count the total no. of keywords in the file - c

Question

I want to count the total no. of keywords in the file but the code counts those keywords that are used to declare the variable.
void main()
{
//2d array used to store the keywords but few of them are used.
char key[32][12]={"int","char","while","for","if","else"};
//cnt is used to count the occurrence of the keyword in the file.
int cnt=0,i;
//used to store the string that is read line by line.
char ch[100];
FILE *fp=fopen("key.c","r");
//to check whether file exists or not
if(fp=='\0')
{
printf("file not found..\n");
exit(0);
}
//to extract the word till it don't reach the end of file
while((fscanf(fp,"%s",ch))!=EOF)
{
//compare the keyword with the word present in the file.
for(i=0;i<32;i++)
{
// compare the keyword with the string in ch.
if(strcmp(key[i],ch)==0) {
//just to check which keyword is printed.
printf("\nkeyword is : %s",ch);
cnt++;
}
}
}
printf("\n Total no. of keywords are : %d", cnt);
fclose(fp);
}
Expected output should be:
Total no. of keywords are : 7
Actual output is coming :
Total no. of keywords are : 3

Answer 1

fscanf(fp,"%s",ch) will match a sequence of non-whitespace characters (see cpp reference), so in your case for, while and if won't be matched as single words - because there's no space after them.

Answer 2

In my opinion, but diverting a little from your intention, you had better to use flex(1) for that purpose, as it will scan the file more efficiently than comparing each sequence with the set of words you may have. This approach will require more processing, as several keywords can be in the same line, and it only filters which lines have keywords on them.
Also, using flex(1) will give you a more efficient C source code, a sample input for flex(1) would be:
%{
unsigned long count = 0;
%}
%%
int |
char |
unsigned |
signed |
static |
auto |
do |
while |
if |
else |
/* ... add more keywords as you want here */
return |
break |
continue |
volatile { printf("keyword is = %s\n", yytext);
count++;
}
\n |
. ;
%%
int yywrap()
{
return 1;
}
int main()
{
yylex();
printf("count = %lu\n", count);
}
The efficiency comes basically from the fact that flex(1) uses a special algorithm that gets the right match with only scanning once the source file (one decision per char, all the patterns are scanned in parallel). The problem in your code comes from the fact that %s format has a special interpretation of what it considers is a word, different as the one defined by the C language (for scanf() a word si something surrounded by spaces, where spaces means \n, \t or only --- it will match as a word something like while(a==b) if you don't put spaces around your keywords). Also, If you need to compare each input pattern with each of the words your algorithm will end doing N passes through each input file character (with each letter meaning N = nw * awl (being N the number of times you compare each character and nw the number of words, awl the average of the list of word lengths in your set) By the way, keywords should not be recognised inside comments, or string literals, It is easy to adapt the code you see above to reject those and do a right scanning. For example, the next flex file will do this:
%{
unsigned long count = 0;
%}
%x COMM1
%x COMM2
%x STRLIT
%x CHRLIT
%%
int |
char |
unsigned |
signed |
static |
auto |
do |
while |
if |
else |
/* ... */
return |
break |
continue |
volatile { printf("kw is %s\n", yytext);
count++;
}
[a-zA-Z_][a-zA-Z0-9_]* |
^[\ \t]*#.* |
"/*"([^*]|\*[^/])*"*/" |
"//".* |
\"([^"\n]|\\")*\" |
\'([^'\n]|\\')*\' |
. |
\n ;
%%
int yywrap()
{
return 1;
}
int main()
{
yylex();
printf("count = %lu\n", count);
}
It allows different regular expressions to be recognised as language tokens, so provision is given to match also C language constructs like identifiers ([a-zA-Z_][a-zA-Z0-9_]*), preprocessor directives (^[\ \t]*#.*), old style C comments ("/*"([^*]|\*[^/])*"*/"), new C++ style comments ("//".*), string literals (\"([^"\n]|\\")*\"), character literals (\'([^'\n]|\\')*\'), where keywords cannot be identified as such.
Flex(1) is worth learning, as it simplifies a lot the input of structured data into a program. I suggest you to study it.
note
you had better to write if (fp == NULL), or even if (!fp)... (You are not doing anything incorrect in your statement if (fp == '\0'), but as \0 is the char representation of the nul character, it's somewhat inconvenient, strange or imprecise to compare a pointer value with a character literal, and suggests you are interpreting the FILE * not as a pointer, but more as an integer (or char) value.) But I repeat, it's something perfectly legal in C language.
note 2
The flex sample code posted above doesn't consider the possibility of running out of buffer space due to input very long tokens (like several line comments overflowing internal buffer space) This is done on purpose, to simplify description and to make the code simpler. Of course, in a professional scanner, all of these must be acquainted for.