
|
+
DLL *xp, *first = x->next;
|
77
|
+
int i = 0;
|
78
|
+
fputs(msg, stdout);
|
79
|
+
printf("length: %d\n", list_length(x));
|
80
|
+
for (xp=x->next; xp->next != first; xp=xp->next) {
|
81
|
+
printf("i:%3d v:%3d n:%3d p:%3d\n", ++i,
|
82
|
+
xp->val, xp->next->val, xp->prev->val);
|
83
|
+
}
|
84
|
+
printf("[last entry points to list head]\n");
|
85
|
+
printf("[val of next of tail is: %d]\n", xp->next->val);
|
86
|
+
}
|
87
|
+
|
88
|
+
DLL *list_new() {
|
89
|
+
DLL *l = (DLL *)malloc(sizeof(DLL));
|
90
|
+
l->next = l;
|
91
|
+
l->prev = l;
|
92
|
+
l->val = 0;
|
93
|
+
return(l);
|
94
|
+
}
|
95
|
+
|
96
|
+
/* inclusive sequence 'from' <-> 'to' */
|
97
|
+
DLL *list_sequence(int from, int to) {
|
98
|
+
int size, tmp, i, j;
|
99
|
+
DLL *l;
|
100
|
+
if (from > to) {
|
101
|
+
tmp = from; from = to; to = tmp;
|
102
|
+
}
|
103
|
+
size = to - from + 1;
|
104
|
+
l = (DLL *)malloc((size+1) * sizeof(DLL));
|
105
|
+
from--;
|
106
|
+
for (i=0, j=1; i<size; ++i, ++j) {
|
107
|
+
l[i].next = &l[i+1];
|
108
|
+
l[j].prev = &l[j-1];
|
109
|
+
l[i].val = from++;
|
110
|
+
}
|
111
|
+
l[0].prev = &l[size];
|
112
|
+
l[size].next = &l[0];
|
113
|
+
l[size].prev = &l[size-1];
|
114
|
+
l[size].val = from;
|
115
|
+
l[0].val = size;
|
116
|
+
return(l);
|
117
|
+
}
|
118
|
+
|
119
|
+
DLL *list_copy(DLL *x) {
|
120
|
+
int i, j, size = list_length(x);
|
121
|
+
DLL *xp, *l = (DLL *)malloc((size+1) * sizeof(DLL));
|
122
|
+
for (i=0, j=1, xp=x; i<size; i++, j++, xp=xp->next) {
|
123
|
+
l[i].next = &l[j];
|
124
|
+
l[j].prev = &l[i];
|
125
|
+
l[i].val = xp->val;
|
126
|
+
}
|
127
|
+
l[0].prev = &l[size];
|
128
|
+
l[size].next = &l[0];
|
129
|
+
l[size].val = list_last(x)->val;
|
130
|
+
return(l);
|
131
|
+
}
|
132
|
+
|
133
|
+
void list_reverse (DLL *head) {
|
134
|
+
DLL *tmp, *p = head;
|
135
|
+
do {
|
136
|
+
tmp = p->next;
|
137
|
+
p->next = p->prev;
|
138
|
+
p->prev = tmp;
|
139
|
+
p = tmp;
|
140
|
+
} while (p != head);
|
141
|
+
}
|
142
|
+
|
143
|
+
int test_lists() {
|
144
|
+
int len = 0;
|
145
|
+
// create a list of integers (li1) from 1 to SIZE
|
146
|
+
DLL *li1 = list_sequence(1, SIZE);
|
147
|
+
// copy the list to li2
|
148
|
+
DLL *li2 = list_copy(li1);
|
149
|
+
// remove each individual item from left side of li2 and
|
150
|
+
// append to right side of li3 (preserving order)
|
151
|
+
DLL *li3 = list_new();
|
152
|
+
// compare li2 and li1 for equality
|
153
|
+
if (!list_equal(li2, li1)) {
|
154
|
+
fprintf(stderr, "li2 and li1 are not equal\n");
|
155
|
+
exit(1);
|
156
|
+
}
|
157
|
+
while (!list_empty(li2)) {
|
158
|
+
list_push_tail(li3, list_pop_head(li2));
|
159
|
+
}
|
160
|
+
// li2 must now be empty
|
161
|
+
if (!list_empty(li2)) {
|
162
|
+
fprintf(stderr, "li2 should be empty now\n");
|
163
|
+
exit(1);
|
164
|
+
}
|
165
|
+
// remove each individual item from right side of li3 and
|
166
|
+
// append to right side of li2 (reversing list)
|
167
|
+
while (!list_empty(li3)) {
|
168
|
+
list_push_tail(li2, list_pop_tail(li3));
|
169
|
+
}
|
170
|
+
// li3 must now be empty
|
171
|
+
if (!list_empty(li3)) {
|
172
|
+
fprintf(stderr, "li3 should be empty now\n");
|
173
|
+
exit(1);
|
174
|
+
}
|
175
|
+
// reverse li1 in place
|
176
|
+
list_reverse(li1);
|
177
|
+
// check that li1's first item is now SIZE
|
178
|
+
if (list_first(li1)->val != SIZE) {
|
179
|
+
fprintf(stderr, "li1 first value wrong, wanted %d, got %d\n",
|
180
|
+
SIZE, list_first(li1)->val);
|
181
|
+
exit(1);
|
182
|
+
}
|
183
|
+
// check that li1's last item is now 1
|
184
|
+
if (list_last(li1)->val != 1) {
|
185
|
+
fprintf(stderr, "last value wrong, wanted %d, got %d\n",
|
186
|
+
SIZE, list_last(li1)->val);
|
187
|
+
exit(1);
|
188
|
+
}
|
189
|
+
// check that li2's first item is now SIZE
|
190
|
+
if (list_first(li2)->val != SIZE) {
|
191
|
+
fprintf(stderr, "li2 first value wrong, wanted %d, got %d\n",
|
192
|
+
SIZE, list_first(li2)->val);
|
193
|
+
exit(1);
|
194
|
+
}
|
195
|
+
// check that li2's last item is now 1
|
196
|
+
if (list_last(li2)->val != 1) {
|
197
|
+
fprintf(stderr, "last value wrong, wanted %d, got %d\n",
|
198
|
+
SIZE, list_last(li2)->val);
|
199
|
+
exit(1);
|
200
|
+
}
|
201
|
+
// check that li1's length is still SIZE
|
202
|
+
if (list_length(li1) != SIZE) {
|
203
|
+
fprintf(stderr, "li1 size wrong, wanted %d, got %d\n",
|
204
|
+
SIZE, list_length(li1));
|
205
|
+
exit(1);
|
206
|
+
}
|
207
|
+
// compare li1 and li2 for equality
|
208
|
+
if (!list_equal(li1, li2)) {
|
209
|
+
fprintf(stderr, "li1 and li2 are not equal\n");
|
210
|
+
exit(1);
|
211
|
+
}
|
212
|
+
len = list_length(li1);
|
213
|
+
free(li1);
|
214
|
+
free(li2);
|
215
|
+
free(li3);
|
216
|
+
// return the length of the list
|
217
|
+
return(len);
|
218
|
+
}
|
219
|
+
|
220
|
+
int main(int argc, char *argv[]) {
|
221
|
+
int n = ((argc == 2) ? atoi(argv[1]) : 1);
|
222
|
+
int result = 0;
|
223
|
+
while(n--) result = test_lists();
|
224
|
+
printf("%d\n", result);
|
225
|
+
return 0;
|
226
|
+
}
|