Using Pointers and strtok()

Question

I\'m building a linked list and need your assistance please as I\'m new to C. I need to input a string that looks like this: (word)_#_(year)_#_(DEFINITION(UPPER CASE))

Answer 1

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

char *strtokByWord_r(char *str, const char *word, char **store){
    char *p, *ret;
    if(str != NULL){
        *store = str;
    }
    if(*store == NULL) return NULL;
    p = strstr(ret=*store, word);
    if(p){
        *p='\0';
        *store = p + strlen(word);
    } else {
        *store = NULL;
    }
    return ret;
}
char *strtokByWord(char *str, const char *word){
    static char *store = NULL;
    return strtokByWord_r(str, word, &store);
}

int main(){
    char input[]="invest_#_1945_#_TRADE";
    char *array[3];
    char *p;
    int i, size = sizeof(array)/sizeof(char*);
    for(i=0, p=input;i<size;++i){
        if(NULL!=(p=strtokByWord(p, "_#_"))){
            array[i]=p;//strdup(p);
            p=NULL;
        } else {
            array[i]=NULL;
            break;
        }
    }
    for(i = 0;i<size;++i)
        printf("array[%d]=\"%s\"\n", i, array[i]);
/* result
array[0]="invest"
array[1]="1945"
array[2]="TRADE"
*/
    return 0;
}

Answer 2

I reckon that your problem is how to extract the three bits of information from your formatted string.

The function strtok does not work as you think it does: The second argument is not a literal delimiting string, but a string that serves as a set of characters that are delimiters.

In your case, sscanf seems to be the better choice:

#include <stdlib.h>
#include <stdio.h>

int main()
{
    const char *line = "invest_#_1945 _#_TRADE  ";

    char word[40];
    int year;
    char def[40];
    int n;

    n = sscanf(line, "%40[^_]_#_%d_#_%40s", word, &year, def);

    if (n == 3) {
        printf("word:   %s\n", word);
        printf("year:   %d\n", year);
        printf("def'n:  %s\n", def);
    } else {
        printf("Unrecognized line.\n");
    }

    return 0;
}

The function sscanf examines a given string according to a given pattern. Roughly, that pattern consists of format specifiers that begin with a percent sign, of spaces which denote any amount of white-space characters (including none) and of other characters that have to be matched varbatim. The format specifiers yield a result, which has to be stored. Therefore, for each specifier, a result variable must be given after the format string.

In this case, there are several chunks:

%40[^_] reads up to 40 characters that are not the underscore into a char array. This is a special case of reading a string. Strings in sscanf are really words and may not contain white space. The underscore, however, would be part of a string, so in order not to eat up the underscore of the first delimiter, you have to use the notation [^(chars)], which means: Any sequence of chars that do not contain the given chars. (The caret does the negation here, [(chars)] would mean any sequence of the given chars.)

_#_ matches the first delimiter literally, i.e. only if the next chars are underscore hash mark, underscore.

%d reads a decimal number into an integer. Note that the adress of the integer has to be given here with &.

_#_ matches the second delimiter.

%40s reads a string of up to 40 non-whitespace characters into a char array.

The function returns the number of matched results, which should be three if the line is valid. The function sscanf can be cumbersome, but is probably your best bet here for quick and dirty input.