Thread: string tokens

I've tried to extract numbers, letters and punctuations etc, from strings but I've never succeeded. For example, how would you 'extract' the words/numbers in; text<space>text<space>number or number<punctuation>number and so on. Can you use strtok() to do that? I don't have any test code, sorry, and I don't even know how to approach this. I looked in the FAQ, but there's only examples created in C++, which I don't know. A push in the right direction would be appreciated.

Thank you, Markallen85, that gave me some idea what I need to do.

To be more specific, this is the problem I have: <number><operator><number>... where number can be any number and opeartor is either +, -, / or *. I want to separate the number and the opeartors, and this is the code I have so far:

Code:

#include <stdio.h>
#include <string.h>

#define BUFFER 1024

int main() {
	
	int loop = 1, len, i;
	char seq[BUFFER];
	char *token;
	
	while(loop == 1) {
	
		printf(": ");
		/* gets the line */
		fgets(seq, sizeof(seq), stdin);
		/* gets the length of the string */
		len = strlen(seq);
		
		token = strtok(seq,"+");
		
		for(i=0; token != NULL; ++i) {
			printf("%s",token);
			token=strtok(NULL,",");
		}
...

Some questions: what type should 'token' be (I use both chars and integers/floats)? Does it have to be a pointer? Is it possble to pre-define the operators? I tried with #define OPS + - / * and put OPS where the "+" is now. I got errors like: error: parse error before ')' token

I'm confused.

I don't think you want to strtok for this. Write a function called lex that returns tokens. A token you can define like this

Code:

#define TOK_PLUS 0
#define TOK_SUB 1
#define TOK_DIV 2
#define TOK_MULT 3
#define TOK_EOF 4

typedef struct Token {
       int type;
       int value;
} Token;

The lex routine will skip space, read the token, and then return the token. In order to do this, look up getc, isspace, isdigit, and isalpha which are in ctype.h and stdio.h.