|
Matt Bishop Office: 2209 Watershed Sciences Phone: +1 (530) 752-8060 Email: mabishop@ucdavis.edu |
Version of April 2, 2024 at 12:13PM
/*
* statistical analysis of file to see if it is binary or text data
*
* every SKIP'th character is read, and checked to see if it is printable
* (white space is considered printable)
*
* if at least THRESHHOLD% characters are printable, it's binary; otherwise
* it's text INPORTANT: THRESHHOLD is a percentage, not a fraction!)
*
* Matt Bishop, ECS 36A
* -- May 22, 2024 Original program
*/
#include <stdio.h>
#include <ctype.h>
#define SKIP 10 /* read 1 out of every SKIP characters */
#define THRESHHOLD 25 /* a percent, *not* a fraction! */
/*
* test to see if this is a binary file
*/
void bintest(char *fname, FILE *fp)
{
int ch; /* input character */
int bin = 0; /* number of binary characters */
int nonbin = 0; /* number of printing characters */
int sample_size = 0; /* number of characters read */
double frac = 0.0; /* percentage of binary characters */
/*
* skip ahead SKIP characters and analyze the character
*/
while(fseek(fp, SKIP, SEEK_CUR) != -1){
/* if done, break out of the loop */
if ((ch = fgetc(fp)) == EOF)
break;
/* read another character */
sample_size += 1;
/* classify the character; note spaces */
/* are considered printable characters */
if (isprint(ch))
nonbin++;
else
bin++;
}
/*
* now do the analysis
*/
/* what percent of the sampled characters are non-printable? */
frac = ((double) bin) / ((double) sample_size) * 100;
/* print out the results */
printf("%s: %d samples,", fname, sample_size);
printf(" %d non-binary, %d (%0.2f%%) binary; a ", nonbin, bin, frac);
printf(frac < THRESHHOLD ? "text" : "binary");
printf(" file\n");
}
/*
* it all begins here
*/
int main(int argc, char *argv[])
{
FILE *fp; /* input file */
char **a; /* pointer in a for loop */
int rv = 0; /* return value; number of file open failures */
/* be sure there's at least one file named */
if (argc < 2){
fprintf(stderr, "Usage: %s file [ ... ]\n", argv[0]);
return(1);
}
/* walk the argument list */
for (a = argv+1; *a != NULL; a++){
/* open the file */
if ((fp = fopen(*a, "r")) == NULL){
perror(*a);
rv++;
continue;
}
/* do the test */
bintest(*a, fp);
/* done with this file */
(void) fclose(fp);
}
/* that's all folks! */
return(rv);
}
|
ECS 36A, Programming & Problem Solving Version of April 2, 2024 at 12:13PM
|
You can get the raw source code here.
|