CODEWITHSUNDEEP
cstringstrncmp
rookie
rookie

Reputation: 1208

better way for string comparison

I have a JSON parser that sends a callback for every parsed key-value pair. I then use strncmp with statically allocated strings in my program to check if the field matches any of the fields I am interested in. I have now ended up with a dozen of strncmp in if-else statements:

if (strncmp(input, "apple", sizeof("apple")-1) == 0) 
    {// do something}
else if (strncmp(input, "banana", sizeof("banana")-1) == 0) 
    {// do something}

I'm not sure if this is an optimal/maintainable way of getting the job done. Is there a better way to do it? I already precompute the string lengths at compile time.

Upvotes: -1

Views: 80

Answers (2)

Julien Thierry
Julien Thierry

Reputation: 661

You could define an array of static strings:

const char *field_names[] = {
    "apple",
    "banana",
    NULL
};

And then loop over the array, looking if the string matches one on the elements of the array:

for (int i = 0; field_names[i] != NULL; i++) {
    if (!strcmp(input, field_names[i]))
        // Do things
}

If you need to limit the size of the comparison, you could have an array a structures combining the name string + the size.

And if you need to do a different action for each one, you could include a function pointer in the structure to indicate what to do.

Upvotes: 1

bipll
bipll

Reputation: 11950

Depends on the size of your set. For large amount of long patterns, your O(MN) lookup is surely not optimal.

You can store them in a hash set (O(M) on average), or in a trie. You can even hardcode one on your own, if you've got a really fixed set of patterns from the Delphic Oracle and first letters are distributed more or less uniformly among patterns:

switch(input[0]) {
case 'a': // check that it's an "apple"
          break;
}

Upvotes: 0

Related Questions