Programming Example for Manipulating Characters

/*
This program is designed to demonstrate the use of "Character
classification and conversion" subroutines. Since we are dealing 
with characters, it is a natural place to demonstrate the use of 
getchar subroutine and putchar subroutine from the stdio library.
  

The program objectives are:

-Read input from "stdin" 

-Verify that all characters are ascii and printable

-Convert all uppercase characters to lowercase

-Discard multiple white spaces

-Report statistics regarding the types of characters

The following routines are demonstrated by this example program: 

-getchar

-putchar

-isascii (ctype) 

-iscntrl (ctype) 

-isspace (ctype)

-isalnum (ctype)

-isdigit (ctype)

-isalpha (ctype)

-isupper (ctype)

-islower (ctype)

-ispunct (ctype) 

-tolower (conv)

-toascii ( conv)

*/
  

#include <stdio.h>  /* The mandatory include file  */
#include <ctype.h>  /* Included for character classification

subroutines */
  

/* The various statistics gathering counters */

int asciicnt, printcnt, punctcnt, uppercnt, lowercnt,  

digcnt, alnumcnt, cntrlcnt, spacecnt, totcnt, nonprntcnt,linecnt, tabcnt ;
  

main()
{
  

int ch ; /* The input character is read in to this */
char c , class_conv() ;
  

asciicnt=printcnt=punctcnt=uppercnt=lowercnt=digcnt==0;
cntrlcnt=spacecnt=totcnt=nonprntcnt=linecnt=tabcnt=0;
alnumcnt=0;
 

while ( (ch =getchar()) != EOF )
{
 

totcnt++;
c = class_conv(ch) ;
putchar(c);
 

}
printf("The number lines of of input were %d\n",linecnt);
printf(" The character wise breakdown follows :\n");
printf(" TOTAL  ASCII  CNTRL   PUNCT   ALNUM  DIGITS UPPER

LOWER   SPACE   TABCNT\n");
 

printf("%5d %5d %5d %5d %5d %5d %5d %5d %5d %5d\n",totcnt,

asciicnt, cntrlcnt, punctcnt, alnumcnt, digcnt, uppercnt,lowercnt, spacecnt, tabcnt );
 

}
 

char class_conv(ch)
char ch;
{
 

if (isascii(ch)) {
 

asciicnt++;
if ( iscntrl(ch) && ! isspace(ch)) {
 

nonprntcnt++ ;
cntrlcnt++ ;
return(' ');
 

}
else if ( isalnum(ch)) {
 

alnumcnt++;
if (isdigit(ch)){

digcnt++;
return(ch);

}
else if (isalpha(ch)){

if ( isupper(ch) ){

uppercnt++ ;
return(tolower(ch));

}
else if ( islower(ch) ){

lowercnt++;
return(ch);

}
else {

/*
We should never be in this situation since an alpha character can only be either uppercase or lowercase.
*/ 

fprintf(stderr,"Classification error for %c \n",ch);
return(NULL);

}

}
else if (ispunct(ch) ){

punctcnt++;
return(ch);

}
else if ( isspace(ch) ){

spacecnt++;
if ( ch == '\n' ){
linecnt++;
return(ch);

}
while ( (ch == '\t' ) || ( ch == ' ' ) ) {
if ( ch == '\t' ) tabcnt ++ ;
else if ( ch == ' ' ) spacecnt++ ;
totcnt++;
ch = getchar();

}
ungetc(ch,stdin);
totcnt--;
return(' ');

}
else {

/*
We should never be in this situation any ASCII character
can only belong to one of the above classifications.
*/
fprintf(stderr,"Classification error for %c \n",ch);
return(NULL);
}

}
else
{

fprintf(stdout,"Non Ascii character encountered \n");
return(toascii(ch));

}

}