bfile_heuristics.c

Thu, 02 Feb 2012 16:55:51 +0100

author
Mike Becker <universe@uap-core.de>
date
Thu, 02 Feb 2012 16:55:51 +0100
changeset 29
fa625066ae52
parent 23
778388400f7b
permissions
-rw-r--r--

fixed author note

/*
 * bfile_heuristics.c
 *
 *  Created on: 20.10.2011
 *      Author: Mike
 */

#include "bfile_heuristics.h"
#include <ctype.h>

bfile_heuristics_t *new_bfile_heuristics_t() {
  bfile_heuristics_t *ret = malloc(sizeof(bfile_heuristics_t));
  ret->level = BFILE_MEDIUM_ACCURACY;
  bfile_reset(ret);
  return ret;
}

void destroy_bfile_heuristics_t(bfile_heuristics_t *def) {
  free(def);
}

void bfile_reset(bfile_heuristics_t *def) {
  def->bcount = 0;
  def->tcount = 0;
}

bool bfile_check(bfile_heuristics_t *def, int next_char) {
  bool ret = false;
  if (def->level != BFILE_IGNORE) {
    def->tcount++;
    if (!isprint(next_char) && !isspace(next_char)) {
      def->bcount++;
    }

    if (def->tcount > 1) { /* empty files are text files */
      switch (def->level) {
      case BFILE_LOW_ACCURACY:
        if (def->tcount > 15 || next_char == EOF) {
          ret = (1.0*def->bcount)/def->tcount > 0.32;
        }
        break;
      case BFILE_HIGH_ACCURACY:
        if (def->tcount > 500 || next_char == EOF) {
          ret = (1.0*def->bcount)/def->tcount > 0.1;
        }
        break;
      default: /* BFILE_MEDIUM_ACCURACY */
        if (def->tcount > 100 || next_char == EOF) {
          ret = (1.0*def->bcount)/def->tcount > 0.1;
        }
        break;
      }
    }
  }

  return ret;
}

mercurial