ken@ihuxq.UUCP (ken perlow) (08/17/84)
/*
To make this work, just cc -O -o<whatever> <this_program>.c
There's no fancy-shmancy archive or makefiles--all the documentation
is in this see-here comment. If you run BSD, read the "portability
considerations" to see what to rehack (essential, but should be trivial).
Author:
Ken Perlow
AT&T Bell Laboratories @ Naperville, IL
..ihnp4!ihuxq!ken
Synopsis:
This program generates random pseudo- (and occasionally real!)
words from 5 to 8 characters in length from a table of pairwise
letter frequencies it generates by scanning through an input
text file. A warning message is printed if the text file does
not contain successors to all letters, and those missing successors
are listed. The program picks a random start letter. Each
successive letter is generated from random access into the
frequency row for the letter just generated. If there is no
successor in the table, it stops generating that particular word
and prints what it has so far (maybe only the 1st letter).
Otherwise it stops when the word has grown to the (random) length
generated for that particular word.
This so-called "order-2" word generation strategy is modified
for readability of output by the following rules:
(1) A "u" which follows a "q" will have a successor chosen (randomly)
from among the vowels [aeio] only.
(2) If the first two letters are identical consonants, the
second will be changed to a random vowel [aeiouy].
(3) If there are 4 consonants in a row, the last will be changed
to a random vowel [aeiouy].
(4) After 2 vowels in a row, the table will be entered for the
2nd vowel until a (random) consonant is selected to succeed
it. This could result in an infinite loop, though it would
require unusual text input in which some vowel has only vowel
successors.
Options:
-f file Text file for table generation.
-i file Binary 26x27 file of integers for fast table generation
(table is read directly in). The 27th column contains
row totals.
-o file Binary 26x27 file of integers to be produced for subsequent
use of "-i" option. Can be invoked irrespective of input
file type, including default.
Default: If neither f, i, nor o invoked, program will prompt for a
text file.
-n nwdsout Number of words to generate. If not specified, program
will loop indefinitely.
-s slpintvl Number of seconds to sleep between successive word
generations. If not specified, no wait will occur.
The random number generator reseeds at random intervals
using the time-of-day in seconds, so this option may
aid in randomization as well as adding drama.
-t If specified, the 26x26 frequency table will be printed,
with each cell's tally scaled from 0 to 9 for ease of
reading. Note that 0 is printed for cell frequencies of
0 only, 1 for frequencies of 1 to (1/9)*max_tally, etc.
After printing the table, the program requests a <CR> to
continue.
-v If specified, perusal of text file will be verbose.
This includes printing "." for every 100 character
pairs scanned (so you know it hasn't died), and giving
summary statistics. Without this option, there will
be no output during the reading of the text file.
This option will be ignored for "-i" binary files.
Portability considerations:
This program is fully portable to any Unix environment except for
the "getrandom" function. The SV_R2 functions "rand", "srand",
and "time" are not standard in all Unix implementations. Any
procedure that returns a random number mod its argument may be
substituted, however. Unix is a trademark of AT&T Bell Laboratories.
Bugs:
None found so far, as the possibility of an infinite loop mentioned
above is now, by definition, a feature. On the other hand, this
algorithm is the acme of non-deterministic programming. So, to
paraphrase Clint Eastwood:
You're probably asking yourself, did it generate 5 characters or
6? Well, in all the excitement, it kinda forgot. But seeing as
it's written in C, the most powerful high-level language in the
world and capable of blowing your kernel clean off, ask yourself
this question, punk: "Do I feel lucky?"
*/
#include <stdio.h>
#include <ctype.h>
#define TRUE 1
#define FALSE 0
#define PROMPT 0
#define TEXTIN 1
#define BININ 2
#define NLETTERS 26
#define NLETPLS1 27
#define MAXWDSIZ 80
FILE *list;
int calls, nvow, ncns, table[NLETTERS][NLETPLS1];
char *letter = {"abcdefghijklmnopqrstuvwxyz"};
char fname[MAXWDSIZ], oname[MAXWDSIZ];
int vowel[6] = {0,4,8,14,20,24};
int binout, fflg, sflg, tflg, vflg, bytes, loops = 0;
extern char *optarg;
extern int optind, opterr;
mktable() {
int i, j, len, ctr, bin, last;
int longest = 0;
double avg = 0.0;
char word[MAXWDSIZ], *ptr;
switch(fflg) {
case PROMPT:
printf("Word file: ");
scanf("%s", fname);
case TEXTIN:
list = fopen(fname, "r");
if (list==NULL) nosoap(fname);
ctr = 0;
while(fscanf(list, "%s", word) != EOF) {
len = strlen(word);
if (notword(word, &len)) continue;
if (longest < len) longest = len;
if (vflg && (++ctr % 100 == 1)) write(1,".",1);
avg += len;
for (i=0;i<len-1;i++) {
table[(int)(word[i]-'a')][(int)(word[i+1]-'a')]++;
table[(int)(word[i]-'a')][NLETTERS]++;
}
}
if (vflg) printf("\n%d words; mean length = %.1f; longest = %d.\n", ctr, avg/ctr, longest);
break;
case BININ:
if (vflg) fprintf(stderr, "Binary input: 'v' option ignored.\n");
if ((bin = open(fname, 0)) == -1) nosoap(fname);
for(i=0;i<NLETTERS;i++) for(j=0;j<NLETPLS1;j++) read(bin,&table[i][j],sizeof(int));
close(bin);
break;
}
if (binout) {
if ((bin = creat(oname, 0644)) != -1)
printf("Binary table file '%s' created.\n", oname);
for(i=0;i<NLETTERS;i++) for(j=0;j<NLETPLS1;j++)
bytes += write(bin,&table[i][j],sizeof(int));
printf("%d bytes written.\n", bytes);
close(bin);
}
word[0] = '\0';
ptr = word;
for(i=0;i<NLETTERS;i++) if(!table[i][NLETTERS]) *ptr++ = *(letter+i);
*ptr = '\0';
if (strlen(word)) fprintf(stderr, "WARNING--no successors found for: %s\n",word);
fclose(list);
}
nosoap(filename) char *filename; {
fprintf(stderr, "Can't open '%s'.\n", filename);
exit(2);
}
notword(string, length) char *string; int *length; {
char *ptr;
if (ispunct(*(string+(*length)-1))) (*length)--;
for (ptr=string; ptr<string+(*length); ptr++) {
if (!isalpha(*ptr)) return(TRUE);
if (isupper(*ptr)) *ptr += 040;
}
return(FALSE);
}
getnext(i,pick) int i,pick; {
int sum,j;
if (i >= 0) {
if (table[i][NLETTERS]==0) return(-1);
sum=j=0;
while (sum<pick) sum += table[i][j++];
}
else
j = getrandom(pick)+1;
switch (--j) {
case 0 :
case 4 :
case 8 :
case 14:
case 20:
case 24: nvow++; ncns=0; break;
default: ncns++; nvow=0; break;
}
return(j);
}
getvowel(set) int set; {
nvow++;
ncns = 0;
return(vowel[getrandom(set)]);
}
getrandom(modulus) int modulus; {
long time();
if (!modulus) return(0);
if (!(calls++ % ((rand() & 0x1f)+1)))
srand((unsigned)time((long *)0) % ((rand() & 0xfff)+1));
return(rand() % modulus);
}
main(argc,argv) int argc; char **argv; {
int i, j, length, pos, opt, snooze;
int iters = 0;
char word[10];
while((opt=getopt(argc,argv,"i:o:f:s:n:tv"))!=EOF)
switch(opt) {
case 'i':
fflg++;
case 'f':
fflg++;
strcpy(fname,optarg);
break;
case 'o':
binout++;
strcpy(oname,optarg);
break;
case 's':
sflg++;
snooze=atoi(optarg);
break;
case 'n':
loops=atoi(optarg);
break;
case 't':
tflg++;
break;
case 'v':
vflg++;
break;
case '?':
fprintf(stderr, "Usage: %s [-f filein][-n nwdsout][-s slpintvl][-t][-v]\n", argv[0]);
exit(2);
break;
}
mktable();
if (tflg) prttbl();
for(;;) {
nvow=ncns=0;
length = getrandom(4) + 5;
i = getnext(-1,NLETTERS);
for (pos=0;pos<length;pos++) {
word[pos] = *(letter+i);
if(pos && (word[pos-1]=='q')) {
j = getvowel(4);
}
else if (nvow <= 2)
j = getnext(i,(getrandom(table[i][NLETTERS])+1));
while (nvow > 2 && j >= 0)
j = getnext(i,(getrandom(table[i][NLETTERS])+1));
if ((j==i && ncns && (ncns>2 || pos<2)) || ncns>3) {
j = getvowel(6);
}
else if (j == -1) {
word[pos] = *(letter+i);
word[pos+1] = '\0';
break;
}
i = j;
}
if (loops) {
iters++;
if (iters > loops) break;
}
word[length] = '\0';
printf("%s\n", word);
if (sflg) sleep(snooze);
}
}
prttbl() {
int i,j,max,floor;
char ch, display[NLETTERS][NLETPLS1];
double agsiz, freq, flpnt;
max=0;
for(i=0;i<NLETTERS;i++) {
display[i][NLETTERS] = '\0';
for(j=0;j<NLETTERS;j++) if (max < table[i][j]) max = table[i][j];
}
agsiz = max/9.0;
printf("Pairwise letter frequencies (scale 1:%.1f) in file '%s'\n\n", agsiz, fname);
for(i=0;i<NLETTERS;i++)
for(j=0;j<NLETTERS;j++) {
freq=table[i][j]/agsiz + 1.0;
flpnt = floor = freq;
if (freq == flpnt) floor--;
display[i][j] = (char)(floor + '0');
}
printf(" abcdefghijklmnopqrstuvwxyz abcdefghijklmnopqrstuvwxyz\n");
for(i=0;i<13;i++)
printf(" %c%s %c%s\n",*(letter+i),display[i],*(letter+13+i),display[13+i]);
printf("\nHit <return> to continue.\n");
read(0,&ch,1);
}
--
*** ***
JE MAINTIENDRAI ***** *****
****** ****** 17 Aug 84 [30 Thermidor An CXCII]
ken perlow ***** *****
(312)979-7261 ** ** ** **
..ihnp4!ihuxq!ken *** ***